close

Вход

Забыли?

вход по аккаунту

?

Investigation of arctic ozone loss using solar occultationand microwave limb sounding instruments

код для вставкиСкачать
Investigation of Arctic Ozone Loss using Solar Occultation and Microwave Limb
Sounding Instruments
by
Cynthia Shaw Singleton
B.S., The Ohio State University, 2001
M.S., University of Colorado, 2004
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Department of Atmospheric and Oceanic Sciences
2006
UMI Number: 3239417
UMI Microform 3239417
Copyright 2007 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, MI 48106-1346
This thesis entitled:
Investigation of Arctic Ozone Loss using Solar Occultation and Microwave Limb
Sounding Instruments
written by Cynthia Shaw Singleton
has been approved for the Department of Atmospheric and Oceanic Sciences
________________________________
Dr. Cora E. Randall
__________________________________
Dr. O. Brian Toon
_____________________________________
Dr. Linnea M. Avallone
____________________________________
Dr. David W. Rusch
____________________________________
Dr. V. Lynn Harvey
Date:_______________________
The final copy of this thesis has been examined by the signatories, and we find that
both the content and the form meet acceptable presentation standards of scholarly
work in the above mentioned discipline.
iii
Singleton, Cynthia S. (Ph.D., Atmospheric and Oceanic Sciences)
Investigation of Arctic Ozone Loss using Solar Occultation and Microwave Limb
Sounding Instruments
Thesis directed by Dr. Cora E. Randall
Unlike the Antarctic, Arctic ozone loss can undergo large interannual variability due
to the changing dynamics and meteorological conditions. In order to explore the
interannual variability, a comprehensive ozone loss analysis will be presented for ten
Arctic winters. Ozone loss calculations will be inferred from observations from the
Upper Atmosphere Research Satellite Microwave Limb Sounder (UARS MLS), Earth
Observing System MLS (EOS MLS), Polar Ozone and Aerosol Measurement
(POAM II/III), Stratospheric Aerosol and Gas Experiment (SAGE II/III), Improved
Limb Atmospheric Spectrometer (ILAS), Halogen Occultation Experiment
(HALOE), Atmospheric Chemistry Experiment Fourier Transform Spectrometer
(ACE-FTS), and Measurement of Aerosol Extinction in the Stratosphere and
Troposphere Retrieved by Occultation (MAESTRO) instruments using the chemical
transport model (CTM) Passive Subtraction technique. In addition, the inferred
ozone loss calculations will be compared to modeled loss from the SLIMCAT CTM.
Theoretical models have historically underestimated ozone loss rates in much of the
lower polar stratosphere compared to inferred loss rates from observations [e.g.,
Becker et al., 2000; Guirlet et al., 2000; Pierce et al., 2003].
It is vital that CTMs
correctly simulate stratospheric O3, because they are an integral part of the
development of chemistry modules in Chemistry Climate Models. Results from this
iv
thesis will be used to improve the treatment of polar ozone in the CTM and thus
CCMs, with a view toward improving our understanding of the coupling between
stratospheric ozone loss and climate.
Dedication
This thesis is dedicated to my parents and my husband, whose love and
support have helped me to obtain my goals.
vi
Acknowledgements
I would like to gratefully acknowledge my advisor, Dr. Cora Randall, for her
support and guidance. I would also like to thank Dr. Lynn Harvey for her
encouragement and many helpful discussions. Thank you to my committee members:
Dr. Brian Toon, Dr. Dave Rusch, and Dr. Linnea Avallone. Thanks to Ketty for
being such a great friend throughout graduate school.
Finally, I am forever indebted to my parents and my husband for their
understanding, endless patience and encouragement. I am truly the luckiest person in
the world to have such a loving family and for this I am forever grateful.
vii
Contents
1
Introduction
1
1.1 The Discovery of Changes in the Stratospheric O3 Layer……….…….…1
1.2 Research Objectives…..…………….……..…..………….………………3
1.3 Thesis Outline…………….....………...………………………………….4
2
The Chemical and Dynamical Processes Involved in Lower Stratosphere
Arctic Polar O3 Loss
7
2.1 Radiative Properties of O3…………..……………………………………7
2.2 Dynamics of the Polar Stratosphere………...…………………………….8
2.2.1 The Brewer-Dobson Circulation…………………………………...8
2.2.2 The Polar Vortex…………………………………………………..9
2.3 Arctic Polar Chemistry……………………………………………….…12
2.3.1 Chapman Reactions……………………………………………….12
2.3.2 Catalytic Processes – HOx, NOx, ClOx, Brx………………………14
2.3.2.1 Hydrogen Oxide Radicals (HOx)………………..…...…15
2.3.2.2 Nitrogen Oxide Radicals (NOx)……...……………...…16
2.3.2.3 Chlorine Radicals (ClOx)……...…………….....………18
2.3.2.4 Bromine Radicals (Brx)………………………………….19
2.3.2.5 Reservoir Species……………………………………..…20
2.3.3 Polar O3 Loss ………………...…………………………….……..22
2.4 Conclusions…………………………………….………………………..26
3. Arctic O3 Observations
28
3.1 Atmospheric Observations………………………………………………28
viii
3.2 Satellite Passive Remote Sensing Instruments…………………...……..29
3.2.1 Occultation Instruments………………….…………….………….30
3.2.2 Limb Emission Instruments …………….………………………...35
3.3 Conclusions…………………………………………………………..….38
4. Techniques to Quantify Chemical O3 Loss
39
4.1 Introduction……………………………………………………………….39
4.2 Implicit Transport: The O3 -Tracer Correlation Technique…………..….39
4.3 Techniques Utilizing Explicit Transport…………………………………41
4.3.1 Match Technique………………………………………………....41
4.3.2 Vortex Average Technique……………………………………….44
4.3.3 Trajectory Ensemble Technique……………………………...…..45
4.3.4 Chemical Transport Model Passive Subtraction Technique…..…..46
4.4 Conclusions………………………………………………………………48
5. The SLIMCAT CTM
49
5.1 Introduction……………………………………………………………….49
5.2 Off-line CTMs……………………………………………………………49
5.3 Model Structure…………………………………………………………..50
5.4 Model Transport…………………………………………………………..51
5.5 Model Chemistry………………………………………………...……….53
5.6 Conclusions……………………………………………………………….56
6. ACP Paper: “2002–2003 Arctic Ozone Loss Deduced from POAM III Satellite
Observations and the SLIMCAT Chemical Transport Model”
57
6.1 Introduction and Objectives………………………………………………...59
ix
6.2 2002-2003 Meteorology………………………..…………………………...63
6.3 POAM III Observations in 2002-2003……………………………………..65
6.4 SLIMCAT 3-D CTM……………………………………………………….69
6.4.1
Model Description………………………………………………….70
6.4.2
Ozone Initialization…………………………………………………71
6.4.3
Pure Passive and Pseudo Passive Runs……………………………..73
6.5 Ozone Loss During 2002-2003……………………………………………..76
6.5.1
Vortex Averaged Inferred Ozone Loss……………………………..81
6.5.2
CTM-PS Modeled Ozone Loss……………………………………..83
6.6 Summary…...……………………………………………………………….86
7. JGR Paper: “Quantifying Arctic Ozone Loss During the 2004-2005 Winter Using
Satellite Observations and a Chemical Transport Model”
88
7.1 Introduction…………………………………………………………………..90
7.2 Data Sets……………………………………………………………………..92
7.2.1
POAM III……………………………………………………………...92
7.2.2
SAGE III………………………………………………………………93
7.2.3
EOS MLS……………………………………………………………...93
7.2.4
ACE-FTS and MAESTRO……………………………………………94
7.2.5
Satellite Comparisons During the 2004-2005 Winter…………………95
7.3 Methods……………………………………………………………………....97
7.4 Meteorology……………………………………………………………….....99
7.5 Results………………………………………………………………………103
7.5.1
Inferred O3 Loss……………………………………………………...103
x
7.5.2
IL and Instrument Sampling…………………………………………108
7.5.3
Column O3 Loss……………………………………………………...110
7.5.4
Comparisons with 1999-2000 IL Calculations………………………111
7.5.5
Modeled O3 Loss……………………………………………………..114
7.5.6
Satellite O3 & CTM Active O3……………………………………….115
7.6 Summary and Conclusions…………………………………………………118
8. Arctic Ozone Loss Climatology from Solar Occultation and Microwave Limb
Sounding Instruments
122
8.1 Introduction………………………………………………………………...122
8.2 Data Sets……………………………………………………………….…..124
8.2.1
POAM II/III……………………………………………………...…124
8.2.2
SAGE II/III ………………………………………………………..125
8.2.3
ILAS………………………………………………………..………126
8.2.4
HALOE……………………………………………………………..127
8.2.5
ACE-FTS and MAESTRO ………………………………………...127
8.2.6
UARS/EOS MLS…………………………………………………...128
8.3 Methods…………………………………………………………………….130
8.4 Meteorology………………………………………………………………..135
8.5 Inferred Loss……………………………………………………………….139
8.6 Comparison with Other Techniques……………………………………….143
8.7 Modeled Loss……………………………………………..……………….145
8.8 Observed and Simulated O3……………………………………………….150
8.9 Conclusions……………………………………………………………...…153
xi
Bibliography
156
xii
Figures
Figure
6.1 Time series of T-Tnat in the Arctic vortex from 1 December 2002 through 15
March 2003 for the 600 K, 550 K, 500 K, and 450 K potential temperature
surfaces vortex wide. Temperatures are the minimum temperatures inside the
polar vortex and were obtained from Met Office analyses. NAT condensation
temperatures were computed using the Hanson and Mauersberger [1988]
expression, assuming 10 ppbv HNO3 and 5 ppmv H2O………………………..64
6.2
Met Office PV (10-5 Km2 kg-1s-1) at the 500 K potential temperature surface for
specific dates during the 2002-2003 winter from 90º N to 30º N. The inner
vortex boundary is denoted by the solid white contour. The black dotted circle
indicates the POAM measurement latitudes……………………………………65
6.3 Northern Hemisphere equivalent latitudes (dots) and geographic latitudes (solid
curve) of POAM measurements on the 500 K potential temperature surface. Red
indicates measurements taken within the inner edge of the vortex boundary, blue
indicates measurements between the outer and inner edges, and black denotes all
measurements taken beyond the outer edge…………………………………….67
6.4 2002/2003 POAM daily average observations on the 650 K, 500 K, 450 K, and
400 K potential temperature surfaces inside the inner vortex edge (blue) and
xiii
outside the outer vortex edge (red)……………………………………………68
6.5 Comparison of the ozone initialization profiles interpolated to the POAM
measurement locations for 30 November (top row), 1 December (middle row),
and 2 December (bottom row). The left column shows the 1 December
initialization profiles (red) interpolated to the POAM measurement locations
(black) on the dates shown. Average differences between the profiles are shown
in ppmv (middle column) and percent (right column). Error bars denote1σ
standard deviation of the distribution…………………………………………..72
6.6 Comparison of the SLIMCAT Passive ozone (ppmv) inside the vortex at the
POAM measurement locations for the Pure Passive (left) and Pseudo Passive
(middle) runs, and for the difference between the two (right; Pseudo minus
Pure. Ozone mixing ratios have been smoothed using a 7-day running
average………………………………………………………………………….75
6.7 Daily average ozone mixing ratios inside the vortex at the POAM
measurement locations for POAM (black) and the SLIMCAT Pseudo Passive
(red) at the six indicated potential temperatures. Error bars denote 1σ standard
deviation of the averages. Points without error bars indicate that only one POAM
observation was made inside the vortex at a given potential temperature level.
Chemical ozone loss is calculated by subtracting the Passive model from the
POAM data. The Pure Passive is shown by the gray line, without error bars
(which are approximately the same size as the error bars for the Pseudo
Passive)…………………………………………………………………………77
6.8 Time series of the inferred ozone loss in 2002-2003 using the SLIMCAT Pure
xiv
Passive (black) and Pseudo Passive (red) (see text for details). Points represent
daily averages of measurements inside the vortex. The dotted black line denotes
0 ppmv. Error bars denote 1σ standard deviation of the differences. Points
without error bars indicate that only one POAM observation was made inside
the vortex at a given potential temperature level………………………………80
6.9 Inferred ozone loss (ppmv) in 2002-2003, as represented by the difference
between POAM and the SLIMCAT Pure Passive (left) or Pseudo Passive
(middle). The solid black line denotes the zero contour. Loss inferred from the
POAM measurements using the vortex average technique initialized with the
inferred 1 January Pure Passive loss profile is shown in the right panel. Data
have been smoothed using a 7-day running average…………………………...81
6.10 2002/2003 CTM-PS modeled ozone difference (ppmv) at the POAM
measurement locations inside the vortex (left), calculated as the Active model
ozone (middle) minus the Pseudo Passive model ozone (see Figure 7). Negative
values indicate modeled ozone loss. For comparison, the POAM ozone
observations (ppmv) are shown in the right panel. Solid black lines in the left
panel denote 0 differences. Ozone mixing ratios have been smoothed using a 7day running average…………………………………………………………….83
6.11 Comparison of the modeled ozone loss inside the vortex (blue) to the inferred
ozone loss using the Pseudo Passive (red). Error bars denote 1σ standard
deviation of the average differences. Points without error bars indicate that only
one POAM observation was made inside the vortex at a given potential
temperature level………………………………………………………………..84
xv
6.12 2002/2003 POAM (black) and SLIMCAT Active (red) In-V daily average ozone
mixing ratios at the potential temperatures indicated in each panel. “Error” bars
represent the standard deviation of the distribution of measurements/model on
each day………………………………………………………………………...85
7.1 Latitudinal coverage for POAM III (red), SAGE III (blue), MAESTRO (black),
and ACE-FTS (black) (Left) during the 2004-2005 Arctic winter. The right plot
shows equivalent latitudes of observations on the 500 K potential temperature
surface, applying the same color arrangement used in the left plot with EOS MLS
(gray). The white line indicates the innermost edge of the polar vortex…….…95
7.2 Area (106 km2) where Northern Hemisphere temperatures fell below TNAT during
the winters from 1978-1979 to 2003-2004 (colors) and for 2004-2005 (black), for
the 550 K, 500 K, and 450 K potential temperature surfaces…………………..99
7.3 Minimum ECMWF operational temperatures minus daily average TNAT (see text)
inside the vortex at the EOS MLS (gray), ACE-FTS and MAESTRO (black),
POAM III (red), and SAGE III (blue) measurement locations on the 600 K, 500
K, 475 K, and 425 K potential temperature surfaces during the 2004-2005 Arctic
winter…………………………………………………………………………..101
7.4 ECMWF operational PV (10-6 K m2 kg-1s-1) on the 500 K surface for
representative days during the 2004-2005 Arctic winter. The inner vortex edge is
denoted by the white contour. The POAM III, SAGE III, ACE-FTS and
MAESTRO measurement locations are indicated with circles, diamonds, and
crosses, respectively. Latitudes range from the equator to the pole, with latitude
circles drawn in 45 degree increments…………………………………………102
xvi
7.5 Differences (ppmv) between passive O3 calculated by the SLIMCAT CTM and
O3 measured by the POAM III, SAGE III, EOS MLS, ACE-FTS, and
MAESTRO instruments. Results correspond to daily averages over the
measurement locations inside the vortex during the 2004-2005 Arctic winter, and
are indicative of inferred photochemical loss throughout the winter. Days with
missing data and days where an instrument did not sample the vortex have been
filled in with a time interpolation. The solid black line denotes the zero contour.
Data have been smoothed with a 7-day running average……………………...103
7.6 Time series of the inferred daily average O3 loss (ppmv) inside the vortex from
EOS MLS (gray), POAM III (red), SAGE III (blue), ACE-FTS (black), and
MAESTRO (green) for the 600 K, 500 K, 475 K, and 450 K surfaces during the
2004-2005 Arctic winter……………………………………………………….106
7.7 As in Figure 5, but using EOS MLS O3 sampled at the POAM III, SAGE III,
ACE-FTS, and MAESTRO instrument locations during the 2004-2005 Arctic
winter. The middle panel is the same as in Figure 5…………………………..109
7.8 Partial column loss for the EOS MLS (gray), POAM III (red), SAGE III (blue),
ACE-FTS (black), and MAESTRO (green) instruments between the 575 K and
400 K potential temperature surfaces during the 2004-2005 Arctic winter. Only
profiles inside the vortex were included in the calculation…………………….111
7.9 Time series of the inferred daily average O3 loss (ppmv) for POAM III during the
2004-2005 (closed circles) and the 1999-2000 (open circles) Arctic winters. The
inferred loss is shown for 1 December through 15 March for the 575 K, 550 K,
500 K, and 475 K surfaces……………………………………………………..113
xvii
7.10 As in Figure 5, but for modeled daily average O3 loss (ppmv) inside the vortex at
the POAM III, SAGE III, EOS MLS, ACE-FTS, and MAESTRO locations
during the 2004-2005 Arctic winter……………………………………………114
7.11 Daily average vortex O3 (ppmv) during the 2004-2005 Arctic winter for POAM
III, SAGE III, EOS MLS, ACE-FTS, and MAESTRO (top row), the SLIMCAT
Active model interpolated to the satellite locations (middle row), and the
SLIMCAT Active model minus the observations (bottom row). Days with
missing data and days where an instrument did not sample the vortex have been
filled in with a time interpolation. Data have been smoothed with a 7-day
running average………………………………………………………………..116
7.12 Observed (black) and modeled (gray) daily average O3 (ppmv) inside the vortex
for POAM III, SAGE III, EOS MLS, ACE-FTS, and MAESTRO on the 600 K
(top row) and 450 K (bottom row) potential temperature surfaces during the
2004-2005 Arctic winter………………………………………………………117
8.1 The equivalent latitude for all vortex observations made during the Arctic winters
of 1994-1995 through 2004-2005 on the 475 K potential temperature surface.
The values of equivalent latitude are color coded by instrument……………...133
8.2 Comparison of the normalization profiles used to correct each instrument. The
normalization profiles are the average ratio of the instrument O3 profile to the
coincident SAGE2 O3 profile. Error bars represent 1 sigma standard deviations
of the averages…………………………………………………………………134
8.3 Area (106 km2) where Northern Hemisphere MetO temperatures fell below TNAT
during the winters from 1994-1995 to 2004-2005 between the 300 K and 700 K
xviii
potential temperature surfaces. The black line indicates the lowest potential
temperature surface (400 K) included in the O3 loss analyses………………...136
8.4 The vortex strength index (m/s) for the Arctic winters from 1994-1995 to 20042005. A higher vortex strength index indicates the vortex is more stable and air
is more likely to be contained within the vortex………………………………137
8.5 Differences (ppmv) between passive O3 calculated by the SLIMCAT CTM and
the combined satellite O3 fields. Results correspond to daily averages over the
measurement locations inside the vortex during the ten Arctic winters between
400 K and 700 K. Days with missing data and days where an instrument did not
sample the vortex have been filled in with a linear time interpolation. The solid
black line denotes the zero contour. Data have been smoothed with a 7-day
running average. White spaces in the contour plots at the end of the winter (e.g.
1999-2000) indicate that vortex observations were no longer made for the
remainder of the analysis period……………………………………………….140
8.6 Time series of the inferred daily average O3 loss (ppmv) inside the vortex from
the combined satellite O3 fields for the 600 K, 500 K, 475 K, and 450 K surfaces
for the ten Arctic winters………………………………………………………142
8.7 As in Figure 5, but for modeled daily average O3 loss (ppmv) inside the vortex at
the combined satellite locations during the ten Arctic winters………………...146
8.8 As in Figure 6, but for modeled daily average O3 loss (ppmv) inside the vortex at
the combined satellite locations during the ten Arctic winters………………...147
8.9 Maximum IL (black) and ML (red) O3 calculations (left panel) and their
corresponding altitude (right panel) for the ten Arctic winters………………...148
xix
8.10 The average O3 loss for the last 14 days of the analysis for the observations
(black) and the model (red) on the 450 K (left) and 600 K (right) potential
temperature surface……………………………………………………………149
8.11 Daily averages of observed O3 at the combined measurement locations inside the
vortex during the ten Arctic winters. Days with missing data and days where an
instrument did not sample the vortex have been filled in with a time
interpolation. Data have been smoothed with a 7-day running average……..151
8.12 As in Figure 11, but for modeled daily average O3 (ppmv) inside the vortex at
the combined satellite locations during the ten Arctic winters………………..152
1
Chapter 1
Introduction
1.1
The Discovery of Changes in the Stratospheric O3 Layer
In a 1974 paper to Nature, scientists Mario Molina and Sherwood Rowland
first hypothesized that commonly used household chemicals could potentially damage
the Earth’s stratospheric ozone (O3) layer [Molina and Rowland, 1974]. Molina and
Rowland [1974] suggested that halogen source gases, called chlorofluorocarbons
(CFCs) were the culprit. At the time CFCs had been widely used since the 1960s as
refrigerants and propellants in aerosol spray cans because the compounds were
chemically stable and nontoxic. Molina and Rowland [1974] theorized that the stable
chemical compounds could remain in the atmosphere for 40-150 years allowing the
compounds to vertically diffuse into the stratosphere. Once in the stratosphere, the
scientists theorized that the CFCs would photodissociate to produce significant
amounts of chlorine atoms, which could then catalytically destroy O3. Molina and
Rowland [1974] stressed that important consequences would result if too much
chlorine was injected into the stratosphere and that scientific studies were needed in
the immediate future to ascertain the possible onset of environmental problems
caused by the CFCs.
2
The implications of Molina and Rowland [1974] were particularly alarming to
the scientific community because of the protective qualities of the stratospheric O3
layer. The O3 layer is a term used to describe a thin layer of the atmosphere that
contains more O3 than other regions and is located between 15 and 30 km in Earth’s
stratosphere. The O3 layer is vital to biological existence on Earth because it absorbs
harmful incident ultraviolet radiation. Ultraviolet radiation, located between 200 nm
and 400 nm of the electromagnetic spectrum, is divided into three regions: UV-A
(320-400 nm), UV-B (280-320 nm), and UV-C (200-280 nm). Of the three types of
UV radiation, UV-C is the only form that is entirely absorbed by O3. This absorption
occurs throughout the upper atmosphere and is complete by approximately 35 km.
Both UV-A and UV-B are moderately absorbed by O3. Of these two types, UV-B is
the most harmful to human and plant life [Leaf, 1993]. It has been hypothesized that
an increase in exposure to UV-B would result in increased incidence of skin cancer
[Abarca and Casiccia, 2002]. Laboratory experiments have indicated plant
hormones and chlorophyl can be damaged with increased exposure to UV light,
resulting in the slowing of plant growth [Liftin, 1994]. Therefore, the thinning of the
stratospheric O3 layer, which would result in the increase in the amount of UV-B
reaching the surface, would have adverse effects on both plant and human life [Leaf,
1993].
The first observations of a drastic reduction in O3 were reported by scientists
from the British Antarctic Survey in 1985. In a classic paper, Farman et al. [1985]
showed that a massive reduction in O3 had occurred during the spring months over
Antarctica. Farman et al. [1985] found that O3 observations taken in 1984 were
3
nearly 40 percent lower than measurements taken 20 years earlier. In their
groundbreaking paper, Farman et al. [1985] concluded that “additional chlorine
might enhance O3 destruction in the cold spring Antarctic stratosphere” and more
observations were needed in the polar night to “improve considerably the prediction
of effects on the O3 layer of future halocarbon releases”.
After the publication of Farman et al. [1985] the international scientific
community went to work to determine the chemical mechanisms causing
stratospheric O3 loss. Today the main chemical loss mechanisms are no longer a
mystery, but now one of the main concerns about the stratospheric O3 layer is how it
will be affected in a changing climate. In order to accurately predict how O3 levels
will respond, we need to accurately model the dynamics and chemistry involved in
polar O3 processes. However, despite much attention paid to the stratospheric O3
problem, numerous theoretical models routinely underestimate O3 loss rates in much
of the lower polar stratosphere compared to “observed” loss rates [e.g., Singleton et
al., 2005; Becker et al., 2000; Guirlet et al., 2000; Deniel et al., 1998; Goutail et al.,
1997; Chipperfield et al., 1996]. WMO [2003] states that, “these uncertainties
prevent reliable predictions of future Arctic O3 losses in a potentially changing
climate.” Therefore, accurately quantifying polar O3 loss with theoretical models is
key to predicting the recovery of O3 as the abundances of O3-destroying chemicals
decrease and new substitutes increase.
1.2
Research Objectives
4
One of the complications to quantifying O3 loss is that no direct observations
of chemical O3 loss rates exist. Rather, chemical loss rates must be inferred from the
measurements with a priori knowledge of, or assumptions about, the O3 variations
resulting from dynamical processes. This thesis will address whether discrepancies
between modeled and inferred O3 loss rates are due to errors inherent in the technique
by which O3 loss is inferred from the observations, to errors in the chemical and/or
dynamical schemes used in the models, or to both. In so doing, the use of a global,
three-dimensional chemical transport model (CTM) will be optimized to both infer O3
loss rates from satellite observations, and to calculate O3 loss rates during winters
with differing meteorological conditions. It is vital that CTMs correctly simulate
stratospheric O3, because CTMs are an integral part of the development of chemistry
modules in Chemistry Climate Models (CCMs). Therefore, improvements made to
CTMs will lead to more accurate calculations of the coupling between climate and
polar O3 loss, once the model is incorporated into a CCM.
1.3
Thesis Outline
This thesis is separated into two sections, a background (Chapters 2 through
5) and a research section (Chapters 6 through 8). Chapter 2 provides an overview of
the important radiative features of O3, followed by the main chemical and dynamical
processes involved in Arctic O3 loss. The satellite instruments used to infer O3 loss
and the technique that was used to infer the loss are discussed in Chapters 3 and 4,
respectively. The final background chapter, Chapter 5, describes the University of
5
Leeds’ SLIMCAT CTM that was employed to quantify both the inferred and modeled
O3 loss results. Chapters 6 through 8 comprise the research section of this thesis.
Chapter 6 is a paper that was published in 2005 in the Journal of Atmospheric
Chemistry and Physics and is entitled, “2002–2003 Arctic O3 loss deduced from
POAM III satellite observations and the SLIMCAT chemical transport model”. As
the name suggests, the objective of the paper was to infer O3 loss results for the 20022003 Arctic winter from POAM III satellite observations. These results were
compared to modeled O3 loss results inferred from the SLIMCAT CTM. In addition,
the effects of gas-phase chemistry on O3 loss inferences were explored and O3 loss
results from different techniques were compared. Chapter 7 is a paper that is
currently in press at the Journal of Geophysical Research and is entitled, “Quantifying
Arctic O3 loss during the 2004-2005 winter using satellite observations and a
chemical transport model”. This paper describes inferred O3 loss calculations for the
2004-2005 Arctic winter using data from four solar occultation satellite instruments,
as well as a microwave limb sounder. Since the instruments had different latitudinal
sampling of the Arctic polar region, the sensitivity of the inferred loss calculation to
the geographic sampling of the instrument was evaluated. Finally, the modeled loss
from the SLIMCAT CTM was compared to the inferred loss calculations to determine
how well the CTM was able to simulate O3 loss during the 2004-2005 Arctic winter.
The last chapter of this thesis, Chapter 8, is entitled, “Arctic Ozone Loss Climatology
from Solar Occultation and Microwave Limb Sounding Instruments”. In this paper a
climatology of the inferred and modeled O3 loss results for ten Arctic winters is
presented. The objective of this chapter is to interpret the interannual variations in
6
inferred O3 loss based on our knowledge of the varying meteorological conditions.
The O3 loss calculated by the SLIMCAT model is compared to the inferred
calculations to determine how well the model was able to simulate O3 loss under
varying meteorological conditions.
7
Chapter 2
The Chemical and Dynamical Processes Involved in Lower Stratosphere Arctic
Polar O3 Loss
In order for O3 loss to take place in the Arctic, important dynamical and
chemical processes must occur. This chapter will provide an overview of the
important radiative features of O3, followed by the main chemical and dynamical
processes involved in Arctic O3 loss.
2.1
Radiative Properties of O3
Unlike molecular nitrogen (N2) and molecular oxygen (O2), which combined
make up approximately 99% of Earth’s atmosphere by volume, O3 is a trace
atmospheric constituent. Although O3 makes up less than 1% of Earth’s atmosphere
by volume, it is one of the most important gases in the atmosphere because it is
radiatively active. Stratospheric O3 absorbs in the infrared (IR), visible, and
ultraviolet (UV) regions of the electromagnetic spectrum. In the IR, O3 has a strong
absorption band within the atmospheric window (8 to 12 µm) near 9.6 µm, due to
vibrational transitions. Since O3 absorbs thermal radiation it is considered a
greenhouse gas. Absorption by O3 in the UV or visible part of the spectrum is due to
electronic transitions. In the UV, O3 absorbs between 200 and 300 nm in the upper
8
stratosphere. This region of the spectrum is called the Hartley band. Also in the UV,
O3 has slightly weaker absorption bands between 300 and 350 nm, called the Huggins
bands. Absorption in the Hartley and the Huggins bands provides the dominant
source of heat in the stratosphere [Salby, 1996]. O3 also absorbs radiation in the
visible part of the spectrum from 440 to 740 nm. This feature, called the Chappuis
band, is weaker than the Hartley and Huggins bands.
2.2
2.2.1
Dynamics of the Polar Stratosphere
The Brewer-Dobson Circulation
Although O3 is produced mainly in the tropics through photodissociation of
oxygen in the UV [WMO, 2003], the maximum observed column abundances of O3
occur at middle and high latitudes. This phenomenon occurs because of the important
role of dynamics in the stratosphere. Early observations of tracers by Brewer [1949]
and Dobson [1956], suggested upwelling of air in the tropics, deflection into the
summer stratosphere and mesosphere, followed by cross-equatorial flow near the
mesopause and downwelling at mid and high latitudes in the winter hemisphere.
These observations explain the main circulation pattern in the stratosphere, which
today is commonly referred to as the Brewer-Dobson circulation. The BrewerDobson circulation is the large-scale diabatic circulation that is responsible for the
gradual meridional overturning and affects the distribution of long-lived chemical
species in the stratosphere. Because of the Brewer-Dobson circulation, O3-rich air
9
from the tropics is transported to the polar regions. As the air descends in the polar
regions it undergoes compression, which results in the greatest O3 density and column
abundances at higher latitudes. This circulation pattern occurs in both the Northern
and Southern Hemisphere; however, it is stronger in the Northern Hemisphere (NH),
which results in larger column O3 in the Arctic [WMO, 2003].
2.2.2 The Polar Vortex
One of the main dynamical features of the winter stratosphere is the polar
vortex. During the autumn months, the polar regions begin to receive less sunlight
and gradually become colder than lower latitudes [e.g., Ray et al., 2002]. By the
winter, a strong temperature gradient develops between the cold polar latitudes and
warmer lower latitudes due to the absence of solar heating in the polar region. As a
result of the large temperature gradient in the winter hemisphere, a strong westerly
flow develops. As the temperature gradient increases, the westerlies intensify to
produce a polar-night jet. The polar-night jet forms the edge of the cyclonically
spinning polar vortex and acts as a barrier to meridional transport, isolating air within
the polar vortex. This occurs because the vortex edge is characterized by regions of
strong potential vorticity, which act as a mixing barrier. The mixing barrier prevents
extra-vortex air to mix with vortex air. However, it is shown below that dynamical
perturbations can cause large-scale mixing events that can lead to exchange of air
across the vortex edge [WMO, 2003]. During the winter the vortex experiences net
radiative cooling because the polar air is warmer than radiative equilibrium [Salby,
10
1996]. This radiative cooling causes descent, which transports O3 and other gases
downward into the vortex throughout the winter [e.g., Rosenfield et al., 1994; 2001].
The polar vortex forms in the winter months in both the Southern Hemisphere
(SH) and the NH. However, the SH polar vortex tends to be very stable and circular,
unlike its NH counterpart [Waugh and Randel, 1999]. The asymmetry in the NH
vortex is due to the topographical features in the NH [WMO, 2003], which cause
strong dynamical forcings at the earth’s surface. These forcings, called planetary
waves, propagate up from the troposphere and are able to penetrate the strong winter
westerlies of the stratosphere. In the NH these planetary waves can drive the
circulation even farther away from radiative equilibrium and consequently can cause
irreversible heat transfer [Salby, 1996]. These disturbances, which occur regularly
during the NH winter, disrupt the zonal flow of the polar vortex, making it weaker.
