close

Вход

Забыли?

вход по аккаунту

?

Passive microwave observations of water vapor profiles during ENSO events

код для вставкиСкачать
INFORMATION TO U SERS
This m anuscript has been reproduced from the microfilm master. UMI films the
text directly from the original or copy submitted.
Thus, som e thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality o f th is rep ro d u ctio n is d ep en d en t upon the quality o f th e copy
su b m itted .
Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleedthrough, substandard margins, and improper alignment
can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and
there are missing pages, th ese will be noted. Also, if unauthorized copyright
material had to be removed, a note will indicate the deletion.
Oversize m aterials (e.g., maps, drawings, charts) are reproduced by sectioning
the original, beginning at the upper left-hand comer and continuing from left to
right in equal sections with small overlaps. Each original is also photographed in
one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white photographic
prints are available for any photographs or illustrations appearing in this copy for
an additional charge. Contact UMI directly to order.
Bell & Howell Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
800-521-0600
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PASSIVE MICROWAVE OBSERVATIONS OF WATER VAPOR PROFILES
DURING ENSO EVENTS
A Dissertation
by
CLAY BRUCE BLANKENSHIP
Submitted to the Office o f Graduate Studies o f
Texas A&M University
in partial fulfillment o f the requirements for the degree of
DOCTOR OF PHILOSOPHY
May 2000
Major Subject: Atmospheric Sciences
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
UMI N um ber 9968884
___
®
UMI
UMi Microform9968884
Copyright 2000 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
PASSIVE MICROWAVE OBSERVATIONS OF WATER VAPOR PROFILES
DURING ENSO EVENTS
A Dissertation
by
CLAY BRUCE BLANKENSHIP
Submitted to the Office o f Graduate Studies o f
Texas A&M University
in partial fulfillment o f the requirements for the degree of
DOCTOR OF PHILOSOPHY
Approved as to style and content by:
/?■
Gerald R. North
(Member)
Thomas T. Wilheit
(Chair o f Committee)
Susan C. Geller
(Member)
Gerald R. North
(Head o f Department)
James P. McGuirk
(Member)
May 2000
Major Subject: Atmospheric Sciences
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
ABSTRACT
Passive Microwave Observations o f Water Vapor Profiles
During ENSO Events. (May 2000)
Clay Bruce Blankenship, B.S., Auburn University;
M.S., Texas A&M University
Chair of Advisory Committee: Dr. Thomas Wilheit
The handling o f the water vapor feedback in general circulation models (GCM’s)
is controversial. Lindzen (1990, 1997) has proposed a mechanism by which water vapor
would cause a negative feedback, rather than the positive feedback which is predicted by
most GCM ’s. His hypothesis posits that warming leads to a more intense convective
circulation, which leads to a drying o f the upper troposphere. In this study, a dataset is
presented that is a relevant test for climate models based on observations o f the response
o f w ater vapor to a regional warming.
Retrievals o f water vapor profiles were performed using a physical relaxation
algorithm on data from the Special Sensor Microwave-Temperature-2 (SSM/T-2) for
November 1995 (during La Nina) and November 1997 (during El Nino). These
retrievals were carried out over the region o f the tropical and subtropical Pacific and
eastern Indian ocean. Water vapor profiles reveal moistening (drying) in the middle and
upper troposphere above areas o f warming (cooling) near the Equator, and drying
(moistening) in the subtropics to the north and south. These observations are consistent
with Lindzen’s hypothesis.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Adjustments were made to the water vapor field to remove the expected water
vapor profile based on both local sea surface temperature and rainfall. After each o f
these adjustments, the water vapor field still shows the aforementioned changes.
Therefore, local effects are unable to completely explain the changes in water vapor; the
large-scale circulation must be considered.
Finally, clear-sky outgoing longwave radiation (OLR) was computed from the
M aryland Terrestrial Radiation Package (MDTERP) using observed mean profiles o f
relative humidity and temperature in a 2.5° grid for each month, ignoring the effects o f
clouds. Increases in OLR o f as much as 10 W m '2 is seen in regions o f drying, and
reduced OLR is seen in regions o f moistening. Patterns o f computed OLR agree closely
with cloud-cleared observations o f OLR. Both observed and calculated OLR plots show
that water vapor changes have a significant impact on the OLR field.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
ACKNOWLEDGMENTS
I would like to thank my advisor Dr. Thomas Wilheit and my committee
members Dr. Sue Geller, Dr. Gerald North, Dr. James McGuirk, and Dr. Timothy
Phillips for their advice and comments. I also appreciate the comments o f Dr. John
Bates o f NOAA.
Special thanks are due to Dr. Robert Ellingson and Dr. Ezra Takara o f the
University o f Maryland for the use o f the MDTERP longwave program and for their
comments.
Data from many different sources were used in this project. Surface temperature
fields are from the NCEP reanalysis provided by the NOAA Climate Diagnostics Center.
SSM/T-2 data were provided by the NASA DAAC. ECMWF analyses used for
temperature profiles were provided by NCAR. OLR observations were taken from the
University o f Maryland’s HIRS data archive. SSM/I monthly rainfall retrievals are from
the Polar Satellite Precipitation Data Center at NASA Goddard Space Flight Center.
Finally, I would like to thank my wife Andrea for her support and encouragement
during m y graduate school career.
Financial support for this research was provided by NASA contract NAS532593.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
vi
TABLE OF CONTENTS
Page
ABSTRACT............................................................................................................
iii
ACKNOWLEDGMENTS....................................................................................
v
TABLE OF CONTENTS.......................................................................................
vi
LIST OF FIGURES................................................................................................
viii
LIST OF TABLES.................................................................................................
x
CHAPTER
I
n
INTRODUCTION......................................................................................
1
1.1
1.2
1.3
1.4
1
1
7
9
THE SPECIAL SENSOR MICROWAVE/TEMPERATURE-2
2.1
2.2
III
12
Instrument Description....................................................................
Water Vapor Profiling Algorithm..................................................
12
13
OBSERVATIONS OF SURFACE TEMPERATURE,
RAINFALL, AND WATER VAPOR.....................................................
15
3.1
3.2
IV
Overview...........................................................................................
Global Warming and the Water Vapor Feedback........................
Observational Studies......................................................................
Focus o f This Study.........................................................................
Observations o f Surface Temperature and Rainfall....................
Observations o f Water Vapor Profiles..........................................
15
21
ADJUSTMENTS IN THE WATER VAPOR FIELD............................
26
4 .1
4.2
4.3
4.4
Introduction......................................................................................
Adjustments for Local Sea Surface Tem perature.......................
Adjustments for Local Rainfall Changes......................................
Combined Approaches....................................................................
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
26
26
32
36
vii
CHAPTER
V
Page
OBSERVATIONS AND CALCULATIONS OF
OUTGOING LONGWAVE RADIATION............................................
39
5.1
5.2
5.3
5.4
Radiative Balance............................................................................
The MDTERP Longwave Model..................................................
Clear-sky OLR Calculations and Observations..........................
Mean SST and OLR Values...........................................................
39
40
40
46
SUMMARY AND CONCLUSIONS......................................................
49
6.1
6.2
Conclusions......................................................................................
Future Work.....................................................................................
49
50
REFERENCES.........................................................................................................
51
APPENDIX ..............................................................................................................
54
V IT A .........................................................................................................................
65
VI
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout perm ission.
viii
LIST OF FIGURES
FIGURE
Page
1
Mean surface temperature from the NCEP reanalysis............................
16
2
Mean surface temperature differences for November 1997 minus
November 1995 from the NCEP reanalysis.........................
17
3
SSM/I-estimated monthly rainfall.............................................................
18
4
Monthly SSM/I-estimated rainfall differences for November 1997
minus November 1995............................................................
19
5
Layer integrated water vapor for November 1995..................................
22
6
Layer integrated water vapor for November 1997..................................
23
7
Layer integrated water vapor difference, November 1997 minus
November 1995.......................................................................
24
Layer integrated water vapor difference predicted by local SST,
November 1997 minus November 1995...............................
30
Observed layer integrated water vapor, minus water vapor predicted
by local SST, for November 1997 minus November
1995..........................................................................................
31
Rain-predicted layer integrated water vapor for November 1997
minus November 1995............................................................
34
Rain-adjusted layer integrated water vapor for November 1997
minus November 1995............................................................
35
Layer integrated water vapor differences for November 1997 minus
November 1995 after making adjustments for local SST
and local rainfall changes......................................................
37
Rain-masked, SST-adjusted layer integrated water vapor differences
for November 1997 minus November 1995........................
38
8
9
10
11
12
13
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
IX
FIGURE
Page
14
Calculated clear-sky OLR for November 1995........................................
42
15
Calculated clear-sky OLR for November 1997........................................
42
16
Observed clear-sky OLR for November 1995.........................................
43
17
Observed clear-sky OLR for November 1997.........................................
43
18
Calculated clear-sky OLR for November 1997 minus November
1995..........................................................................................
45
Observed clear-sky OLR for November 1997 minus November
1995..........................................................................................
45
19
R ep ro d u ced with p erm ission of th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
X
LIST OF TABLES
TABLE
Page
1
Calculated outgoing longwave radiation for three atmospheres
2
SSM/T-2 channel characteristics..................................................................
13
3
Correlations between an exponential function o f SST and layerintegrated water vapor..................................................................................
28
Correlations between rainfall change and layer-integrated water vapor
change.............................................................................................................
33
Mean SST, ctTs4, and OLR values for November 1995 and November
1997.................................................................................................................