Planetary waves also penetrate the SH vortex; however, they are much weaker and
produce smaller departures from local radiative equilibrium, which leaves the
Antarctic vortex considerably colder than the Arctic [Salby, 1996]. In addition, the
weaker planetary wave activity in the SH results in a weaker Brewer-Dobson
circulation in the SH, making the Antarctic polar vortex much colder [WMO, 2003].
As a result of these topographical differences, there is stronger descent in the NH,
which leads to larger amounts of O3 transported into the Arctic vortex [WMO, 2003].
Unlike the formation of the vortex, which is caused by radiative events, the
vortex breakup is a dynamical phenomenon that can occur on a much shorter
timescale [Staehelin et al., 2001]. The event that causes the breakup of the Arctic
polar vortex is called a final stratospheric warming event. During both midwinter and
11
final stratospheric warming events, temperatures can increase in the polar region by
as much as 50 K [Salby, 1996] in a few days, and the event is generally accompanied
by an increase in planetary waves that originate in the troposphere [Harris et al.,
2002]. The increase in temperature in the polar region lessens the equatorward
temperature gradient, resulting in a weakening of the zonal flow [e.g., Limpasuvan et
al., 2004]. Generally, the stratospheric circulation will become highly asymmetric,
which can cause the polar vortex to be displaced from the pole and may lead to the
splitting of the vortex [e.g., Limpasuvan et al., 2004]. During the final stratospheric
warming event, the planetary waves lead to the deceleration of the westerlies or the
reversal of the zonal-mean winds to easterlies [Salby, 1996]. Once the final breakup
of the polar vortex has occurred in the spring, the zonal winds in the stratosphere
resume their summer easterly pattern [Staehelin et al., 2001].
In the Arctic, stratospheric warming events can also occur prior to the final
breakup of the vortex. These midwinter events, if strong enough, can dramatically
weaken the vortex and also cause it to split. If temperatures remain cold after a
midwinter warming, the vortex can recover, but is typically less stable than prior to
the warming. One of the main consequences of having a midwinter stratospheric
warming event is that mixing can occur between air from inside and outside the
vortex, which can impact the amount of O3 loss that occurs during a given winter.
The NH vortex is very difficult for models to simulate, partly due to the variability
associated with the stratospheric warming events. In the NH, stratospheric warming
events can change annually in terms of number, strength, and timing of the events
[e.g., Kodera et al., 2000].
12
In the NH, the characteristics of the vortex, e.g., whether it is cold and stable
(like the SH) or dynamically perturbed and warmer, have large consequences for the
amount of O3 loss that is observed during the winter and early spring. Chapters 6
through 8 will show how winters with different meteorological conditions can lead to
changes in Arctic O3 loss.
2.3 Arctic Polar Chemistry
2.3.1 Chapman Reactions
In 1930 British scientist Sydney Chapman proposed that the formation of O3
was initiated in the stratosphere by photolysis of O2 [Chapman, 1930] and that this
reaction led to the formation of the stratospheric O3 layer. Although Chapman’s
theory was not complete, it provided the basic foundation for stratospheric O3
chemistry. Today, we refer to this foundation as the Chapman mechanism.
The first reaction of the Chapman mechanism involves the photodissociation of
molecular oxygen at wavelengths less than 240 nm (UV) to create two O atoms:
O2 + hv → O + O
(1)
Here the O atoms are in the ground-level triplet state O (3P). Next, the O atoms
combine with O2 in the presence of a third molecule (where M is either N2 or O2) to
form O3 [Chapman, 1930]. Reaction (2) occurs quickly because the O atoms are
highly reactive as a result of their unpaired electrons [Jacob, 1999]:
O + O2 + M → O3 + M
(2)
13
O3 molecule absorbs photons at wavelengths between 240 and 320 nm to decompose
back to O2 and an excited singlet state O atom, O(1D).
O3 + hv → O2 + O
(3)
The final sink for O3 requires the O atom to be converted to O2, which is reaction (4)
in the Chapman mechanism:
O3 + O → 2O2
(4)
Reactions 2 and 3 both proceed quickly; therefore, O and O3 are rapidly interconverted. Chapman [1930] was the first to predict the rapid conversion between O
and O3. Because of this rapid conversion it is difficult to separate these two species,
so the sum of O and O3, often referred to as the odd oxygen chemical family ([Ox] =
[O] + [O3]), is used instead. Below approximately 50 km, O3 is the dominant form
of Ox. Since reactions 2 and 3 occur quickly, the rate of Ox production is not
dependent on reaction 2; therefore, it is only dependent on the rate of production of
atomic O in reaction 1 [Seinfeld and Pandis, 1998]. Likewise, the rate of Ox
destruction is governed by reaction 4.
The rate of O3 formation, which is governed by reaction 1, depends on the
amount of incoming solar radiation, which varies by latitude, altitude, and season
[Seinfeld and Pandis, 1998]. Based upon the Chapman mechanism, the regions with
the highest rates of O3 production are found at the equator at altitudes above 40 km
[Dessler, 2000]. However, it is important to point out that although the rates of O3
production are the largest in the tropics, the lowest column O3 values are actually
found in the tropics and the largest column O3 is found in the high latitudes during the
14
winter. This apparent inconsistency is the result of the stratospheric transport that
was discussed in the preceding section.
The Chapman mechanisms were believed to be the governing set of chemical
reactions controlling stratospheric O3 until the late 1950s. During this time
comparisons to observations showed that the theory overestimated O3 concentrations
in the upper atmosphere by a factor of two or more. This discrepancy could be
explained if the source of O3 was too large, which was not likely (because reaction (1)
was well constrained by spectroscopic data), or if there was a missing sink that was
not accounted for in the Chapman mechanism [Jacob, 1999].
2.3.2 Catalytic Processes – HOx, NOx, ClOx, Brx
It was not until the 1970s that the atmospheric scientific community
determined that catalytic loss processes were also responsible for stratospheric O3
loss. Early studies [e.g., Bates and Nicolet, 1950; Crutzen, 1970; Stolarski and
Cicerone, 1974; McElroy et al., 1986] demonstrated that catalytic reactions were an
important part of the stratospheric O3 budget. The O3 loss mechanism for the
catalytic cycle has the following form:
X + O3 → XO + O2
(5)
XO + O → X + O2
(6)
Net: O3 + O → 2O2
where X represents a free-radical catalyst, which, in the stratosphere, can be OH, NO,
Cl, or Br. XO represents the intermediate of the catalytic cycle. The importance of
15
the cycle is dependent on the concentration of X and the rate of each reaction
involved in the cycle [Seinfeld and Pandis, 1998]. The following paragraphs will
discuss the main catalytic cycles involved in the destruction of stratospheric O3.
2.3.2.1 Hydrogen Oxide Radicals (HOx)
The first of the catalytic cycles to be postulated involved hydrogen radicals
[Bates and Nicolet, 1950] and is now referred to as the HOx cycle. HOx is a chemical
family comprised of OH (hydroxyl radical) and HO2 (hydroperoxy radical), and H
(atomic hydrogen). H is generally considered a negligible component in the
stratosphere; therefore, catalytic reactions involving H will not be discussed here. In
the stratosphere, there are three main HOx cycles. The first HOx cycle, cycle 1, is:
OH + O3 → HO2 + O2
(7)
HO2 + O → OH + O2
(8)
Net: O3 + O → 2O2
OH, HO2, and O increase with altitude; however, O3 decreases with altitude. Cycle 1
is important at altitudes below 40 km [Seinfeld and Pandis, 1998]. Both cycles 2 and
3 have a slightly different format than the traditional catalytic cycle outlined in
reactions 5-7. Cycle 2 is
OH + O → H + O2
(9)
H + O2 + M → HO2 + M
(10)
HO2 + O → OH + O2
(8)
Net: O + O + M → O2 + M
16
where reaction 9 depends on the concentration of O. As a result, cycle 2 is only
important in the upper stratosphere [Seinfeld and Pandis, 1998]. In cycle 3, the only
form of Ox involved is O3:
OH + O3 → HO2 + O2
(11)
HO2 + O3 → OH + O2 + O2
(12)
Net: O3 + O3 → O2 + O2 + O2
As a result, cycle 3 is only important at altitudes below 30 km [Seinfeld and Pandis,
1998].
The primary source of HOx in the stratosphere is the oxidation of H2O by
O(1D) [Dessler , 2000]. O(1D), as shown in reaction 3, is formed by the photolysis of
O3. The products of the oxidation of H2O are 2 OH radicals. Another important
source of HOx in the lower stratosphere is from the oxidation of CH4 (methane) by
O(1D), where one of the products is a OH molecule [Seinfeld and Pandis, 1998].
Since the main sources of HOx require O(1D), which is formed by UV radiation, the
abundance of HOx goes to zero at night in the middle atmosphere [Dessler, 2000].
2.3.2.2 Nitrogen Oxide Radicals (NOx)
The second important set of catalytic reactions involves the NOx chemical
family. Like HOx, NOx is shorthand for a family of short-lived, O3-destroying
species. For stratospheric reactions, NOx is typically defined as the sum of NO (nitric
oxide) and NO2 (nitrogen dioxide). The potential negative impact of NOx species on
the stratospheric O3 layer was first theorized by Crutzen [1970].
17
The main source of NOx into the stratosphere is through transport of N2O
(nitrous oxide) from the troposphere. N2O is a stable molecule that is produced both
naturally (e.g., from forest soils and the ocean) and anthropogenically (e.g., from
combustion and biomass burning). Because of the stable nature of the molecule it can
be easily transported to the stratosphere where it will either photolyze or react with
O(1D). The main reactions that lead to the production of NOx are:
N2O + O(1D) →NO + NO (13)
N2O + hν →NO + N(4S)
(14)
Although these reactions do not account for the main loss of N2O, which is the
photolysis of N2O to N2 and O(1D), they are the most important suppliers of NOx in
the stratosphere. Of the two reactions, reaction 13 is the most important [Dessler,
2000]. Reaction 14 is mostly confined to the tropics because of the larger insolation
and subsequently larger photolysis rates.
The second source of odd nitrogen in the stratosphere is energetic particle
precipitation [e.g., Randall et al., 2006a, b]. Routinely, energetic particles provide a
continuous source of NOx to the lower thermosphere and mesosphere [Randall et al.,
2006a, b]. The NOx so produced can descend during the polar night to supply NOx to
the stratosphere. In addition, but more rarely, very energetic electrons and protons
can produce NO directly in the stratosphere. Once in the stratosphere, the NOx can
take part in the catalytic NOx cycle.
The main NOx catalytic cycle involved in the destruction of O3, which is
analogous to the HOx cycle, is:
NO + O3 → NO2 + O2 (15)
18
NO2+ O → NO + O2
(16)
Net: O3 + O → O2 + O2
The net result of this cycle is to convert two odd oxygen species into two oxygen
molecules. The NOx catalytic cycle plays a dominant role in the catalytic destruction
of O3 between 22 and 40 km. There are additional reactions in which NOx and other
stratospheric species jointly destroy O3; these reactions will be discussed later in this
section.
2.3.2.3 Chlorine Radicals (ClOx)
The research of Molina and Rowland [1974] and Stolarski and Cicerone
[1974] led to the realization that human-produced CFCs (chlorofluorocarbons) could
photolyze in the stratosphere where ample UV light of appropriate energy (between
185 and 210 nm) could release Cl (chlorine atoms). CFCs were first manufactured in
the 1930s and became increasingly popular for both industrial and home use (e.g.,
refrigerants and aerosol spray propellants). The two most prevalent CFCs were
CFCl3 and CF2Cl2, which are more commonly referred to as CFC-11 and CFC-12,
respectively. Once in the stratosphere, CFC-11 and CFC-12 can undergo the
following reactions
CFCl3 + hv → CFCl2 + Cl
(17)
CF2Cl2 + hv → CF2Cl + Cl (18)
Both of these reactions are generally confined to the tropics and occur between 25
and 35 km and near 23 km for CFC-12 and CFC-11, respectively [Seinfeld and
19
Pandis, 1998]. Another source of stratospheric chlorine is from the photolysis of
CH3Cl (methyl chloride). Unlike CFCs, CH3Cl is a natural source of stratospheric
chlorine. Molina and Rowland [1974] and Stolarski and Cicerone [1974] suggested
that once Cl was released in the stratosphere, it could destroy stratospheric O3. These
important studies led to the discovery of the catalytic loss mechanisms involving
active chlorine. The active chlorine species make up the ClOx family, which is
defined as the sum of Cl and ClO (chlorine monoxide). The majority of ClOx
throughout the stratosphere is in the form of ClO [Dessler, 2000].
The main catalytic cycle involving ClOx and Ox is [Stolarski and Cicerone,
1974]:
Cl + O3 → ClO + O2
(19)
ClO+ O → Cl + O2
(20)
Net: O3 + O → 2O2
This cycle occurs primarily in the upper stratosphere because of the larger
concentration of atomic oxygen relative to the lower stratosphere. The second
reaction is the rate-limiting step for this catalytic cycle. The net result of this cycle is
to convert two Ox to two molecules of diatomic oxygen.
2.3.2.4 Bromine Radicals (Brx)
The last important family involved in the catalytic destruction of O3 is
bromine. The Brx family is comprised of Br (atomic bromine) and BrO (bromine
monoxide). Stratospheric bromine comes from tropospheric sources of Br that are
20
produced from natural and anthropogenic sources [Dessler, 2000]. The main sources
of bromine to the stratosphere are CH3Br (methyl bromide), CBrClF2 (Halon-1211),
and CBrF3 (Halon 1301). In the upper stratosphere, where O is more abundant, the
following reaction becomes important:
BrO + O → Br + O2 (21)
Br + O3 → BrO + O2 (22)
Net: O3 + O → 2 O2
In the lower stratosphere the majority of Brx is BrO; however in the upper
stratosphere there are roughly equal amounts of Br and BrO [Dessler,2000].
2.3.2.5 Reservoir Species
The aforementioned catalytic cycles will keep destroying O3 until the reactive
forms of ClOx, HOx, NOx, and Brx are converted to reservoir species. Reservoir
species are the members of a chemical family which are relatively inert and, as a
result, have much longer lifetimes. Once an active species is tied up in its reservoir
form it can no longer participate in catalytic destruction of O3, unless and until it is
converted back into the active form. The HOx, NOx, and ClOx families all have
reservoir species. Brx also has reservoir species; however, the reservoirs are very
short lived because they are easily converted back to reactive forms of bromine
through photolysis.
The HOx catalytic cycles are disrupted when the following reactions occur
OH + HO2 → H2O + O2 (23)
21
OH + NO2 +M → HNO3 + M (24)
HO2 + NO2 + M → HNO4 + M (25)
These reactions convert species of the HOx family into their reservoir form H2O,
HNO3 (nitric acid), and HNO4 (peroxynitric acid). If the reservoir species are not
transported into the troposphere they can eventually be converted back to HOx. H2O
can react with O(1D) to produce two OH radicals. HNO3 and HNO4 can undergo
photolysis to release HOx.
The main reservoir species for NOx are HNO3, HNO4, ClONO2 (chlorine
nitrate), and N2O5 (dinitrogen pentoxide). In order to form HNO3 and HNO4, NOx
reacts with HOx in the termolecular reactions 24 and 25. N2O5 is formed by:
NO3 + NO2 + M → N2O5 + M (26)
NOx reacts with ClOx to form ClONO2 in the following reaction:
ClO + NO2 + M→ClONO2 + M (27)
These reservoir species can be converted back to NOx through photolysis.
ClOx species can also undergo reactions that will convert them to reservoir
form. In addition to reaction 27, the other reactions involving ClOx are:
Cl + CH4 → HCl + CH3 (28)
ClO + HO2 → HOCl + O2 (29)
The reservoir species for ClOx are HCl (hydrochloric acid), HOCl (hypochlorous
acid), and ClONO2. Like the NOx reservoir species, all three ClOx reservoirs can be
converted back to their active form by photolysis (however, HCl is more likely to be
converted back by reacting with OH or the other ClOx reservoir species).
22
2.3.3 Polar O3 Loss
In 1985 the scientific community was taken by surprise when a massive
decrease in Antarctic O3 was first reported by Farman et al. [1985]. This major
decrease in stratospheric O3 over the Antarctic was later referred to as the “O3 hole”.
Two years later Hofmann et al. [1987] showed that the O3 loss was occurring below
the middle stratosphere where most known catalytic cycles peaked. The scientific
community realized that these known gas-phase reactions alone would not lead to the
amount of or location of loss that was observed in the Antarctic, thus a race ensued to
find the missing chemistry responsible for the decrease in O3.
After the 1985 discovery, observations that followed [e.g., Anderson et al.,
1989] indicated that there were high ClO concentrations in the polar region that were
at times anticorrelated with O3 in the polar regions during the winter. As a result of
this discovery, the two main catalytic cycles involved in lower stratospheric O3
depletion were discovered [WMO, 2003]. Both cycles involve halogen oxide radicals.
The first cycle, proposed by Molina and Molina [1987], involves a termolecular selfreaction of ClO (where M is N2 or O2).
ClO + ClO + M → ClOOCl + M
(30)
ClOOCl + hv → ClOO + Cl
(31)
ClOO + M → Cl + O2 + M
(32)
2[Cl + O3 → ClO + O2]
(33)
Net:
2O3 → 3O2
23
Because reaction 30 involves M, it occurs more readily at higher pressures. In order
for this cycle to be significant it would require high concentrations of ClOx. The
second cycle that was discovered involved both bromine and chlorine [McElroy et al.,
1986; Tung et al., 1986].
BrO + ClO → Br + Cl + O2 (34)
→ BrCl + O2
(35)
BrCl + hv → Br + Cl
(36)
Cl + O3 → ClO + O2
(37)
Br + O3 → BrO + O2
(38)
Net: 2 O3 → 3 O2
For the above cycles to destroy a sufficient amount of O3 to account for that
observed over Antarctica, they require more ClOx than the gas-phase reactions
produce [Seinfeld and Pandis, 1998]. In order to account for the additional ClOx,
Crutzen et al. [1986], McElroy et al. [1986], and Solomon et al. [1986] suggested that
chlorine could be converted (or “activated”) from reservoir form (HCl and ClONO2)
to ClOx by chemical reactions occurring on the surfaces of polar stratospheric clouds
(PSCs) and that the enhanced ClOx resulting from these reactions would lead to
enhanced lower stratosphere O3 losses. Today we refer to the reactions that occur on
the surfaces of the PSCs (and other particles) as heterogeneous reactions. The key
heterogeneous reactions that are important for polar O3 loss are
ClONO2 + HCl → Cl2 + HNO3
(39)
ClONO2 + H2O → HOCl + HNO3
(40)
HOCl + HCl → H2O + Cl2
(41)
24
HOBr + HCl → BrCl + H2O
(42)
The chlorine-containing products (Cl2, HOCl, and BrCl) of the heterogeneous
reactions photolyze with the return of sunlight in the spring. The active chlorine
released will then undergo the catalytic cycles to continue destroying O3 [Solomon,
1999].
PSCs are separated into two categories, type I and type II [Turco et al., 1989].
Type I PSCs are composed of particles containing nitric acid (HNO3). Lidar
depolarization studies have indicated that the HNO3 PSCs can exist in solid and liquid
phase; therefore, type I PSCs are further subdivided based on the phase of the particle
[Browell et al., 1990]. Type Ia are believed to be solid particles made of nitric acid
trihydrate (NAT, or HNO3·3H2O) [Hanson and Mauersberger, 1988; Voigt et al.,
2000]. This type of PSC typically forms at temperatures below which NAT can
condense. This threshold temperature occurs approximately at 195 K in the lower
stratosphere and is commonly referred to as TNAT. Type Ib PSCs are believed to be
composed of liquid particles of supercooled ternary solutions (STS) of
HNO3/H2SO4/H2O [Carslaw et al., 1994] and also are observed at temperatures
similar to those of type Ia PSCs [Drdla et al., 2003]. Type Ia and Ib particles have
characteristic radii of 0.5 to 1 micron [Dessler, 2000]. However, low concentrations
(<10-2 cm-3) of large, solid-HNO3 particles have been observed during cold Arctic
winters [WMO, 2003]. Aircraft observations made by Fahey et al. [2001] taken
during the 1999-2000 Arctic winter indicate that solid particles with diameters of 10
to 20 microns were present during the winter. Fahey et al. [2001] referred to these
large particles as “NAT rocks”. Additional studies have suggested that the large,
25
solid HNO3 particles have been present during other Arctic winters [Poole et al.,
2003; Larsen et al., 2004; Voigt et al., 2005].
The second PSC category is referred to as type II. Type II PSCs are
composed of ice particles that form at temperatures below the ice frost point (~188 K)
in the lower stratosphere. Temperatures within the NH vortex rarely fall below the
ice frost point; therefore, type II PSCs are less frequently observed in the NH than in
the SH [e.g., Massoli et al., 2006]. However, temperatures in the NH often fall below
195 K; thus type I PSCs occur annually in the Arctic [e.g., Poole et al., 2003; Voigt et
al., 2005]. Type II particles are generally much larger than type I particles and have
characteristic radii of 5-20 microns [e.g., Drdla and Turco, 1991].
Since both type I and II PSCs require such low temperatures to form they are
only present during the polar winter. In the Arctic, temperatures can vary from year
to year; therefore, the number of PSCs and hence the amount of O3 loss varies
interannually [e.g., Pawson and Naujokat, 1999]. Although it has been well
understood that PSCs play an important role in stratospheric O3 chemistry, there are
many unresolved issues concerning the formation processes of type I and II PSC
particles and this is presently an area of active research.
In addition to activating chlorine, PSCs affect O3 by removing active nitrogen
from the lower stratosphere [WMO, 2003]. Once a PSC particle grows sufficiently,
the particle can sediment. Typically only solid particles can grow to a large enough
size to sediment [Larsen et al., 2004]. If HNO3 is permanently removed from the
stratosphere by the sedimenting particle, the process is referred to as irreversible
denitrification [Drdla et al., 2003]. By removing HNO3 from the stratosphere,
26
denitrification helps to exacerbate O3 loss by preventing chlorine from being tied up
in reservoir form (ClONO2) [Solomon, 1999]. The winter temperatures are not as
cold in the Arctic compared to the Antarctic, because of the enhanced wave activity
[e.g., Manney et al., 1996]. As a result, the Arctic generally experiences fewer PSCs
and PSC events last half as long as the Antarctic [WMO, 2003]; therefore, the
denitrification in the Arctic is much less severe [Tabazadeh et al., 2000].
Observations have shown that denitrification in the Arctic only removes from 17 to
50% of HNO3, with larger amounts of HNO3 removal occurring during the colder
winters [Dessler et al., 1999; Irie et al., 2001; Santee et al., 2000, 2002], and is
confined to a few kilometers (approximately 18 to 21 km) [e.g., WMO, 2003]. If
temperatures are low enough for ice particles (type II PSCs) to form, a water removal
process called dehydration can occur. Unlike denitrification, dehydration can serve to
lessen O3 loss [Chipperfield and Pyle, 1998] because in drier conditions
heterogeneous reaction rates slow down and PSCs do not form as readily [WMO,
2003]. However, the temperatures in the Arctic (even during cold winters) are
typically too warm for type II PSCs clouds to form and for dehydration to take place
[WMO, 2003].
2.4 Conclusions
The stratospheric dynamical and chemical processes that affect O3 during an
Arctic winter are very complex. These are further complicated by the large
interannual variability that occurs in the Arctic. Unlike the Antarctic winter, which is
27
typically characterized as having a cold and very stable polar vortex, the
meteorological conditions in the Arctic can fluctuate. The Arctic vortex can be cold
and stable, like the Antarctic vortex, or warm and disturbed. Because of this, there is
large interannual variability in the amount of O3 loss that can occur in the Arctic
during any given winter. This chapter provided an overview of the key processes that
are important for Arctic O3 loss. The impact of the changing meteorological
conditions in the Arctic on O3 loss will be discussed in depth in Chapters 6 through 8.
28
Chapter 3
Arctic O3 Observations
3.1 Atmospheric Observations
Atmospheric remote sensing is a way of determining information about
properties of our atmosphere without coming in direct contact with the volume of air
[Stephens, 1994]. There are two types of remote sensing methods: active and passive.
Active remote sensing involves sending out an electromagnetic signal to a target and
then monitoring the interaction between the target and signal [Stephens, 1994].
Unlike the active remote sensing method, passive remote sensing does not send out an
energy source; instead this method measures radiation emitted by the target or
measures radiation originating from other sources after it has been scattered or
absorbed by the target.
A benefit to remote sensing platforms is that they can cover many locations of
the globe (and depending on the orbit and type of measurement, can provide near
global coverage) on a single day. In addition, many satellite instruments make
observations on long temporal scales, often lasting over 5 years. The combination of
the broad temporal and spatial sampling makes remote sensing observations ideal for
atmospheric studies. The work presented in this dissertation will use data from
satellite based passive remote sensors. This chapter will provide an overview of the
29
instruments whose data are included in this work and the observation techniques
employed by each of the instruments.
3.2 Satellite Passive Remote Sensing Instruments
Data from ten different satellite instruments were analyzed in this work to calculate
Arctic O3 loss. These include observations from the Upper Atmosphere Research
Satellite Microwave Limb Sounder (UARS MLS), Earth Observing System
Microwave Limb Sounder (EOS MLS), Polar Ozone and Aerosol Measurement
(POAM II/III), Stratospheric Aerosol and Gas Experiment (SAGE II/III), Improved
Limb Atmospheric Spectrometer (ILAS), Halogen Occultation Experiment
(HALOE), Atmospheric Chemistry Experiment Fourier Transform Spectrometer
(ACE-FTS), and Measurement of Aerosol Extinction in the Stratosphere and
Troposphere Retrieved by Occultation (MAESTRO) instruments. Although there are
many differences between these instruments, they all rely on the spectroscopic
behavior of O3 in order to quantify the amount of O3 in the stratosphere. In other
words, the instruments are not directly sampling O3, but instead they are examining
how O3 interacts with radiation (e.g., infrared, visible, ultraviolet, or microwave).
The satellite instruments applied in this work utilize either the solar occultation
technique or the limb emission technique to retrieve information about atmospheric
species. Occultation and limb emission will be discussed in sections 3.2.1 and 3.2.2,
respectively. The satellite instruments that utilize each of the viewing techniques will
be discussed in their respective sections.
30
3.2.1 Occultation Instruments
Occultation instruments retrieve profiles of atmospheric species by measuring
absorption of radiation at different wavelengths (UV, visible, or infrared). Since the
occultation technique is a passive technique it relies on energy from another source,
with the three common sources being solar, lunar, and stellar. The datasets from
eight solar occultation instruments were used here; therefore, the remaining
discussion will focus on the solar occultation technique.
The instruments from this study that utilize solar occultation are POAM II/III,
SAGE II/III, ILAS, HALOE, ACE-FTS, and MAESTRO. The objective of the solar
occultation technique is to retrieve a profile of an atmospheric species by quantifying
the attenuation of solar radiation by that species. Each solar occultation instrument
performs two types of occultation events each orbit - a sunrise and a sunset. These
terms refer to whether the sun is rising or setting behind the Earth from the
perspective of the satellite. During a sunset event the instrument first scans the Sun
above the atmosphere to measure the unattenuated intensity of solar radiation [Lucke
et al., 1999]. As the satellite line of sight descends through the atmosphere, the
intensity from the sun is attenuated by the Earth’s atmosphere and the instrument
measures the attenuated solar radiation at different tangent altitudes. In order to
retrieve profiles of atmospheric species, the telemetry data and the raw data as a
function of tangent altitude are converted to profiles of transmission versus tangent
altitude for each instrument channel or wavelength.
31
There are several important advantages of the solar occultation technique.
One advantage is that the observations can have very high vertical resolution for a
remote sensor. Typically solar occultation instruments operating in the UV to visible
spectral range have a vertical resolution of approximately 1-2 km. The vertical
resolution of solar occultation instruments operating in the IR, where the solar signal
is less intense than in the visible, tend to have vertical resolution around 2-4 km.
Another important benefit of the technique is that the measurements are selfcalibrating since the instruments measure the unattenuated solar intensity while
observing the sun “above” the atmosphere. The solar occultation data also tend to
have high precision and accuracy, both because of the self-calibrating technique and
strong source signal (the sun). The main disadvantage of the solar occultation
technique is the limited spatial coverage it offers. Typically most solar occultation
instruments make about 30 profile measurements each day, which is many fewer than
other passive remote sensing techniques. These observations occur around two
circles of latitude, typically one in each hemisphere, with measurements spaced
evenly in longitude by about 25 degrees; the exact number of occultation events and
thus the longitude spacing depends on the altitude of the satellites (length of the
orbit), which varies from about 600-1020 km. Due to the advancements in stellar
occultation techniques the problem of limited geographic sampling will likely be
eliminated for future occultation instruments. Stellar occultation instruments, which
rely on bright stars as light sources, provide approximately the same vertical
resolution as solar occultation instruments, but make many more observations than
solar occultation instruments because of the increased number of light sources.
32
POAM II [Glaccum et al., 1996] and POAM III [Lucke et al., 1999] are ninechannel photometers that were designed to study chemistry in the polar regions.
POAM II was launched on September 1993 and made observations until November
1996 [Lucke et al., 1999]; POAM III was launched in March 1998 and operated until
December of 2005. Both instruments measured vertical profiles of O3, NO2, and
aerosol extinction; POAM III also measured vertical profiles of H2O. The O3
observations have a vertical resolution of 1 km in the stratosphere [e.g., Bevilacqua et
al., 1997; Randall et al., 2003]. The POAM wavelength channels vary only slightly
between the instruments. The nine channels of the POAM II instrument range in
wavelength from 0.352 to 1.06 µm. O3 is measured in the Chappuis band with a
channel centered at 0.601µm. POAM III channels range between 0.353 and 1.02 µm.
The POAM III O3 channel is also centered in the Chappuis band at 0.603µm. Both
instruments were launched in the same sun-synchronous near-polar orbit (98.7°
inclination). Because of the inclination, POAM II and III measurement latitudes vary
slowly between 54° N and 71° N in the Northern Hemisphere. During the Arctic
winter the instruments both sampled inside and outside the polar vortex.
The SAGE II and III instruments were grating spectrometers [e.g., Mauldin et
al., 1985; Brogniez et al., 2002; Wang et al., 2002; Thomason and Taha, 2003; Wang
et al., 2006]. SAGE II began taking atmospheric observations in October 1984, and
ceased observations in August 2005. The follow-on instrument, SAGE III, started
taking measurements in February 2002 and operations were terminated in March
2006. Both instruments measured stratospheric profiles of O3, NO2, H2O, and aerosol
extinction. SAGE II and III sampled O3 with a vertical resolution of approximately
33
0.5 to 1km [e.g., Brogniez et al., 2002; Wang et al., 2002; 2006]. The spectral
coverage varied little between the two instruments; SAGE II and SAGE III channels
range between 0.385 and 1.02 µm and 0.290 and 1.03 µm, respectively. Both
instruments measured O3 near 0.6µm. One of the main differences between the two
instruments was their orbit. SAGE II was launched in a 57° mid-inclination orbit. In
a period of about a month, it measured from a latitude of about 60° in the winter
hemisphere to about 80° in the summer hemisphere. Thus the coverage was
temporally sparse in the high northern latitudes [Wang et al., 2002]. SAGE III was
launched into a 99° sun-synchronous polar orbit and had extended northern latitude
coverage up to 80° N. The increased polar coverage by SAGE III allows for more
detailed studies of Arctic ozone loss.
ILAS began taking observations in November 1996. Due to spacecraft
problems, ILAS observations ceased in June 1997 [Sasano et al., 1999a]. ILAS was
launched in a 98.6 ° sun-synchronous polar oribit. Because of the inclination, ILAS
made measurements between 57°N and 72°N in the Northern Hemisphere [Nakajima
et al., 2002], latitudes that were very similar to those of the POAM measurements on
any given day. The ILAS instrument consisted of an IR and a visible spectrometer.
The IR spectrometer operated between 6.21 and 11.76 µm to measure vertical profiles
of O3, HNO3, NO2, N2O, CH4, and H2O. The O3 channel is centered near 9.6 µm.
The visible spectrometer covered the 753–784 nm spectral band to measure
temperature and aerosol extinction. The vertical resolution of each retrieved species
is approximately 1.9 to 3.5 km in the lower stratosphere [Nakajima et al., 2002].
34
The HALOE instrument was launched onboard the Upper Atmosphere Research
Satellite (UARS) and started taking observations in October 1991. Observations
ceased in November 2005. HALOE measured absorption bands of many polar
species important to polar chemistry, including O3, CH4, H2O, NO, NO2, and aerosol
extinction. The vertical resolution of O3 is approximately 2 km [e.g., Randall et al.,
2003]. The HALOE channels are in the infrared between 2.45 and 10.04 µm [Russell
et al., 1993], with the O3 channel centered at 9.6 µm [e.g., Morris et al., 2002]. Like
SAGE II, HALOE was launched in a mid-inclination orbit (57°), so its latitude
excursions are very similar to those of SAGE II. This means that in about a month it
covers latitudes from about 60° latitude in the winter hemisphere to about 80° latitude
in the summer hemisphere.
The ACE-FTS and MAESTRO instruments were launched onboard the ACE
satellite in August 2003 and are currently operational. ACE-FTS is a Fourier
Transform Spectrometer that operates in the 2 to 13 µm spectral region and measures
a large number of atmospheric species which include O3, H2O, CH4, NO, NO2,
HNO3, HCl, N2O5, and ClONO2 [Walker et al., 2005]. Each species is retrieved with
a vertical resolution of approximately 4 km in the lower stratosphere [Bernath et al.,
2005]. MAESTRO is a dual optical spectrophotometer [e.g., Bernath et al., 2005]. It
measures vertical profiles of O3, NO2, and aerosol extinction within the 0.400 to 1.03
µm spectral region. The vertical resolution of MAESTRO is approximately 1 km.
The ACE spacecraft was launched into a relatively high inclination orbit (74°). Thus
like SAGE II and HALOE, ACE-FTS and MAESTRO measure from the polar to low
35
latitudes; but unlike SAGE II and HALOE, they have extended (a month or more)
periods of observations in the polar regions.
3.2.2 Limb Emission Instruments
Data from two limb emission instruments were used in this study - UARS
MLS [Barath et al., 1993] and EOS MLS [Waters et al., 2006]. The UARS
instrument was the first of the two MLS instruments to be launched. UARS MLS
began taking observations in September 1991 and was decommissioned in December
2005. Between 1997 and 2005 the instrument did not take observations at a full level
of operation because of problems with spacecraft power. EOS MLS was launched in
July 2004 and is currently operational.
There are many benefits to microwave limb sounding. Unlike the occultation
instruments, which require measurements of a background energy source (e.g., the
sun, the moon, or stars), the limb emission technique only relies on emission of
radiation by species in the atmosphere. Therefore, it is not limited to looking only at
a single energy source, and can obtain near-global sampling. Observations in the
microwave are also useful because many chemical species that react with O3 have
spectral lines in this region of the electromagnetic spectrum (e.g., ClO) [Waters,
1993]. Another benefit to the microwave limb emission technique is that
observations can be made at any time of the day (since the technique does not rely on
the sun). As a result, limb emission instruments can observe both the sunlit and dark
sides of the Earth each day. Finally, another important benefit of the technique is that
36
the instruments can penetrate heavy aerosol or cloud layers (e.g., polar stratospheric
clouds and cirrus), which are difficult for UV or visible techniques to see through
[Waters et al., 1999; Waters et al., 2006].
Although there are many positives to limb emission instruments, the high
sampling rate of the instruments does come at a cost. Unlike the solar occultation
instruments, which are very small and relatively inexpensive to construct, limb
emission instruments are generally much larger instruments and require more power.
As a result the instruments typically are much more expensive to build and maintain
than solar occultation instruments. Also, another main drawback of the limb emission
technique is that the observations have lower vertical resolution than the solar
occultation observations.
UARS and EOS MLS are similar instruments, thus they retrieve information
about atmospheric composition in an analogous process. In order to make an
atmospheric observation, microwave radiation from the atmospheric limb is received
by an antenna. The vertical resolution of the observations is determined by the
vertical dimension of the antenna [Waters et al., 1999]. Both instruments have an
antenna that is approximately 1.6 m in length, which corresponds to a vertical
resolution of approximately 3 km [Waters et al, 1989; 1999]. Once radiation is
received by the antenna, it is then sent to radiometers for processing. UARS MLS
had three separate radiometers operating at different spectral channels: 63-GHz, 183GHz, and 205-GHz. The main species measured by UARS MLS are ClO, O3, SO2,
HNO3, H2O, temperature, and pressure [Waters, 1993; Waters et al., 1989; Livesey et
al., 2003]. EOS MLS has 5 radiometers, which operate in different frequencies than
37
those on UARS MLS. The radiometers are centered near: 118 GHz, 190 GHz, 240
GHz, 640 GHz, and 2.5 THz [Waters et al., 2006]. EOS MLS measures the same
species as UARS MLS and measures additional stratospheric species, which include
HCl, BrO, and N2O [Waters et al., 2006]. After the signals are processed by each
instrument, the power of each spectral channel is quantified and digitized. During
normal operations the instruments take observations in both daylight and nighttime
conditions. The instruments are calibrated with each limb scan. During calibration a
switching mirror rotates to either accept radiation from space or an internal
calibration target. This process is important in order to optimize radiometric
calibration of the instrument [Waters, 1993].
The main difference between the two MLS instruments is their orbits. Like
HALOE, UARS MLS was launched onboard the UARS spacecraft [Reber, 1993].
UARS MLS was launched in a mid-inclination orbit (57°); as a result the instrument
only took measurements from approximately 34o on one side of the equator to 80o on
the other. In order to have increased global coverage, the UARS spacecraft
performed a yaw maneuver to switch viewing orientation by 180 o. This process
would occur approximately every 36 days. As a result, the instrument did not sample
the polar regions continuously throughout the polar winter. EOS MLS was launched
in 2004 onboard the Aura satellite in a near-polar (98° inclination), sun-synchronous
orbit. EOS MLS observations span from 82° S to 82° N [Waters et al., 2006] on
every orbit; therefore, EOS MLS samples the northern polar latitudes continuously
throughout the winter while UARS MLS did not.
38
3.3 Conclusions
This chapter discusses the satellite instruments applied in this work as well as
the observation techniques employed by each of the instruments. The two
measurement techniques that were discussed in detail were the occultation technique
and the limb emission technique. Both of these techniques have benefits for
atmospheric observations. The solar occultation instruments provide high quality
atmospheric observations with high vertical resolution; however, the observations are
geographically sparse. The converse is true for the microwave limb emission
instruments which have increased geographic coverage, but often have lower vertical
resolution. The geographic sampling and the vertical resolution of the instruments,
and how these instrument characteristics relate to the inferred O3 loss calculations,
will be examined in Chapters 6 through 8.
39
Chapter 4
Techniques to Quantify Chemical O3 Loss
4.1 Introduction
One of the difficulties in quantifying chemical O3 loss is that direct
observations of chemical O3 loss do not exist. In order to infer chemical loss from the
measurements, an a priori knowledge of the O3 variations due to dynamical processes
is required. These variations must be separated from the O3 observations in order to
isolate the chemical loss. WMO [2003] and Harris et al. [2002] state the O3 loss
techniques can be split into two categories based on how the techniques isolate the
dynamical variation of O3. WMO [2003] states the first category includes techniques
that “allow for transport effects implicitly by using the relation between O3 and a
long-lived chemical tracer”, while the second category includes techniques that “take
into account the effect of transport explicitly by using transport calculations based on
meteorological analyses”. In this chapter the main techniques from both categories
will be described.
4.2 Implicit Transport: The O3 -Tracer Correlation Technique
40
The main O3 loss technique that utilizes implicit transport is the O3-tracer
correlation technique [e.g., Tilmes et al., 2004]. This technique was first defined by
Proffitt et al. [1990] and utilizes the relationship between O3 and a long-lived tracer
(e.g. CH4, N2O, or HF) to quantify chemical changes in O3. The technique is based
on the premise that prior to the onset of chemical O3 loss, vortex O3 has a compact
relationship with long-lived tracers [WMO, 2003]. The technique assumes that within
an isolated airmass the relationship between O3 and the long-lived tracer will remain
unaltered and will only change if chemical loss has occurred [Lemmen et al., 2006].
In order to quantify the amount of chemical loss, the difference between the original
O3-tracer relationship and the final relationship is compared.
Although this is a popular O3 loss technique, there are three areas of
uncertainty related to the technique. The first concerns defining a compact
relationship between O3 and a tracer. In order to accurately quantify O3 loss using
this technique, it is imperative that the compact relationship be determined just prior
to the onset of chemical loss [e.g. Tilmes et al., 2004; Lemmen et al., 2006]. WMO
[2003] indicates that previous studies have had difficulty defining a compact
relationship during select Arctic winters. The second area of uncertainty is due to the
impact of mixing on the O3 loss calculations. O3 and long-lived tracers have a
different relationship outside the vortex than inside, due to the difference in the
photochemical lifetime of O3 (where photochemical lifetimes of O3 are shorter outside
the vortex). Therefore, horizontal mixing can alter the vortex correlations [e.g. Plumb
and Ko, 1992; Proffitt et al., 1992; Michelsen et al., 1998; Plumb et al., 2000, 2003].
Müller et al. [2005] have shown that mixing with extravortex air can increase O3
41
values, which would result in an underestimation of loss. Finally, the third area of
uncertainty for the O3-tracer correlation technique is due to differential descent within
the vortex. If descent rate is not uniform, which is often the case in the Arctic vortex,
errors can be introduced into the O3-tracer relationship and the loss calculations [Ray
et al., 2002; Salawitch et al., 2002; Konopka et al., 2004]. In order to accurately
quantify O3 loss using the O3-tracer correlation technique, each of these areas must be
considered.
4.3 Techniques Utilizing Explicit Transport
The techniques that use an explicit transport scheme can be divided into two
subcategories. The first subcategory utilizes Lagrangian O3 measurements (where O3
measurements are taken following specified air parcels) and evaluates the chemical
change of an air parcel. The main technique from this subcategory is the match
technique. The second subcategory includes explicit transport schemes that use bulk
advection. These techniques, as described by WMO [2003], “use explicit transport
calculations to advect bulk quantities like vortex averages or gridded O3 fields and
compare these with later measurements of O3.” There are three main techniques that
fit in this subcategory, which include: the vortex average technique, the trajectory
ensemble technique, and the chemical transport model passive subtraction technique.
4.3.1
Match Technique
42
The match technique utilizes a pseudo-Lagrangian approach to quantify the
chemical O3 loss in an air parcel that has been observed two or more times by
ozonesondes [e.g., Rex et al., 2003, and references therein]. Forward trajectory
calculations are run from the location where an initial O3 observation was made to
determine if the path of the air parcel was intercepted by a second ozonesonde
observation [Rex et al., 1999; 2002]. In order to accurately simulate the path of a
parcel, the trajectories are run with radiative heating rates to account for diabatic
motions.
To be considered a “match event” and to be included in the O3 loss
calculations, the air parcel must satisfy some quality checks. In order to eliminate
statistical error connected with the uncertainty in the vertical position of the
trajectories, the difference in the maximum and minimum O3 values between the two
observations cannot be larger than 15 % (25 %) within a 2 K (5 K) vertical range [Rex
et al., 1999]. In addition to O3, the potential vorticity (PV) of the air parcel is also
monitored to eliminate trajectories in which PV is not conserved. The maximum
change in PV along the trajectory cannot be larger than 25 % of the mean PV along
the trajectory [Rex et al., 1999]. Another way trajectories are rejected in the match
analysis is by running a large number of trajectories. Trajectories are run in a cluster
to find match pairs. If, over the entire life of the trajectory, the trajectories diverge
more than 1200 km on the same potential temperature surface, or more than 1300 km
after undergoing descent, the match is rejected. One of the main quality criteria of
the match technique is the match radius. The match radius is the horizontal distance
between the location of the ozonesonde at the altitude of the match and the position of
43
the trajectory [Rex et al., 1998]. Previous studies have used a maximum radius
between 400 and 500 km to reduce the statistical uncertainty in the O3 loss
calculations. Finally, all match events must occur within 10 days, to reduce errors
that result from the divergence in trajectory calculations. However, most match
events occur within a 5-7 day period [WMO, 2003].
The error associated with a single match event can be on the same order as the
amount of chemical change; therefore, chemical loss must be quantified from a
statistical analysis of many match events [Becker et al., 1998]. The loss is calculated
by a linear regression of the differences between the initial and final observation for
all match events. The primary quantity derived by the match technique is the O3 loss
rate per sunlit hour [Rex et al., 1999]. The O3 loss rate per day can be computed by
multiplying the O3 loss per sunlit hour by the number of sunlit hours experienced by
the trajectory on a given day [Rex et al., 1999].
As previously mentioned, one of the largest sources of uncertainties in the
match technique is due to systematic errors in the trajectory calculations and diabatic
descent rates [e.g., Rex et al., 1999; WMO, 2003]. In order to reduce these errors
sufficiently, many match events are required to produce statistically significant O3
loss estimates. Another important issue for the match technique is how completely
the vortex is sampled. In order for match O3 loss calculations to represent vortexaverage conditions, the combined match events must sample the vortex
homogeneously [Harris et al., 2002]. However, if the matches are not well
distributed within the vortex, the results could be biased. The final uncertainty in the
match technique concerns the assumption that the sampled air parcel does not mix
44
with its surroundings along a trajectory. This assumption is applied because the
trajectories are only run for short time intervals. Nonetheless, small-scale mixing
along the trajectory is inevitable. Because of this, the no-mixing assumption causes
errors in O3 loss calculations, especially during dynamically active winters.
4.3.2
Vortex Average Technique
The objective of the vortex average technique is to isolate the chemical
variation of an average ozone profile inside the vortex by subtracting the change in
the profile resulting from vortex average descent. This technique has been applied to
O3 observations from remote sensing instruments [e.g., Bevilacqua et al., 1997,
Hoppel et al. 2002, Singleton et al., 2005; Manney et al., 2006] and to observations
from ozonesondes [e.g., Rex et al., 1998]. The average vortex descent rates are
generally calculated from a radiative transfer model, which is forced with
temperatures from meteorological analyses [WMO, 2003]. In order to determine the
O3 loss, a vortex average O3 profile is calculated by averaging all vortex observations
prior to the onset of chemical loss. The profile is then descended throughout the
winter using the calculated diabatic descent rates to quantify the dynamical variation
in vortex O3. This profile is then subtracted from the observations to isolate the
chemical change in O3. The vortex average technique assumes that the dynamical
change in vortex O3 is dominated by uniform diabatic descent, and that mixing
between vortex and extra-vortex air is negligible [Hoppel et al., 2002]. Because of
this assumption, only vertical transport is considered.
45
There are potential problems with applying the vortex average technique to
quantify Arctic chemical O3 loss. Like the match technique, the vortex average
technique assumes that horizontal mixing does not occur. This assumption produces
larger uncertainty for the vortex average technique because it is applied over an entire
winter, compared to a short (~ 10 day) time period. Uncertainty also results from the
diabatic descent calculations. Only one vortex average O3 profile is descended in the
vortex to quantify the diabatic descent, thus the technique does not account for the
variable descent within the vortex. As mentioned above, previous studies have shown
that the descent within the Arctic vortex is often not uniform [e.g., Ray et al., 2002].
Finally, uncertainty also arises from incomplete O3 sampling. If the O3 observations
that are used for the analysis do not uniformly sample the vortex, the O3 loss
calculations will not represent the true vortex average conditions [WMO, 2003;
Manney et al., 2006].
4.3.3
Trajectory Ensemble Technique
The trajectory ensemble technique was developed by the work of Manney et
al. [1995 a, b, c; 1996 a, b; 1997]. In this technique, trajectories are initialized at
locations where there are O3 observations. The trajectories are initialized with the O3
from the observations and then run passively (without chemistry) forward in time for
approximately two weeks. The short timescale is chosen to reduce impact from
horizontal mixing. Diabatic descent rates, generally calculated from a radiative
transfer model, are incorporated in the trajectories to account for vertical motion. At
46
the end of the two week period, O3 observations are interpolated to the end position of
the trajectories [Manney et al., 1995a]. A vortex average of the O3 from the
observations and the trajectories is calculated to reduce the sensitivity to transport
features of the individual trajectories [WMO, 2003]. Finally, the difference between
the fields is calculated to infer the chemical change in O3 [Manney et al., 1995b].
The results from this O3 loss technique are stated as the vortex averaged O3 loss
[WMO, 2003].
The trajectory ensemble technique suffers from similar problems as the match
technique. The main cause of uncertainty is errors in the trajectory calculations and
errors in diabatic descent rates [WMO, 2003]. In order to minimize these problems,
trajectories are only run for short periods of time.
4.3.4
Chemical Transport Model Passive Subtraction Technique
The final technique discussed here is the chemical transport model passive
subtraction (CTM-PS) technique [e.g., Goutail et al., 1997; Deniel et al., 1998;
Hoppel et al., 2002; Singleton et al., 2005; 2006]. The CTM-PS technique is the
basis of this thesis and will be discussed in detail in Chapters 6 through 8. The CTMPS technique was developed from the work of Manney et al. [1995b], where, as
previously mentioned, a passive O3 field was subtracted from observations to isolate
the chemical change in O3. Unlike the work of Manney et al. [1995b], the CTM-PS
technique does not rely on trajectories to determine the dynamical variation of O3, but
does rely on a three-dimensional CTM. A full description of the CTM that was
47
applied in this thesis will be discussed in Chapter 5. Prior to the onset of O3 loss, a
passive O3 tracer is initialized in a CTM. The model is then forced daily by
meteorological analyses in order to advect the passive O3 field throughout the winter.
The passive O3 field is then interpolated to the observation locations and the
difference between the observations and the passive field represents the chemical
change in O3.
Like the other O3 loss techniques, the CTM-PS technique also suffers from
uncertainties. Because CTM-PS relies on accurate horizontal and vertical transport of
O3, any problems in the transport will cause errors in the O3 loss calculations. An
additional potential problem for the CTM-PS technique is the initialization of the O3
fields. If the model passive O3 field is not exactly equal to the observations on the
first day of the analysis, this difference will be propagated through the O3 loss
calculations [e.g. Singleton et al., 2005; 2006]. For example, if the CTM passive field
is greater than the observations on the first day of the analysis, the inferred O3 loss
calculations will overestimate the O3 loss. In order to handle this problem the CTM
O3 field can be initialized with observations from satellites.
One of the main values of CTM-PS, compared to the other techniques, is that
it can be used to evaluate present understanding of atmospheric processes and can
assist in climate studies. CTMs have detailed chemical and dynamical modules;
therefore, they can be compared with observations to test present understanding of
atmospheric processes. By comparing O3 loss inferred from observations to the
CTM-modeled loss, gaps in knowledge of polar processes can be identified. CTMs
are also important because they play an integral part in climate studies. CTMs are
48
off-line models; therefore, they cannot be used to make projections about climate.
However, they often serve as the chemistry modules for coupled Chemistry Climate
Models (CCMs).
4.4 Conclusions
In this chapter an overview of the five main techniques for calculating O3 loss
was presented. Although each of the techniques has been widely used for O3 loss
studies, each has areas of weakness. The two main areas of uncertainty for the
techniques are transport (both horizontal and vertical) and mixing. Unlike the
Antarctic vortex, which is relatively stable, the Arctic vortex is dynamically active.
Therefore, the meteorological conditions in the Arctic are very complex. In addition,
the conditions can vary considerably from year to year. Because of these two
characteristics, it is imperative that the uncertainties in the O3 loss technique be
considered when Arctic O3 loss results are analyzed.
49
Chapter 5
The SLIMCAT CTM
5.1 Introduction
There are two main objectives for this thesis. The first objective is to infer O3
loss from satellite observations from the 1994-1995 through the 2004-2005 Arctic
winters. The second objective is to examine how well a chemical transport model
(CTM) can reproduce Arctic O3 loss that was inferred from satellite observations. In
order to quantify the inferred and modeled loss the Chemical Transport Model
Passive Subtraction (CTM-PS) technique was applied. This technique is described in
detail in Chapters 4, 6, 7, and 8. The CTM that was employed to quantify both the
inferred and modeled loss was the University of Leeds’ SLIMCAT CTM. This
chapter will provide an overview of the structure and the dynamical and chemical
modules in the SLIMCAT CTM.
5.2 Off-line CTMs
The SLIMCAT CTM is an off-line model; that is, the model relies on
meteorological analyses (either from a global climate model or from other
meteorological analyses) as input. Because CTMs are dependent on meteorological
50
analyses they cannot be used for predictions of future climate, in the way that coupled
chemistry climate models (CCMs) can. However, CTMs are valuable tools because
they can simulate the present-day atmosphere when they are forced with observations.
Thus, simulations from CTMs can be directly compared with observations and/or
forced with observations in order to evaluate current modeling capabilities, which is
vital to determine our present understanding of atmospheric phenomena. In addition,
CTMs play an integral part in the development of CCMs, because the chemical
modules in CCMs are often developed from the detailed chemical modules first
incorporated in CTMs.
5.3 Model Structure
The SLIMCAT CTM was first described in Chipperfield et al. [1996]. Since
the model was first introduced, it has undergone many changes, one of which was the
vertical domain of the model. The first version of SLIMCAT only included the
stratosphere and the lower boundary of the model was at approximately 9 km
(potential temperature of ~330 K). This early version of the CTM only used pure
isentropic levels [Feng et al., 2005]. In the present version of the CTM, the lower
boundary has been extended to the surface. The vertical coordinate in SLIMCAT is
now a hybrid σ-Ө coordinate, with terrain-following levels near the surface and
isentropic surfaces in the stratosphere [Chipperfield, 2006]. The transition from σ
layers to pure isentropic levels in the stratosphere occurs at approximately 350 K.
The top vertical boundary in the model is set by the meteorological analyses used as
51
input. In total, SLIMCAT has 24 levels from the surface to approximately 55 km
[e.g., Feng et al., 2005]. SLIMCAT has a vertical resolution that is comparable to
satellite-based limb sounding observations in the lower stratosphere and is
approximately 1.5-2 km.
The global horizontal grid is an Eulerian grid, which extends to both poles.
For the work presented here, SLIMCAT was run using two different horizontal grids,
7.5°×7.5° and 2.8°×2.8°. Before a simulation was run for a particular winter a low
resolution model run was started on 1 January 1977. On 1 December of each
simulation year, the high resolution model O3 fields were initialized with the O3 fields
from the low resolution run. The coarse resolution run was started in 1977 in order to
spin up the model and minimize the effects of model initialization.
5. 4 Model Transport
As was mentioned above, off-line CTMs do not calculate their own winds,
thus they require meteorological fields as inputs. For this analysis, all SLIMCAT
runs were forced with European Centre for Medium-Range Weather Forecasts
(ECMWF) analyses. SLIMCAT runs prior to 1 January 2000 were forced with
ECMWF reanalyses (ERA-40) while runs after this date were forced with the
ECMWF operational analyses. There is a bias in ERA-40 temperature with respect to
other analyses; however, previous studies have shown that ERA-40 correctly handles
interannual variability [e.g., Manney et al., 2005; Tilmes et al., 2006]. The ECMWF
winds are input into the model as spectral coefficients of streamfunction and velocity
52
potential and then converted to grid-point fields by a spectral transform using
Legendre coefficients [Chipperfield, 2006]. The winds are then interpolated linearly
in potential temperature from the analysis levels to the SLIMCAT levels. During a
run, SLIMCAT is forced with 6-hourly meteorological fields.
Vertical transport is typically a difficult process for CTMs to handle. Vertical
winds are provided with meteorological analyses, but they are not used to quantify the
vertical transport in SLIMCAT. The reason for this is the vertical winds are provided
on a different vertical grid than the model’s and inconsistencies can be introduced
between the horizontal and vertical winds if the vertical winds are interpolated to the
model grid [Chipperfield, 2006]. Therefore, another approach is taken to calculate
vertical transport. In the analyses presented here, the vertical advection in SLIMCAT
was calculated from diabatic heating rates [Feng et al., 2005]. In this approach, net
diabatic heating rates are calculated using a radiation scheme and then converted to a
vertical mass flux. The radiation scheme is based on the NCAR CCM radiation
scheme [Briegleb, 1992], which uses a δ-Eddington approximation [Joseph et al.,
1976] and can be extended from 1000 hPa to the top of the atmosphere [Feng et al.,
2005]. The CCM radiation scheme was used for model levels above 350 K.
In order to properly handle the transport of chemical species, a tracer
advection scheme is required for CTMs. Eulerian models do not prescribe mixing
because it is an intrinsic function of the advection scheme. If the advection in an
Eulerian model is not handled properly the model will simulate too much mixing.
The desired properties of an advection scheme are that it should conserve mass, have
low numerical diffusion and low dispersion, should preserve tracer-tracer
53
correlations, and preserve chemical family abundances [Chipperfield, 2006]. For the
runs used in this work, chemical tracers were advected using the Prather [1986] finite
volume advection scheme and a 30-minute timestep. The Prather [1986] scheme
conserves second-order moments, which helps to reduce excess mixing in SLIMCAT.
The scheme handles advection in the x, y, and z directions separately and is based on
mass fluxes averaged over the interfaces of the SLIMCAT grid. The scheme not only
advects tracer mass from a grid box, but it also advects the total mass of the grid box.
In order to make sure that mass is conserved, the mass on each isentropic surface is
calculated at every time step. Adjustments in the mass of a tracer are made if the
mass within the gridbox is not correct. The benefit of this advection scheme is that it
has low numerical diffusion, it conserves mass, and it is able to maintain strong tracer
gradients [e.g. Chipperfield, 1999; 2006]. A disadvantage of this scheme is that care
needs to be taken to make sure that tracer-tracer correlations are handled properly.
5.5 Model Chemistry
The detailed gas-phase stratospheric chemistry scheme included in SLIMCAT
was first described by Chipperfield [1999]. A limited number of chemical families
are used in the SLIMCAT. Only species that are in rapid photochemical steady-state
and that can be computed at night are included in a chemical family [Chipperfield,
1999]. All other chemical species are integrated separately in SLIMCAT. To limit
computational costs, different schemes are used to integrate the species depending on
whether they are long-lived, short-lived, or a chemical family. For long-lived
54
chemical species (i.e., tropospheric source gases) a simple Euler integration scheme is
applied. Chemical families and species that are short-lived are integrated with the
more computationally expensive semi-implicit scheme [Ramaroson et al., 1992]. In
order to resolve the diurnal cycles of some chemical species, a 15-minute timestep
was applied.
Photolysis rates in the model are calculated from lookup tables that contain
information on solar zenith angle (SZA), pressure, temperature, and column O3. This
scheme was based on the work of Lary and Pyle [1991]. During a model run,
photolysis rates from the table are interpolated to a specific location and time.
Photolysis rates are calculated up to a SZA of 96° and SLIMCAT accounts for
multiple scattering and spherical geometry [Chipperfield, 1999]. Photochemical
information for the runs was based on JPL 2003 [Sander et al., 2003]; however, cross
sections for Cl2O2 were taken from Burkholder et al. [1990] and extrapolated to 450
nm [Feng et al., 2005]. Values for tropospheric source gas abundances (CH4, N2O,
and halocarbons) were based on WMO [2003].
To handle heterogeneous chemical reactions, SLIMCAT has an equilibrium
treatment for the reactions on liquid aerosols, nitric acid trihydrate (NAT), and ice
[Chipperfield et al., 1999; Feng et al., 2005]. For heterogeneous reactions to occur
on liquid aerosols, H2SO4 loading is required. Liquid sulfate aerosols are specified
from a climatology, from 1979 through 1999, of monthly fields from WMO [2003]
[Feng et al., 2006]. After 1999, the liquid sulfate aerosols in SLIMCAT were set to
1999 values. In order to calculate the composition of the liquid
H2O/H2SO4/HNO3/HCl aerosols, the analytical scheme of Carslaw et al. [1995a, b] is
55
applied [Chipperfield et al., 1999]. The analytic scheme, which is based on the
thermodynamic model of Carslaw et al. [1995a] computes the composition of ternary
aerosols based on the model temperature, H2SO4, HNO3, and H2O. The liquid aerosol
reactions in SLIMCAT are activated based on the equilibrium vapor pressure of
H2SO4, which is determined by the expression of Ayers et al. [1980]. The chemical
scheme in SLIMCAT assumes that only liquid aerosols are present at temperatures
greater than TNAT [Davies et al., 2002].
Ice PSCs form in SLIMCAT when they are thermodynamically possible
[Chipperfield, 1999]. Ice particles are assumed to have a radius of 10 microns and to
have a NAT coating. The expression of Hanson and Mauersberger [1988] is applied
in the model to determine the amount of HNO3 that should be removed from the gas
phase in the presence of ice particles [Davies et al., 2002].
For the model simulations presented in this work, a simple NAT-based
denitrification scheme was activated. The scheme is based on the work of Davies et
al. [2002] and assumes that NAT particles have a bimodal distribution. This
assumption is based on the observations of Fahey et al., [2001]. The two modes have
radii values of 0.5 and 6.5 microns and fall velocities of 1 and 1100 meters per day,
respectively. To calculate values of NAT in the model, the expression of Hanson and
Mauersberger [1988] is applied using the model temperature, H2O, and HNO3 fields
[Chipperfield et al., 1999; Davies et al., 2003]. NAT particles are assumed to form in
the model when the model temperature reaches TNAT. The condensed HNO3 mass is
first assigned to the smaller particles. Once a number density of 1 cm-3 is reached,
any remaining condensed HNO3 is allocated to the 6.5 micron particles [Davies et al.
56
2002]. After the particles form, they are sedimented from the model, but they can
evaporate at lower levels in the model based on model temperatures [Chipperfield,
1999]. Davies et al. [2002] compared the NAT-based denitrification scheme in
SLIMCAT to observations of denitrification taken during the 1999-2000 Arctic
winter. The 1999-2000 Arctic winter was very cold, had a strong and stable vortex,
and experienced widespread denitrification. The results indicate that the NAT
denitrification scheme was able to reproduce the denitrification because the model
odd nitrogen was in good agreement with observations taken with instruments aboard
the ER-2. However, agreement between observations and the model denitrification is
highly dependent on the temperature analyses that are used in the model simulation.
5.6 Conclusions
This chapter provides an overview of the main features of the SLIMCAT
CTM. Presently, SLIMCAT is one of the premier CTMs because of the sophisticated
chemical and dynamical modules. In order to evaluate the chemistry and dynamics
within SLIMCAT, the model will be compared with atmospheric observations. These
comparisons will be one focus of Chapters 6 through 8.
57
Chapter 6
2002-2003 Arctic ozone loss deduced from POAM III satellite observations and
the SLIMCAT chemical transport model
C. S. Singleton1, C. E. Randall1, M. P. Chipperfield2, S. Davies2, W. Feng2, R.M.
Bevilacqua3, K.W. Hoppel3, M.D. Fromm4, G.L. Manney5,6, V.L. Harvey1
1
Laboratory for Atmospheric and Space Physics, UCB 392, University of Colorado,
Boulder, CO 80309-0392, USA
2
Institute for Atmospheric Science, School of Earth and Environment, University of
Leeds, Leeds LS2 9JT, UK
3
Naval Research Laboratory, Remote Sensing Physics Branch, Naval Research
Laboratory, Washington, D.C., 20375-5351, USA
4
Computational Physics, Inc., Springfield, VA 22151, USA
5
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA
6
Department of Natural Sciences, New Mexico Highlands University, Las Vegas,
NM, 87701, USA
Abstract
The SLIMCAT three-dimensional chemical transport model (CTM) is used to infer
chemical ozone loss from Polar Ozone and Aerosol Measurement (POAM) III
58
observations of stratospheric ozone during the Arctic winter of 2002-2003. Inferring
chemical ozone loss from satellite data requires quantifying ozone variations due to
dynamical processes. To accomplish this, the SLIMCAT model was run in a
“passive” mode from early December until the middle of March. In these runs, ozone
is treated as an inert, dynamical tracer. Chemical ozone loss is inferred by subtracting
the model passive ozone, evaluated at the time and location of the POAM
observations, from the POAM measurements themselves. This “CTM Passive
Subtraction” technique relies on accurate initialization of the CTM and a realistic
description of vertical/horizontal transport, both of which are explored in this work.