47
Mean SST, ctTs4, and OLR values for the entire region o f study for
four Novembers.............................................................................................
48
Mean SST, o T s4, and OLR values for the East Pacific region for four
Novembers.....................................................................................................
48
4
5
6
7
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
6
1
CHAPTER I
INTRODUCTION
1.1 Overview
The topic o f how accurately general circulation models (GCM ’s) forecast future
clim ate is controversial, particularly with regard to how well the water vapor feedback is
modeled. The traditional view is that the presence o f water vapor amplifies any
warming which is forced by increases in other greenhouse gases (a positive feedback
effect). This is consistent with the temperature changes predicted by most G CM ’s.
However, Lindzen (1990, 1997) has proposed a mechanism by which water vapor would
cause a negative feedback, reducing any radiatively forced temperature increases. In this
study, a dataset is presented that is a relevant test for climate models. Changes in the
water vapor distribution in the Pacific and Eastern Indian Oceans in response to a large
El Nino warming are presented. The changes in water vapor are suggestive that
Lindzen’s proposed mechanism may be correct. GCM’s should be able to reproduce the
general features o f the observed water vapor changes for an El Nino warming if they are
to be expected to accurately model the water vapor feedback.
1.2 Global Warming and the Water Vapor Feedback
a. The importance o f water vapor
Proper understanding o f the role o f water vapor in the atmosphere is essential for
accurate weather and climate modeling. Knowledge o f the distribution o f water vapor
and the processes that affect its distribution is necessary for the production o f the
This dissertation follows the style and format o f the Bulletin o f the American Meteorological Society.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
2
familiar daily weather forecasts. This knowledge is equally important for understanding,
monitoring, and predicting global climate change.
Water in all its phases plays an important role in weather and climate in many
different ways. The presence of water vapor warms the Earth due to its greenhouse
properties. Because the atmosphere is largely transparent to visible radiation
(Otherwise, we couldn’t see!), most (51%) o f the solar radiation impinging on the top o f
the atmosphere is absorbed at the Earth’s surface (Wallace and Hobbs, 1977). (The rest
is absorbed in the atmosphere, backscattered, or reflected.) This energy is re-radiated as
infrared radiation. W ater vapor and other gases such as carbon dioxide absorb much o f
this infrared radiation, trapping energy that would otherwise escape directly to space,
and warming the Earth. Water vapor is the most important o f these greenhouse gases in
terms o f the amount o f infrared radiation it absorbs.
W ater vapor also plays a role in cooling the Earth due to its phase transitions. In
the absence o f an atmosphere, radiative energy balance calculations determine the mean
surface temperature o f the Earth would be 255 K. In a non-convective atmosphere with
assumed temperature and humidity distributions, Moller and Manabe (1961) calculated
the m ean surface temperature as 350 1C. W hen the role o f convection is included, this
number is reduced to the observed value o f 288 K. Convective adjustment reduces the
greenhouse effect by about 75% because o f the latent heat transfer involved in
evaporation o f water at the surface and condensation above the surface. Transport o f
latent heat past the strongest absorbing layer in the atmosphere cools the surface more
efficiently than radiation alone. It is worth noting that the scale height o f w ater vapor in
R ep ro d u ced with p erm ission of the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
3
the atmosphere is only about 2 km (Stephens, 1990) so the convection does not have to
be deep at all to bypass the bulk o f the infrared absorbers in the atmosphere.
b. The positive water vapor feedback
The concentrations o f carbon dioxide, methane, and other greenhouse absorbers
in the atmosphere have been increasing due to human activity since the Industrial
Revolution. The atmospheric concentration for carbon dioxide was 358 ppmv in 1992,
compared to the preindustrial level o f 280 ppmv. It is expected to reach approximately
500 ppmv by the end o f the twenty-first century (Houghton et al., 1996). The direct
result o f increased greenhouse gases is a warming. Current general circulation models
predict a 1.2 °C global mean temperature change as a direct response to doubled CO 2
(Houghton etal., 1990).
The Clausius-Clapeyron equation gives the relationship between saturation vapor
pressure and temperature. It is:
de5 _
dT
L
T (a 2 -o r,)
where es is the equilibrium vapor pressure o f water at temperature T, L is the latent heat
o f vaporization o f water, and a i and
012
are the specific volumes o f liquid water and
water vapor respectively. This means that the amount o f water vapor an air mass can
hold increases with temperature. If relative humidities remain constant, a warming will
lead to an increase in atmospheric water vapor. Since water vapor is a greenhouse gas,
this increase will lead to a further warming. This positive feedback effect has been
known since the work o f Arrhenius (1896).
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
4
Cess et al. (1990) investigated the feedbacks in 19 different general circulation
models and found similar results for all the models: a roughly 70% enhancement in
sensitivity due to water vapor. This is consistent with the early model study by Manabe
and W etherald (1967) which held relative humidity constant. According to the 1990
report o f the Intergovernmental Panel on Climate Change (Houghton et al., 1990), the
water vapor feedback amplifies the direct effect o f CO 2 doubling from 1.2 °C by a factor
of 1.4 to l.7°C, based on climate models. The total change is increased to 1.9 °C (a
factor o f 1.6) when the warming effect o f the (relatively small) absorption o f solar
radiation by water vapor is also considered. Recently, Hall and Manabe (1999) found a
stronger amplification: a 1.05 °C globally averaged surface warming for a 500-year
integration o f a Geophysical Fluid Dynamics Laboratory coupled ocean-atmosphere
model with doubled CO 2 and no water vapor feedback became a 3.38 °C warming with
feedback active.
c. Upper tropospheric humidity and the negative water vapor feedback
Lindzen (1990, 1997) has proposed a contrary view that the presence o f water
vapor causes a negative global warming feedback. He proposed that increasing
temperatures would cause an enhanced convective cycle. The tops o f convective clouds
would be higher, causing air detraining from the cloud tops to be drier. This effect
would cause the subsiding air in the upper troposphere to be drier. A drier upper
troposphere would allow more infrared radiation to escape to space, cooling the Earth.
In this way, water vapor could have a regulating effect on climate.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
5
Because o f the nonlinear nature o f the absorption o f radiation (absorption varies
roughly with the logarithm o f the amount o f absorber due to saturation effects), it is
possible to decrease the greenhouse effect while increasing the net amount o f water
vapor. This can be done by making the dry areas drier and the humid areas more humid.
This is exactly what would happen with an enhanced convective cycle, according to the
model o f Lindzen (1990). Even if this effect does not reverse the sign o f the overall
water vapor feedback, it could still account for a reduction in the strength o f the positive
amplification.
Table 1 illustrates the importance o f upper tropospheric water vapor. Three
different atmospheres were used as inputs to the Maryland Terrestrial Radiation Package
(MDTERP) (Ellingson and Gille, 1978; Warner and Ellingson, 2000), which is described
in section 5.2.
The first is a standard tropical atmosphere (Air Force Cambridge
Research Laboratory, 1965). The second (“dry bottom”) is the same standard tropical
atmosphere with water vapor below 3.5 km (about 672 mb) reduced by 50%. The final
atmosphere (“dry top”) is a standard tropical atmosphere with water vapor above 3.5 km
reduced by 50%. The table gives the upward flux o f longwave radiation at the top o f the
atmosphere (outgoing longwave radiation or OLR). Reducing the water vapor above 3.5
km by 50% produces an increase o f 9.18 W m'2 in OLR, while halving the water vapor
below 3.5 km causes an increase o f 6.28 W m '2 in OLR. The change in upper
tropospheric water vapor has a larger effect, even though only 20% o f the w ater vapor in
the atmosphere is above the 3.5 km layer (given the 2 km scale height). In spite o f its
relatively low concentrations, water vapor at high levels is important because a very
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
6
small absolute change in water vapor can allow more infrared radiation from lower
levels to pass through. This is especially important in the driest regions. Spencer and
Braswell (1997) presented data showing the extreme dryness o f the upper troposphere
relative to what earlier datasets have indicated.
The increase in water vapor at low levels with temperature is not
disputed. The change in upper tropospheric water vapor is more controversial. Even
with a positive feedback due to water vapor at low levels, a negative feedback due to
w ater vapor at upper levels in the troposphere could reduce the magnitude o f the positive
feedback, or reverse it altogether due to the nonlinear absorption o f infrared radiation by
w ater vapor.
Table 1. Calculated outgoing longwave radiation for three atmospheres. “Dry bottom” is
a standard tropical atmosphere with water vapor reduced by 50% below 3.5 km. “Dry
top” is a standard tropical atmosphere with water vapor above 3.5 km reduced by 50%.
Atmosphere
Standard tropical
OLR
(W m '2)
287.71
Change from
standard tropical
0.0
Dry bottom
293.99
6.28
Dry top
296.89
9.18
d. Other feedbacks
Clouds also play a major role in regulating climate. Three cloud parameters
which are determined by climate also regulate the climate, and thus allow for feedbacks.
They are cloud amount, cloud altitude, and cloud w ater content. Another feedback
R ep ro d u ced with p erm ission of the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
7
mechanism involves changes in the lapse rate. Many other complex physical and
chemical feedback mechanisms between the atmosphere, land, ice, ocean, and biosphere
exist. Only the role o f water vapor is covered in this study.
1.3 Observational Studies
a. Importance
Models can be used to predict future climate, but they will never be perfect.