The analysis suggests that chemical ozone loss during the 2002-2003 winter began in
late December. This loss followed a prolonged period in which many polar
stratospheric clouds were detected, and during which vortex air had been transported
to sunlit latitudes. A series of stratospheric warming events starting in January
hindered chemical ozone loss later in the winter of 2003. Nevertheless, by 15 March,
the final date of the analysis, ozone loss maximized at 425 K at a value of about 1.2
ppmv, a moderate amount of loss compared to loss during the unusually cold winters
in the late-1990s. SLIMCAT was also run with a detailed stratospheric chemistry
scheme to obtain the model-predicted loss. The SLIMCAT model simulation also
shows a maximum ozone loss of 1.2 ppmv at 425 K, and the morphology of the loss
calculated by SLIMCAT was similar to that inferred from the POAM data. These
results from the recently updated version of SLIMCAT therefore give a much better
quantitative description of polar chemical ozone loss than older versions of the same
model. Both the inferred and modeled loss calculations show the early destruction in
59
late December and the region of maximum loss descending in altitude through the
remainder of the winter and early spring.
6.1 Introduction and Objectives
Knowing and understanding the factors that control halogen-catalyzed ozone loss
in the polar lower stratosphere is fundamental to our understanding of how the
stratosphere is affected by anthropogenic influences. In spite of attention placed on
ozone loss in the polar regions, numerous theoretical models routinely underestimate
ozone loss rates in much of the lower polar stratosphere (between about 400 and 550
K) compared to “observed” loss rates [e.g., Chipperfield et al., 1996; Goutail et al.,
1997; Deniel et al., 1998; Becker et al., 2000; Guirlet et al., 2000]. Even with the
most recent Arctic field campaign results (e.g., SOLVE I/II, the SAGE III Ozone
Loss and Validation Experiment; THESEO-2000, the Third European Stratospheric
Experiment on Ozone; and VINTERSOL, Validation of International Satellites and
Ozone Loss) this long-standing problem has yet to be resolved [e.g., Pierce et al.,
2003]. Rex et al. [2002a] identified two main areas of uncertainty in modeling Arctic
ozone loss: quantifying denitrification and chlorine activation, and understanding
early winter ozone loss at high solar zenith angles. Although the early winter loss
does not account for a large fraction of the total loss, Rex et al. [2002a, 2003] noted
that a full understanding is required for reliable predictions of future ozone levels in
the Arctic Stratosphere.
60
One of the complications in quantifying ozone loss is that no direct
observations of chemical ozone loss rates exist. Rather, chemical loss rates must be
inferred from the measurements with a priori knowledge of, or assumptions about, the
ozone variations due to dynamical processes. As noted by Manney et al. [2003a],
uncertainties in these dynamical processes are large and poorly quantified, and thus
can lead to large uncertainties in the “measurements” of ozone loss. In order to
determine the variation of ozone due solely to chemical processes the dynamical and
chemical variations must be separated in the observed ozone fields. Four methods
have primarily been used to isolate photochemical loss [e.g., Harris et al., 2002; Rex
et al., 2002b; Newman and Pyle, 2003]:
1. The “Match” technique quantifies photochemical ozone loss by measuring
the difference in ozone in an air parcel sampled at different times [Rex et al.,
2003, and references therein]. “Matches” occur when trajectories indicate that
the same air parcel is observed multiple times by one or more instruments
(either ozone sondes or satellites), within some prescribed tolerance limits. If
the vortex is sampled homogeneously, the ozone loss result reflects vortex
average conditions [Harris et al., 2002].
2. The “Tracer Correlation” technique removes the effect of transport by
comparing the pre-winter and post-winter relations between ozone volume
mixing ratio and an inert tracer, such as nitrous oxide (N2O) or methane
(CH4), inside the vortex [Proffitt et al., 1990; Müller et al., 1997, 2001]. This
method assumes that in the absence of ozone production or loss, the
61
ozone/tracer relationship remains constant; thus, any post-winter deviations
from the pre-winter relationship are interpreted as chemically induced.
3. The “Vortex Average” technique quantifies dynamical variation for an
average ozone profile inside the vortex by calculating vortex average descent
rates from a radiative transfer model. This technique assumes that the
dynamical contribution to ozone change inside the vortex is dominated by
diabatic descent, and that mixing between vortex and extra-vortex air is
minimal; therefore, only vertical transport is considered [Hoppel et al., 2002].
4. The “Passive Subtraction” technique requires ozone to be simulated as a
passive tracer. The passive ozone is then subtracted from ozone
measurements to quantify the change in ozone due to chemistry [e.g., Manney
et al., 1995a, 2003b]. In this work we use a 3-D chemical transport model
(CTM) to simulate ozone as a passive tracer [e.g., Goutail et al., 1997; Deniel
et al., 1998; Hoppel et al., 2002] and will refer to this technique as the “CTM
Passive Subtraction” (CTM-PS) technique.
As mentioned by Guirlet et al. [2000] and Harris et al. [2002], quantitative
comparisons of the different ozone loss calculations can be difficult since each
method considers different altitudes, time periods, and area averages of the vortex.
When comparing ozone loss results it is critical to understand these differences as
well as the weaknesses of each method. Two large sources of uncertainty in the
Match method are errors in the trajectory calculations [Rex et al., 1999] and neglect
of mixing. Many Match pairs are required in order to reduce errors sufficiently to
produce statistically significant ozone loss estimates, and the Match technique
62
assumes that the sampled air parcel does not mix with its surroundings along a
trajectory. The Tracer Correlation technique quantifies the variation of ozone due to
transport using the correlation between ozone and an inert tracer. In order to define
the tracer correlations adequately, data is needed throughout the stratosphere. Since
ozone tracer correlations are often different outside the vortex than inside, processes
such as descent and horizontal mixing can alter the correlations in ways that can
mimic ozone loss [Michelsen et al., 1998]. Mixing across the vortex edge or
differential descent and mixing within the vortex may disrupt the compactness of
ozone/tracer relationships and can result in anomalous relationships; such effects
must be considered before estimates of ozone loss can be made reliably from tracer
relationships [Plumb et al., 2000; Ray et al., 2002]. The Vortex Average method as
applied by Hoppel et al. [2002] uses vortex-averaged descent rates, tantamount to
assuming uniform descent within the vortex, and does not account for lateral mixing
across the vortex edge. Lateral mixing across the vortex edge is particularly
important to consider in winters when the vortex is disturbed. The CTM-PS
technique includes horizontal transport, but it also has several areas of uncertainty.
Most importantly, it is dependent on the proper initialization of the CTM ozone
fields, correct representation of transport in the model, and proper gas phase
chemistry to isolate heterogeneous induced ozone loss.
The main purpose of this paper is to describe CTM-PS ozone loss results for
the Arctic 2002-2003 winter using observations from the third Polar Ozone and
Aerosol Measurement (POAM) instrument [Lucke et al., 1999] and the SLIMCAT
CTM [Chipperfield, 1999]. Comparisons between CTM-PS results and Vortex
63
Average results are also shown, but detailed analysis of these comparisons, as well as
comparisons with the Match and Tracer Correlation ozone loss calculations, are the
subject of future work. The CTM-PS technique, depending on the sophistication and
accuracy of the CTM, is in some sense the most complete method for determining
ozone loss. That is, if the chemistry and dynamics are accurate within the CTM, all
the processes needed to deduce chemical ozone loss are included and few
assumptions are required. The CTM-PS technique is an integral part of the
development of coupled Chemistry Climate Models (CCMs), the framework of which
relies on accurate treatment of ozone loss processes in the chemical calculations used.
Investigations such as those described below will thus result in a more accurate
investigation of the coupling between global climate change and polar ozone loss.
6.2 2002-2003 meteorology
The 2002-2003 winter can be characterized as an unusually cold early winter
and dynamically active and warm mid to late winter [Manney et al., 2005]. Figure 1
shows the minimum Met Office temperatures inside the Arctic polar vortex with
respect to Nitric Acid Trihydrate (NAT) condensation temperatures (TNAT) at four
different potential temperature levels from 450 K (about 18 km) to 600 K (about 22
Km). TNAT values were computed using the expression given by Hanson and
Mauersberger [1988], Met Office pressure, and by assuming 10 ppbv HNO3 and 5
ppmv H2O. Vortex wide, minimum temperatures were below TNAT until mid-January,
with a few exceptions at 600 K. Throughout the lower stratosphere temperatures
64
Figure 6.1: Time series of T-TNAT in the Arctic vortex from 1 December 2002
through 15 March 2003 for the 600 K, 550 K, 500 K, and 450 K potential temperature
surfaces vortex wide. Temperatures are the minimum temperatures inside the polar
vortex and were obtained from Met Office analyses. NAT condensation temperatures
were computed using the Hanson and Mauersberger [1988] expression, assuming 10
ppbv HNO3 and 5 ppmv H2O.
increased rapidly in late January, as a major stratospheric warming occurred.
Temperatures were just recovering toward pre-warming levels when a strong minor
warming occurred in February. Although temperatures began to decrease after the
warming, the vortex was never again as cold as in December. After early February,
minimum vortex temperatures reached TNAT or fell below TNAT on a few occasions at
600 and 550 K. At 500 and 450 K vortex wide minima fell below TNAT after
February.
Although the polar vortex was very cold in December and January, it was
neither circular nor centered on the pole. Figure 2 shows maps of the Met Office PV
fields on the 500 K potential temperature surface for specific days during the 2002-
65
2003 winter. In December and January the vortex was often elongated, allowing air
within it to make frequent excursions into the sunlight at lower latitudes. As
described below, the very low temperatures and prolonged solar exposure led to
ozone loss as early as late December. However, the major warming in January
followed by the strong minor warming in February caused the vortex to shrink and
split, as indicated by the maps for 21 January and 17 February. The series of
warming events also caused temperatures to increase, limiting the total amount of
ozone loss over the winter [Manney et al., 2004].
Figure 6.2: Met Office PV(10-5 Km2 kg-1s-1) at the 500 K potential temperature
surface for specific dates during the 2002-2003 winter from 90°N to 30°N. The inner
vortex boundary is denoted by the solid white contour. The black dotted circle
indicates the POAM measurement latitude.
6.3 POAM III observations in 2002-2003
POAM III [Lucke et al., 1999] is a nine-channel solar occultation photometer
with wavelength channels ranging from 0.353 to 1.02µm to measure profiles of
66
ozone, nitrogen dioxide, water vapor, and aerosol extinction. During one day POAM
makes 14-15 measurements around a circle of latitude in each hemisphere, with
successive measurements separated in longitude by about 25°. The POAM
measurement latitude varies smoothly and slowly over the course of a year between
55°N and 73°N in the northern hemisphere (NH) and between 63°S and 88°S in the
southern hemisphere (SH). The POAM measurement latitude variation over the NH
winter (the measurement coverage is the same each year) is shown in Figure 3. Also
shown in Figure 3 is the equivalent latitude [Butchart and Remsberg, 1986]
(equivalent latitude is the latitude that would enclose the same area between it and the
pole as does the PV contour) at 500 K of each POAM measurement obtained during
the 2002-2003 winter. The PV fields used in the equivalent latitude calculation were
obtained using the Met Office meteorological analysis. In this figure the
measurements are color-coded according to their position with respect to the vortex
(outside: outside the outer edge, edge: between the inner and outer edge, and inside:
inside the inner edge), which is defined using the discrimination algorithm of Nash et
al. [1996] and, as for the equivalent latitudes, the Met Office-derived PV. Figure 3
shows that although only a relatively narrow range of latitudes is sampled by POAM,
a much larger range of equivalent latitudes was sampled during the 2002-2003 winter
because the vortex was often elongated and displaced from the pole. Thus, POAM
sampled inside, outside, and on the edge of the vortex on a nearly daily basis
throughout the winter.
The POAM ozone data set used in this study is version 3.0 [Lumpe et al.,
2002]. Version 4.0 POAM data became available after the analysis for this work had
67
Figure 6.3: Northern Hemisphere equivalent latitudes (dots) and geographic latitudes
(solid curve) of POAM measurements on the 500 K potential temperature surface.
Red indicates measurements taken within the inner edge of the vortex boundary, blue
indicates measurements between the outer and inner edges, and black denotes all
measurements taken beyond the outer edge.
been completed. Comparisons between version 3.0 and version 4.0 POAM ozone
data indicate differences of less than 1% on average, so the results presented here are
not expected to change significantly with the new version. The vertical resolution of
the version 3.0 retrievals is approximately 1 km in the stratosphere, and the random
error is < 10% above 10 km (<5% above 15 km) [Lumpe et al., 2002]. This data set
has undergone extensive validation and intercomparison with other remote sensing
data sets and balloon-borne ozonesondes [Lumpe et al., 2003; Randall et al., 2003;
Prados et al., 2003]. Randall et al. [2003] show that on average, NH POAM ozone
profiles agree to within about 5% with ozonesonde and other satellite data from 13 to
60 km. Below 13 km the POAM measurements appear to be biased increasingly high
with decreasing altitude reaching values of about 40% (0.1 ppmv) higher then
ozonesondes at 10 km [Randall et al., 2003; Prados et al., 2003].
68
Figure 4 shows the evolution of ozone measured by POAM throughout the
2002-2003 winter from 400 K (about 15 km) to 650 K (about 25 km). The
measurements are color-coded according to their position with respect to the vortex
edge. Lower stratospheric ozone in the polar region generally increases throughout
the winter due to descent of ozone-rich air from higher altitudes. At 650 K ozone
outside the outer edge of the vortex is significantly higher than ozone inside the inner
Figure 6.4: 2002/2003 POAM daily average observations on the 650 K, 500 K, 450
K, and 400 K potential temperature surfaces inside the inner vortex edge (blue) and
outside the outer vortex edge (red).
edge of the vortex primarily because poleward transport of ozone rich tropical and
subtropical air is limited to the vortex exterior [e.g., Manney et al., 1995a; Randall et
al., 1995]. Enhanced diabatic descent causes an overall increase in vortex ozone,
69
from about 3 ppmv in December to 4.5 ppmv in March. At 500 K vortex and extravortex ozone are nearly identical in early December. This is because enhanced
diabatic descent increases 500 K ozone mixing ratios sampled by POAM inside the
vortex by about the same amount that mixing with subtropical extra-vortex air
increases 500 K mixing ratios sampled by POAM outside the vortex. However, from
late December to late January vortex ozone diverges from extra-vortex ozone,
declining from about 3 ppmv to 2.3 ppmv. A gradual increase is then observed
during February and March inside the vortex. At 400 and 450 K, enhanced diabatic
descent causes vortex ozone to exceed extra-vortex ozone in early December. At 450
K in late January, however, vortex ozone declines to values comparable to those
observed outside the vortex. We interpret the ozone declines at 500 and 450 K as
evidence of chemical ozone loss. This interpretation is consistent with the
meteorology of the 2002-2003 winter described in Sect. 2. Vortex air was cold
enough in the early winter to support PSC formation, and had experienced significant
solar exposure as it was drawn to lower latitudes. Further evidence that conditions
were primed for ozone loss is seen in the POAM measurements of PSCs (not shown).
In December of 2002 the proportion of POAM observations in which a PSC was
detected was larger than previously observed in December by either POAM III or its
predecessor, POAM II, which operated from October 1993 to November 1996 [Alfred
et al., 2005]. PSC occurrence frequencies decreased substantially after the January
17 warming, with only sporadic observations of PSCs in February and March.
6.4 SLIMCAT 3-D CTM
70
Here we summarize the main details of the SLIMCAT 3-D CTM and describe
the initialization that was performed specifically for the study of the 2002-2003
Arctic winter.
6.4.1 Model description
SLIMCAT is a 3-D off-line chemical transport model described in
Chipperfield [1999]. The model has a detailed treatment of stratospheric chemistry,
which includes all of the species believed to be important in the chemistry of the
polar stratosphere, and a description of heterogeneous chemistry on solid and liquid
aerosols. The model temperatures and horizontal winds are specified from analyses
and the vertical transport in the stratosphere is diagnosed from radiative heating rates.
Radiative heating rates are used because they provide the best simulation of
stratospheric transport. SLIMCAT uses the Prather [1986] advection scheme which
has very low numerical diffusion. In the stratosphere the model uses an isentropic
coordinate and this has recently been extended down to the surface using hybrid
sigma-theta levels.
The setup of the model runs for the winter 2002-2003 simulations used here is
described in detail in Feng et al. [2005]. They summarize recent changes in the
model to improve the treatment of chemistry and transport relevant to the high
latitude lower stratosphere aimed at improving the model performance. For the runs
71
used here SLIMCAT was initialized on 1 January 1989 and integrated at low
horizontal resolution (7.5x7.5º) for ~14 years using European Centre for MediumRange Weather Forecasts (ECMWF) analyses [Feng et al., 2005]. The model has 24
levels from the surface to ~55 km and the resolution in the lower stratosphere is ~ 2
km. Output from this low resolution run was interpolated to a higher horizontal
resolution (2.8x2.8º) in mid-November 2002. This model was then integrated
through the 2002-2003 Arctic winter in a series of experiments.
6.4.2 Ozone initialization
A large source of uncertainty in the CTM-PS method is errors in the CTM
initialization, thus special attention was paid to the initial model ozone field. Satellite
observations of ozone were used to reinitialize the SLIMCAT ozone fields (both the
chemically integrated and passive fields) on 1 December 2002. Only the ozone
model fields were reinitialized in the model because of the lack of global observations
of other constituents. There is thus the potential for inconsistencies in runs when
ozone is not treated as a passive tracer, because the other constituents were
determined from the multiannual run as described above. This needs to be considered
when interpreting model and measurement differences; however, we feel that it is
better to use the ozone fields to constrain the model. The ozone fields were
constructed from 2002 Northern Hemisphere observations from POAM and the
Halogen Occultation Experiment (HALOE) using PV-mapping, as described in
Randall et al. [2002, 2004]. For the 1 December initialization date, the ozone
72
reconstruction included data acquired between 21 November and 11 December.
Based on statistical analyses of a year of reconstructions (not shown), on average the
ozone reconstructions agree with the satellite data comprising them to within 1%
above about 1000 K, but exhibit a 5% (0.1 ppmv) positive bias below 800 K. Figure
5 shows the POAM ozone profiles from 30 November through 2 December as well as
the 1 December mapped initialization fields interpolated to the POAM measurement
locations on these dates (30 November and 2 December are
Figure 6.5: Comparison of the ozone initialization profiles interpolated to the POAM
measurement locations for 30 November (top row), 1 December (middle row), and 2
December (bottom row). The left column shows the 1 December initialization
profiles (red) interpolated to the POAM measurement locations (black) on the dates
shown. Average differences between the profiles are shown in ppmv (middle
column) and percent (right column). Error bars denote 1σ standard deviation
distribution.
73
shown because POAM only made four measurements on 1 December). The
initialization fields overall compare well with the POAM observations, but are higher
than the POAM observations at 500 K. When combining satellite data and model
results to infer ozone loss, it is critical that the model faithfully represent the satellite
data prior to any ozone loss. If there is an offset between the model ozone and
satellite data, ozone loss (or production) will be inferred even on the initial date of
calculations. Such an initialization error will be carried through the calculations,
affecting modeled ozone changes due to both horizontal and vertical transport. The
500 K discrepancy shown in Figure 5 will lead to an overestimate in the modeled
ozone increase due to descent even at lower potential temperature levels, and hence
an overestimate in the chemical loss inferred by subtracting the modeled passive
ozone from the POAM ozone. These differences are considered when results are
interpreted.
6.4.3 Pure passive and pseudo passive runs
For this study each SLIMCAT run contained two ozone fields. In addition to
the chemically integrated “Active” ozone, which is coupled to the heating rate
calculation, the model contained a “Pure Passive” ozone tracer. The “Pure Passive”
ozone tracer was advected using identical transport to the other chemical species but
with no chemical change. The model results were then interpolated to the POAM
measurement locations. Chemical ozone loss was calculated by subtracting the Pure
Passive model ozone from the POAM measurements (“inferred” loss) or from the
74
Active model ozone (“modeled”loss). Conventionally, both gas phase and
heterogeneous chemistry are turned off in the passive model with the CTM-PS
technique. One concern with this and other ozone loss methods (e.g., tracer
correlations) is that passively transported ozone is not expected to be accurate if
transported for periods longer than approximately one month [Manney et al., 1995a,
b]. The main source of stratospheric ozone is from production in the middle
stratosphere at low latitudes [Brasseur and Solomon, 1984]. Manney et al. [1995b]
noted that if air is passively advected for long periods of time, this low-latitude ozone
source will not be maintained.
As a result, air passively transported poleward and downward may be
deficient in ozone. On the other hand, at polar latitudes local NOx (NO+NO2)
chemistry results in a net destruction of ozone in the middle stratosphere, so not
accounting for this chemistry would result in the downward transport of too much
ozone. Other ozone-destroying catalytic cycles can also be important in the lower
and upper stratosphere [Lary, 1997]. Thus, whether the net effect of gas phase
chemistry is to increase or decrease ozone depends on a number of parameters that
will vary in season, latitude, and altitude. For the December-March time period at
high latitudes, photochemistry is expected to be important mainly above about 650 K
[e.g., Garcia and Solomon, 1983; Randall et al., 1995], but these processes can also
contribute appreciably at lower altitudes, as shown below. To explore the effects of
gas phase chemistry on the ozone loss inferences, the model runs were first done with
the conventional “Pure Passive” calculation, and then repeated with a “Pseudo
Passive” calculation, in which gas phase reactions remained activated, but cold,
75
chlorine-activating heterogeneous chemical reactions on solid and liquid PSCs were
switched off. Results from both calculations are shown below to quantify the net
change in ozone (production-loss, with the caveat that this could be influenced by
errors in the transported ozone as described above) as well as the change due to
heterogeneous processes alone.
Figure 6 compares model results inside the vortex at the POAM locations for
the Pure Passive (no chemistry) and the Pseudo Passive (activated gas phase
chemistry) runs. Differences between the Pseudo and Pure Passive ozone mixing
ratios increase gradually in time at all altitudes, with the Pseudo Passive lower than
the Pure Passive. Differences between the Pseudo Passive and Pure Passive reach
about 0.6 ppmv in mid-March near 600 K. We attribute this to increasing catalytic
Figure 6.6: Comparison of the SLIMCAT Passsive ozone (ppmv) inside the vortex at
the POAM measurement locations for the Pure Passive (left) and Pseudo Passive
(middle) runs, and for the difference between the two (right; Pseudo minus Pure).
Ozone mixing ratios have been smoothed using a 7-day running average.
ozone destruction as sunlight returns to the polar region. It is interesting that
differences between the Pure and Pseudo Passive calculations decrease in magnitude
76
above 600 K in late February and March. This results from an increase in
competition between catalytic ozone loss at high latitudes and production at lower
latitudes (which was then followed by advection to high latitudes), so that the net
effect of photochemistry is less significant. Differences are smaller at the lowest
altitudes, consistent with the expectation that photochemistry, either through direct or
indirect (via descent of chemically processed air) mechanisms, should be less
important at these altitudes. Nevertheless, Figure 6 shows that even at potential
temperatures as low as 450-500 K, gas phase chemistry, in the absence of chlorineactivating heterogeneous reactions on solid and liquid PSCs, can contribute to ozone
loss by as much as 0.4 ppmv by mid-March. It is also important to consider that there
may be a potential inconsistency in the Pseudo Passive, since only the ozone fields
were reinitialized from observations. The Pure Passive ozone would not be affected
since ozone is treated as a completely passive tracer. Additional work will be
required to quantify the potential inconsistency.
6.5 Ozone loss during 2002-2003
In this section we apply the CTM-PS technique using both the Pure and
Pseudo Passive SLIMCAT CTM results to infer the magnitude of ozone loss inside
the Arctic vortex during the 2002-2003 winter from the POAM observations. CTMPS results are then compared with those calculated using the Vortex Average
technique.
77
The Passive technique is illustrated in Figure 7, which shows 2002-2003 time
series of POAM ozone inside the vortex and the co-located passive modeled ozone at
different potential temperatures. From 600 to 700 K (about 23 to 26 km), the overall
character of the Pseudo Passive model and POAM time series in Figure 7 is similar,
showing generally increasing ozone mixing ratios throughout the winter. We
attribute this overall increase to enhanced diabatic descent within the vortex.
Agreement between the Pseudo Passive and POAM data is often within the error bars,
although the model is systematically higher than POAM in December and January by
up to 0.2 ppmv. Because this bias first appears within the first week in December, we
attribute it to errors in the initialization field that cannot be checked due to lack of
global measurements. The necessity of including gas-phase chemistry in the Pseudo
Figure 6.7: Daily average ozone mixing ratios inside the vortex at the POAM
measurement locations for POAM (black) and the SLIMCAT Pseudo Passive (red) at
the six indicated potential temperatures. Error bars denote 1σ standard deviation of
the averages. Points without error bars indicate that only one POAM observations
was made inside the vortex at a given potential temperature level. Chemical ozone
loss is calculated by subtracting the Passive model from the POAM data. The Pure
Passive is shown by the gray line, without error bars (which are approximately the
same size as the error bars for the Pseudo Passive).
78
Passive model is apparent, as the agreement between the Pseudo Passive and POAM
data is better than between the Pure Passive and POAM data throughout most of the
winter. An exception to this occurs at 700 K in late February and March, when the
Pseudo Passive model underestimates POAM ozone mixing ratios, whereas the Pure
Passive is in agreement. Close inspection of the comparisons from 600-700 K
indicates that in late February and March, ozone in the Pseudo Passive model
systematically declines sooner than observed by POAM.
At 500 K, POAM measures increasing ozone in the first half of December,
followed by decreasing ozone mixing ratios into late January, and then increasing
ozone through mid-March. Model passive ozone at these altitudes steadily increases
throughout the winter due to enhanced diabatic descent in the vortex. Initialization
errors cause the Pseudo Passive model to exceed the POAM observations in early
December. From 2-6 December, for instance, the average difference between the
POAM data and Pseudo Passive model at 500 K is 0.31 ± 0.11 ppmv. Nevertheless,
it is clear that the decline starting in late December and continuing through January
represents a divergence of the observations from the passive model that on average
exceeds the initialization differences, an indication of chemical processes. Indeed,
the average difference between the POAM data and Pseudo Passive model at 500 K
from 2-6 January is 0.63 ± 0.10 ppmv, which is larger by a factor of 2 than the
difference obtained at the beginning of December. That chemical loss started in late
December and became increasingly statistically significant with time suggests that the
air parcels at 500 K sampled by POAM at this time had been exposed to PSC
formation temperatures and sunlight for prolonged periods of time. PSCs were
79
observed by POAM between 650 K and 500 K from late November through mid
January, with a few sightings in early February [Alfred et al., 2005]. Trajectory
calculations (not shown) confirm that air at the POAM measurement locations inside
the vortex at 500 K in late December had been exposed to temperatures below the
NAT condensation temperature and to as much as 50 h of sunlight in the previous 10
days.
The 450 K ozone increase through mid-December in both the Pseudo Passive
model and POAM observations is a signature of enhanced diabatic descent inside the
vortex and the absence of chemical loss. The observations begin to diverge from the
Pseudo Passive model in late December, as chemical ozone loss evidently begins,
even though POAM ozone is still increasing. POAM ozone decreases by about 0.4
ppmv in January, but then remains relatively constant or declines slightly through
mid-March, perhaps and indication of diabatic descent of air from above that has
experienced heterogeneous loss. Very low ozone (< 2 ppmv) is observed on several
occasions in early March, at a time when PSCs were observed at the POAM
measurement locations. It is thus possible that heterogeneous processing led to
localized ozone loss. More analysis is required to determine if heterogeneous
processing caused the localized ozone loss on such a short time scale.
Chemical ozone loss inside the vortex inferred from the POAM observations
in 2002-2003 is depicted as differences between the observations and the Passive
models in Figures 8 and 9, where negative differences signify loss. These figures
show results from both the Pure Passive and Pseudo Passive calculations. There is
80
Figure 6.8: Time series of the inferred ozone loss in 2002-2003 using the SLIMCAT
Pure Passive (black) and Psuedo Passive (red) (see text for details). Points represent
daily averages of measurements inside the vortex. The dotted black line denotes 0
ppmv. Error bars denote 1σ standard deviation of the differences. Points without
error bars indicate that only one POAM observation was made inside the vortex at a
given potential temperature level.
little difference between the two model calculations in December at any potential
temperature shown here. In both calculations, ozone loss (compared to the initial
differences on 1 December) begins in late December from about 450 to 550 K. Loss
this early in the winter is unusual, and as noted above, occurred after cold vortex air
was drawn equator ward to sunlit latitudes. After the major stratospheric warming on
17 January, ozone loss at 500 K ceases and begins to recover due to diabatic descent,
however, ozone loss continues at 450 K. Despite the January and February warming
events vortex temperatures at 450 K still fell below TNAT as (shown in Figure 1),
consequently ozone loss persisted. The region of maximum loss gradually descends
in altitude from about 500-550 K in late December to 400-450 K in mid March.
Because of initialization errors, the difference plots are somewhat misleading,
indicating more ozone loss than would otherwise be inferred had the initialization
81
been more accurate. Even at 425 K, where the initialization error is insignificant, an
overestimate in the loss would result from propagation of errors as the air from higher
altitudes that contained an initial bias descends. We conservatively estimate this
calculated loss bias to be on the order of 0.3 ppmv. Thus, Figure 9 shows that by
mid-March the maximum ozone loss due to halogen-catalyzed ozone destruction after
heterogeneous processing occurred near 425 K at a (corrected) value of
approximately 1.2 ppmv. Gas phase chemistry occurring in the absence of
heterogeneous processing contributed an additional 0.4 ppmv of loss from 400-500 K.
Figure 6.9: Inferred ozone loss (ppmv) in 2002-2003, as represented by the
difference between POAM and the SLIMCAT Pure Passive (left) or Pseudo Passive
(middle). The solid black line denotes the zero contour. Loss inferred from the
POAM measurements using the vortex average technique initialized with the inferred
1 January Pure Passive loss profile is shown in the right panel. Data have been
smoothed using a 7-day running average.
6.5.1 Vortex average inferred ozone loss
Ozone loss inferred using the CTM-PS approach is now compared to that
calculated using the vortex average technique (see Hoppel et al., 2002) applied to
82
POAM observations inside the vortex (Figure 9, right panel). Heating rates from the
radiative transfer model of Rosenfield et al. [1994] are used to calculate vortex
averaged diabatic descent as a function of potential temperature. These descent rates
are then used to estimate the vortex average ozone variation due to dynamics. Since
the vortex average technique shows loss due to all chemical processes, it should only
be compared to the Pure Passive CTM-PS result. The vortex average calculation does
not start until 1 January, in order to minimize the errors due to cross-vortex mixing
while still capturing the start of significant ozone loss. To account for the later
initialization date, the 1 January loss calculated by the Pure Passive CTM-PS
approach was added to the initialization of the vortex average calculation.
The vortex average loss is similar to the CTM-PS results. The region of
maximum ozone loss descends gradually throughout the winter in the lower
stratosphere, and is located at approximately the same theta level as in the CTM-PS
inference. However, more ozone loss occurs in the vortex average calculation near
400 K than in the CTM-PS calculation. A likely explanation is horizontal transport or
mixing across the vortex edge, which is not included in the vortex average approach
[Hoppel et al., 2002]. During highly disturbed winters, this can be a large source of
error in the vortex average calculation. At 400 K, the Northern Hemisphere vortex is
never very impermeable at these lower potential levels [Manney et al., 1994] and any
mixing with extra-vortex air will decrease ozone mixing ratios inside the vortex. By
omitting this effect the vortex average method will overestimate the ozone loss (the
dynamical component subtracted from the observations will be too high, so the
difference will be too large). Just the opposite will occur above 500 K, where extra-
83
vortex ozone mixing ratios are larger than those inside the vortex. Whether this effect
was large enough in 2002-2003 to cause the discrepancies shown in Figure 9 is a
subject of future investigation.
6.5.2 CTM-PS modeled ozone loss
In this section we compare the inferred ozone loss to the SLIMCAT modeled
loss using the Active and Pseudo Passive ozone fields. SLIMCAT modeled ozone
loss inside the vortex is shown in Figure 10, and is compared to the inferred loss in
Figure 11. Similar to the inferred loss, the region of maximum modeled loss
descends from about 500-550 K in late December to 425-450 K by mid March as
shown in Figure 11. The magnitude of the modeled loss in mid-March at 450 K and
500 K is about 0.2-0.3 ppmv less than that inferred from the observations. This is
Figure 6.10: 2002/2003 CTM-PS modeled ozone difference (ppmv) at the POAM
measurement locations inside the vortex (left), calculated as the Active model ozone
(middle) minus the Pseudo Passive model ozone (see Figure 7). Negative values
indicate modeled ozone loss. For comparison, the POAM ozone observations (ppmv)
are shown in the right panel. Solid black lines in the left panel denote 0 differences.
Ozone mixing ratios have been smoothed using a 7-day running average.
84
Figure 6.11: Comparison of the modeled ozone loss inside the vortex (blue) to the
inferred ozone loss using the Pseudo Passive (red). Error bars denote 1σ standard
deviation of the average differences. Points without error bars indicate that only one
POAM observation was made inside the vortex at a given potential temperature level.
consistent with the initialization error in the inferred loss calculations discussed
above. Similar to the inferred loss calculation, the maximum modeled loss occurs at
425 K. Additionally, the magnitude of the maximum modeled loss is approximately
1.2 ppmv by 15 March. These results suggest that SLIMCAT reliably simulates the
observations of ozone during 2002-2003. This is shown clearly in the two right
panels of Figure 10 and in Figure 12, which show contour plots and time series,
respectively, of the POAM measurements and the Active model ozone. At 450 K, the
model and observations generally agree within the standard deviations of the data,
with a small systematic bias between the two that is largely due to initialization
85
errors. There is an indication that the model might underestimate the loss at 450 K in
March, but variations in the distributions are too large to ascribe quantitative
Figure 6.12: 2002/2003 POAM (black) and SLIMCAT Active (red) In-V daily
average ozone mixing ratios at the potential temperatures indicated in each panel.
“Error” bars represent the standard deviation of the distribution of
measurements/model on each day.
significance to this. At 500 K and 600 K, the modeled ozone loss and inferred ozone
loss start to diverge in late January, with SLIMCAT underestimating ozone loss at
both levels. The disagreement is manifested as a failure of the model to maintain
ozone loss as long as is observed, which as shown in Figure 12 is caused by an
overestimate of ozone at these levels by the model. This may indicate that the model
incorrectly simulates the effects of the late January major warming at these levels,
allowing too much mixing with extra-vortex air or too much diabatic descent inside
the vortex. However, above 600 K in February and March model ozone is too low,
possibly suggesting an underestimate of descent rates or an underestimate of mixing.
86
6.6 Summary
We have presented an overview of the 2002-2003 Arctic ozone loss results
computed from the POAM satellite observations and the SLIMCAT CTM using the
CTM-PS technique. PSC occurrences peaked in December when the 2002-2003
stratospheric temperatures were at their lowest. Dynamical activity led to stretching
of the vortex to lower latitudes, which increased the amount of solar exposure
received by the vortex early in the winter, leading to a late December onset of ozone
loss. Stratospheric warming events limited PSC formation in late winter and early
spring. As a result, the maximum ozone loss inferred from POAM data for the 20022003 winter was moderate compared to other cold Arctic winters in the late-1990s.
Ozone loss results inferred from POAM observations and a Pseudo Passive
(activated gas phase chemistry) model were compared with those from a Pure Passive
(no chemistry) model to determine the influence of gas phase chemistry on CTM-PS
ozone loss calculations. The largest differences in the two passive fields occurred
above 450 K at a value of .6 ppmv and can be attributed to NOx chemistry included in
the Pseudo Passive run. After accounting for initialization errors, the maximum
ozone loss inferred from POAM observations and both CTM-PS calculations was
approximately 1.2 ppmv by mid March between 450 and 425 K.
The CTM-PS calculations were compared to Vortex Average ozone loss
calculations. Ozone loss from the Vortex Average technique was similar to the CTMPS technique, except that more loss was inferred near 400 K and less loss was
87
inferred at 500 K. Additional work is required to understand the differences between
the two techniques.
The SLIMCAT model was also run with the full chemistry in order to
compare model ozone loss with inferred ozone loss from the POAM observations.
Earlier studies have shown CTMs have had difficulty reproducing the extent of
denitrification and chlorine activation observed during cold Arctic winters and, as a
result, CTMs have typically underestimated ozone loss under these conditions.
Recent changes made in SLIMCAT, as discussed in Feng et al. [2005], have
improved the model’s ability to reproduce polar dynamical and chemical processes.
Consequently, the SLIMCAT model produces similar ozone loss morphology to the
inferred results for the 2002-2003 winter, with loss occurring in late December near
550 K and descending throughout the winter, maximizing near 425 K by 15 March at
around 1.2 ppmv. SLIMCAT’s ability to simulate ozone loss in Arctic winters with
different meteorological conditions will be the subject of future work. Initialization
remains an issue for the CTM-PS technique. Future near global ozone observations
from NASA’s Earth Observing System Aura spacecraft will be used to initialize CTM
ozone fields and to calculate ozone loss with the CTM-PS technique, resulting in
improved inferred and modeled ozone loss calculations.
88
Chapter 7
Quantifying Arctic Ozone Loss During the 2004-2005 Winter Using Satellite
Observations and a Chemical Transport Model
C. S. Singleton1, C. E. Randall1, V.L. Harvey1, M. P. Chipperfield2, W. Feng2, G.L.
Manney3,4, L. Froidevaux3, C.D. Boone5, P. F. Bernath5, K. A. Walker5, C.T.
McElroy6,7, K.W. Hoppel8
1
Laboratory for Atmospheric and Space Physics, UCB 392, University of Colorado,
Boulder, CO 80309-0392, USA
2
Institute for Atmospheric Science, School of Earth and Environment, University of
Leeds, Leeds LS2 9JT, UK
3
Jet Propulsion Laboratory, California Institute of Technology, Pasadena CA 91109
4
Department of Physics, New Mexico Institute of Mining and Technology, Las
Vegas, NM, 87701, USA
5
Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1,
Canada
6
Meteorological Service of Canada, Environment Canada, 4905 Dufferin St.,
Downsview, ON,
Canada, M3H 5T4
7
Department of Physics, University of Toronto, 60 St. George St., Toronto, ON,
Canada,
89
M5S1A7
8
Naval Research Laboratory, Remote Sensing Physics Branch, Code 7220, Building
2, 4555 Overlook Ave., SW, Washington, D.C. 20375-5351, USA
Abstract:
During the last decade much attention has been placed on quantifying and modeling
Arctic stratospheric O3 loss. At issue in particular is the reliability of models for
simulating the loss under variable dynamical conditions in the Arctic region. This
paper describes inferred O3 loss calculations for the 2004-2005 Arctic winter using
data from four solar occultation satellite instruments, as well as the Earth Observing
System Microwave Limb Sounder (EOS MLS). O3 loss is quantified with the
“Chemical Transport Model (CTM) Passive Subtraction” approach, using a passive
O3 tracer field from the SLIMCAT CTM. The 2004-2005 Arctic winter was
moderately active dynamically, but was still one of the coldest Arctic winters on
record, with prime conditions for O3 loss. Loss estimates inferred from all of the
different satellite instruments peaked in mid March at 450 K between 2-2.3 ppmv,
slightly less than similar estimations for the cold 1999-2000 winter. The SLIMCAT
CTM was also used to simulate O3 for the 2004-2005 winter. In March near 450 K,
the model O3 was 0.3 ppmv (~10-15%) lower than the observations, leading to a
maximum O3 loss that was 10-15% larger than that inferred from observations, using
the Passive Subtraction approach. Modeled loss maximized around the same time as
that inferred from observations. Although some discrepancies between the observed
and modeled O3 remain, the level of agreement presented here shows that the
90
SLIMCAT CTM was able to satisfactorily simulate O3 and polar O3 loss during the
dynamically active 2004-2005 Arctic winter.
7.1 Introduction
Since the discovery of the Antarctic O3 hole in 1985 [Farman et al., 1985],
modeling O3 loss has been an important focus of research in the atmospheric science
community. At issue most recently is the feedback between climate change and
stratospheric O3 levels. In order to predict changes in climate, scientists have
employed coupled Chemistry Climate Models (CCMs). An integral part of
developing CCMs is defining the most appropriate atmospheric chemistry modules,
which should ideally be derived from chemical transport models (CTMs). CTMs are
forced with winds and temperatures derived from analyses of observed
meteorological parameters; therefore, they cannot be used to make projections about
future climate. However, CTMs are used to simulate present day atmospheric
conditions and can be compared to observations to test our understanding of
atmospheric phenomena. Previous studies have shown that CTMs have
underestimated Arctic chemical O3 loss compared to observed loss [Chipperfield et
al., 1996; Goutail et al., 1997; Deniel et al., 1998; Becker et al., 2000; Guirlet et al.,
2000], which implies a gap in our understanding of O3 loss processes. WMO [2003]
states that “global CTMs reproduce a large fraction (60 to 100 %, depending on the
winter) of the observed O3 loss in the Arctic and its variability”; however,
uncertainties exist due to the “current unrealistic representation of denitrification
processes in 3-D CTMs and unexplained O3 losses during cold Arctic Januarys”. The
91
report also states that, “these uncertainties prevent reliable predictions of future Arctic
O3 losses in a potentially changing climate [WMO, 2003].” Therefore, in order to
correctly develop CCMs to make accurate predictions about future climate, it is
imperative for CTMs to simulate changes in the stratospheric O3 layer accurately.
Recent studies have shown that changes made to CTMs have now improved
their ability to simulate Arctic O3 loss, even during complex, dynamically active
winters [e.g. Feng et al., 2005]. Singleton et al. [2005] have shown that during the
2002-2003 Arctic winter the SLIMCAT CTM was able to simulate O3 loss that was
inferred from Polar Ozone and Aerosol Measurement (POAM) III observations. In
order to rigorously evaluate SLIMCAT or any other CTM for reliability in simulating
Arctic O3 loss processes, it is necessary to investigate multiple Arctic winters, since
there is large interannual variability due to complex dynamical activity [WMO, 2003].
This paper describes inferences of chemical O3 loss from observations (hereafter
referred to as the inferred O3 loss) inside the polar vortex for the 2004-2005 Arctic
winter (defined here as the time period from 1 December 2004 – 1 April 2005) using a
version of the well-validated “Passive Subtraction” technique [e.g. Harris et al., 2002;
WMO, 2003; Manney et al., 1995a, b; 2003, Singleton et al., 2005]. Calculations of
inferred O3 loss in the lower stratosphere are shown at discrete levels and for the
integrated partial column. Data from five different satellite instruments are used,
including POAM III, the Stratospheric Aerosol and Gas Experiment (SAGE) III, the
Earth Observing System Microwave Limb Sounder (EOS MLS), and the Atmospheric
Chemistry Experiment Fourier Transform Spectrometer (ACE-FTS) and
Measurement of Aerosol Extinction in the Stratosphere and Troposphere Retrieved by
92
Occultation (MAESTRO). Inferred O3 loss during the 2004-2005 winter is compared
to loss inferred during the cold 1999-2000 Arctic winter. These inferred O3 loss
calculations, as well as the observations of O3 itself, are compared to simulations from
the SLIMCAT CTM to deduce how well the dynamics and chemistry were simulated
for the 2004-2005 Arctic winter.
7.2 Data Sets
In this section the five satellite data sets that were analyzed for the 2004-2005
Arctic winter are described.
7.2.1
POAM III
POAM III (hereafter referred to as POAM) is a nine-channel solar occultation
photometer that was launched in March 1998; the instrument ceased operations in
December of 2005 due to an instrument anomaly. POAM has channels ranging from
0.353 to 1.02 µm and measures vertical profiles of O3, NO2, H2O, and aerosol
extinction [Lucke et al., 1999]. Because of the sun-synchronous polar orbit, 14-15
POAM observations occur around a circle of latitude in each hemisphere each day,
with Northern Hemisphere latitudinal coverage varying slowly between 55° N and
73° N. During the Arctic winter, POAM sampled both inside and outside the polar
vortex. For this analysis POAM version 4.0 retrievals, which have a vertical
resolution of approximately 1 km in the stratosphere, are used. Version 4.0 O3 data
have changed little from version 3.0, which was validated by Randall et al. [2003];
93
POAM O3 measurements agree to within ±5% with correlative ozonesonde and
satellite data between 13 and 60 km.
7.2.2 SAGE III
SAGE III (hereafter referred to as SAGE) was launched in December 2001,
and ceased operations in March of 2006. It utilizes solar occultation to measure
vertical profiles of O3, NO2, H2O, temperature, pressure, and aerosol extinction [Chu
et al., 2002; Thomason and Taha, 2003; Wang et al., 2006]. SAGE uses a grating
spectrometer with spectral channels ranging from 280 to 1545 nm. SAGE was
launched into a sun-synchronous orbit and its Northern Hemisphere observations
range between 50° and 80° N. Atmospheric profiles of O3 are sampled with ~0.5 km
vertical resolution in the lower stratosphere. As shown by Wang et al. [2006], “the
agreement between SAGE and correlative measurements is approximately 5% down
to 17 km”. For this work, version 3.0 SAGE data have been used in the
comparisons.
7.2.3
EOS MLS
EOS MLS was launched in July 2004. The EOS MLS instrument is
comprised of heterodyne radiometers operating in 5 spectral regions: 118 GHz, 190
GHz, 240 GHz, 640 GHz, and 2.5 THz [Waters et al., 2006]. EOS MLS measures
limb emission at these wavelengths to obtain vertical profiles of a number of species
94
that are relevant to polar studies including temperature, H2O, HNO3, O3, HCl, ClO,
and N2O. The NASA Aura satellite that hosts the EOS MLS instrument is in a nearpolar, sun-synchronous orbit. On each orbit, EOS MLS observations span from 82° S
to 82° N [Waters et al., 2006]. The vertical resolution varies for each species and is
approximately 2.7 km for O3 in the lower stratosphere [Froidevaux et al., 2006].
7.2.4
ACE-FTS and MAESTRO
The Atmospheric Chemistry Experiment (ACE) satellite was launched in
August 2003. Two solar occultation instruments are included on ACE, the ACE-FTS
and MAESTRO instruments. For this work ACE-FTS version 2.2 O3 update (which
compared to versions 1.0 and 2.2 has improved agreement with SAGE, POAM, and
ozonesondes near the profile peak) and MAESTRO version 1.1 data are used. ACEFTS is a high-resolution, Fourier transform infrared spectrometer [Walker et al.,
2005] that operates in the 2 to 13 micron spectral region. It measures the vertical
distribution of many constituents relevant to polar studies, including temperature, O3,
H2O, CH4, NO, NO2, HNO3, HCl, N2O5, and ClONO2 with a vertical resolution of
approximately 4 km in the lower stratosphere [Bernath et al., 2005]. MAESTRO is
an optical spectrometer covering the 400 to 1030 nm spectral region with a vertical
resolution of approximately 1km; it measures vertical profiles of O3, NO2, and aerosol
extinction.
As indicated above each instrument has a different vertical resolution for the
O3 observations. For this study, all observations and the CTM have been interpolated
95
to a standard potential temperature grid corresponding to a vertical resolution of about
1 km. Before conducting the analysis, the original profiles were compared to the
interpolated profiles to ensure that the vertical structure of the profile was not
compromised by the interpolation. Implications of the different vertical resolutions
for the conclusions drawn here are discussed below.
7.2.5 Satellite Comparisons During the 2004-2005 Winter
The left panel in Figure 1 shows the latitude sampling for POAM, SAGE,
MAESTRO, and ACE-FTS between 50° and 90° N during the 2004 – 2005 winter.
EOS MLS has been omitted from this panel because of the large number of
observations made by the instrument. In the Northern Hemisphere, EOS MLS (not
shown), SAGE, and POAM made observations poleward of 60° N from December
through mid April, while MAESTRO and ACE-FTS only sampled air poleward of
60° N from January through early February and from late February through early
April. The right panel in Figure 1 shows the equivalent latitudes of all the
Figure 7.1: Latitudinal coverage for POAM III (red), SAGE III (blue), MAESTRO
(black), and ACE-FTS (black) (Left) during the 2004-2005 Arctic winter. The right
plot shows equivalent latitudes of observations on the 500 K potential temperature
surface, applying the same color arrangement used in the left plot with EOS MLS
(gray). The white line indicates the innermost edge of the polar vortex.
96
observations made between 50° and 90° N on the 500 K potential temperature surface
during the same period. Equivalent latitude, originally defined by Butchart and
Remsberg [1986], is a vortex-centered coordinate system where 90° is always in the
center of the vortex. The Arctic vortex is often displaced from the pole; therefore,
conventional zonal means obscure dynamical features by averaging air that is inside
and outside the vortex; thus equivalent latitude is used here. The equivalent latitude
was calculated from potential vorticity (PV), which was computed from the European
Centre for Medium-Range Weather Forecasts (ECMWF) operational analyses. The
equivalent latitude was then interpolated from the ECMWF model grid to the
observation locations. The vortex algorithm of Harvey et al. [2002] was applied to
calculate the edge of the Arctic vortex. This algorithm integrates Q, a strain/rotation
parameter around streamfunction isopleths in each hemisphere. Q is negative
(positive) in flows dominated by rotation (shear). Q is negative inside the center of
the vortex and becomes positive toward the edge with large gradients near the polar
night jet. Neighboring streamfunction contours where Q changes sign are vortex
edge candidates. Of these candidates, the streamfunction contour with the largest
integrated wind speed is defined as the vortex edge (see Harvey et al. [2002] for
further details). This vortex edge definition agrees well with algorithms by Waugh
and Randel [1999] and Nash [1996]. The white line in the right panel of Figure 1
denotes the vortex edge as determined using this criterion. The right panel of Figure
1 shows that SAGE, EOS MLS, and POAM sampled air within the vortex throughout
most of the Arctic winter. SAGE and POAM observations inside the vortex were
located mostly near the edge in mid to late December, and SAGE measurements
97
inside the vortex in February and March were more sporadic than in January. ACEFTS and MAESTRO only sampled vortex air (i.e. air inside the vortex) from early
January through late March, with fewer measurements inside the vortex in early
February.
7.3 Methods
To quantify O3 loss from satellite observations, the “Passive Subtraction”
technique has been used. In this technique, a passive O3 tracer field, which represents
the change in O3 due to dynamics (horizontal mixing and descent), is subtracted from
the observations to quantify the change in O3 due to chemistry [Manney et al.,
1995a,b; 2003; Goutail et al., 1997, 2005; Deniel et al., 1998; Hoppel et al., 2002;
Singleton et al., 2005]. This technique, which has been widely used and compared
favorably with other O3 loss methods [e.g. Harris et al.,2002; WMO, 2003], was
applied to the five data sets discussed above. Here we used the University of Leeds
SLIMCAT CTM [see Feng, W., et al., Large chemical ozone loss in 2004/05 Arctic
winter/spring, submitted to Geophys. Res. Lett., 2006, hereafter referred to as FE;
Feng et al., 2005] to simulate a global, passive O3 tracer. The passive O3 tracer field
was then linearly interpolated from the model grid (2.8° × 2.8°) to each satellite
measurement location. The inferred O3 loss (IL) at each measurement location is
calculated as the difference between the satellite observation and the modeled passive
O3; hereafter this technique is referred to as “CTM Passive Subtraction” (CTM-PS)
[Singleton et al., 2005]. Unlike the study of Singleton et al. [2005], which utilized
two different methods to quantify the chemical O3 loss, only one method is used here.
98
The full chemical model was also sampled at each observation location and the
difference between the full model and the passive tracer is referred to as the modeled
or simulated chemical O3 loss (ML). The IL and ML were calculated at the satellite
observation locations that were inside the vortex, as determined using the vortex
criteria of Harvey et al. [2002]. Loss calculations were not extended below 400 K
because of uncertainties in the identification of the vortex edge due to contamination
from the presence of the subtropical jet [Harvey et al., 2002] and the increased
influence of mixing [e.g., McIntyre, 1995].
The SLIMCAT CTM is a 3-D offline model with detailed stratospheric
chemistry, which includes heterogeneous chemistry on solid and liquid aerosols. A
full description of the model can be found in Chipperfield [1999] and detailed
adjustments to the model chemistry and transport are discussed in Feng et al. [2005]
and FE. The SLIMCAT CTM uses a hybrid sigma-theta grid, with isentropic
coordinates in the stratosphere [see Chipperfield 2006]. The run used here extends
from the Earth’s surface to approximately 55 km and has a vertical resolution of
approximately 2 km in the lower stratosphere. For the 2004-2005 simulation, the
model temperatures and horizontal winds were forced by ECMWF operational
analyses. The NCAR CCM radiation scheme [Briegleb, 1992] was used to handle
vertical transport above 350 K.
For this work, a low resolution (7.5 x 7.5°) run was started on 1 January 1977
and forced with ECMWF ERA-40 and operational analyses. A high resolution (2.8 x
2.8°) model run was initialized in November 2004 from the low resolution run. The
same SLIMCAT CTM initialization of FE has been applied in this work. Values of
99
tropospheric source gases (e.g., CH4, N2O, and halocarbons) have been set based on
WMO [2003] [FE]. The model was then integrated through the 2004-2005 Arctic
winter with daily meteorological input files from the ECMWF operational analyses.
7.4 Meteorology
The 2004-2005 Arctic winter was one of the coldest winters recorded in the
Arctic [e.g., Manney et al., 2006]. Figure 2 shows the area in the Northern
Hemisphere from 1978 through 2005 during the months from December through
February where temperatures fell below the nitric acid trihydrate (TNAT) formation
temperatures. Temperatures prior to 2000 were taken from ECWMF ERA-40
reanalysis, while temperatures after this year were taken from ECMWF operational
Figure 7.2: Area (106 km2) where Northern Hemisphere temperatures fell below
TNAT during the winters from 1978-1979 to 2003-2004 (colors) and for 2004-2005
(black) for the 550 K, 500K, and 450 K potential temperature surfaces.
100
analyses. Although ERA-40 has a bias in temperature with respect to other analyses,
it correctly handles interannual variability [e.g., Manney et al., 2005; Tilmes et al.,
2006]. A daily averaged TNAT value was computed using the Hanson and
Mauersberger [1988] expression, where average vortex HNO3 and H2O values were
taken from the 2004-2005 SLIMCAT CTM run. These calculations were conducted
for the 550 K, 500 K, and the 450 K potential temperature surfaces. Throughout most
of the winter, the 2004-2005 TNAT areas are at the high end of the range shown here.
At all three potential temperature levels 2004-2005 had the largest area with
temperatures below TNAT in late January. There was also another period at 450 K in
mid February where the 2004-2005 winter had the largest possible PSC formation
area since 1978. The 2004-2005 Arctic winter vortex enclosed a volume of air where
temperatures fell below TNAT that was larger than previously observed [e.g., Rex, M.,
et al., Large stratospheric ozone loss during Arctic winter 2004/2005, submitted to
Geophys. Res. Lett., 2006, hereafter referred to as RE; Tilmes et al., 2006]. This
diagnostic is a measure of the probability of PSC formation, which is positively
correlated with column O3 loss.
To examine the 2004-2005 meteorology more closely, Figure 3 shows the
difference between the minimum ECMWF operational temperatures found inside the
vortex at the satellite measurement locations and the daily average TNAT (calculated
same as above) for the 600 K, 500 K, 450 K, and 425 K potential temperature
surfaces. Of all of the instruments, EOS MLS sampled air that experienced the
lowest temperatures. Because of the large sampling of the instrument, the EOS MLS
temperatures are most representative of the overall minimum temperatures found
101
inside the vortex. Since the other instruments sampled air masses with higher
temperatures, this suggests that the other instruments did not sample air in the center
of the vortex cold pool. From December through the end of February temperatures at
the instrument locations often fell below TNAT, with the exception of the ACE
instruments, which either did not sample inside the vortex or were at lower
geographic latitudes (in early January). Temperatures dramatically increased by
Figure 7.3: Minimum ECMWF operational temperatures minus daily average TNAT
(see text) inside the vortex at the EOS MLS (gray), ACE-FTS and MAESTRO
(black), POAM III (red), and SAGE III (blue) measurement locations on the 600 K,
500 K, 475 K, and 425 K potential temperature surfaces during the 2004-2005 Arctic
winter.
March at all of the measurement locations and never returned to the TNAT levels due
to the onset of a “major final warming” [Manney et al., 2006].
Even though the Arctic region was unusually cold in 2004-2005, the polar
vortex was dynamically active [Manney et al., 2006; Schoeberl et al., 2006]. Figure 4
102
shows ECMWF PV and the vortex edge computed by the vortex determination
criteria of Harvey et al. [2002] on the 500 K potential temperature surface for
selected representative days. SAGE, POAM, ACE-FTS, and MAESTRO
measurement locations are superimposed. From December through mid February the
vortex remained intact. On 24 February, a warming occurred in the lower
stratosphere (see Figure 2) that caused a localized split in the vortex; by 1 March the
vortex had reformed. However, by mid March the major final warming caused the
vortex to split throughout the depth of the stratosphere and it remained split
throughout March. Between January and March the vortex often traversed to lower
latitudes, which allowed for more sunlight exposure and thus greater potential for O3
loss.
Figure 7.4: ECMWF operational PV (10-6 K m2 kg-1s-1) on the 500 K surface for
representative days during the 2004-2005 winter. The inner vortex edge is denoted
by the white contour. The POAM III, SAGE III, ACE-FTS and MAESTRO
measurement locations are indicated with circles, diamonds, and crosses,
respectively. Latitudes range from the equator to the pole, with latitude circles drawn
in 45 degree increments.
103
7.5 Results
7.5.1 Inferred O3 Loss
In order to quantify the IL, the passive model field was interpolated to the
satellite measurement locations inside the polar vortex. Daily average profiles of IL
for each of the satellite instruments are shown in Figure 5. For qualitative purposes,
days when the instrument did not sample inside the vortex and days with missing
data (see Figure 1) were filled in by a time interpolation. Data in all contour plots
presented in this paper have been smoothed using a seven-day running average.
When applying the CTM-PS technique, it is important to analyze the
agreement between the model and the observations on the first day of the analysis, 1
December 2004. Any differences at this time will descend in the model during the
Figure 7.5: Differences (ppmv) between passive O3 calculated by the SLIMCAT
CTM and O3 measured by the POAM III, SAGE III, EOS MLS, ACE-FTS, and
MAESTRO instruments. Results correspond to daily averages over the measurement
locations inside the vortex during the 2004-2005 Arctic winter, and are indicative of
inferred photochemical loss throughout the winter. Days with missing data and days
where an instrument did not sample the vortex have been filled in with a time
interpolation. The solid black line denotes the zero contour. Data have been
smoothed with a 7-day running average.
104
run in accordance with the model vertical transport. A positive offset (an indication
of O3 production) or a negative offset (an indication of O3 loss) on 1 December would
falsely mask or enhance IL calculations on lower potential temperature surfaces at a
later date [Singleton et al., 2005]. On 1 December only POAM, SAGE, and EOS
MLS were taking observations inside the polar vortex. The IL from these instruments
falsely indicates a small loss of approximately 0.2 ppmv on 1 December above 450 K
because of an initial mismatch between the model and observations. Below 450 K
there was a more significant (false) loss of 0.5 ppmv at the start of the analysis.
Typically, air starting below 450 K on 1 December would descend too far (below 400
K, the lower limit of this analysis) during the winter to affect air near levels of peak
O3 loss in March; however, air above 450 K would likely have an impact on O3 loss
calculations inferred at the end of the winter. In this case the offset could lead to an
overestimate of loss by up to 0.2 ppmv.
Similar morphology is observed in all of the plots in Figure 5. In particular,
loss (relative to the initial offset) begins in early January at the highest potential
temperature levels depicted here. In order to determine if this loss is truly chemical
or a problem with the modeled dynamics, SLIMCAT and ACE-FTS vortex CH4
fields were compared (not shown). The CH4 fields overall show good agreement.
During late December through mid January, the rate of descent in SLIMCAT is
somewhat faster than the observations above 550 K. At the same altitudes, signatures
consistent with mixing are apparent in both the model and the observations in mid to
late January; e.g., CH4 increases in time, whereas descent alone would lead to
monotonic decreases. Manney et al. [2006] also point out that vortex intrusions had
105
occurred during this time as indicated by EOS MLS N2O observations. SLIMCAT
shows a slightly sharper rise in CH4 than ACE-FTS, which may indicate the model
slightly overestimates mixing compared to the observations. Both of these effects
could increase the SLIMCAT Passive O3, resulting in too much loss at these
altitudes. However, Manney et al. [2006] show that chlorine activation was observed
by MLS above 600 K. Our understanding of this high-altitude loss is thus not
complete, and will be a subject for further study. Loss gradually increases at lower
potential temperature levels, as depleted air descends in time and additional local
chemical depletion occurs. By late February, loss greater than 1.0 ppmv is inferred
throughout the altitude range below 575 K, persisting throughout the month of
March. Maximum losses occur near 450 K in mid-March, at values near 2 ppmv.
Note that other studies have shown that significant loss occurred during the 20042005 Arctic winter below 400 K [RE; von Hobe et al., 2006], a region not
investigated in this study.
A more quantitative diagnosis can be conducted by examining time series of
daily averaged IL, which are shown in Figure 6 for the 600 K, 500 K, 475 K, and 450
K surfaces. Unlike Figure 5, days where the instrument did not sample inside the
vortex and days with missing data have not been filled in with a time interpolation.
Neither ACE-FTS nor MAESTRO sampled vortex air until January. At all levels
shown here, IL calculations for the different instruments agree very well, although
the solar occultation results are noisier because not as many data points are averaged
together compared to EOS MLS (see the right panel of Figure 1 for instrument
sampling). Over the range of potential temperature levels considered here, the
106
maximum amount of IL during the 2004-2005 winter occurred between 475 K and
450 K at approximately 2 – 2.3 ppmv. The IL calculations presented here are
comparable to those computed from other techniques. Manney et al. [2006]
computed EOS MLS IL by analyzing EOS MLS N2O observations and vortexaveraged descent rates and found a maximum vortex averaged loss of 1.2-1.5 ppmv
between 450 K and 500 K. The same analysis applied to the outer edge of the vortex
indicated a maximum loss of ~2 ppmv in the same potential temperature region
[Manney et al., 2006]. Jin et al. [2006] inferred loss from ACE-FTS observations
from correlations of O3 and CH4, correlations between O3 and an artificial tracer, and
Figure 7.6: Time series of the inferred daily average O3 loss (ppmv) inside the vortex
from EOS MLS (gray), POAM III (red), SAGE III (blue), ACE-FTS (black), and
MAESTRO (green) for the 600 K, 500 K, 475 K, and 450 K surfaces during the
2004-2005 Arctic winter.
107
the profile descent technique. The profile descent technique has also been applied in
previous O3 loss studies, but has been has been described with different nomenclature
[e.g., Larsen et al.,1994; Urban et al., 2004; Raffalski et al., 2005; Manney et al.
2006]. Jin et al. [2006] found that the maximum IL occurred between 475 and 500 K
and ranged from 1.6-2.3 ppmv. RE applied the vortex average descent approach to
SAGE and POAM data to calculate the accumulated loss between 5 January and 25
March. Compared to the CTM-PS IL results, RE indicates that the maximum loss
occurred at a lower altitude (between 400 and 450 K) and was much less than the
CTM-PS. In particular, the vortex average descent approach shows a loss of 1-1.2
ppmv at 500 K for SAGE and POAM, whereas the CTM-PS approach indicates a
larger loss of 1.7 – 2 .0 ppmv at this altitude. The vortex average descent approach
also shows a larger amount of loss at 450 K compared to 500 K, which is not as
pronounced in the CTM-PS IL calculations. However, it is important to point out that
accumulated loss calculated in RE started on a later date than the CTM-PS IL
calculations (5 January compared to 1 December). In Figure 6 there is a positive
slope in the differences at 600 K between the observations and modeled passive O3
during December, most evident in the SAGE data. Since O3 production at 600 K in
December is not expected, this suggests a slight error in the passive model
calculations. At this level O3 mixing ratios are generally greater outside the vortex
than inside because of poleward transport of O3 rich subtropical air [Manney et al.,
1995b; Randall et al., 1995; Singleton et al., 2005]. Thus, the slope could result from
horizontal transport of air from outside the vortex that is not captured by the model.
SAGE measurement locations moved equatorward during December, closer to the
108
vortex edge than the other instruments, possibly magnifying any error in the model
transport near the vortex edge.