There is no substitute for observations in furthering our understanding o f how the
climate system works. It may be argued that the satellite record o f detailed global
observations is not long enough to capture a significant climate change signal relevant to
global warming. There are, however, very significant interannual and interseasonal
differences which can be observed. Observations o f these changes can be used to better
understand the climate system and to validate and improve climate models.
b. The E l Nino-Southern Oscillation
Perhaps the largest interannual climate signal is the El Nino-Southern Oscillation
(ENSO). El Nino, also known as the warm phase o f ENSO occurs about every 3 to 4
years on average (albeit irregularly) and lasts for 12-18 months. It is characterized by a
shift in the “warm pool” from the western to the eastern Equatorial Pacific, high
pressures in the western tropical Pacific, and low pressures in the eastern tropical
Pacific. La Nina, or the cold phase o f ENSO, often follows El Nino, and is characterized
by the opposite pressure pattern, and colder temperatures in the eastern Pacific. El Nino
events are generally more extreme than La Nina events in terms o f deviation from mean
weather patterns. Both o f these phenomena alter weather patterns over much o f world.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
8
c. Previous work
Many studies o f the relationship between w ater vapor and surface temperature by
looking at interannual changes have been performed. Yang and Tung (1998) found that
correlations in the tropics between relative humidity (RH) and surface temperature could
not be established, but found statistically significant correlations between specific
humidity and surface temperature in the upper troposphere, using humidity retrievals
from the TIROS Operational Vertical Sounder (TOVS) and the Special Sensor
Microwave Imager (SSM/I). Similarly, Sun and Oort (1995) found that specific
humidity at 500 mb increases with temperature, but relative humidity does not.
Houghton et al. (1996) reviewed several studies o f the role o f upper tropospheric water
vapor. They found that the upper tropospheric water vapor feedback seemed to be
positive, but that this was not convincingly established. Chaboureau et al. (1998) found
that in the middle to upper troposphere, water vapor content does not systematically vary
w ith SST but depends on the dynamics.
Evidence for a negative water vapor feedback was found by Chou (1994) by
studying data from April 1985 and April 1987, the latter from an El Nino event. Chou
found a reduction o f 4.0 W m'2 in the greenhouse effect during the warmer month o f
April 1987 by comparing changes in both incoming solar radiation and outgoing
longwave radiation (OLR) from the Earth Radiation Budget Experiment (ERBE). This
was attributed primarily to a reduction in the clear-sky greenhouse effect in the Northern
Hemisphere subsidence region.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
9
In contrast, Soden (1997) found the opposite result for May 1985 and May 1987.
Further evidence for an overall positive feedback was found by Inamdar and
Ramanathan (1998). They found global sensitivity to water vapor increases consistent
with the positive feedback given by GCM’s by looking at the annual cycles o f
temperature and water vapor.
1.4 Focus of This Study
a. Goal o f this study
Since we have had global satellite coverage for a relatively short time, there is
not a large observable global mean temperature change. However there is a large
regional interannual temperature signal due to ENSO. During November 1997, sea
surface temperatures (SST’s) in the eastern Equatorial Pacific were 5-6 °C warmer than
November 1995. There was also a large decrease in SST in an area west o f Indonesia.
These temperature changes were accompanied by changes in the water vapor field.
Changes in the water vapor field are investigated by using water vapor profiles
retrieved from the Special Sensor Microwave Temperature-2 (SSM/T-2). Almost all
previous studies o f upper tropospheric water vapor used infrared sensors. The use o f
microwave measurements will provide a valuable additional way to measure this
parameter. The SSM/T-2 is sensitive to water vapor at altitudes below about 200 mb so
it is well suited to measuring middle and upper tropospheric humidity. Additionally,
microwave measurements offer the advantage o f working in most cloudy situations,
removing much o f the bias due to observing only clear areas.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
10
The goal o f this study is to investigate the behavior o f the water vapor field in
response to temperature changes, and to investigate the resulting changes in OLR. In
this study, data from two phases o f ENSO are used to address the following questions
relevant to the water vapor feedback problem:
1) Does the upper troposphere dry in response to warming?
2) What is the response o f the OLR field to changes in water vapor and
temperature?
In this study, changes in sea surface temperature (SST), rainfall, and humidity in
the Pacific and eastern Indian Oceans between November 1995 (during La Nina) and
November 1997 (during a very strong El Nino) are presented and analyzed. SST and
water vapor fields are used to answer the first question. Rainfall fields together with the
SST distribution are used to gain insight into the general circulation.
While this situation is clearly not the same as a global warming scenario with
greatly increased CO 2 , we can observe the response o f the water vapor field to changes
in the surface temperatures. This provides some insight into whether the circulation is
indeed intensified in response to warming. In addition, adjustments are performed to see
how much o f the water vapor change can be explained by local changes in temperature
and rainfall.
As a final step, the question o f the impact o f the water vapor changes on the OLR
field is addressed. Observed mean temperature and humidity profiles are used to
calculate the resulting clear-sky OLR field for each month. Direct observations o f clearsky OLR are also used for intercomparison and validation.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
11
b. Region o f study
The area between 40° N and 40° S, east o f 75° E, and west o f 75° W is used in
this study. This includes the ocean regions with the largest SST change due to ENSO,
i.e. the Equatorial Pacific and eastern Indian Ocean. A poleward limit o f 40° latitude is
sufficient to include the subsiding branch o f the Hadley circulation. Lindzen et al.
(1995) suggested that a surrogate for global climate change must consider averages over
major circulation systems, since climate is not locally determined.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
12
CHAPTER n
THE SPECIAL SENSOR MICROWAVE/TEMPERATURE-2
2.1. Instrument Description
The SSM/T-2 is the first operational microwave radiometer to be placed in orbit
for the purpose o f water vapor sounding. It was first launched in November 1991 aboard
the Defense Meteorological Space Program (DMSP) F-l 1 spacecraft. The F-l 1 orbits
the earth in a near-polar (98.8° inclination), sun-synchronous orbit with a period o f 101.8
minutes and an altitude o f 833 km.
The SSM/T-2 has five channels whose characteristics are summarized in Table 2
(Falcone et al., 1992). Three o f the channels are centered around the strong peak in the
w ater vapor absorption spectrum at 183.31 GHz, giving the instrument its sensitivity to
humidity at different levels in the atmosphere and enabling the retrieval o f w ater vapor
profiles. These channels actually consist o f two sidebands each, situated symmetrically
about the 183.31 GHz frequency. For example, the 183.31+/-1 GHz channel has two
passbands centered at 182.31 and 184.31 GHz. The remaining two channels are known
as window channels because o f their tendency to have significant surface contribution.
The SSM/T-2 is a cross-track scanner. Every 8 seconds, the sensor makes 28
measurements across a scan line perpendicular to the spacecraft velocity vector centered
at nadir with 3 degree increments (from -40.5 to 40.5 degrees about nadir), as well as
four calibration measurements of a hot-load target o f known temperature (approximately
300 K) and the cosmic background (2.7 K). Data from the two beam positions on either
end may be unusable due to interference from the glare obstruction panel, or “glob”. To
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
13
avoid this problem, all data from these beam positions have been eliminated. The swath
width o f one complete scan is about 1400 km, or about 1200 km after removal o f the
data from the extreme beam positions.
Table 2. SSM/T-2 channel characteristics.
Frequency
(GHz)
183.31+1
183.31±3
183.31±7
150.0
91.655
Nadir resolution
(km)
48
48
48
54
84
NEAT*
(K)
0.8
0.6
0.6
0.6
0.6
*noise equivalent temperature uncertainty
The SSM/T-2 has a parabolic reflector which rotates to scan along the swath.
This rotation o f the reflector rotates the plane o f polarization o f the radiation which is
measured, so that each beam position along the scan has a different polarization
associated with it. This polarization is the same for all channels.
2.2. W ater Vapor Profiling Algorithm
Water vapor profiles were retrieved from SSM/T-2 microwave radiometer data.
The retrieval method is described in Blankenship et al. (2000) and in Blankenship
(1997). The version o f the algorithm which solves for the natural logarithm o f relative
humidity was used. This method was shown to be more accurate in terms o f agreement
with radiosonde and analysis profiles because the sensitivity functions for water vapor
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
14
are more nearly linear w ith respect to In (RH) than RH. Also, the version which uses
four SSM/T-2 channels only was used rather than the version which also uses SSM/I
data, because SSM/I and SSM/T-2 data were not available from the same satellite for the
month o f November 1995. Details o f the algorithm are given in the appendix.
Water vapor profile retrievals must be initialized with a temperature profile.
European Centre for M edium Range Weather Forecasting (ECMWF) analyses are used
for this purpose. The ECMWF analyses are reported twice daily, at 0 and 1200 UTC, on
a grid o f resolution 2.5 degrees latitude by 2.5 degrees longitude. They include height,
temperature, and humidity fields reported at 13 pressure levels. ECMWF data are
interpolated in space to match the location o f satellite observations.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
15
CHAPTER IU
OBSERVATIONS OF SURFACE TEMPERATURE,
RAINFALL, AND WATER VAPOR
3.1. Observations of Surface Temperature and Rainfall
Figure 1 shows monthly mean surface temperature fields from November 1995
(during La Nina) and November 1997 (during El Nino) for the area between latitudes 40
°N and 40°S and from longitude 75° E eastward to 75° W. The difference in surface
temperatures between these two months (November 1997 minus November 1995) is
shown in Figure 2. These temperatures are from the NCEP reanalysis provided by the
NOAA Climate Diagnostics Center. The most significant change is the spreading o f the
area o f warmest temperatures from the region north o f Australia in 1995 to its location in
1997 when it is spread out across the Equator towards South America. The greatest sea
surface temperature changes are a warming o f 5-6 °C in the Equatorial eastern Pacific
and a cooling o f around 4 °C over a smaller region near the Equator in the eastern Indian
Ocean.