7.5.2 IL and Instrument Sampling
As previously mentioned the EOS MLS instrument makes near-global
observations on a daily basis, whereas the geographic sampling of the solar
occultation instruments is limited to a single latitude circle on any given day.
Therefore, the EOS MLS IL should more adequately represent true vortex average
conditions. However, the right panel of Figure 1 indicates that although the solar
occultation instruments have limited geographic sampling, they observed a wide
range of equivalent latitudes during the winter; thus in terms of vortex space the solar
occultation instruments were not very limited. To examine the sensitivity of the solar
occultation IL calculations to limited geographic coverage, we sampled the EOS MLS
instrument at the other instrument locations and then calculated the IL. The results of
this test are shown in Figure 7. Qualitatively, the IL morphology derived from EOS
MLS sampled at the solar occultation locations is the same as derived from the solar
occultation instruments themselves. Inferred O3 loss begins earliest at the highest
potential temperature levels, followed by increasing loss at lower levels, with peak
loss occurring near 450 K in March. The peak magnitude of the loss inferred from
EOS MLS data sampled at the solar occultation locations is only slightly less than
that inferred from the solar occultation data themselves.
109
Figure 7.7: As in Figure 5, but using EOS MLS O3 sampled at the POAM III, SAGE
III, ACE-FTS, and MAESTRO instrument locations during the 2004-2005 Arctic
winter. The middle panel is the same as in Figure 5.
The similarity between Figures 7 and 5 confirms that the individual differences
between each of the panels in Figure 5 are due primarily to differences in geographic
sampling. This is expected since O3 data from most of the instruments have been
well validated [e.g., Randall et al., 2003; Wang et al., 2005; Froidevaux et al., 2006]
(although see discussion below regarding ACE-FTS versus MAESTRO). EOS MLS
has the lowest vertical resolution of all of the instruments tested. Therefore, if the
vertical resolution had been significant there would have been large discrepancies
between Figure 7 and 5. That IL variations from instrument to instrument are
generally quite minor (e.g., see Figure 6) leads to the conclusion that even though
solar occultation geographic coverage is limited, the sampling of equivalent latitude
can be broad enough to adequately define O3 loss inside the vortex. RE also showed
little difference in the O3 loss estimates from sondes, POAM, and SAGE, although
the sampling of the instruments in latitude space was quite different. RE concluded
110
that the similarities in the results increase the confidence that sampling issues do not
have a significant impact on the IL calculations.
The ACE-FTS vs. MAESTRO comparisons require further explanation. Only
preliminary validation has been completed for ACE-FTS version 1.0 data [Walker et
al., 2005]; however, detailed studies are in progress for the current versions of both
ACE-FTS and MAESTRO data that we use here. Because these instruments are on
the same satellite, their geographic sampling is identical. Therefore, differences
between IL results from these instruments can be attributed to differences in the solar
occultation measurements themselves. Indeed, the only reason the Figure 7 EOS
MLS plots for the ACE-FTS and MAESTRO sampling are not identical is that days
where there is missing ACE-FTS or MAESTRO data (as with all the other
instruments) have been filled in with a time interpolation. That the MAESTRO data
show more apparent IL in Figure 5 is attributed to the fact that O3 mixing ratios
retrieved from MAESTRO are on average ~5-10% lower than those retrieved from
ACE-FTS, even for coincident data (not shown). However, because of the higher
variability in the MAESTRO data, few O3 measurements near 600 K caused the
average vortex MAESTRO O3 to be slightly higher than the ACE-FTS data in
January.
7.5.3
Column O3 Loss
The partial column is the vertical integral of the difference between the
observed O3 and the modeled passive O3 field [Rex et al., 2002]. The partial column
111
O3 loss for 2004-2005 is shown in Figure 8. Here the partial column was computed
between the potential temperature surfaces of 400 K and 575 K, and only
observations where the instruments were sampling within the vortex at all potential
temperature levels were used in the calculation. Figure 8 shows that, overall, the
partial column O3 loss is in strong agreement for the five instruments, with maximum
loss at the end of March near 100-120 Dobson Units (DU), with an uncertainty of
~15 DU. There is an initialization offset between the model and the observations on
the first day of the analysis, representing about 15% of the maximum loss observed.
The partial column loss results, once corrected for the initialization offset, agree well
with the 90 DU loss calculated by RE between 400 K and 550 K from ozonesondes
(M. Rex, private communication).
Figure 7.8: Partial column loss for the EOS MLS (gray), POAM III (red), SAGE III
(blue), ACE-FTS (black), and MAESTRO (green) instruments between the 575 K
and 400 K potential temperature surfaces during the 2004-2005 Arctic winter. Only
profiles inside the vortex were included in the calculation.
7.5.4 Comparisons with 1999-2000 IL Calculations
112
The Arctic O3 loss which occurred during 2004-2005 was large compared to
most previous Arctic winters. One of the coldest observed Arctic winters took place
during 1999-2000 [e.g. WMO, 2003]. In order to compare the two Arctic winters
directly, the CTM-PS technique was applied using the SLIMCAT CTM for the 19992000 winter. The same version of the SLIMCAT CTM was used with the 1999-2000
run. Unlike the 2004-2005 run; however, the SLIMCAT CTM was only forced with
ECMWF operational winds from January through the remainder of the winter; during
December 1999 the CTM was forced with ECMWF ERA-40 winds. Since POAM
was the only instrument taking observations in both 2004-2005 and 1999-2000 and
because there are in general very few discrepancies in the IL for the five instruments,
we compare the two years only using POAM data.
Figure 9 shows the time series of the daily averaged IL inside the vortex for the
POAM instrument during both of the cold Arctic winters for the 575 K, 550 K, 500
K, and 475 K potential temperature surfaces from 1 December until 15 March. The
analysis was only conducted through mid-March because after this time in 2000 the
vortex was too close to the pole for POAM to sample [Hoppel et al., 2002; Rex et al.,
2002]. Unfortunately, the analysis was not conducted for levels below 475 K because
100 hPa was the lower limit for the 1999-2000 ECMWF operational wind fields that
were used to determine the location of the vortex edge. These comparisons are
therefore at potential temperature levels above where the maximum loss (in terms of
mixing ratio) occurred in both winters [RE]. RE applied the vortex average descent
approach between 5 January and 25 March to SAGE, POAM, and ozonesondes in
2006 and to ozonesondes in 2000, and showed that the 2005 loss below 475 K was as
113
Figure 7.9: Time series of the inferred daily average O3 loss (ppmv) for POAM III
during the 2004-2005 (closed circles) and the 1999-2000 (open circles) Arctic
winters. The inferred loss is shown for 1 December through 15 March for the 575 K,
550 K, 500 K, and 475 K surfaces.
large as or larger than the loss that occurred in 2000. Above 475 K, differences
between CTM-PS and RE are expected because of the differences in the time period
over which the loss calculations were made. Figure 9 reveals that on the 575 K and
550 K potential temperature surfaces the 2004-2005 calculations show slightly more
loss than the 1999-2000 calculations during March. RE also show that there was
more loss in March during 2005 than in 2000 at 550 K, but indicate up to 1 ppmv less
accumulated loss occurred in both years compared to the CTM-PS results. The
reverse occurs on the 500 K and 475 K surfaces, as shown in Figure 9, where the
1999-2000 winter shows slightly more loss in March, by approximately 0.2 and 0.5
114
ppmv, respectively. RE indicate a larger difference between the years at 475 K, with
0.9 ppmv more loss during 2000 than in 2005. Figure 9 indicates that early in the
winter (mid December through the end of January) more loss occurred during 19992000 than in 2004-2005 between 475 K and 575 K (the upper reaches of chemical O3
loss). We speculate that this is due to the fact that 2004-2005 was more dynamically
active than 1999-2000 [Manney et al., 2006; Salawitch et al., 2002]. Our
understanding of these differences is not complete, and will be a subject for further
study. These results agree with those presented in RE and Manney et al. [2006],
which show that more loss occurred during 2000 at 500 K and 475 K than in 2005.
7.5.5 Modeled O3 Loss
The SLIMCAT CTM ML was analyzed to determine how well the model was
able to reproduce the 2004-2005 Arctic O3 loss inferred from the observations. A
contour plot of the daily average ML is shown in Figure 10. In order to compute the
Figure 7.10: As in Figure 5, but for modeled daily average O3 loss (ppmv) inside the
vortex at the POAM III, SAGE III, EOS MLS, ACE-FTS, and MAESTRO locations
during the 2004-2005 Arctic winter.
115
ML, the SLIMCAT CTM was sampled at each of the satellite measurement locations.
As in Figure 5, data on days where there was missing satellite data and on days where
the instrument did not sample inside the vortex are filled in by a time interpolation.
Both the IL and ML show strong descent of O3 loss starting near 700 K in December
and peaking below 500 K by mid March. Above 500 K the ML slightly
underestimates the magnitude of the IL starting in February. The amount by which
the model underestimates the loss varies for each instrument. The opposite is true
below 500 K, where the ML calculations show a peak loss that is larger in magnitude
by about 0.3 ppmv than the peak IL values at these potential temperature levels. At
450 K the ML diverges from the IL calculations starting in late February, when the
major final warming occurred. This suggests the possibility that the model had
difficulty simulating the effects of the stratospheric warming event. Despite these
differences, qualitatively the ML agrees well with the IL calculations.
7.5.6
Satellite O3 & CTM Active O3
The differences between the ML and the IL calculations are due to differences
between the satellite O3 observations and the Active (full chemistry and dynamics)
model field. If the chemistry, physics and the meteorology in the SLIMCAT CTM
were accurate, the Active model O3 should match the observations, within the
uncertainties introduced by initialization errors. Figure 11 shows the evolution of
daily average vortex O3 for the observations, the corresponding Active model sampled
at the observation locations, and the difference (Active model O3 minus observed O3).
Time series of the Active model and the observations at 600 K and 450 K are shown
116
in Figure 12. Qualitatively, the model simulates the observed O3 quite well. At these
altitudes, O3 mixing ratios increase with increasing altitude. Between 550 K and 700
K, O3 mixing ratios increase in time from December to March because of poleward
transport of ozone-rich air into the vortex. From about 450-550 K, O3 mixing ratios
decrease in time, as halogen-activated O3 destruction becomes prevalent. Below 450
K, O3 mixing ratios increase in December and early January as O3-rich air descends;
mixing ratios then decrease as halogen-activated chemistry takes over.
Figure 7.11: Daily average vortex O3 (ppmv) during the 2004-2005 Arctic winter for
POAM III, SAGE III, EOS MLS, ACE-FTS, and MAESTRO (top row), the
SLIMCAT Active model interpolated to the satellite locations (middle row), and the
SLIMCAT Active model minus the observations (bottom row). Days with missing
data and days where an instrument did not sample the vortex have been filled in with
a time interpolation. Data have been smoothed with a 7-day running average.
117
One notable feature in Figure 11 is the small model underestimate of O3 that
descends in time throughout the winter; this is surrounded at lower and higher
altitudes by model overestimates of the observed O3 field. The overestimates at the
beginning of the winter are correlated with initialization errors, and cannot be
interpreted as indicating any errors in the model chemistry or physics. The precise
magnitude and altitude of the underestimate (blue region in Figure 11) varies with
instrument and time, but in general is on the order of 0.3 ppmv or less, and descends
from about 550-600 K in December to about 400-450 K in mid-March. The
underestimate is on average smaller when compared to EOS MLS data, which may in
part be due to the fact that averaging large numbers of data points masks any
initialization errors that might be present at given latitudes. Since the model O3
underestimates occur near the primary altitudes/times of halogen-activated O3 loss,
they most likely result from small discrepancies between the simulated and observed
relative contributions of transport and chemistry to the overall O3 variations. The
Figure 7.12: Observed (black) and modeled (gray) daily average O3 (ppmv) inside
the vortex for POAM III, SAGE III, EOS MLS, ACE-FTS, and MAESTRO on the
600 K (top row) and 450 K (bottom row) potential temperature surfaces during the
2004-2005 Arctic winter.
118
results from Figure 12 are particularly interesting at 450 K, where the model
overestimates O3 in December and January, but then underestimates O3 in March. As
noted above, the initial overestimate is clearly related to an initialization error, as it
occurs even in early December. The fact that this then transitions to an underestimate
explains the larger ML compared to the IL. The differences between the model and
observations in March are not well understood. They may be due to the fact that the
model did not properly simulate the impact of the final stratospheric warming event.
That is, it is possible that the model was too diffusive in the lower stratosphere and
allowed low O3 from outside the vortex to mix into the vortex, resulting in O3 mixing
ratios that were too low. SLIMCAT and ACE-FTS vortex averaged CH4 mixing
ratios (not shown) increase in time during March below 500K, consistent with
increased mixing. EOS MLS N2O observations presented in Manney et al. [2006] also
confirm that mixing was occurring during this time. As with the higher altitudes, CH4
mixing ratios in SLIMCAT increase more than those in ACE-FTS, which may be a
signature of mixing that is not accounted for properly in the model. The differences
in March may also be due to the model overestimating the chemical loss because of
too strong denitrification, which would allow high ClOx to persist longer [M.P.
Chipperfield, private communication].
7.6 Summary and Conclusions
We have presented an overview of the O3 loss during the 2004-2005 Arctic
winter using the CTM-PS technique. The 2004-2005 winter had lower temperatures
119
than most previously recorded Arctic winters [Manney et al., 2006; FE; Jin et al.,
2006; RE] and had a dynamically active vortex. During the winter the vortex
experienced two warming events. The first event occurred in late February and split
the vortex in the lower stratosphere. In addition to weakening the vortex, the
warming event caused the vortex to be stretched and brought down to lower latitudes.
The vortex reformed in a couple of days and remained intact until the major final
warming event occurred in early to mid March [Manney et al., 2006] and split the
vortex through the depth of the stratosphere. After the first warming event the
temperatures warmed and never fell below TNAT for the remainder of the winter.
The 2004-2005 Arctic winter was unique because of the large number of
satellite observations made. Here we have employed observations from the POAM,
SAGE, EOS MLS, ACE-FTS, and MAESTRO satellite instruments. The low
temperatures experienced during the winter allowed for prime conditions for chemical
O3 loss. The inferred loss calculations show similar morphology for all of the
instruments. The peak inferred loss occurred for each instrument at approximately
450 K in mid March at a value of approximately 2–2.3 ppmv. A slight offset between
the passive model and SAGE, POAM, and EOS MLS data on 1 December may cause
an overestimate of the loss by approximately 0.2 ppmv. Maximum O3 loss
calculations are comparable to calculations of FE and to outer vortex calculations of
Manney et al. [2006], but are slightly larger than RE. Partial column loss results
agree well with ozonesonde calculations of RE between 400 K and 550 K (M. Rex,
private communication). Because each instrument had different latitudinal sampling
of the vortex, we were able to evaluate the sensitivity of the inferred loss calculation
120
to the geographic sampling of the instrument.
The results suggest that although the
solar occultation instruments have limited geographic sampling they do observe a
wide range of equivalent latitudes during the winter and therefore are able to
adequately sample the vortex for O3 loss studies.
In this study we calculated the modeled loss from the SLIMCAT CTM and
compared the results to IL calculations to determine how well the CTM was able to
simulate O3 loss during the 2004-2005 Arctic winter. The SLIMCAT CTM modeled
loss had a morphology similar to that of the inferred loss calculations. In the past,
CTMs have underestimated O3 loss compared to loss inferred from observations
[Chipperfield et al., 1996; Goutail et al., 1997; Deniel et al., 1998; Becker et al.,
2000; Guirlet et al., 2000]. Unlike earlier versions of the SLIMCAT CTM, this
version no longer underestimates the loss. On the contrary, the CTM slightly
overestimates the loss by approximately 0.3 ppmv (~10-15%) after 1 March between
450 and 500 K, but overall the modeled loss agrees very well with the inferred loss
calculations. Future work will involve comparing other retrieved satellite species to
the model in order to fully test the model’s ability to simulate the chemistry and
dynamics of the 2004-2005 Arctic winter. This study and the study of Singleton et al.
[2005] have shown that the SLIMCAT CTM was able to reproduce the maximum
loss inferred from satellite observations for both the 2002-2003 and 2004-2005 Arctic
winters using the CTM Passive Subtraction technique to within 15 %. Although
some discrepancies between the observed and modeled O3 remain (such as O3 loss
above 550 K), the level of agreement has improved from previous studies. Thus,
121
these results help to increase confidence in our understanding of Arctic O3 loss
processes.
Acknowledgments:
This research is supported by the NASA Earth System Science Fellowship Program,
the NASA SOSST Program, and the Aura/HIRDLS Program. Work at the Jet
Propulsion Laboratory, California Institute of Technology was done under contract
with the National Aeronautics and Space Administration. The ACE mission is
supported by funding from the Canadian Space Agency, the Natural Sciences and
Engineering Research Council of Canada and the Canadian Foundation for Climate
and Atmospheric Sciences. The SLIMCAT modeling work was supported by the EU
SCOUT-O3 Project. We thank two anonymous reviewers for helpful comments on
this manuscript. CSS thanks Patricia Weis-Taylor, Ph.D. for helpful discussions.
122
Chapter 8
Arctic Ozone Loss Climatology from Solar Occultation and Microwave Limb
Sounding Instruments
8.1 Introduction
Unlike the Antarctic, which experiences little interannual variability in
meteorological conditions, the conditions in the Arctic vary from year to year.
Because polar O3 loss is dependent on temperature, the amount of O3 loss in the
Arctic can also vary [e.g., Manney et al., 2003a; Rex et al., 2004; Tilmes et al., 2004].
In order to gain a proper understanding about Arctic O3 loss, it is necessary to
examine many years of data to study winters with different meteorological conditions.
To explore the interannual variability in Arctic O3 loss, we have inferred O3 loss
from satellite observations from the 1994-1995 through the 2004-2005 Arctic winters.
Data from both solar occultation and microwave limb sounding instruments have
been applied in this study. These instruments include the Upper Atmosphere
Research Satellite Microwave Limb Sounder (UARS MLS), Earth Observing System
Microwave Limb Sounder (EOS MLS), Polar Ozone and Aerosol Measurement
(POAM II/III), Stratospheric Aerosol and Gas Experiment (SAGE II/III), Improved
Limb Atmospheric Spectrometer (ILAS), Halogen Occultation Experiment
(HALOE), Atmospheric Chemistry Experiment Fourier Transform Spectrometer
123
(ACE-FTS), and Measurement of Aerosol Extinction in the Stratosphere and
Troposphere Retrieved by Occultation (MAESTRO). Since the majority of the
instruments did not consistently sample the Arctic vortex throughout the duration of
the winter, we have combined the data sets to make one merged O3 field for each
winter. In order to infer O3 loss from the combined O3 field, the Chemical Transport
Model Passive Subtraction (CTM-PS) [e.g., Deniel et al., Singleton et al., 2006;
2005] technique was applied. The CTM-PS technique has been adapted from the
well-validated Passive Subtraction approach [e.g., Harris et al., 2002; WMO, 2003;
Manney et al., 1995a, b; 2003, Singleton et al., 2005, 2006].
Because the meteorological conditions and the amount of O3 loss in the Arctic
can vary from year to year, Arctic O3 processes are much more difficult to simulate
than Antarctic processes. In order to adequately validate the ability of a model to
simulate Arctic O3 loss processes, the modeled O3 loss must be tested during
numerous winters to span the range of meteorological conditions. We present
modeled Arctic O3 loss results from the University of Leeds’ SLIMCAT CTM. The
results are then compared to the inferred O3 loss calculations to determine how well
the SLIMCAT CTM was able to simulate Arctic O3 loss during various winters. In
addition, the simulated O3 is compared to the O3 observations. Although previous
studies have indicated that CTMs underestimate Arctic chemical loss [Chipperfield et
al., 1996; Goutail et al., 1997; Deniel et al., 1998; Becker et al., 2000; Guirlet et al.,
2000], adjustments have been made to the SLIMCAT CTM to improve polar O3 loss
processes in the model [e.g., Feng et al., 2005; Chipperfield, 2006]. We show below
that these adjustments lead to much better agreement between model and
124
observations than previously found. Since developing chemistry modules in CTMs is
integral to developing chemistry modules in Chemistry Climate Models, it is vital that
CTM accurately simulate O3 loss during winters with varying meteorological
conditions.
8.2 Data Sets
In this section the ten satellite data sets that were analyzed are described.
8.2.1
POAM II/III
The Polar Ozone and Aerosol Measurement (POAM) instruments, POAM II
[Glaccum et al., 1996] and POAM III [Lucke et al., 1999], were nine-channel
photometers. The instruments were designed to study chemistry in the polar regions
and measure vertical profiles of O3, NO2, H2O, and aerosol extinction. The O3
observations have a vertical resolution of approximately 1 km in the stratosphere
[e.g., Bevilacqua et al., 1997; Randall et al., 2003]. POAM II was launched on the
Satellite Pour l’Observation de la Terre (SPOT) 3 satellite in September 1993 and
successfully made observations until November 1996 when the satellite failed [Lucke
et al., 1999]. SPOT 3 was launched into a sun-synchronous near-polar orbit (98.7°
inclination) at an altitude of 833 km, which allowed for approximately 14-15
observations around a circle of latitude in each hemisphere each day [Randall et al.,
1995]. The measurement latitude varied slowly between 54° N and 71° N in the
125
Northern Hemisphere, and during the Arctic winter the instruments sampled both
inside and outside the polar vortex. POAM II wavelength channels range from .352
to 1.06 µm. For this analysis POAM II version 6.0 O3 data are used. Rusch et al.
[1997] showed that an earlier version of the POAM II data agreed with correlative
measurements to within 5-7%, with a low bias below 22 km that reached 20% at 15
km; version 6.0 comparisons are similar.
POAM III was launched in March 1998 on the SPOT 4 satellite; the instrument
ceased operations in December of 2005 due to an instrument anomaly. POAM III
was launched in the same orbit as POAM II in order to continue polar observations.
As a result, POAM III sampled the same number of observations each day and over
the same latitude range as POAM II [Lucke et al., 1999]. POAM III channels were
slightly different than POAM II and range between 0.353 to 1.02 µm. For this
analysis POAM III version 4.0 data are used, which varies little from version 3.0 data.
Randall et al. [2003] show that version 3.0 POAM III O3 measurements agree to
within ±5% with correlative ozonesonde and satellite data between 13 and 60 km.
8.2.2 SAGE II/III
The Stratospheric Aerosol and Gas Experiment (SAGE) II first started taking
atmospheric observations in October 1984, and ceased operating in August 2005.
SAGE II was launched in a 57° mid-inclination orbit aboard the Earth Radiation
Budget Satellite (ERBS). Because of its orbit, SAGE II alternated from
approximately 60° in the winter hemisphere to 80° in the summer hemisphere in one
126
month, so coverage is temporally sparse in the polar regions, particularly during
winter [Wang et al., 2002]. SAGE II utilized solar occultation to measure vertical
profiles of O3, NO2, H2O, and aerosol extinction. SAGE II channels ranged between
0.385 and 1.02 µm. Here we use version 6.2 O3 data, which varies on the order of
0.5% from version 6.1. Version 6.1 O3 data agrees within 10% with ozonesondes
down to the tropopause, as shown by Wang et al. [2002] and has a vertical resolution
of approximately 1 km.
SAGE III was launched in December 2001 on the Meteor 3M spacecraft.
SAGE III began taking measurements in February 2002 and operations were
terminated in March 2006. SAGE III also utilized solar occultation and measured the
same species as SAGE II [Chu et al., 2002; Thomason and Taha, 2003; Wang et al.,
2006]. Unlike SAGE II, SAGE III was launched into a sun-synchronous polar orbit
and its Northern Hemisphere observations range between 50° and 80° N. The
increased northern latitude coverage by SAGE III allows for more detailed Northern
Hemisphere polar studies. The spectral coverage for SAGE III ranged between 0.290
and 1.03 µm. SAGE III sampled O3 with ~ 0.5 km vertical resolution. SAGE III
version 3.0 data are applied in this work; these data agree to within 5% with
correlative measurements down to 17 km [Wang et al., 2006].
8.2.3 ILAS
The Improved Limb Atmospheric Spectrometer (ILAS) began taking
observations in November 1996 and ceased observations in June 1997 [Sasano et al.,
127
1999a]. ILAS consists of an IR spectrometer that operated between 6.21 and 11.76
µm to measure vertical profiles of O3, HNO3, NO2, N2O, CH4, and H2O. The vertical
resolution of each retrieved species was approximately 1.6 km [Nakajima et al.,
2002]. ILAS was launched in a sun synchronous orbit, with a 98.6 ° inclination and
made measurements between 57°N and 72°N in the Northern Hemisphere [Nakajima
et al., 2002]. The data applied in this analysis was ILAS version 6.10.
8.2.4 HALOE
The Halogen Occultation experiment (HALOE) instrument started taking
observations in October 1991 and ceased observations in November 2005. HALOE
was launched in a mid-inclination orbit (57°) onboard the Upper Atmosphere
Research Satellite (UARS) [Reber, 1993]. HALOE used solar occultation to measure
absorption bands of many species in the infrared between 2.45 and 10.04 µm [Russell
et al., 1993]. Some of the species included O3, CH4, H2O, NO, NO2, and aerosol
extinction. In approximately one month, HALOE covered latitudes from about 60°
latitude in the winter hemisphere to about 80° latitude in the summer hemisphere.
The O3 vertical resolution is approximately 2 km [e.g., Randall et al., 2003]. HALOE
version 19 data was used in this analysis. This version has been compared with an
earlier version of SAGE II data (6.0) by Morris et al. [2002]. Morris et al. [2002]
indicates the differences between the datasets varied from 4–12% throughout most of
the stratosphere.
8.2.5 ACE-FTS and MAESTRO
128
The Atmospheric Chemistry Experiment Fourier Transform Spectrometer
(ACE-FTS) and the Measurement of Aerosol Extinction in the Stratosphere and
Troposphere Retrieved by Occultation (MAESTRO) instruments were launched
onboard the ACE satellite in August 2003 and are currently operational. Both
instruments utilize solar occultation to measure polar relevant species. The ACE
spacecraft was launched in a high inclination orbit (74°), thus samples from the polar
to low latitudes. ACE-FTS operates in the 2 to 13 micron spectral region and
measures many constituents at a vertical resolution of approximately 4 km in the
lower stratosphere [Bernath et al., 2005]. The species include O3, H2O, CH4, NO,
NO2, HNO3, HCl, N2O5, and ClONO2 [Walker et al., 2005]. ACE-FTS version 2.2
O3 update was used for this analysis, which has improved agreement with correlative
observations near the profile peak compared to versions 1.0 and 2.2. MAESTRO
makes observations in the 400 to 1030 nm spectral region. MAESTRO measures
vertical profiles of O3, NO2, and aerosol extinction with a vertical resolution of
approximately 1 km. We use version 1.2 MAESTRO data for this analysis. Kar et
al. [Initial comparison of ozone and NO2 profiles from ACE-MAESTRO with balloon
and satellite data, submitted to J. Geophys. Res., 2006] show that v1.2 MAESTRO O3
profiles agree with correlative measurements from ACE-FTS, SAGE III and POAM
III to within about 15% in the lower stratosphere.
8.2.6 UARS/EOS MLS
129
Data was analyzed from two microwave limb sounding (MLS) instruments in
this study, the Upper Atmosphere Research Satellite Microwave Limb Sounder
(UARS MLS) [Barath et al., 1993] and the Earth Observing System Microwave Limb
Sounder (EOS MLS) [Waters et al., 2006]. Like HALOE, UARS MLS was launched
on the UARS satellite. The mid-inclination orbit allowed the instrument to take
measurements from approximately 34° on one side of the equator to 80° on the other.
Approximately every 36 days UARS performed a yaw maneuver and switched
viewing orientation by 180° degrees. Because of the yaw cycle, UARS MLS
coverage of the high northern latitudes was limited during the winter. UARS MLS
measured emission spectra near 63, 205, and 183 GHz using three radiometers
[Livesey et al., 2003]. Some of the relevant polar species include ClO, O3, and H2O.
UARS MLS started taking observations in September 1991 and continued operating
at a full level of operation until December 1993. After this time, measurement
coverage started to become more intermittent, until the instrument was
decommissioned in December 2005. UARS MLS data version 5, which has been
validated by Livesey et al. [2003], was applied in this analysis. Livesey et al. [2003]
state the agreement with SAGE II O3 is typically within 5%; however, larger
differences (~30%) are observed at low latitudes (30°S- 30°N). O3 has a vertical
resolution of approximately 3.5 – 4 km in the lower stratosphere [Manney et al.,
2003]
EOS MLS was launched in July 2004 onboard the Aura satellite. Aura is in a
near-polar, sun-synchronous orbit; thus every orbit EOS MLS observations span from
82° S to 82° N [Waters et al., 2006]. As a result of the orbit, EOS MLS samples the
130
northern polar latitudes more continuously throughout the winter than UARS MLS.
EOS MLS has 5 radiometers which operate in the following spectral regions: 118
GHz, 190 GHz, 240 GHz, 640 GHz, and 2.5 THz [Waters et al., 2006]. The main
species measured by EOS MLS which are relevant to polar studies include H2O,
HNO3, O3, HCl, ClO, and N2O. In the lower stratosphere the vertical resolution for
O3 is approximately 2.7 km [Froidevaux et al., 2006]. The MLS data applied in this
study is version 1.51, which has been validated by Froidevaux et al. [2006].
Froidevaux et al. [2006] found that the overall agreement between EOS MLS O3 and
stratospheric profiles from SAGEII, HALOE, POAMIII, and ACE-FTS was
approximately 5% to 10%.
8.3 Methods
The CTM-Passive Subtraction (CTM-PS) technique was applied to infer O3
loss from both solar occultation and limb sounding observations. The CTM-PS
technique was developed from the passive subtraction technique, which was first
described by Manney et al. [1995a, b], and has since been applied in many O3 loss
studies [e.g., Deniel et al., 1998; Singleton et al., 2005; 2006]. In order to infer O3
loss calculations from observations, the CTM-PS technique requires that a passive O3
field (i.e., a field with no chemical variation in O3) be generated by a CTM. The
passive O3 field is then interpolated to the measurement locations and the difference
between the observations and the passive field is the inferred O3 loss (IL). The
passive O3 field can also be subtracted from the “active” model O3 field (referred to
131
here as the modeled or simulated O3), in which both the chemistry and dynamics are
activated, to quantify the modeled O3 loss (ML). The ML was computed at the
observation locations in order to be directly compared with the IL calculations.
The University of Leeds’ SLIMCAT CTM was used for this analysis.
SLIMCAT is a 3-D offline model with detailed stratospheric chemistry, which
includes heterogeneous chemistry on solid and liquid aerosols and a NAT-based
denitrification scheme (see Davies et al. [2002]). The model is described in detail in
Chipperfield [1999] and recent adjustments to the model are discussed in Feng et al.
[2005; 2006]. The vertical domain of the model extends from the surface to
approximately 55 km. The vertical grid in SLIMCAT is a hybrid sigma-theta grid,
which is described in Chipperfield [2006], and has isentropic coordinates in the
stratosphere. The vertical resolution in the stratosphere is comparable to remote
sensing observations and is approximately 2 km. For the runs used in this study,
SLIMCAT was forced with daily temperatures and horizontal winds from the
European Centre for Medium-Range Weather Forecasts (ECMWF) analyses. Runs
prior to 1 January 2000 were forced with the reanalyses (ERA-40), and runs after this
date were forced with the operational analyses. The ERA-40 temperatures have a
slight bias compared to those in other meteorological analyses; nevertheless, studies
have shown that ERA-40 correctly handles interannual variability [e.g., Manney et
al., 2005; Tilmes et al., 2006]. The vertical transport in the stratosphere is determined
based on calculations from the NCAR CCM radiation scheme [Briegleb, 1992].
Prior to the winter simulations, a low resolution (7.5 x 7.5°) run was started on
1 January 1977. The results of this run were then used to initialize the high resolution
132
(2.8 x 2.8°) model runs on 1 December of each year from 1994 to 2004. The only
exceptions were for the 2002-2003 and 1999-2000 runs, where the modeled O3 was
initialized from O3 fields constructed from Northern Hemisphere observations from
POAM III and HALOE using PV-mapping, as described in Randall et al. [2002,
2005] (see Singleton et al. [2005]). The impact of the model O3 initialization will be
discussed in more detail in Section 5. The model runs presented here have the same
setup as those described in Feng et al. [2005]. The tropospheric source gases in
SLIMCAT are based on WMO [2003] values. The only additional difference between
the model runs analyzed here is the bromine loading for the 2004-2005 run. The
2004-2005 simulation includes an extra 6 pptv of lower stratospheric Bry from shortlived species, which is based on the findings of Salawitch et al. [2005]. This change
is not expected to have an impact on the short-term winter simulations, but would
have an impact on the mid-latitude O3 trend for a long-term run (W. Feng, private
communication).