Monthly rainfall totals and differences for the same pair o f months are shown in
Figures 3 and 4. These data were provided by the Polar Satellite Precipitation Data
Center at NASA Goddard Space Flight Center and are based on SSM/I measurements
(W ilheitet al., 1991).
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
16
a) November 1995
Nov 95 S u rface Tem perature
Longitude
5
10
15
20
deg C
25
30
b) November 1997
Nov 97 S urface Tem perature
Longitude
5
10
15
20
deg C
25
30
Figure 1. Mean surface temperature from the NCEP reanalysis.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
17
Latitude
S urface TemptNov 9 7 — Nov 9 5
Longitude
Figure 2. Mean surface temperature differences for November 1997 minus November
1995 from the NCEP reanalysis.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
a) November 1995
Monthly Rain R ate:f13nav95
m
2S0
Lanqituda
100
200
10
300
mny month
400
S00
b) November 1997
Monthly Rain R ate:f14nov97
Longitudt
mm/moi
Figure 3. SSM/I-estimated monthly rainfall.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
19
N av 9 7 — N ov 9 5 rain dfff
100
^
—400
150
200
Longitude
---- • ' 1-.538—
^
M
-2 0 0
.0
200
mm/month
250
400
Figure 4. Monthly SSM/I-estimated rainfall differences for November 1997 minus
November 1995.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
20
Rain fields give some indication o f the vertical air speed during this time period.
The presence o f rainfall in the Intertropical Convergence Zone near the Equator and in
the South Pacific Convergence Zone indicates rising air in those regions. Some rainfall
is also seen in the north Pacific storm tracks. The almost complete lack o f rainfall in the
subtropical highs indicates areas o f subsidence.
The change in rainfall largely follows what is expected from the local
temperature changes. The warming in the eastern Pacific ITCZ is coincident with
increased rainfall and the cooling in the eastern Indian ITCZ is coincident with
decreased rainfall. The entire ITCZ in the Pacific is observed to move southward. The
eastward shift in the warmest water has been accompanied by an eastward shift in the
SPCZ, as evidenced by the dipole structure seen in the south Pacific at about 150° W (or
210° E). A similar eastward shift in the area o f maximum rainfall is also seen in the
north Pacific. It is noteworthy that in the subtropical highs in the eastern Pacific (north
and south) and the subtropical high west o f Australia, the change in rainfall is very small
because the total rainfall was very small in both months.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
21
3.2 Observations o f Water Vapor Profiles
W ater vapor profiles were performed for the region o f study for November 1995
and November 1997. Only data from even numbered days o f the year were used
because o f disk space and processing time issues. For these days, every available
SSM/T-2 observation from the F -12 (1995) or F-14 (1997) satellite over the ocean
portions o f these areas were used. Retrieved profiles were averaged in 1° by 1° bins,
retaining the full vertical resolution.
Average retrieved water vapor amounts were divided into four layers (surface to
800 mb, 800 to 600 mb, 600 to 400 mb, and 400 to 200 mb) for ease o f display. Layer
integrated water vapor for November 1995 and November 1997 are plotted in Figures 5
and 6 respectively. Layer integrated water vapor differences (November 1997 minus
November 1995) are plotted in Figure 7. Here, grey indicates areas where the water
vapor change was insignificant compared to the accuracy o f the retrieval.
In the lowest layer, water vapor increased all along the Equator. It is not apparent
why the surface to 800 mb layer o f the Equatorial eastern Indian Ocean moistened as it
cooled. For higher layers, water vapor near the Equator increased with increased
surface temperature in the Eastern Pacific and decreased with surface cooling in the
western Pacific and Indian Ocean. This is expected, as a warmer surface leads to more
convection and a more moist atmosphere.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
b)
Avq 8 0 0 - 6 0 0 m p tWV:Nov 9 5 ( e v e n doys)
Ungitwdt
Ungoud*
Langiwdt
LangUudt
Figure 5. Layer integrated water vapor for November 1995 a) Surface to 800 mb, b) 800 to 600 mb, c) 600
to 400 mb, d) 400 to 200 mb.
to
to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
b)
Unpitudc
Ungritadt
Avq 6 0 0 —4 0 0 mD iVW:Nov 9 7 (even dovs)
LcngiWdt
UngiUid*
Figure 6. Layer integrated water vapor for November 1997. a) Surface to 800 mb, b) 800 to 600 mb, c) 600
to 400 mb, d) 400 to 200 mb.
K>
U>
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ava sfc-B O O m b IW :N ov 9 7 -N o v 9 5 (« v tn d a v t
j». j Am
Le'lVbM*
1.0
- 0 .5
0.0 r
La*alW<«
, 0.5
Total IWV [ c m ]
1.0
“ 0.5
0 .0 ,
,0 .5
Totol IWV [ c m ]
Avg 6Q 0-40D mb lWV:Nov 9 7 -h o v 95 (e*«n doy»)
lanptadt
- 1 .0
- 0 .5
0 .0 ,
,0 .5
T otal IWV [ c m ]
0 .1 5 - 0 .1 0 - 00.05
.0 5 0 0 0 , 0 05
T otal IWv [ c m ]
0.10
0.1?
Figure 7. Layer integrated water vapor difference, November 1997 minus November 1995. a) Surface to 800
mb, b) 800 to 600 mb, c) 600 to 400 mb, d) 400 to 200 mb. Grey indicates an insignificant change.
to
25
The large changes in water vapor in the subtropics are notable. In the central
Pacific subtropics, there are regions o f drying to the north and south o f the Equator.
This is consistent with Lindzen’s hypothesis applied to the Hadley circulation. Warmer
surface temperatures at the Equator cause enhanced convection, moistening in the
Equatorial areas o f rising air, and drying in the subtropical areas o f subsidence. This
pattern is less evident in the eastern Pacific, probably due to the eastward shift in the
South Pacific Convergence Zone and a similar north Pacific feature, evidenced by two
diagonal bands o f blue (moistening) in Figure 7d at about 140° W (220° E) .
Even stronger evidence o f a subtropical response to Equatorial temperature
changes is seen in the eastern Indian Ocean. At all levels except for the lowest, there is
pronounced Equatorial drying coincident with surface temperature decreases. This
feature is accompanied by strong moistening in the subtropics directly to the south (and
perhaps to the north near India). This is also consistent with Lindzen’s hypothesis.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
26
CHAPTER IV
ADJUSTMENTS IN THE WATER VAPOR HELD
4.1 Introduction
It has already been noted that ENSO warming exhibits a different signature from
global warming. Lau et al. (1996) suggested that regional interannual variability should
not be used as a surrogate for global warming unless proper corrections are made for the
effect o f large scale circulation. They suggested averaging over a domain which
included both rising and subsiding branches o f the Walker and Hadley circulations. The
region o f study already includes regions o f subsiding and rising air. Averaging water
vapor profiles is undesirable because o f the nonlinear effect o f water vapor on OLR.
There is a correction that can be performed by attempting to separate local and
large-scale effects. This can be done by subtracting the part o f the water vapor field that
can be predicted by local SST or rainfall observations. The residual field will be the
water vapor change that cannot be explained by local changes in SST or rainfall.
4.2 Adjustments for Local Sea Surface Temperature
Stephens (1990) derived an expression for total integrated water vapor as a
function o f sea surface temperature, assuming a constant surface relative humidity, r,
and a constant ratio o f the water vapor scale height to the atmospheric scale height, W.
Since the water vapor scale height is typically around 2 km and the atmospheric scale
height is typically 7 km, the value o f W is about 3.5. Based on the Clausius-Clapeyron
equation, the relationship is:
w = 10.82
exp[a(Tt - 288K)]
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
27
where a= 0.0064 K '1 and w, in cm, is the expected integrated water vapor based on SST.
Rather than using prescribed values o f r and L, Stephens estimated the r/(l+L ) factor
from a least square fit o f observational data. This approach assumes there is an
approximate equilibrium process linking total integrated water vapor and SST through
vertical mixing. If we ignore the heat capacity o f the atmosphere, then the atmosphere
must respond to radiative heat input by developing a convective circulation. According
to this approach, the departure of the observed integrated water vapor from w is due to
the contribution o f atmospheric dynamics (Serke, 1996).
This same procedure can be used to predict the integrated water vapor for each
layer as a function o f SST by finding a best fit value o f r/(l+L) for each layer. A best fit
was performed using data from both November 1995 and November 1997.
Table 3 shows the correlations between each layer o f integrated water vapor and
the SST for both November 1995 and November 1997, and for both the West Pacific
(defined here as all ocean areas between 110° E and 180° E) and East Pacific (east of
180° latitude). As expected, lower layers o f water vapor are generally more closely
correlated with SST. Correlations are generally stronger in the West Pacific where
temperatures are warmer, leading to more convection which provides a physical linkage
between SST and upper-level humidity. Similarly, in 1997 correlations increased in the
East Pacific and decreased in the West Pacific due to the shift o f warm water eastward
associated with El Nino.
Although the degree of correlation between SST and humidity is related to the
SST, the correlation coefficients are fairly similar for both months and both areas.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
28
Table 3. Correlations between an exponential function o f SST and layer-integrated
water vapor.