Throughout the 1994-1995 through the 2004-2005 Arctic winters (defined
here as the time period from 1 December – 1 April), there exists a wealth of
stratospheric observations from solar occultation and microwave limb sounding
instruments. Although there are many datasets available during this time period, not
all of the instruments made observations throughout the duration of each winter. To
illustrate this point, Figure 1 shows the equivalent latitude (a vortex-centered
coordinate system, where 90° is always in the center of the vortex) of all vortex
observations [Butchart and Remsberg, 1986]. Since ECMWF potential vorticity
(PV) fields were not available throughout the entire time period of this study, the
133
equivalent latitude was calculated from Met Office (MetO) PV. The equivalent
latitude was calculated at the MetO model grid and then was interpolated to the
observation locations. The only year that will not be discussed in this paper is the
1997-1998 winter, because there were so few vortex observations made during this
winter (as indicated by Figure 1). There were only slightly more vortex observations
during the 2000-2001 winter; therefore, the O3 loss results computed for this winter
will only be discussed qualitatively.
Figure 8.1: The equivalent latitude for all vortex observations made during the Arctic
winters of 1994-1995 through 2004-2005 on the 475 K potential temperature surface.
The values of equivalent latitude are color coded by instrument.
In previous studies by Singleton et al. [2005; 2006], inferred loss (IL) and
modeled loss (ML) calculations were computed separately at each instrument location
during the Arctic winters of 2002-2003 and 2004-2005. In order to utilize all of the
datasets in the current study, the O3 observations were combined to make one O3 field
for each winter. To do so, all the instruments were normalized to eliminate offsets in
the O3 field due individual instrument biases. Because SAGE II was one of the only
134
instruments that made observations throughout the 1994-1995 through 2004-2005
time period, all instruments were corrected to SAGE II using a normalization factor.
For all instruments except UARS and EOS MLS, comparisons between coincident
measurements that were within 2 hours and 500 km were used to determine the
normalization factor. In order to find an adequate number of coincidences with both
MLS instruments the time criterion was extended to 12 hours. The average ratio of
SAGE II to the coincident profiles was used to compute the normalization factor.
This factor was then applied to each observation. Figure 2 shows the normalization
profiles that were applied to each instrument. The normalization profiles indicate that
the differences between the instruments are well within 20%. The only exception is
for UARS MLS, which has differences that are greater than 50% below 425 K, which
is below the altitude levels validated by Livesey et al. [2003]. Therefore, UARS MLS
data below 425 K has not been included in the analyses. Once the satellite datasets
Figure 8.2: Comparison of the normalization profiles used to correct each instrument.
The normalization profiles are the average ratio of the instrument O3 profile to the
coincident SAGE2 O3 profile. Error bars represent 1 sigma standard deviations of the
averages.
135
were normalized, all vortex observations were combined for each winter. The
SLIMCAT (passive and modeled) O3 was then sampled at each observation location
and combined. In order to determine the vortex edge, the vortex criteria of Harvey et
al. [2002] were applied. IL and ML calculations were computed between 400 and
700 K for the vortex observations. The calculations were not extended below 400 K
due to uncertainties in identifying the vortex edge (see Singleton et al., 2006 for
additional details).
8.4 Meteorology
Since the meteorological conditions in the Arctic winter are so variable, the
number of polar stratospheric clouds (PSCs) that form during a winter can also vary.
During cold winters, Arctic temperatures typically drop below PSC formation
temperatures and more O3 loss will occur. PSCs typically form at temperatures below
which nitric acid trihydrate can condense (TNAT). The area in the Northern
Hemisphere where winter temperatures fell below TNAT are shown as contour plots in
Figure 3 for all winters from 1994-1995 through 2004-2005. TNAT values were
computed using the expression from Hanson and Mauersberger [1988] using HNO3
and H2O from SLIMCAT and MetO temperature analyses. Figure 3 indicates the
1995-1996, 1999-2000, and the 2004-2005 winters had the largest areas with T<
TNAT. The 1994-1995, 2000-2001, and 2002-2003 winters had large areas with
temperatures below TNAT early in the winter; however, temperatures increased by late
136
January. Unlike the other winters which had temperatures less than TNAT in
December, the 1996-1997 winter stayed warm until January. However, the 19961997 winter was unique because it had temperatures that remained below TNAT later
into the winter (approximately the end of March) than the other years. The warmest
winter was the 1998-1999 winter, which only had two brief periods (in early
December and February) where temperatures fell below TNAT.
Figure 8.3: Area (106 km2) where Northern Hemisphere MetO temperatures fell
below TNAT during the winters from 1994-1995 to 2004-2005 between the 300 K and
700 K potential temperature surfaces. The black line indicates the lowest potential
temperature surface (400 K) included in the O3 loss analyses.
Another important diagnostic of the meteorological conditions is the strength
of the polar vortex. Unlike the Antarctic vortex, which is almost always cold and
stable, the Arctic vortex is continuously disturbed by planetary waves that propagate
up from the troposphere. When the Arctic vortex is disturbed, mixing is more likely
to occur across the vortex edge. Mixing across the vortex edge complicates O3 loss
137
calculations. In order to assess the strength of the Arctic vortex and the timing of
potential mixing events during the years included in the climatology, the following
diagnostic for vortex strength is examined. Here we use the average wind speed at
the vortex edge multiplied by a normalized PV gradient (which is quantified by
dividing the PV gradient field by the maximum value, resulting in a unitless index
ranging from 0 to 1). Both fields are in equivalent latitude coordinates. This vortex
strength diagnostic is shown in Figure 4 for the years 1994-1995 through 2004-2005
as a function of altitude and time. If the maximum PV gradient is exactly collocated
Figure 8.4: The vortex strength index (m/s) for the Arctic winters from 1994-1995 to
2004-2005. A higher vortex strength index indicates the vortex is more stable and air
is more likely to be contained within the vortex.
with the polar jet, then this diagnostic is reduced to the average wind speed at the
vortex edge. If the maximum PV gradient is not collocated with the jet then the
138
magnitude of this diagnostic decreases. A high vortex strength index indicates that
the vortex is more stable and that air is more likely to be contained within the vortex
and not mix with extra-vortex air.
Figure 4 indicates that there was a large amount of variability in the strength
of the vortex during the Arctic winters. The winters that are characterized by a strong
vortex from December until March are the 1994-1995, 1995-1996, and 1999-2000
winters [Rex et al., 1997; Becker et al., 2000]. Each of these winters also experienced
extended periods where temperatures were below TNAT, and as a result these winters
had the highest potential for significant O3 loss. The 2004-2005 winter had a high
vortex strength index, although it was not as high as the 1994-1995, 1995-1996, and
1999-2000 winters. This is because the 2004-2005 vortex was dynamically active
during the winter, which led to mixing of extravortex air into the vortex [Manney et
al., 2006; Schoeberl et al., 2006]. Even though the 2004-2005 winter experienced
very low temperatures (similar to 1999-2000), mixing would have likely limited the
amount of O3 loss during the winter. Also, the winter experienced a major final
warming early in March (which drastically decreased the vortex strength index),
which would have limited the amount of loss for the remainder of the winter. The
1996-1997 winter was disturbed until mid January after which time the vortex
strengthened and persisted until late April [Müller et al., 1997]. At the same time the
vortex strengthened, temperatures fell below TNAT; therefore, O3 loss would likely
occur later during this winter. The 1998-1999 and the 2001-2002 winters both started
off with a weak vortex due to early major warmings in December [Naujokat et al.
2002], and as a result both winters (in particular 1998-1999) have a low vortex
139
strength index. In addition to having a weak vortex, the 1998-1999 winter also
experienced very warm temperatures making the probability for significant O3 loss
during the winter very low. Unlike the 1998-1999 winter, 2001-2002 had a large area
where temperatures were below TNAT throughout December until the major warming
occurred at the end of the month. Therefore, it is likely that of these two winters,
2001-2002 would experience more loss. Both the 2002-2003 and the 2003-2004
winters experienced major warmings in January [Manney et al., 2005]. The 20022003 winter was very cold early and had extensive areas with temperature below
TNAT; however, the major warming in late January followed by two nearly major
warmings (in mid-February and early March) limited O3 loss potential for the
remainder of the winter [Manney et al., 2005]. After the January warming the 20032004 vortex remained weak and warm in the lower stratosphere, reducing the
potential for large O3 losses.
8.5 Inferred Loss
The vortex daily average IL for the ten Arctic winters analyzed here is shown
as a contour plot in Figure 5. As mentioned above, the IL is the difference between
the observed O3 field and the passive O3 interpolated to the observation locations.
For qualitative purposes, all contour plots in this paper have been filled in by a time
interpolation on days when a vortex observation was not made. In addition, data in
all of the contour plots have been smoothed using a seven-day running average.
140
Singleton et al. [2005; 2006] discuss the importance of model initialization
when applying the CTM-PS technique to calculate O3 loss. In order to correctly infer
O3 loss from the observations, it is necessary that the passive O3 is equal to the
observations on the first day of the analysis (1 December). If there are any
differences between the passive O3 and the observations at the start of the analysis,
the differences will be propagated in the passive O3. Consequently, these differences
Figure 8.5: Differences (ppmv) between passive O3 calculated by the SLIMCAT
CTM and the combined satellite O3 fields. Results correspond to daily averages over
the measurement locations inside the vortex during the ten Arctic winters between
400 K and 700 K. Days with missing data and days where an instrument did not
sample the vortex have been filled in with a linear time interpolation. The solid black
line denotes the zero contour. Data have been smoothed with a 7-day running
average. White spaces in the contour plots at the end of the winter (e.g. 1999-2000)
indicate that vortex observations were no longer made for the remainder of the
analysis period.
141
can affect the IL calculations at a later date in the analysis. For example, a negative
offset on the first day of the analysis would falsely indicate that chemical loss has
occurred [Singleton et al., 2005; 2006]. In this analysis the model passive O3 was not
initialized with the combined O3 fields, thus there is an offset on 1 December in some
of the years. In order to correct for this offset, the model passive O3 field has been
corrected to account for any initialization differences. This same correction profile
has been applied to the simulated O3, since the passive and the simulated O3 are equal
on 1 December.
As previously mentioned, since there is large interannual variability in the
meteorological conditions in the Arctic, the amount of O3 loss can also change from
year to year. This interannual variability is evident in Figure 5. It is important to
note that the smoothness of the contours does vary based upon the number of
observations that were included in the average. The number of vortex observations
during a given winter is shown in Figure 1. For example, during the 2004-2005
winter, observations from 7 satellite instruments were included in the average,
compared to just 3 during the 2001-2002 winter. There were very few observations
taken during the 2000-2001 winter; therefore, conclusive statements will not be made
about this winter. To examine the IL in a more quantitative manner, a time series of
the daily average IL is shown in Figure 6 for the 600, 500, 475, and 450 K surfaces.
Days when observations were not made inside the vortex have not been interpolated
in the time series figures.
The CTM-PS technique indicates that the largest amount of loss occurred
during the 1994-1995, 1995-1996, and 1999-2000 winters. As mentioned above,
142
these winters experienced extended periods of very low temperatures (T<TNAT) and
had strong vortexes. As a result, these winters had the highest potential for
Figure 8.6: Time series of the inferred daily average O3 loss (ppmv) inside the vortex
from the combined satellite O3 fields for the 600 K, 500 K, 475 K, and 450 K surfaces
for the ten Arctic winters.
significant O3 loss of the ten winters analyzed here. Within the altitude range
considered here (400 and 700 K), the maximum loss for these cold winters occurred
between 450 and 500 K and ranged from approximately 2.4-2.8 ppmv. 2004-2005
also experienced a significant amount of loss and had a peak loss between 450 and
500 K. Even though the winter was extremely cold, the loss during 2004-2005 was
not as large as the losses that occurred during 1994-1995, 1995-1996, and 1999-2000.
143
We think that this is due to the dynamically active 2004-2005 vortex. The 20012002 and the 2003-2004 winters had a substantial amount of loss, although less than
the 1994-1995, 1995-1996, and 1999-2000 winters. Both winters had similar
morphology, and experienced more loss above 550 K than the other winters. The
peak O3 loss for 2001-2002 and the 2003-2004 occurred between approximately 575
and 600 K. Both of these winters had areas where temperatures were below TNAT at
the beginning of the winter at the higher potential temperature levels, but not later in
the winter when the lower altitudes would have experienced the lower temperatures.
In other words, PSCs probably formed efficiently only in the upper altitudes, which
led to loss in those regions, but not below. The maximum loss for the 2001-2002 and
the 2003-2004 winters was approximately 1.7- 2 ppmv. The 2002-2003 and the
1996-1997 winters had a moderate amount of loss. The 2002-2003 winter had a
maximum loss near 450 K, while the peak loss in the 1996-1997 winter spanned a
much broader altitude region between 450 K and 550 K. Both winters had a
maximum loss of approximately 1.5 ppmv. Of all the winters examined here, the
1998-1999 winter experienced the least amount of loss. This winter was much
warmer than the other winters and had a very unstable vortex. A maximum loss of
approximately 0.9 ppmv occurred near 550 K. The 2000-2001 winter also
experienced very minimal loss; however, because so few observations were made
during this winter conclusive statements cannot be made about this winter.
8.6 Comparison with Other Techniques
144
In this section CTM-PS IL results during the winters with the largest O3 loss
(1994-1995, 1995-1996, and 1999-2000) are compared to other inferred calculations
which have been computed using many different O3 loss techniques. Each study was
run for different time periods during the winter and with different datasets; therefore,
some differences are expected between the calculations. The CTM-PS IL results for
the 1994-1995 winter are comparable to the results of Rex et al. [1999]. Rex et al.
[1999] applied the Match technique using ozonesonde observations to calculate an
accumulative loss at 450 K of 2.0 ppmv. These loss results are slightly lower than the
2.4 - 2.7 ppmv loss computed by CTM-PS in late March at 450 K. However, the
differences can be attributed to the time duration of the analyses. Rex et al. [1999]
computed the accumulative O3 loss from 1 January through 20 March and the CTMPS analysis was started on 1 December. By 1 January, some O3 loss had already
occurred in the CTM-PS results.
For the 1995-1996 winter, the CTM-PS technique indicates that a maximum loss
of approximately 2.1 to 2.4 ppmv occurred at 450 K. These results are comparable to
the results of Rex et al. [1997] and Müller et al. [1997]. Rex et al. [1997] applied the
Match technique to ozonesonde observations and found a maximum cumulative loss
of approximately 2.4 ppmv between 20 January and 9 April. These results were
calculated for a different time period than the CTM-PS analysis; therefore, the
differences are expected. Müller et al. [1997] inferred O3 loss from HALOE
observations using the O3-tracer technique. The results indicate that the peak loss
during the 1995-1996 winter occurred near 450 K and was approximately 3.3 ppmv.
Differences between CTM-PS and Müller et al. [1997] are likely due to sampling of
145
the instruments. The combined satellite O3 fields (and ozonesonde data) are more
representative of vortex-average conditions compared to using data from just one
satellite instrument.
For the 1999-2000 winter, the CTM-PS IL results indicate a maximum loss
occurred at 450 K and was between 2.4 and 2.8 ppmv. These results are larger than
the maximum cumulative loss calculated by Hoppel et al. [2002] using the vortex
average approach with POAM III data. Hoppel et al. [2002] found the maximum loss
occurred near 475 K and was approximately 1.5-2 ppmv. The accumulative loss
calculations where computed for the period of 1 December to 15 March. The
comparisons for the 1994-1995, 1995-1996, and 1999-2000 winters indicate, overall
the IL results computed from the CTM-PS technique are comparable to other
inferences of O3 loss from different techniques.
8.7 Modeled Loss (ML)
The daily average vortex ML for the ten Arctic winters is shown as contour
plots in Figure 7. The model has been sampled at the different instrument locations
and then combined as was done with the IL calculations. Figure 7 shows that
SLIMCAT was able to reproduce the general pattern of the IL calculations shown in
Figure 5 in most winters. The vertical region of maximum loss in SLIMCAT is
generally the same as in the observations. In every winter except 2004-2005,
however, the model shows less loss than is inferred from the observations. During
2004-2005 SLIMCAT slightly overestimates the maximum loss near 450 K. This is
146
the same result that was found in Singleton et al. [2006] where the loss was inferred
separately from the POAM III, SAGE III, EOS MLS, MAESTRO, and ACE-FTS
instruments.
Figure 8.7: As in Figure 5, but for modeled daily average O3 loss (ppmv) inside the
vortex at the combined satellite locations during the ten Arctic winters.
The time series of the ML is shown in Figure 8. At all theta levels there is
much less variability in the ML compared to the IL results, but this is particularly true
at 600 K. At this level, SLIMCAT overestimates the amount of loss during the 19981999 and 1999-2000 winters, but underestimates it in all other years. In fact, the
model was unable to capture the maximum loss for the 2001-2002 and 2003-2004
winters at 600 K. For both of these winters, SLIMCAT underestimated the loss by
approximately 1 ppmv. At the 500, 475, and 450 K surfaces SLIMCAT has the
largest difficulty simulating loss during the cold 1994-1995 and the 1995-1996
147
winters starting in late January. During the 1994-1995 winter, SLIMCAT
underestimates the maximum loss at each of these levels by up to approximately 1
ppmv. However, the discrepancies are not as large for the 1995-1996 winter. The
Figure 8.8: As in Figure 6, but for modeled daily average O3 loss (ppmv) inside the
vortex at the combined satellite locations during the ten Arctic winters.
largest difference in 1995-1996 occurs on the 500 and 475 K levels, where the model
underestimates the maximum loss by approximately 0.4 ppmv. Unlike the 1994-1995
and the 1995-1996 winters, SLIMCAT slightly overestimates the loss during 20042005 on the 475 K and 450 K surfaces. At these levels SLIMCAT indicates that the
2004-2005 winter had the largest amount of loss and overestimates the loss compared
to the IL by approximately 0.3-0.4 ppmv. Although there were some discrepancies
between the IL and ML results during the 1994-1995, 1995-1996, and 2004-2005
148
winters, SLIMCAT did not have trouble simulating the maximum loss for the cold
1999-2000 winter below 500 K
The extent of agreement between the maximum IL and ML results at the end
of the analysis time period is shown for each year in Figure 9. The left panel is a time
series of the average maximum IL (black) and ML (red) over the last 14 days of the
analysis for the ten Arctic winters. The right panel of Figure 9 shows the altitude at
which the maximum loss occurred. The figure indicates that although there are some
differences in the amount and the location of the maximum loss, overall SLIMCAT is
able to capture the interannual variability experienced in the Arctic. Figure 9
suggests that the model had less difficulty determining the altitude of the maximum
loss than the amount of loss during a given Arctic winter. As mentioned above, the
maximum loss in the observations occurred during the 1994-1995, 1995-1996, and
1999-2000 winters. These were the winters with low temperatures (T<TNAT) and
prolonged periods of high vortex strength index. For these years the average loss was
approximately 2.4 ppmv. The model indicates that the maximum loss occurred
Figure 8.9: Maximum IL (black) and ML (red) O3 calculations (left panel) and their
corresponding altitude (right panel) for the ten Arctic winters.
149
during the 2004-2005 winter, the year with maximum PSC formation probability, at
approximately 2.2 ppmv; however, the 1995-1996 and 1999-2000 winters also had a
large amount of loss around 2 ppmv.
Figure 10 shows the average loss for the last 14 days of the analysis on the
450 K and 600 K potential temperature surfaces. We have chosen these levels
because maximum loss tends to occur at the lower altitudes near 450 K in cold
years, whereas maximum loss tends to occur at the higher altitudes during warm
years. In general, the model does a good job simulating the loss at both levels, but
not quite as well at 600 K as at 450 K. The largest discrepancy at 450 K occurs
during the cold 1994-1995 winter, where the model underestimates the loss by
approximately 1 ppmv. In all other winters, both warm and cold, the agreement at
450 K between the ML and IL is excellent, although there is a clear tendency for the
Figure 8.10: The average O3 loss for the last 14 days of the analysis for the
observations (black) and the model (red) on the 450 K (left) and 600 K (right)
potential temperature surface.
150
model to underestimate the loss compared to the observations, as discussed above.
The level of agreement at 600 K is not as satisfying, with the model underestimating
the loss compared to the IL in both 2001-2002 and 2003-2004 by 0.8-1.0 ppmv. It is
likely that the IL in these two warm winters resulted from heterogeneous chemistry
that occurred early in the winter when the TNAT area was relatively small but centered
at the higher potential temperatures. This could imply that heterogeneous processing
at the higher altitudes is not handled correctly by the model.
8.8 Observed and Simulated O3
The differences between the IL and the ML are due to differences between the
observed and the simulated O3. These differences may be caused by an error in the
chemistry and/or dynamics of the model. The daily average O3 for the ten Arctic
winters is shown as contour plots for the observations and the modeled O3 in Figures
11 and 12, respectively. Overall, the similarities between the model and observations
are remarkable. This is true for the large scale features such as the relatively larger
O3 mixing ratios in 1998-1999 compared to other years and the temporal variation in
low O3 mixing ratios each year below 500 K. At 600 K and below dynamical effects
along with chemistry that occurs after chlorine activation on polar stratospheric
clouds can alter vortex O3 values. At 600 K the largest differences between the
observations and the modeled O3 took place during the 2001-2002 and 2003-2004
winters, when SLIMCAT overestimates the amount of O3. As a result the model
151
Figure 8.11: Daily averages of observed O3 at the combined measurement locations
inside the vortex during the ten Arctic winters. Days with missing data and days
where an instrument did not sample the vortex have been filled in with a time
interpolation. Data have been smoothed with a 7-day running average.
underestimates the maximum loss observed during these winters (as shown in the
right panel of Figure 10). At 600 K, O3 mixing ratios are generally larger outside of
the vortex; therefore, vortex O3 will be changed if mixing with extravortex air occurs.
As mentioned above, the 2001-2002 and 2003-2004 winters were both dynamically
active and experienced extended periods where the vortex was disturbed (see Figure
4). Therefore, the discrepancies between the observations and the modeled O3 are
likely the result of the model overestimating the amount of mixing that took place.
For every year, with the exception of the 2004-2005 winter, the model overestimates
O3 at 450 K, which results in an underestimation of ML. Similar to the 600 K level,
mixing with extravortex air at 450 K can lead to changes in O3. At 450 K O3 mixing
152
Figure 8.12: As in Figure 11, but for modeled daily average O3 (ppmv) inside the
vortex at the combined satellite locations during the ten Arctic winters.
ratios are generally smaller outside the vortex than inside; therefore, mixing of
extravortex air into the vortex would give the appearance of chemical loss. At 450 K,
the largest differences between the observations and the modeled O3 occurred during
the 1994-1995 winter, where the model overestimates the amount of O3. This
overestimation results in too little ML (as shown in the left panel of Figure 10).
During February 1995 the vortex was displaced to lower latitudes due to a minor
warming [e.g., Naujokat et al., 1995; Müller et al, 1996]. However, the vortex was
not as disturbed as in other years, so it is unclear why there is such a large difference
between the observations and the model O3 during this year. Future work will
investigate these differences. The 2004-2005 winter was the only winter in which
SLIMCAT underestimates the O3 at 450 K, which corresponds to the overestimation
of ML. Manney et al. [2006] indicate that EOS MLS N2O observations confirm that
153
mixing was occurring during this time, so it is likely that the model was too diffusive
in the lower stratosphere.
8.9 Conclusions
Unlike the Antarctic, Arctic ozone loss can undergo large interannual
variability due to the changing dynamics and meteorological conditions. In order to
explore the interannual variability, we have presented a climatology of O3 loss during
ten Arctic winters using the chemical transport model (CTM) Passive Subtraction
technique. O3 loss calculations were inferred from O3 data fields which were formed
from observations made by the UARS MLS, EOS MLS, POAM II/III, SAGE II/III,
ILAS, HALOE, ACE-FTS, and MAESTRO instruments. Results indicate that in
order to have significant O3 loss the vortex must be stable. In addition, temperatures
must be below TNAT during time periods when the vortex is stable. The winters that
experienced the largest amount of O3 loss were the 1994-1995, 1995-1996, and 19992000 winters. These were also the winters that had the strongest and most extensive
(in altitude and time) vortex when considered over the entire season, but not
necessarily the largest probability of PSC formation. The average loss during these
winters was approximately 2.4 ppmv, which was found to be comparable to results
from other O3 loss techniques. Each of these winters had a very strong vortex during
periods when the temperature fell below TNAT.
In this study we presented modeled O3 loss results from the SLIMCAT CTM
and compared the results to loss inferred from observations to determine how well the
154
CTM was able to simulate O3 loss during the ten Arctic winters with varying
meteorological conditions. Overall, the morphology in the modeled O3 loss
calculations was similar to that of the inferred loss calculations during the winters
with varying meteorological conditions. The winters with a significant amount of
loss in the model were 1994-1995, 1995-1996, 1999-2000, and 2004-2005. The
largest amount of loss occurred during 2004-2005 and the average loss during the last
14 days of the analysis was approximately 2.2 ppmv, which was greater than the
observations by approximately 0.3-0.4 ppmv. The 2004-2005 winter was the only
winter in which SLIMCAT overestimated the maximum observed loss. In every
other winter SLIMCAT underestimates the loss. Future work will involve comparing
other observations to the model in order to fully test the model’s ability to simulate
the chemistry and dynamics for each of these ten Arctic winters. Overall, SLIMCAT
was able to reproduce the morphology and the interannual variability of loss inferred
from Arctic observations during ten Arctic winters.
This study shows that while knowledge of polar O3 loss has progressed since
the discoveries of Farman et al.[1985], there is still progress that needs to be made in
the modeling of stratospheric O3 loss processes. Stratospheric modeling capabilities
have improved since previous studies and results presented here indicate that the
CTMs are able to simulate interannual variability in stratospheric O3. However,
CTMs still have difficulty reproducing the maximum loss that is inferred from
observations. With very few stratospheric missions planned for the future, the
necessity to understand the differences between CTMs and observations is even more
urgent. Since CTMs play an integral part in developing atmospheric chemistry
155
modules in Chemistry Climate Models, improving the agreement between inferred
and modeled O3 loss results is important for reliable predictions of future Arctic O3
losses.
156
Bibliography
Abarca, J. F. and C.C. Casiccia, (2002), Skin cancer and ultraviolet-B radiation under
the Antarctic ozone hole: southern Chile, 1987-2000. Photodermatology,
Photoimmunology & Photomedicine, 18, 294-302. doi: 10.1034/ j.16000781.2002.02782.x.
Alfred, G., et al.(2005), Observations and analysis of polar stratospheric clouds
detected by POAM III and SAGE III during the 2002/2003 northern hemisphere
winter, Atmos. Chem. Phys. Discuss., in preparation.
Anderson, J. G., W. H. Brune, and M. H. Proffitt (1989), Ozone destruction by
chlorine radicals within the Antarctic vortex: The spatial and temporal evolution of
ClO-O3 anticorrelation based on in situ ER-2 data, J. Geophys. Res., 94, 11,465–
11,479.
Ayers, G.P., R.W. Gillett, and J.L. Gras (1980), On the vapor pressure of sulfuric
acid, Geophys. Res. Lett., 7, 433-436.
Barath, F.T., et al. (1993), The Upper Atmosphere Research Satellite microwave limb
sounder instrument, J. Geophys. Res., 98, 10,751–10,762.
Bates, D.R., and M. Nicolet (1950), Atmospheric hydrogen, Publ. Astron. Soc. Pac.,
62, 106–110.
Becker, G., R. Müller, D. S. McKenna, M. Rex, and K. S. Carslaw (1998), Ozone loss
rates in the Arctic stratosphere in the winter 1991/92: Model calculations compared
with Match results, Geophys. Res. Lett., 25, 4325–4328.
Becker, G., R. Müller, D.S. McKenna, M. Rex, K. Carslaw, and H. Oelhaf (2000),
Ozone loss rates in the Arctic stratosphere in the winter 1994-1995: Model
simulations underestimate results of the Match analysis, J. Geophys. Res., 105,
15,175-15,184.
Bernath, P. F., et al. (2005), Atmospheric Chemistry Experiment (ACE):Mission
overview, Geophys. Res. Lett., 32, L15S01, doi:10.1029/2005GL022386.
Bevilacqua, R. M., et al. (1997), POAM II ozone observations in the Antarctic ozone
hole in 1994, 1995, and 1996, J. Geophys. Res., 102, 23,643–23,658.
Brewer, A.W., (1949), Evidence for a world circulation provided by measurements of
helium and water vapor distribution in the stratosphere. Q.J.R. Meteorol. Soc.,
157
75, 351-363.
Briegleb, B. P. (1992), Delta-Eddington Approximation for Solar Radiation in the
NCAR Community Climate Model, J. Geophys. Res., 97, 7603–7612.
Brogniez, C., A. Bazureau, J. Lenoble, and W. P. Chu (2002), Stratosopheric Aerosol
and Gas Experiment (SAGE) III measurements: A study on the retrieval of ozone,
nitrogen dioxide, and aerosol extinction coefficients, J. Geophys. Res., 107, 4758,
doi:10.1029/2001JD001576.
Browell, E. V., S. Ismail, A. F. Carter, N. S. Higdon, C. F. Butler, P. A. Robinette, O.
B. Toon, M. R. Schoeberl, and A. F. Tuck (1990), Airborne lidar observations in the
wintertime Arctic Stratosphere: Polar stratosphere clouds, Geophys. Res. Lett., 17,
385–388.
Burkholder J. B., J.J. Orlando, and C.J. Howard (1990), Ultraviolet absorption crosssections of Cl2O2 between 210 and 410 nm, J. Phys. Chem., 94, 687–695.
Butchart, N., and E.E. Remsberg (1986), The area of the stratospheric polar vortex as
a diagnostic for tracer tracer transport on an isentropic surface, J. Atmos. Sci., 43,
1319-1339.
Carslaw, K. S., B. P. Luo, S. L. Clegg, Th. Peter, P. Brimblecombe, and P. J. Crutzen
(1994), Stratospheric aerosol growth and HNO3 gas phase depletion from coupled
HNO3 and water uptake by liquid particles, Geophys. Res. Lett., 21, 2479–2482.
Carslaw, K.S., B. Luo, and T. Peter (1995a), An analytic expression fro the
composition of aqueous HNO3-H2SO4 stratospheric aerosols including gas phase
removal of HNO3, Geophy. Res. Lett., 22, 1877-1880.
Carslaw, K.S., S. L. Clegg, and P. Bringblecome (1995b), A thermodynamic model
of the system HCl-HNO3-H2SO4-H2O, including solubilites of HBr, from < 200 K
to 328 K, J. Phys. Chem., 99, 11,557-11,574.
Chapman, S., (1930), On ozone and atomic oxygen in the upper atmosphere. Philos.
Mag, 10, 369-383.
Chipperfield, M.P., A.M. Lee, and J.A. Pyle (1996), Model calculations of ozone
depletion in the Arctic polar vortex for 1991/92 to 1994/95, Geophys. Res. Lett., 23,
559-562.
Chipperfield, M. P., and J. A. Pyle (1998), Model sensitivity studies of Arctic ozone
depletion, J. Geophys. Res., 103, 28,389–28,404.
158
Chipperfield, M. P. (1999), Multiannual simulations with a three-dimensional
chemical transport model, J. Geophys. Res., 104, 1781–1805.
Chipperfield, M. P. (2006), New version of the TOMCAT/SLIMCAT off-line
chemical transport model: Intercomparsion of stratospheric tracer experiments, Q.J.R.
Meteorol. Soc., 132, 1179-1203.
Chu, W.P. C.R. Trepte, R.E. Veiga, M.S. Cisewski, and G. Taha (2002), SAGE III
measurements, Proc. SPIE Int. Soc. Opt. Eng., 48, 457-464.
Crutzen, P. J. (1970), The influence of nitrogen oxide on the atmospheric ozone
content, Q. J. R. Meteorol. Soc., 96, 320–327.