Region
West Pacific
Nov 1995
West Pacific
Nov 1997
East Pacific
Nov 1995
East Pacific
Nov 1997
Layer
Total
(Surface to 200 mb)
.920
.924
.873
.917
Surface to 800 mb
.936
.945
.927
.957
800 to 600 mb
.893
.872
.777
.824
600 to 400 mb
.705
.678
.590
.689
400 to 200 mb
.682
.712
.665
.740
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
29
Changes in SST explain between 34.8% and 91.6% o f the variance in the layer
integrated water vapor for the situations described above. The remaining variability in
the water vapor field is due to some other mechanism. Figure 8 shows the part o f the
water vapor difference field for four layers which can be explained by the local
thermodynamic response to SST. Figure 9 shows the residual water vapor difference for
each layer, which is attributable to large scale dynamics.
M ost o f the features seen in the unmodified water vapor field are still present
after accounting for the local effect o f temperature on water vapor. For the layers above
the surface layer, the eastern Equatorial Pacific moistened in response to warming.
Except for two small regions o f moistening, there is a small but consistent decrease in
the water vapor to the north and south of the eastern Pacific ITCZ. The signal in the
Indian Ocean is still quite strong with drying in the ITCZ in the region o f surface
cooling, and moistening to the south consistent with a weakened Hadley circulation. In
light o f this, the changes in the water vapor field can not solely be explained by local
changes in the temperature field. The observed water vapor changes are, however,
explainable by changes in the strength o f the Hadley circulation in response to changes
in SST in the ITCZ.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Avq
T>iarmodyn o m t e l W |N o v f l 7 - N o v 9 5 ( e v a n doy«)
Avg flO Q -^ Q m b T > > e rm o O v r o m jc f ^ V ^ to ^ 9 7 - N o v 9 5 ( e v f ln doys]
Unpwoi
LOfpWtf*
-1.0
- 0 .5
0.0
Q.5r
.
T h e rm o d y n a m ic Layer IWv l e m j
1.0
Avg B 00-4Q Q m b Tharm oflvnom tc lWV:Nov 9 7 - N o v 9S> (e» « n d a y s)
UfipUrf*
-1.0
- 0 .5
0.0
1.0
- 0 .5
.0.0
Q.5
.
T h e rm o d y n a m ic Layer IWv [ c m j
1.0
Avg 4 0 0 - 2 0 0 i r ^
Longfertt
fl.5.
,
T h e rm o d y n a m ic L ayer IWV [ c m ]
1.0
- 0 .1 5 - 0 . 1 0 - 0 .0 5 0.00
0.05
0.10
T h e rm o d y n a m ic Loyer IWV [ c m ]
0.15
Figure 8. Layer integrated water vapor difference predicted by local SST, November 1997 minus November
1995. a) Surface to 800 mb, b) 800 to 600 mb, c) 600 to 400 mb, d) 400 to 200 mb.
OJ
o
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
b)
Avo afc-B O O m b Dynam ic IWV.Nov 9 7 -N o v 9 5 (»*en d a * s)
"r 'P' iutfw ~‘r
1
LofigiUM*
D ynam ic L ayer IWV
l£*Q>totfa
longffajtf*
- 1 .0
- 0 .5 ,
0.0
0 .5 ,
D y n am ic L ayer IWV i c m ]
1.0
- 0 .1 5 - 0 . 1 0 - 0 .0 5
0.00
0.05
0,10 0.15
D ynom ic Loyer IWv [ c m ]
Figure 9. Observed layer integrated water vapor, minus water vapor predicted by local SST, for November
1997 minus November 1995. a) Surface to 800 mb, b) 800 to 600 mb, c) 600 to 400 mb, d) 400 to 200 mb.
32
4.3 Adjustments for Local Rainfall Changes
Rainfall observations give an indication o f the presence o f rising air, since rising
air cools, which leads to saturation. In this way, the rainfall field can be used as a proxy
for the vertical component o f the circulation, and hence a predictor o f water vapor. In a
sim ilar manner to the temperature correction, we can use the observed monthly rainfall
change to predict the change in water vapor. A linear regression was performed on the
w ater vapor differences for each layer versus the rainfall differences. (Differences were
used rather than individual values because correlations between water vapor differences
and rainfall differences were larger than those between individual water vapor and
rainfall values.) Table 4 gives correlations between the rainfall change and the layer
integrated water vapor changes for the entire region o f study and for specific areas.
Although changes in rainfall explain much less o f the variability in the temperature field
than SST’s do, between 19.6% and 34.6% o f the variance is explained for the three
layers above 800 mb.
Figure 10 shows the water vapor difference for each layer that is predicted from
the observed rainfall difference. This represents the changes in water vapor that are
explainable due to the locally observed changes in rainfall. Figure 11 shows the
difference between this field and the observed change in water vapor. The pattern o f
moistening in the ITCZ and drying in the subtropics in the Pacific is still evident after
making these changes, as is the opposite pattern in the eastern Indian Ocean.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
33
Table 4. Correlations between rainfall change and layer-integrated w ater vapor change.
Layer
Correlation
Total (Surface to 200 mb)
.504
Surface to 800 mb
.206
800 to 600 mb
.443
600 to 400 mb
.574
400 to 200 mb
.588
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nov 97 - Nov 95 rain—predicted wv diff:800 to 600 mb
Nov 97 — Nov 95 rain-predicted wv diff:Sfc to 800 mb
o
-20
-to
L o n g itu d e
L o n g itu d e
•
1.0
- 0 .5
Nov 97 - Nov 95
0.0
cm
0.5
1.0
- 0 . 6 - 0 . 4 - 0 . 2 0.0 0.2
cm
redicted wv diff:600 to 400 mb
-2 0
0.4 0.6
Nov 97 - Nov 95 roin-predicted wv diff:40Q to 200 mb
-2 0
-40
-4 0
Longltud*
- 0 .4
-0 .2
0.0
cm
L on g itu d e
0.2
0.4
-0.10
-0 .0 5
0.00
cm
0.05
0.10
Figure 10. Rain-predicted layer integrated water vapor for November 1997 minus November 1995. a)
Surface to 800 mb, b) 800 to 600 mb, c) 600 to 400 mb, d) 400 to 200 mb.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nov 97 — Nov 95 roin—adjusted wv diff:Sfc to 800 mb
Nov 97 — Nov 95 roin—adjusted wv diff:80D to 600
L on g itu d e
L on g itu d e
-1 .5 -1 .0 -0 .5
Nov 97 — Nov 95 ram —odiusted wv diff:600 to 400 mb
1.0
-0.5
0.0
cm
0.5
I.O
1.5
Nov 97 — Nov 95 roin—odiusted wv ditf:400 to 200
Longitudi
-
0.0
L o n g itu d e
0.5
1.0
-
0.20
-
0.10
0.00
cm
0.10
0.20
Figure 11. Rain-adjusted layer integrated water vapor for November 1997 minus November 1995. a)
Surface to 800 mb, b) 800 to 600 mb, c) 600 to 400 mb, d) 400 to 200 mb.
U>
L*
36
4.4. Combined Approaches
Both the SST and rainfall adjustment can be taken into account simultaneously.
Figure 12 shows the residual water vapor change for the four layers after applying both
o f these corrections in series. Once again, drying is observed in the Pacific subtropics
(mostly in the central and eastern Pacific) to the north and south o f areas o f warming,
and moistening is observed in the Indian Ocean to the south o f the area o f cooling.
Another way o f viewing this data is used in Figure 13, which shows the SSTadjusted water vapor field, with regions o f large rainfall change masked out. Areas
where rainfall increased more than 250 mm are shown in black, and areas where rainfall
decreased more than 250 mm are shown in red. This shows the relationship o f the
changes in temperature-adjusted water vapor in areas o f relatively small absolute change
in rainfall to large increases (black) or decreases (red) in convection. This method
shows SST-independent changes in water vapor in areas o f relatively little rainfall
change. There is a clear signal in the Indian Ocean where the subtropical upper
tropospheric water vapor (adjusted for SST changes) increased as the Hadley circulation
weakened (as evidenced by less rainfall at the ITCZ). It also shows that much o f the
drying in the Pacific subtropics was not coincident with large rainfall decreases,
although this is partly explained by an eastward shift in the SPCZ.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a )' i A.Nov
97
— Nov 95 Roin/SST—adjusted wv
diff:Sfc toi —800
mb b )7
ii
i
i ~ i~ ~i~ ■ l —
m ii■ ■■■ ■ ■ ■ i
r— r •■■in
Nov 97
— iNov■ ■95
Roin/SST—adjusted wv■ —
diff:800
to 600i mb
l
i
■■■
— i
i —i
mi
in r i
150
Longltud
-
2
-
1
0
cm
1
2
2
C l .Nov 97 — Nov 95 Roin/SST—adjusted wv diff:600 to 400 mb
7 40 r ’— '— i—" —
L ongltud*
— r^-i — ■— Tl
Nov 97 — Nov 95 Rai
justed wv diff:400 to 200 mb
7 40 m — >— i— ■— u m i m b
: 1
-2 0
lo n g iti.a
-
1.0
L ongltud*
-
0.20
-
0.10
0.00
cm
0.10
0.20
Figure 12. Layer integrated water vapor differences for November 1997 minus November 1995 after making
adjustments for local SST and local rainfall changes, a) Surface to 800 mb, b) 800 to 600 mb, c) 600 to 400
mb, d) 400 to 200 mb.
U>
-4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nov97—Nov95 Moiked SST—adjusted wv diff:Sfc to BOO mb
Ndv97-N ov95 Masked SST—adjusted wv diff:BOO to 600 m b
Longitude
Longitude
-1
0
cm
Nov97-Nav95 Masked SST—odiusted wv diff:600 to 400 mb
Nov97-Nov95 Masked SST—adjusted wv diff:400 to 200 mb
L ongitude
L ongitude
-
0.20
-
0.10
0.00
cm
0.10
0.20
Figure 13. Rain-masked, SST-adjusted layer integrated water vapor differences for November 1997 minus
November 1995. a) Surface to 800 mb, b) 800 to 600 mb, c) 600 to 400 mb, d) 400 to 200 mb Black
indicates a rainfall change o f over 250 mm/month, while deep red indicates a negative rainfall change o f 250
mm/month or greater. Other colors represent layer water vapor as indicated by the color bar.
U)
00
39
CHAPTER V
OBSERVATIONS AND CALCULATIONS OF
OUTGOING LONGWAVE RADIATION
5.1 Radiative Balance
The difference between longwave emission from the surface (uTs4 from the
Stefan-Boltzmann law, where cris the Stefan-Boltzmann constant o f 5.67x1 O'8 W m '2 K"4
and Ts is the surface temperature) and longwave emission at the top o f the atmosphere
(OLR) goes into heating the atmosphere and is known as the atmospheric greenhouse
effect, G„. All that is needed to calculate the greenhouse effect is the surface
temperature and OLR. The change in greenhouse effect with surface temperature,
dGa/dTs is a sensitivity parameter which is indicative o f the total feedback between
surface temperature and greenhouse effect. If this parameter is lower than about 2.2
W m‘2 K '1, there is a negative feedback which acts to regulate temperature. If it is higher
enough, there is a positive feedback which enhances temperature changes. This cutoff
was calculated by Inamdar and Ramanathan (1998) based on mean surface temperature
values and the predictions o f models with a fixed relative humidity (i.e. no water vapor
feedback).
We have seen by comparing w ater vapor profiles from 1995 and 1997 that the
water vapor was redistributed, with some areas becoming more humid and other areas
becoming drier. The actual effect o f these changes in the radiative balance o f the earth
can be ascertained by looking at the OLR changes.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
40
5.2 The MDTERP Longwave Model
The Maryland Terrestrial Radiation Package (MDTERP) was written by Dr. Ezra
Takara and Dr. Robert Ellingson o f the University o f Maryland. This program calculates
upward and downward radiances o f infrared radiation for 305 spectral intervals up to
wavenumber (2 tt/A.) 3000 cm"1. It assumes a plane parallel atmosphere, local
thermodynamic equilibrium, and a black surface. Gaseous absorption by water vapor,
carbon dioxide, ozone, methane, and nitrous oxide are calculated based on the 1992
high-resolution transmission (HITRAN) data base and fit to line by line calculations as
described in Warner and Ellingson (2000). Upward and downward radiances are
calculated at the four Gaussian angles (21.48, 47.93, 70.73, and 86.02 degrees) and nadir
(Ellingson and Gille, 1978).
MDTERP was used to calculate the OLR that would result from the observed
profiles o f temperature and water vapor. As a simplifying assumption, clouds were
ignored, since we are focusing on the effects o f water vapor. For a scenario with
enhanced convection in which the humid ascending region gets more humid and the dry
descending region gets drier (as Lindzen suggested), ignoring clouds will cause the
amplitude o f OLR decreases in the rising branch to be overestimated, since this region is
cloudy in both situations.
5.3 Clear-sky OLR Calculations and Observations
Average observed water vapor profiles (from the satellite retrievals) and
temperature profiles (from the ECMWF analyses used to initialize the retrievals) in 2.5°
by 2.5° grid boxes for each month were used as inputs to MDTERP. Default mixing
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
41
ratios o f carbon dioxide, methane and nitrous oxide and the standard ozone profile were
used. The input profiles were stored in 100 layers o f 200 meters height each. Figures 14
and 15 show the computed clear-sky OLR for November 1995 and 1997 respectively.
The observed changes o f clear-sky OLR for the same months are plotted in
Figures 16 and 17. These data are taken from the University o f Maryland HIRS data
archive. OLR is calculated from measurements at four spectral intervals using the
method o f Ellingson et al. (1989). This is in contrast to, e.g. the NOAA operational
OLR algorithm which infers OLR (generally defined as the total top-of-atmosphere
upward flux from 5 to 50 microns) from only one spectral interval (currently 10.5 to 11.5
microns) (Gruber and Krueger, 1984). Therefore, the Maryland OLR should capture
more o f the variability in OLR associated with water vapor changes. This is the clearsky product, which uses only pixels flagged as being cloud-free.
The overall patterns o f the calculated and observed clear-sky OLR
products are in agreement. In each o f these figures, the ITCZ is clearly visible as a band
o f low OLR with adjacent regions o f higher OLR in the subtropics. The observed clearsky OLR is greater, in general, than the calculated clear-sky OLR. Part o f this bias may
be due to cloud clearing errors in the observed OLR product. There will also be errors
due to averaging all profiles before computing OLR.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
100
150
_
200
Longitude
BBra-
220
240
280
OLR
Figure 14. Calculated clear-sky OLR for November 1995.
Latitude
Nov 9 7 C om puted OLR
-2 0
-4 0
100
220
150
240
.
200
Longitude
OLR
280
Figure 15. Calculated clear-sky OLR for November 1997
R ep ro d u ced with p erm ission of th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
43
Nov 9 5 Observed Clear Sky OLR
401
201
©
-O
3
"o
01_____________________________
g p iH H M M W H liB
~T t ?^33E &
*•
-4 0 1
100
22U
_________
150
2W
.
200
Longitude
OLR W a r t )
280
Figure 16. Observed clear-sky OLR for November 1995.
Nov 9 7 Observed Clear Sky OLR
©
3
T7
-2 0
-4 0
150
2+ 0
Longitude
200
EE
OLR
Figure 17. Observed clear-sky OLR for November 1997.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
44
Clear-sky OLR differences for November 1997 minus November 1995 are
plotted in Figures 18 (computed) and 19 (observed). In both OLR fields, drier areas
increase the amount o f OLR (i.e., the infrared radiation escaping to space), and
moistening reduces OLR, as expected. Amplitudes o f changes are generally larger in the
observed OLR. This systematic bias may be due to the larger maximum values (and
hence larger range) o f absolute observed OLR.
Note the increased OLR (by as much as 20 W m '2) in regions o f drying in the
eastern Pacific subtropics and the Equatorial Indian Ocean. It is important to note that
the changes in the w ater vapor field do show up in the OLR, which directly affects the
temperature by radiative forcing. The overall change in actual (all-sky) OLR in the
Pacific should be more negative than is suggested in either Figure 18 or 19, because the
mostly cloud-free subtropics will have little difference between clear-sky and all-sky
OLR, while the increases in clear-sky OLR in the ITCZ will be partially compensated
for by the effects o f clouds in the cloudy pixels.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Figure 18. Calculated clear-sky OLR for November 1997 minus November 1995.
Nov 9 7 —9 5 Observed Clear Sky OLR
401
20
4>
T?
3
'. 'i T T
.
r im ,
"-3S6*
. J>V
;.j
o
-2 0
-401
100
150
. 2 0 0
Longitude
250
OLR (W /m -2)
Figure 19. Observed clear-sky OLR for November 1997 minus November 1995.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
46
5.4 Mean SST and OLR Values
M ean values o f SST, aT s4 and OLR for November 1995 and November 1997 are
given in Table 5. These are given both for the entire region o f study and for the East
Pacific (east o f 180° longitude) only. The East Pacific has larger net changes because
the entire region o f study has partially compensating cooling and warming signals.
SST’s are used without land surface temperatures because o f the lower uncertainties
(about I °C) in SST (Chou, 1994). The sign o f the OLR change depends on whether the
observed or computed value is used. In any case, the OLR increase is less than the
increase in upward longwave flux at the surface (aTs4), indicating that the greenhouse
effect is larger for November 1997. Taken alone, this is suggestive o f a positive
feedback. However, given the uncertainties in these values as well as the unknown
effects o f changes in solar absorption and feedbacks other than water vapor, drawing any
conclusions from these mean values is difficult. The real value in these data is in the
patterns revealed in Figures 18 and 19.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
47
Table 5. Mean SST, cyT4, and OLR values for November 1995 and November 1997.
Whole area Whole area East Pacific East Pacific
Nov. 95
Nov. 97
Nov. 95
Nov. 97
Mean SST [°C]
23.39
23.66
23.33
24.04
Mean ctTs4 [W m '2]
440.4
442.2
439.0
443.5
Computed Clear-sky OLR
(Ocean) [W m '2]
267.1
265.4
267.8
267.2
Observed Clear-sky OLR
(Ocean) [W m '2]
277.0
277.1
276.9
277.1
Observed Clear-sky OLR
(Land and Ocean) [W m '2]
276.8
277.4
276.6
277.0
It is interesting that the clear-sky OLR values change very little compared to the
changes in upward infrared flux at the surface. In order to pursue this further, Tables 6
and 7 give mean values o f SST, ctTs4 and OLR for each November from 1994 to 1997
for the entire region o f study and the East Pacific, respectively. Observed OLR values
are from the HIRS data archive at the University o f Maryland. The observed clear-sky
OLR over ocean is remarkably constant from November 1994 to November 1997
compared to the changes in upward longwave flux at the surface. A larger dataset is
needed to determine whether this is coincidence and how this relates to the total OLR.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
48
Table 6. Mean SST, oT s4, and OLR values for the entire region o f study for four
Novembers. “NA” indicates data not available.
SST
[°C]
<crTs4>
[W m '2]
Observed
Clear-sky
Ocean OLR
[W m '2]
Observed
All-sky
Ocean OLR
[W m'2]
Observed
Clear-sky
OLR (Land
+ Ocean)
[W m'2]
Observed
All-sky
OLR (Land
+ Ocean)
[W m '2]
Nov. 1994
23.50 441.0
277.2
253.4
276.8
253.0
Nov. 1995
23.39 440.5
277.0
251.9
276.8
251.1
Nov. 1996
23.21 439.4
277.8
251.4
277.5
251.3
Nov. 1997
23.66 442.2
277.1
NA
277.4
NA
Table 7. Mean SST, <tTs4, and OLR values for the East Pacific region for four
Novembers. “NA” indicates data not available.
SST
[°C]
< aT s4>
[W m’2]
Observed
Clear-sky
Ocean OLR
[W m'2]
Observed
All-sky
Ocean OLR
[W m '2]
Observed
Clear-sky
OLR (Land
+ Ocean)
[W m’2]
Observed
All-sky
OLR (Land
+ Ocean)
[W m ‘2]
Nov. 1994
23.68 441.2
277.3
250.2
277.0
249.8
Nov. 1995
23.33 439.0
276.9
254.5
276.6
254.3
Nov. 1996
23.22 438.5
277.7
254.6
277.3
254.2
Nov. 1997
24.04 443.5
277.1
NA
277.0
NA
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
49
CHAPTER VI
SUMMARY AND CONCLUSIONS
6.1 Conclusions
Water vapor profiles, retrieved from SSM/T-2 microwave observations by a
physical retrieval method, together with temperature observations for the months o f
November 1995 and November 1997 were used. These observations reveal that
warming at the Equator is accompanied by moistening in the middle and upper
troposphere directly above the warming and a drying in the adjacent subtropical
subsidence zones o f the Hadley circulation. This is consistent with Lindzen’s model o f
enhanced convection caused by wanning.
Adjustments to the water vapor field were made by subtracting out the part o f the
field which could be explained by changes in local SST and rainfall. Moistening in the
ITCZ and drying in the subtropics in response to ITCZ warming were still observed.
Local changes were not sufficient to explain the changes in water vapor; large-scale
dynamic effects are involved.
These observed changes demonstrate that the mechanism for drying the upper
troposphere as proposed by Lindzen (1990) does operate in the real atmosphere for a
regional warming. Even though a regional wanning does not reflect the expected
temperature pattern o f global warming, these observations can be used to verify whether
models can reproduce this mechanism. In particular, models should be able to reproduce
the observed drying in the upper and middle troposphere o f the subtropical East Pacific
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
50
during El Nino if they are to be expected to accurately model the effects o f the water
vapor feedback.
Current climate models predict a positive feedback due to water vapor consistent
with that found by holding atmospheric relative humidity constant (Cess et al., 1990).
The results o f this study suggest that the mechanism proposed by Lindzen which causes
a negative water vapor feedback is active. While the overall sign o f the water vapor
feedback is uncertain from this limited data set, this effect may serve to reduce the total
positive water vapor feedback if it does not completely counteract it.
6.2 Future Work
This study could be expanded to include data from a longer time period and a
larger portion o f the globe. This study only looked at water vapor over the ocean; the
behavior o f water vapor over land may be different. In addition, climate model
predictions could be compared to observations such as to see how well the models
reproduce the observed drying in the upper troposphere subtropics during an El Nino
event.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
51
REFEREN CES
A ir Force Cambridge Research Laboratory, 1965: Handbook o f Geophysics and Space
Environments. McGraw-Hill, 432 pp.
Al-Khalaf, A. K., 1995: Retrieval o f atmospheric water vapor profiles from the special
sensor microwave temperature-2 (SSM/T-2). Ph.D. Dissertation, Texas A&M
University, 145 pp.
Arrhenius, S., 1896: On the influence o f carbonic acid in the air upon the temperature o f
the ground, Philos. Mag., 41, 237-276.
Blankenship, C. B., 1997: Retrieval o f water vapor profiles using SSM/I and SSM/T-2.
M. S. Thesis, Texas A&M Univ., 76 pp.
Blankenship, C. B., A. K. Al-Khalaf, and T. T. Wilheit, 2000: Retrieval o f water vapor
profiles using SSM/T-2 and SSM/I data. J. Atmos. Sci., in press.
Cess, R. D., G. L. Potter, J. P. Blanchet, G. J. Boer, A. D. Delgenio, et al., 1990:
Intercomparison and interpretation o f climate feedback processes in 19
atmospheric general circulation models. J. Geophys. Res., 95, 16601-16615.
Chaboureau J.-P., A. Chedin, and N. A. Scott, 1998: Relationship between sea surface
temperature, vertical dynamics, and the vertical distribution o f atmospheric water
vapor inferred fromTOVS observations,/. Geophys. Res., 103, 23173-23180.
Chou, M.-D., 1994: Coolness in the tropical Pacific during an El Nino episode. J.
Climate. 7, 1684-1692.
Ellingson, R. G., and J. C. Gille, 1978: An infrared radiative transfer model: Part I Model description and comparison o f observations with calculations. J. Atmos.
Sci., 35,523-545.
Ellingson, R. G., D. J. Yanuk, H-T. Lee, and A. Gruber, 1989: A technique for
estimating outgoing longwave radiation from HERS radiance observations. /.
Atmos. Ocean. Tech., 6, 706-711.
Falcone, V. J., M. K. Griffin, R. G. Isaacs, J. D. Pickle, J. F. Morrissey et al., 1992:
DMSP FI 1 SSM/T-2 calibration and validation data analysis, Phillips
Laboratory, Hanscom Air Force Base, MA, 432 pp.
Gruber, A. and A. F. Krueger, 1984: The status o f the NOAA outgoing longwave
radiation data set. Bull. Amer. Meteor. Soc., 65,958-962.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
52
Hall, A. and S. Manabe, 1999: The role o f water vapor feedback in unperturbed climate
variability and global warming. J. Climate, 12,2327-2346.
Houghton, J. T., G. J. Jenkins, and J. J. Ephraums, Eds., 1990: Climate Change: The
IP C C Scientific Assessment. Cambridge University Press, 366 pp.
Houghton, J. T., L. G. Filho, B. Callander, N. Harris, A. Kattenburg, and K. Maskell,
Eds., 1996: Climate Change 1995: The Science o f Climate Change. Cambridge
University Press, 572 pp.
Inamdar, A. K. and V. Ramanathan, 1998: Tropical and global scale interactions among
w ater vapor, atmospheric greenhouse effect, and surface temperature. J.
Geophys. Res., 103, 32,177-32,194.
Lau, K..-M., C.-H. Ho, and M.-D. Chou, 1996: Water vapor and cloud feedback over the
tropical oceans: Can we use ENSO as a surrogate for climate change? Geophys.
Res. Lett., 23, 2971-2974.
Lindzen, R. S., 1990: Some coolness concerning global warming. Bull. Amer. Meteor.
Soc., 71, 288-299.
Lindzen, R. S., 1997: Can increasing carbon dioxide cause climate change?, Proc. Natl.
Acad. Sci. USA 94, 8335-8342.
Lindzen, R. S., B. Kirtman, D. Kirk-Davidoff, and E. K. Schneider, 1995: Seasonal
surrogate for climate. J. Climate., 8, 1681-1684.
Manabe, S., and R. T. Wetherald, 1967: Thermal equilibrium o f the atmosphere with a
given distribution o f relative humidity. J. Atmos. Sci., 24, 241-259.
Meeks, M. L., and A. E. Lilley, 1963: The microwave spectrum o f oxygen in the earth’s
atmosphere. J. Geophys. Res., 68, 1683-1703.
Moller, F., and S. Manabe, 1961: Uber das Strahlungsgleichgewicht der Atmosphare, Z
fu r Meteorol., 15, 3-8.
Schaerer, G., and T. T. Wilheit, 1979: A passive microwave technique for profiling o f
atmospheric water vapor. Radio Sci., 14, 371-375.
Serke, D. J., 1996: Evolution o f the southeast Pacific ITCZ in boreal spring as viewed
from SSM/I and SSM/T-2. M. S. Thesis, Texas A&M Univ., 159 pp.
Soden, B. J., 1997: Variations in the tropical greenhouse effect during El Nino. J.
Climate, 10, 1050-1055.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
53
Spencer, R. W., and W. D. Braswell, 1997: How dry is the tropical free troposphere?
Implications for global warming theory. Bull. Amer. Meteor. Soc., 78, 10971106.
Stephens, G.L., 1990: On the relationship between water vapor over the oceans and sea
surface temperature. J. Climate, 3,634-645.
Sun, D.-Z. and A. H. Oort, 1995: Humidity-temperature relationships in the tropical
troposphere. J. Climate, 8, 1974-1987.
Wallace, J. M. and P. V. Hobbs, 1977: Atmospheric science: An introductory survey.
Academic Press, 467 pp.
Warner, J., and R. G. Ellingson, 2000: A narrowband longwave radiation model based
on parameters fitted to LBLRTM. J. Atmos. Sci. (in press)
Wilheit, T. T., 1990: An algorithm for retrieving water vapor profiles in clear and
cloudy atmospheres from 183 GHz radiometric measurements: Simulation
studies. J. Appl. Meteor., 29, 508-515.
Wilheit, T. T., and A. K. Al-Khalaf, 1994: A simplified interpretation o f the radiances
from SSM/T-2. Meteorol. Atmos. Phys, 54, 203-212.
Wilheit, T. T., A. T. C. Chang, and L. S. Chiu, 1991: Retrieval o f monthly rainfall
indices from microwave radiometric measurements using probability distribution
functions. J. Atmos. Oceanic Technol., 8, 118-136.
Yang, H. and K. K. Tung, 1998: Water vapor, surface temperature, and the greenhouse
effect— a statistical analysis o f tropical-mean data, J. Climate, 11,2686-2697.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
54
APPENDIX
WATER VAPOR PROFILING ALGORITHM
1. Relative Humidity Sensitivity Functions
Given the physical state o f the atmosphere and the surface, it is straightforward
to compute the observed brightness temperature (Equation A l, below). We are
interested in the inverse problem o f finding the atmospheric state from a set of observed
brightness temperatures. We approach the water vapor profiling problem by using an
observed temperature profile and surface temperature (interpolated from the ECMWF
analyses.). We are interested in solving for the moisture profile. The oxygen content o f
the atmosphere is constant (except for a small correction due to the humidity). We will
use the retrieved humidities to constrain the amount and distribution o f cloud liquid
water. Ice particles in clouds will be ignored. Furthermore we will assume a field o f
view with no precipitation and realize that significant amounts o f precipitation will lead
to an inability to retrieve the correct moisture profile. With these constraints, we can
then solve for a set o f retrieved humidities and use it to construct a water vapor profile.
The theory o f water vapor profiling using the 183.31 GHz line was initially
developed by Schaerer and Wilheit (1979). The upwelling brightness temperature at a
frequency v observed by a microwave radiometer in space, assuming a specular surface
and making the Rayleigh-Jeans approximation, can be expressed as:
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
55
«
00
r f l t (o o ) =
«
~ ) 7 (h')dh
y (h )e h
A
-jr(h')dh'
y(h)dh + e
*
-J/(A')rfA'
R^T(Ji)e 0
°
y(h)dh
o
2
-\y(h)dh
+ e 0
(l- /? ) r ,+
e 0
(A l),
where y is the absorption coefficient o f the atmosphere at height h and frequency v, T the
thermodynamic temperature o f the atmosphere at height h, Ts the surface temperature,
and TCb is 2.7 K, the brightness temperature o f the cosmic background radiation. This
equation may be expressed in the form (Meeks and Lilley, 1963):
30
Tb1 = \ K v(h)T(h)dh + A
0
(A2),
where A and K(v,h) are substantially independent o f the temperature profile. These Kv
are weighting functions which are used for retrieval o f atmospheric temperature profiles.
We can express equation A2 in an analogous form, solving for y(h) rather than
T(h) (since the amount o f water vapor (or any absorber) affects the observed T at
through y):
09
Tat = \G '(h)r(h)dh + D
0
(A3),
where D represents terms independent o f y. Unfortunately, G '(h) is a function o f y at all
heights. We can linearize the problem by assuming a humidity profile and dealing with
small perturbations to it. The change in brightness temperature due to a small change in
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
56
the absorption as a function o f height, 5y(h), can be expressed as:
«57;t =
J c ? (A )
Sr(h )d h
(A4).
0
The change in brightness temperature is equal to the integral o f the change in absorption
coefficient times a sensitivity function, G(h). This sensitivity function can be derived to
be:
G ( h ,y ( h ) , T ( h ) ) = e
-jjlh)dh
f y(h)dh
J0
Q
- ( y(h)dh
»0
(A5),
where Tet(h) and Tei(h) are the upwelling and downwelling brightness temperatures,
respectively, at height h. The first term o f this function gives the effect o f the difference
o f the thermodynamic temperature o f the atmosphere at a height h from the upwelling
brightness temperature at that level, and includes effects o f attenuation due to absorption
between h and the top o f the atmosphere. The second term o f G(h) gives the
corresponding effect o f the downwelling radiation, including attenuation from h down to
the surface and back to the top o f the atmosphere, taking the surface reflectivity into
account. Note that this function makes sense physically. For example, at a level where
the brightness temperature o f the upwelling radiation is equal to the thermodynamic
temperature o f the atmosphere, the first term o f the sensitivity function will be zero, for
changes in the amount o f water vapor at that height will have no direct effect on the
upwelling radiance.
These sensitivity functions give the effect o f a change in optical depth on the
observed brightness temperatures. They can be reformulated w ith respect to relative
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout perm ission.
57
humidity by:
G
'w'«i)WiWi"GW
Wip-w
< A 6 ) -
These new sensitivity functions relate the brightness temperature change to a relative
humidity change by:
A7flt = j G K(h)AR(h)dh
(A7).
Given profiles o f temperature and absorbing constituents (including the water
vapor) and surface temperature and reflectivity, we can calculate the GR(h). The desired
AT's are the differences in the observed and calculated brightness temperatures. From
this information, we can solve for a set o f AR's which satisfy Equation A7.
Al-Khalaf (1995) calculated sensitivity functions for various frequencies and
atmospheric temperature and humidity profiles over land, ocean, and mixed
backgrounds. Sensitivity functions for one particular atmosphere are shown in Figure
A l. Note that channels near the 183.3 GHz peak (where the absorption is large) have
peaks higher in the atmosphere than the other channels. As water vapor is added to the
atmosphere, the peaks o f all channels will shift upwards. Also, the shape o f the 150
GHz sensitivity function will start to resemble the shape o f the other channels' functions
by crossing over to negative values at intermediate altitudes. Conversely, for drier
atmospheres, the functions for channels near the line center will begin to cross over to
positive values for low altitudes. It is these nonlinearities in the sensitivity functions
which necessitate an iterative approach to the retrieval.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
58
18-
161412
O)
-
10- 183+/-1
183+/-3
6
-
4 - 183+/-7
2
150
-
-15
-5
5
15
25
Sensitivity Functions (K/km or cm/km)
Figure A l. Relative humidity sensitivity functions for 4 SSM/T-2 channels and the total
integrated water vapor pseudochannel for a US standard atmosphere with constant RH o f
50%. Sensitivity functions (SF’s) for the SSM/T-2 channels are expressed in units o f K
k m '1. The SF for TIWV is expressed in units o f centimeters o f water per kilometer (cm
k m '1), and has been doubled.
2. The Physical Relaxation Method
The upwelling radiances at the top o f the atmosphere can be computed using
Equation A l given sufficient information about the atmospheric temperature profile and
concentrations o f absorbers, the surface temperature, and the surface reflectivity for each
frequency. We are interested in solving the inverse problem or retrieval problem: what
water vapor profile could have produced a set o f observed radiances?
In the physical relaxation method, a first guess humidity profile is chosen. The
resulting radiances are computed and compared with the observed radiances. If they do
R ep ro d u ced with p erm ission of th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
59
not match closely enough, a new profile is chosen using the sensitivity functions in the
previous section. The process is then repeated with the new profile until the observed
and computed (retrieved) brightness temperatures agree closely enough.
3. Retrieval Algorithm
a. Algorithm overview
The retrieval algorithm used is based on the model o f Al-Khalaf (1995). The
humidity profile is represented by an array o f relative humidities {Rj} at a set o f five to
ten levels. RH is assumed to be linear with height between these levels and constant
below the lowest level. The RH at 20 km is fixed at zero with a linear interpolation to
the highest retrieved level. Note that since the temperature profile is held constant
throughout the retrieval, specifying the relative humidity is equivalent to specifying any
other moisture parameter, such as mixing ratio.
A flowchart for this algorithm is shown in Figure A2. For an SSM/T-2
observation over ocean, a known temperature profile is taken from ECMWF analyses. A
first guess humidity profile is obtained using the method o f Section 3.c o f the appendix.
Brightness temperatures corresponding to this atmosphere are calculated for each o f the
channels. These are compared to the observed brightness temperatures by calculating a
normalized brightness temperature error C =
, where M is the number
o f channels used, Tjret and T ° s are the retrieved and observed brightness temperatures,
respectively, for the i'th channel, and a; is the NEAT (noise equivalent delta-temperature)
for the i'th channel.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
60
Known temperature profile
First guess humidity profile
Radiative transfer calculation of
brightness temperature
Check for
Yes
Cj<.l or j=25
C=min {Cj}
No
No
Yes
j=j+l
Update humidity profile,
Is C<1 ?
minimizing E
Check for supersaturation
Retrieval is “good’
(Use profile with
Retrieval is
“bad”
and add clouds
best Cj)
End
Figure A2. Flowchart for the retrieval algorithm.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
61
If C<1, the measurements agree with the observations, on average, to within the
noise level o f the instrument, and the humidity profile is taken to be correct. If not, a
new moisture profile is calculated solving for a set o f changes to the relative humidities
8Rj which minimizes the error E in
(A8),
where N is the number o f retrieved levels, bj is the standard deviation o f relative
humidity for the j'th level, Rj is the retrieved relative humidity for the j'th level, and
Rj is the mean relative humidity for the j'th level. If the new humidity profile is
supersaturated at any level, clouds are added to the model atmosphere as described in
Section 3.d o f the appendix. Relative humidity statistics were derived by Al-Khalaf
(1995) from a global dataset o f radiosondes.
At this point, the formal uncertainty o f each component o f5 R (and R) can be
determined by taking the root sum