Crutzen, P. J., and F. Arnold (1986), Nitric acid cloud formation in the cold Antarctic
stratosphere: A major cause for the springtime ‘ozone hole,’ Nature, 324, 651–655.
Davies, S., M. P. Chipperfield, K. S. Carslaw, B. M. Sinnhuber, J. G. Anderson, R.
M. Stimpfle, D. M. Wilmouth, D. W. Fahey, P. J. Popp, E. C. Richard, P. von der
Gathen, H. Jost, and C. R. Webster (2002), Modeling the effect of denitrification on
Arctic ozone depletion during winter 1999/2000, J. Geophys. Res.,
doi:10.1029/2001JD000445.
Deniel, C., R.M. Bevilaqua, J.P. Pommereau, and F. Lefèvre (1998), Arctic chemical
ozone depletion during the 1994-1995 winter deduced from POAM II satellite
observations and the REPROBUS three-dimensional model, J. Geophys. Res., 103,
19,231 – 19,244.
Dessler, A. E., J. Wu, M. L. Santee, and M. R. Schoeberl (1999), Satellite
observations of temporary and irreversible denitrification, J. Geophys. Res., 104,
13,993–14,002.
Dobson, G. M. B., (1956), Origin and distribution of the polyatomic molecules in the
atmosphere. Proc. Roy. Soc. London, 236A, 187–193.
Drdla, K, and R.P. Turco (1991), Denitrification through PSC formation: A 1-D
model incorporating temperature oscillations, J. Atmos. Chem., 12, 319-366.
Drdla, K., M.R. Shoeberl, and E.V. Browell (2003), Microphysical modeling of the
1999-2000 Arctic winter: 1. Polar stratospheric clouds, denitrification, and
dehydration, J. Geophys. Res., 107, 8312, doi:10.1029/2001JD000782.
Fahey, D.W., et al. (2001), The detection of large HNO3-containing particles in the
winter arctic stratosphere, Science, 291, 1026-1031.
Farman, J. C., B. G. Gardiner, and J. D. Shanklin (1985), Large losses of total ozone
in Antarctica reveal seasonal ClOx/NOx interaction, Nature, 315, 207–210.
159
Feng, W., M.P. Chipperfield, S. Davies, B. Sen, G. Toon, J.F. Blavier, C.R. Webster,
C.M. Volk, A. Ulanovsky, F. Ravegnani, P. von der Gathen, H. Jost, E.C. Richard
and H. Claude (2005), Three-Dimensional Model Study of the Arctic Ozone Loss in
2002/03 and Comparison with 1999/2000 and 2003/04, Atmos. Chem. Phys., 5, 139152.
Feng, W., M.P. Chipperfield, M. Dorf, K. and Pfeilsticker (2006), Mid-latitude ozone
changes: Studies with a 3-D CTM forced by ERA-40 analyses, Atmos. Chem. Phys.
Discuss., 6, 6695-6722.
Froidevaux, L., et al. (2006), Early validation analyses of atmospheric profiles from
EOS MLS on the Aura satellite, IEEE Trans. Geosci. Remote Sensing 44, no. 5, 11061121.
Garcia, R. R. and S. Solomon, (1983), A numerical model of the zonally averaged
dynamical and chemical structure of the middle atmosphere, J. Geophys. Res., 88,
1379-1400.
Glaccum, W., et al., (1996), The Polar Ozone and Aerosol Measurement instrument,
J. Geophys. Res., 101, 14,479-14,487.
Goutail, F., N.R.P. Harris, I. Kilbane-Dawe, and G.T. Amanatidis (1997), Ozone loss:
the global picture, Proc. European Ozone Meeting, Schliersee, Germany.
Goutail, F., et al. (2005), Early unusual ozone loss during the Arctic winter
2002/2003 compared to other winters, Atmos. Chem. Phys., 5, 665-677.
Grooß, J.-U., et al. (2002), Simulation of ozone depletion in spring 2000 with the
Chemical Lagrangian Model of the Stratosphere (CLaMS), J. Geophys. Res.,
107(D20), 8295, doi:10.1029/2001JD000456.
Guirlet, M., M.P. Chipperfield, J.A. Pyle, F. Goutail, J.P. Pommereau, and E. Kyrö
(2000), Modeled Arctic ozone depletion in winter 1997/1998 and comparison with
previous winters, J. Geophys. Res., 105, 22,185-22,200.
Hanson, D. and K. Mauersberger (1988), Laboratory studies of the nitric acid
trihydrate: Implications for the south polar stratosphere, Geophys. Res. Lett., 15, 855858.
Harris, N. R. P., M. Rex, F. Goutail, B. M. Knudsen, G. L. Manney, R. Müller, and P.
von der Gathen (2002), Comparison of empirically derived ozone losses in the Arctic
vortex, J. Geophys. Res., 107(D20), 8264, doi:10.1029/2001JD000482.
Harvey, V.L., R.B. Pierce, T.D. Fairlie, and M.H. Hitchman (2002), A climatology of
stratospheric polar vortices and anticyclones, J. Geophys. Res., 107, 4442,
doi:10.1029/2001JD001471.
160
Hofmann, D. J., J. W. Harder, S. R. Rolf, and J. M. Rosen (1987), Balloon-borne
observations of the development and vertical structure of the Antarctic ozone hole in
1986, Nature, 326, 59–62.
Hoppel, K., R.M. Bevilaqua, G. Nedoluha, C. Deniel, F. Lefèvre, J. Lumpe, M.
Fromm, C.E. Randall, J. Rosenfield, and M. Rex (2002), POAM III observations of
arctic ozone loss for the 1999/2000 winter, J. Geophys. Res., 107,
doi:10.1029/2001JD000476.
Irie, H., M. Koike, Y. Kondo, G. E. Bodeker, M. Y. Danilin, and Y. Sasano (2001),
Redistribution of nitric acid in the Arctic lower stratosphere during the winter of
1996–1997, J. Geophys. Res., 106, 23,139–23,150.
Jacob, D. J. (1999), Introduction to atmospheric chemistry, 266 pp, Princeton, New
Jersey.
Joseph, J. H., W. J. Wiscombe, and J. A. Weinman (1976), The Delta- Eddington
approximation for radiative flux transfer, J. Atmos. Sci., 33, 2,452–2,459.
Jin, J. J., et al. (2006), Severe Arctic ozone loss in the winter 2004/2005: observations
from ACE-FTS, Geophys. Res. Lett., 33, L15801, doi:10.1029/2006GL026752.
Kodera, K., Y. Kurodo, and S. Pawson (2000), Stratospheric sudden warmings and
slowly propagating zonal-mean zonal wind anaomalies, J. Geophys. Res., 105,
12,351-12,359.
Kondo, Y., H. Irie, M. Koike, and G. E. Bodeker (2000), Denitrification and
nitrification in the Arctic stratosphere during the winter of 1996–1997, Geophys. Res.
Lett., 27, 337–340.
Larsen, N., B. Knudsen, I. S. Mikkelsen, T. S. Jørgensen, and P. Eriksen (1994),
Ozone depletion in the arctic stratosphere in early 1993, Geophys. Res. Lett., 21,
1,611–1,614.
Larsen, N. B.M Knudsen, S.H. Svendsen, T. Deshler, J.M. Rosen, R. Kivi, C.
Weisser, J. Schreiner, K. Mauerberger, F. Cairo, J. Ovarlez, H. Oelhaf, and R. Spang
(2004), Formation of solid particles in synoptic-scale Arctic PSCs in early winter
2002/2003, Atmos. Chem. Phys., 4,2001-2013, SRef-ID:1680-7324/acp/2004-4-2001.
Lary, D. J., and J. A. Pyle (1991), Diffuse radiation, twilight and photochemsistry, J.
Atmos. Chem., 13, 373-392.
Leaf, A., Loss of stratospheric O3 and health effects of increased ultraviolet radiation,
In Critical condition: Human Health and environment, ed. E. Chivian, M. McCally,
H. Hu, and A. Haines, 139-50. Cambridge, MA: The MIT Press.
161
Lemmen, C., R. Müller, P. Konopka, and M. Dameris (2006), Critique of the tracertracer correlation technique and its potential to analyze polar ozone loss in chemistryclimate models, J. Geophys. Res., 111, D18307, doi:10.1029/2006JD007298.
Limpasuvan, V., D.W. J. Thompson, and D. L. Hartmann (2004), The life cycle of the
Northern Hemisphere sudden stratospheric warmings, J. Climate, 17, 2584-2596.
Livesey, N. J., W. G. Read, L. Froidevaux, J. W. Waters, M. L. Santee, H. C.
Pumphrey, D. L. Wu, Z. Shippony, and R. F. Jarnot (2003), The UARS Microwave
Limb Sounder version 5 data set: Theory, characterization, and validation, J.
Geophys. Res., 108(D13), 4378, doi:10.1029/2002JD002273.
Lucke, R. L., et al. (1999), The polar ozone and aerosol measurement (POAM) III
instrument and early validation results, J. Geophys. Res., 104, 18,785-18,799.
Lumpe, J. D., et al. (1997), POAM II retrieval algorithm and error analysis, J.
Geophys. Res., 102, 23,593–23,614.
Lumpe, J. D., R. M. Bevilacqua, K. W. Hoppel, and C. E. Randall (2002), POAM III
retrieval algorithm and error analysis, J. Geophys. Res., 107, 4575,
doi:10.1029/2002JD002137.
Lumpe, J. D., et al. (2003), Comparison of POAM III ozone measurements with
correlative aircraft and balloon data during SOLVE, J. Geophys. Res., 107, 8316,
doi:10.1029/2001JD000472.
Manney, G. L., R.W. Zurek, A. O’Neil, and R. Swinbank (1994), On the motions of
air through the stratosphere polar vortex, J. Atmos. Sci., 51, 2,973-2,994.
Manney, G. L., R. W. Zurek, W. A. Lahoz, R. S. Harwood, J. C. Gille, J. B. Kumer,
J.L Mergenthaler, A. E. Roche, A. O’Neill,R. Swinbank, and J. W. Waters (1995a),
Lagrangian transport calculations using UARS data. Part I: Passive Tracers, J.
Atmos. Sci., 52, 3,049-3,068.
Manney, G. L., R. W. Zurek, L. Froidevaux, J. W. Waters, A. O’Neill, and R.
Swinbank (1995b), Lagrangian transport calculations using UARS data. Part II:
Ozone, J. Atmos. Sci., 52, 3069-3081.
Manney, G. L., R. W. Zurek, L. Froidevaux, and J. W. Waters (1995c), Evidence for
Arctic ozone depletion in late February and early March 1994, Geophys. Res. Lett.,
22, 2,941–2,944.
Manney, G. L., M. L. Santee, L. Froidevaux, J. W. Waters, and R. W. Zurek (1996a),
Polar vortex conditions during the 1995–96 Arctic winter: Meteorology and MLS
ozone, Geophys. Res. Lett., 23, 3,203–3,206.
162
Manney, G. L., L. Froidevaux, J. W. Waters, M. L. Santee, W. G. Read, D. A.
Flower, R. F. Jarnot, and R. W. Zurek (1996b), Arctic ozone depletion observed by
UARS MLS during the 1994–95 winter, Geophys. Res. Lett., 23, 85–88.
Manney, G. L., L. Froidevaux, M. L. Santee, R. W. Zurek, and J. W. Waters (1997),
MLS observations of Arctic ozone loss in 1996–97, Geophys. Res. Lett., 24, 2,697–
2,700.
Manney, G. L., L. Froidevaux, M. L. Santee, N. J. Livesey, J. L. Sabutis, and J. W.
Waters (2003a), Variability of ozone loss during Arctic winter (1991–2000) estimated
from UARS Microwave Limb Sounder measurements, J. Geophys. Res., 108, 4149,
doi:10.1029/2002JD002634.
Manney, G. L., J. L. Sabutis, S. Pawson, M. L. Santee, B. Naujokat, R. Swinbank, M.
E. Gelman, and W. Ebisuzaki (2003b), Lower stratospheric temperature differences
between meteorological analyses in two cold Arctic winters and their impact on polar
processing studies, J. Geophys. Res., 108(D5), 8328, doi:10.1029/2001JD001149.
Manney, G. L., K. Krüger, J. L. Sabutis, S. A. Sena, and S. Pawson (2005), The
remarkable 2003–2004 winter and other recent warm winters in the Arctic
stratosphere since the late 1990s , J. Geophys. Res., 110, D04107,
doi:10.1029/2004JD005367.
Manney G.L., M.L. Santee, L. Froidevaux, K. Hoppel, N.J. Livesey, J.W. Waters
(2006), EOS MLS observations of ozone loss in the 2004–2005 Arctic winter,
Geophys. Res. Lett., 33, L04802, doi:10.1029/2005GL024494.
Massoli, P., M. Maturilli, and R. Neuber (2006), Climatology of Arctic polar
stratospheric clouds as measured by lidar in Ny-Alesund, Sptisbergen (79°N, 12°E),
J. Geophys. Res., 111, D09206, doi:10.1029/2005JD005840.
Mauldin, L. E., III, N. H. Zaun, M. P. McCormick, J. H. Guy, and W. R. Vaughn
(1985), Stratospheric Aerosol and Gas Experiment II instrument: A functional
description, Opt. Eng., 24, 307–312.
McElroy, M. B., R. J. Salawitch, S. C. Wofsy, and J. A. Logan (1986), Reductions of
Antarcit ozone due to synergistic interactions of chlorine and bromine, Nature, 321,
759-762.
McIntyre, M.E. (1995), The stratospheric polar vortex and sub-vortex: fluid dynamics
and Midlatitude ozone loss, Phil. R. Soc. Lond. A, 35, 227-240.
Michelsen, H. A., G. L. Manney, M. R. Gunson, and R. Zander (1998), Correlations
of stratospheric abundances of NOy, O3, N2O, and CH4 derived from ATMOS
measurements, J. Geophys. Res., 103, 28,347– 28,359.
163
Molina, M. J., and F. S. Rowland (1974), Stratospheric sink for
chlorofluoromethanes: Chlorine atom catalyzed destruction of ozone, Nature, 249,
820–812.
Molina, L. T., and M. J. Molina (1987), Production of Cl2O2 from the self-reaction
of the ClO radical, J. Phys. Chem., 91, 433–436.
Morris, G. A., J. F. Gleason, J. M. Russell III, M. R. Schoeberl, and M. P.
McCormick (2002), A comparison of HALOE V19 with SAGE II V6.00 ozone
observations using trajectory mapping, J. Geophys. Res., 107, 4177,
doi:10.1029/2001JD000847.
Müller, R., J. Grooß, D. S. McKenna, P. J. Crutzen, C. Brühl, J. M. Russell III, and A.
F. Tuck (1997), HALOE observations of the vertical structure of chemical ozone
depletion in the Arctic vortex during winter and early spring 1996–1997, Geophys.
Res. Lett., 24, 2,717–2,720.
Müller, R., U. Schmidt, A. Engel, D.S. McKenna, and M.H. Proffitt (2001), The
O3/N2O relationship from balloon-borne observations as a measure of Arctic ozone
loss in 1991-1992, Q.N.R. Meteorol. Soc., 127, 1,389-1,412, 2001.
Müller, R., S. Tilmes, P. Konopka, J.-U. Grooβ, and H.-J. Jost (2005), Impact of
mixing and chemical change on ozone-tracer relations in the polar vortex, Atmos.
Chem. Phys.,5, 3,139-3,151.
Nakajima, H., et al. (2002), Characteristics and performance of the Improved Limb
Atmospheric Spectrometer (ILAS) in orbit, J. Geophys. Res., 107, 8213,
doi:10.1029/2001JD001439.
Nash, E.R., P.A. Newman, J.E. Rosenfield, and M.R. Schoeberl (1996), An objective
determination of the polar vortex using Ertel’s potential vorticity, J. Geophys. Res.
101, 9,471-9,478.
Naujokat, B., K. Krüger, K. Matthes, J. Hoffmann, M. Kunze, and K. Labitzke
(2002), The early major warming in December 2001 – exceptional?, Geophys. Res.
Lett., 29, 2023, doi:10.1029/2002GL015316.
Newman, P. A., Pyle, J. A. (Lead Authors), Austin, J., Braathen, G. O., Canziani, P.
O., Carslaw, K. S., de F. Forster, P. M., Godin-Beekmann, S., Knudsen, B. M.,
Kreher, K., Nakane, H., Pawson, S., Ramaswamy, V., Rex, M., Salawitch, R. J.,
Shindell, D. T., Tabazadeh, A., and Toohey, D. W. (2003), Polar Stratospheric
Ozone: Past and Future, Chapter 3 in “Scientific Assessment of Ozone Depletion:
2002, Global Ozone Research and Monitoring Project - Report No. 47”, Geneva.
Pawson, S., and B. Naujokat (1999), The cold winters of the middle 1990s in the
northern lower stratosphere, J. Geophys. Res., 104, 14,209 – 14,222.
164
Pierce, R. B., et al. (2003), Large-scale chemical evolution of the Arctic vortex during
the 1999/2000 winter: HALOE/POAM III Lagrangian photochemical modeling for
the SAGE III Ozone Loss and Validation Experiment (SOLVE) campaign, J.
Geophys. Res., 108, 8317, doi:10.1029/2001JD001063.
Plumb, R. A., and M. K. W. Ko (1992), Interrelationships between mixing ratios of
long-lived stratospheric constituents, J. Geophys. Res., 97, 10,145–10,156
Plumb, R. A., D. W. Waugh, and M. P. Chipperfield (2000), The effects of mixing on
tracer relationships in the polar vortices, J. Geophys. Res., 105, 10,047–10,062.
Plumb, R. A., W. Heres, J. L. Neu, N. M. Mahowald, J. del Corral, G. C. Toon, E.
Ray, F. Moore, and A. E. Andrews (2003), Global tracer modeling during SOLVE:
High-latitude descent and mixing, J. Geophys. Res., 107, 8309,
doi:10.1029/2001JD001023.
Poole, L.R, C.R. Trepte, V.L. Harvey, G.C. Toon, and R.L. VanValkenburg, (2003),
SAGE III observations of Arctic polar stratospheric clouds – December 2002,
Geophys. Res. Lett., 30, 2216, doi:10.1029/2003GL018496.
Prados, A. I., G.E. Nedoluha, R.M. Bevilacqua, D.R. Allen, K.W. Hoppel, and A.
Marenco (2003), POAM III ozone in the upper troposphere and lowermost
stratosphere: Seasonal variability and comparisons to aircraft observations, J.
Geophys. Res., 108, 4218-4228.
Prather, M. J. (1986), Numerical advection by conservation of second order moments,
J. Geophys. Res., 91, 6,671–6,681.
Proffitt, M. H., et al. (1990), Ozone loss in the Arctic polar vortex inferred from high
altitude aircraft measurements, Nature, 347, 31-36.
Proffitt, M. H., S. Solomon, and M. Loewenstein (1992), Comparison of 2-D model
simulations of ozone and nitrous oxide at high latitudes with stratospheric
measurements, J. Geophys. Res., 97, 939– 944.
Ramaroson, R., M. Pirre, and D. Cariolle (1992), A box model for on-line
computations of diurnal varations in a 1D model: Potential applications to
multidimensional cases, Ann. Geophys., 10, 416-428.
Raffalski, U., et al. (2005), Evolution of stratospheric ozone during winter 2002/2003
as observed by a ground-based millimetre wave radiometer at Kiruna, Sweden,
Atmos. Chem. Phys., 5, 1,399-1,407.
Randall, C.E., D.W. Rusch, R.M. Bevilacqua, J. Lumpe, T.L. Ainsworth, D.
Debrestian, M. Fromm, S.S. Krigman, J.S. Hornstein, E.P. Shettle, J.J. Olivero, and
165
R.T. Clancy, (1995), Preliminary results from POAM II: stratospheric ozone at high
northern latitudes, Geophys. Res. Lett., 22, 2,733-2,736.
Randall, C. E., et al. (2002), Reconstruction of three-dimensional ozone fields using
POAM III during the SOLVE, J. Geophys. Res., 107, 8299,
doi:10.1029/2001JD000471.
Randall, C. E., et al. (2003), Validation of POAM III ozone: Comparisons with
ozonesonde and satellite data, J. Geophys. Res., 108, 4367,
doi:10.1029/2002JD002944.
Randall, C. E., et al. (2005), Reconstruction and simulation of stratospheric ozone
distributions during the 2002 austral winter, J. Atmos. Sci., 62, 748-764.
Randall, C.E., V.L. Harvey, C.S. Singleton, P.F. Bernath, C.D. Boone, J.U. Kozyra
(2006a), Energetic particle precipitation effects on the Arctic middle atmosphere in
2006, Geophy. Res. Letters, in press.
Randall, C.E., V.L. Harvey, C.S. Singleton, S.M. Bailey, P.F. Bernath, M. Codrescu,
H. Nakajima (2006b), Energetic particle precipitation effects on the southern
hemisphere stratosphere in 1992-2005, J. Geophy. Res., submitted.
Ray, E. A., F. L. Moore, J. W. Elkins, D. F. Hurst, P. A. Romashkin, G. S. Dutton,
and D. W. Fahey (2002), Descent and mixing in the 1999–2000 northern polar vortex
inferred from in situ tracer measurements, J. Geophys. Res., 10,
8285,doi:10.1029/2001JD000961.
Reber, C. A. (1993), The upper atmosphere research satellite (UARS), Geophys. Res.
Lett., 20, 1,215–1,218.
Rex, M., et al. (1998), In situ measurements of stratospheric ozone depletion rates in
the Arctic winter 1991/1992: A Lagrangian approach, J. Geophys. Res., 103, 5,843–
5,854.
Rex, M., P. von der Gathen, G.O. Braathen, N.R.P. Harris, et al. (1999), Chemical
ozone loss in the Arctic winter 1994/1995 as determined by the Match technique, J.
Atmos. Chem., 32, 35-39.
Rex, M., D. Toohey, and N.R.P. Harris (2002a), Report on the Arctic Ozone Loss
Workshop, SPARC (Stratospheric Processes and their Role in Climate) Newsletter
19, 26-29.
Rex, M., et al. (2002b), Chemical depletion of Arctic ozone in winter 1999/2000, J.
Geophys. Res., 107, 8276, doi:10.1029/2001JD000533.
166
Rex, M., R. J. Salawitch, M. L. Santee, J. W. Waters, K. Hoppel, and R. Bevilacqua
(2003), On the unexplained stratospheric ozone losses during cold Arctic Januaries,
Geophys. Res. Lett., 30, 1008, doi:10.1029/2002GL016008.
Rosenfield, J. E., P. A. Newman, M. R. Schoeberl (1994), Computations of diabatic
descent in the stratospheric polar vortex, J. Geophys. Res., 99, 16,677-16,690,
10.1029/94JD01156.
Rosenfield, J. E., and M. R. Schoeberl (2001), On the origin of polar vortex air, J.
Geophys. Res., 106, 33,485–33,498.
Rusch, D. W., et al. (1997), Validation of POAM ozone measurements with
coincident MLS, HALOE, and SAGE II observations, J. Geophys. Res., 102, 23,615–
23,628.
Russell, J. M., III, L. L. Gordley, J. H. Park, S. R. Drayson, W. D. Hesketh, R. J.
Cicerone, A. F. Tuck, J. E. Frederick, J. E. Harries, and P. J. Crutzen (1993), The
Halogen Occultation Experiment, J. Geophys. Res., 98, 10,777–10,797.
Salawitch, R. J. et al. (2002), Chemical loss of ozone during the Arctic winter of
1999/2000: An analysis based on ballone-borne observations, J. Geophys. Res., 107,
8269, doi:10.1029/2001JD000620.
Salby, M. L. (1996), Fundamentals of atmospheric physics, 627pp, Academic Press,
San Diego.
Sander, S. P., Friedl, R. R., Ravishankara, A. R., et al. (2003), Chemical kinetics and
photochemical data for use in atmospheric studies, Evaluation no. 14, JPL Publ., 0225.
Santee, M. L., G. L. Manney, W. G. Read, L. Froidevaux, and J. W. Waters (1996),
Polar vortex conditions during the 1995–96 Arctic winter: MLS ClO and HNO3,
Geophys. Res. Lett., 23, 3,207–3,210.
Santee, M. L., G. L. Manney, N. J. Livesey, and J. W. Waters (2000), UARS
Microwave Limb Sounder observations of denitrification and ozone loss in the 2000
Arctic late winter, Geophys. Res. Lett., 27, 3,213–3,216.
Santee, M. L., A. Tabazadeh, G. L. Manney, M. D. Fromm, R. M. Bevilacqua, J. W.
Waters, and E. J. Jensen (2002), A Lagrangian approach to studying Arctic polar
stratospheric clouds using UARS MLS HNO3 and POAM II aerosol extinction
measurements, J. Geophys. Res., 107, 4098, doi:10.1029/2000JD000227.
Santee, M.L., G.L. Manney, N.J. Livesey, L. Froidevaux, I.A. MacKenzie, H.C.
Pumphrey, W.G. Read, M.J. Schwartz, J.W. Waters, and R.S. Harwood (2005), Polar
processing and development on the 2004 Antartic ozone hole: First results from MLS
on Aura, Geophys. Res. Lett., 32, L12817, doi:10.1029/2005GL022582.
167
Sasano, Y., M. Suzuki, T. Yokota, and H. Kanzawa (1999a), Improved Limb
Atmospheric Spectrometer (ILAS) for stratospheric ozone layer measurements by
solar occultation technique, Geophys. Res. Lett., 26, 197–200.
Sasano, Y., H. Nakajima, H. Kanzawa, M. Suzuki, T. Yokota, H. Nakane, H.
Gernandt, A. Schmidt, A. Herber, V. Yushkov, V. Dorokhov, T. Deshler (1999b),
Validation of ILAS Version 3.10 ozone with ozonesonde measurements, Geophys.
Res. Lett., 26, 831-834, 10.1029/1999GL900101.
Schoeberl, M. R., S. R. Kawa, A. R. Douglass, and T. J. McGee (2006), Chemical
observations of polar vortex intrusions, J. Geophys. Res., in press.
Seinfeld J. H., and S. N. Pandis (1998), Atmospheric chemistry and physics, from air
pollution to climate change, 1326 pp, John Wiley and Sons, Inc., New York.
Singleton, C., C. E. Randall, M. P. Chipperfield, S. Davies, W. Feng, R. M.
Bevilacqua, K. W. Hoppel, M. D. Fromm, G. L. Manney, and V. L. Harvey (2005),
2002-2003 Arctic Ozone Loss Deduced from POAM III Satellite Observations and
the SLIMCAT Chemical Transport Model, Atmos. Chem. Phys., 5, 597-609, SRef-ID:
1680-7324/acp/2005-5-597.
Singleton,C.S., C.E. Randall, V.L. Harvey, M.P. Chipperfield, W. Feng, G.L.
Manney, L. Froidevaux, C.D. Boone, P. F. Bernath, K. A. Walker, C. T. McElroy,
K.W. Hoppel (2006), Quantifying Arctic Ozone Loss During the 2004-2005 Winter
Using Satellite Observations and a Chemical Transport Model, J. Geophy. Res., in
press.
Solomon, S., Garcia, R. R. Garcia, F. S. Rowland, and D. J. Wuebbles (1986), On the
depletion of Antarctic ozone, Nature, 321, 755-758.
Solomon, S. (1999), Stratospheric ozone depletion: A review of concepts and history,
Rev. Geophys., 37, 275–316.
Staehelin, J., N. R. P. Harris, C. Appenzeller, and J. Eberhard (2001), Ozone trends:
A review, Rev. Geophys., 39, 231–290.
Stephens, G. (1994), Remote Sensing of the lower atmosphere: An introduction.
Oxford Univ. Press.
Stolarski, R. S., and R. J. Cicerone (1974), Stratospheric chlorine: A possible sink for
ozone, Can. J. Chem., 52, 1,610–1,615.
Tabazadeh, A., M. L. Santee, M. Y. Danilin, H. C. Pumphrey, P. A. Newman, P. J.
Hamill, and J. L. Mergenthaler (2000), Quantifying denitrification and its effect on
ozone recovery. Science 288:1407-1411.
168
Thomason, L. W., and G. Taha (2003), SAGE III aerosol extinction measurements:
Initial results, Geophys. Res. Lett., 30, 1631, doi:10.1029/2003GL017317.
Walker, K. A., C. E. Randall, C. R. Trepte, C. D. Boone, and P. F. Bernath (2005),
Initial validation comparisons for the Atmospheric Chemistry Experiment (ACEFTS), Geophys. Res. Lett., 32, L16S04, doi:10.1029/2005GL022388.
Thomason, L.W., and G. Taha, SAGE III aerosol extinction measurements: Initial
results, Geophys. Res. Lett., 30, 1631, doi:1029/2003GL017317, 2003.
Tilmes, S., R. Müller, A. Engel, M. Rex, and J. M. Russell III (2006), Chemical
ozone in the Arctic and Antarctic stratosphere between 1992 and 2005, Geophys. Res.
Lett., in press.
Tilmes S., R. Müller, J.-U. Grooβ, and J.M. Russell III (2004), Ozone loss and
chlorine activation in the Arctic winters 1991-2003 derived with the tracer-tracer
correlations, Atmos. Chem. Phys.,4, 2181-2213.
Tilmes, S., et al. (2006), Chemical ozone in the Arctic and Antarctic stratosphere
between 1992 and 2005, Geophys. Res. Lett., submitted.
Tung, K. K., M. K. W. Ko, J. M. Rodriguez, and N. D. Sze (1986), Are Antarctic
ozone variations a manifestation of dynamics or chemistry?, Nature, 333, 811–814.
Turco, R. P., O. B. Toon, and P. Hamill (1989), Heterogeneous physicochemistry of
the polar ozone hole, J. Geophys. Res., 94, 16,493–16,510.
Urban, J., et al. (2004), The northern hemisphere stratospheric vortex during the
2002–03 winter: Subsidence, chlorine activation and ozone loss observed by the Odin
Sub-Millimetre Radiometer, Geophys. Res. Lett., 31, L07103,
doi:10.1029/2003GL019089.
Voigt, C., et al. (2000), Nitric acid trihydrate (NAT) in polar stratospheric clouds,
Science, 290, 1756-1758.
Voigt, C., et al., (2005), Nitric acid trihydrate (NAT) formation at low NAT
supersaturation in polar stratospheric clouds (PSCs), Atmos. Chem. Phys., 5,1,3711,380.
von Hobe, M., et al. (2006), Severe ozone depletion in the cold Arctic winter 20042005, Geophys. Res. Lett., in press.
Walker, K.A., C.E. Randall, C.R. Trepte, C.D. Boone, and P.F. Bernath (2005), Initial
validation comparisons for the Atmospheric Chemistry Experiment (ACE-FTS),
Geophys. Res. Lett., 32, L16S04, doi: 10.1029/2005GL022388.
169
Wang, H. J., D. M. Cunnold, L. W. Thomason, J. M. Zawodny, and G. E. Bodeker
(2002), Assessment of SAGE version 6.1 ozone data quality, J. Geophys. Res.,
107(D23), 4691, doi:10.1029/2002JD002418.
Wang, H.-J., D. M. Cunnold, C. Trepte, L. W. Thomason, and J. M. Zawodny (2006),
SAGE III solar ozone measurements: Initial results, Geophys. Res. Lett., 33, L03805,
doi:10.1029/2005GL025099.
Waters, J.W. (1993), “Microwave limb sounding,” in Atmospheric RemoteSensing by
Microwave Radiometry, ch. 8, Ed. New York: Wiley.
Waters, J.W., et al. (1999), The UARS and EOS Microwave Limb Sounder (MLS)
experiments, J. Atmos. Sci., vol. 56, pp. 194–217.
Waters, J.W., et al. (2006), The Earth Observing System Microwave Limb Sounder
(EOS MLS) on the Aura satellite, IEEE Trans. Geosci. Remote Sens.,44, 1,075–
1,092.
Waugh, D. W., and W. J. Randel (1999), Climatology of Arctic and Antarctic polar
vortices using elliptical diagnostics, J. Atmos. Sci., 56, 1,594– 1,613.
WMO (World Meteorological Organization), Scientific Assessment of Ozone
Depletion: 2002, Global Ozone Research and Monitoring Project – Report No. 47,
498 pp., Geneva, 2003.
Документ
Категория
Без категории
Просмотров
0
Размер файла
3 346 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа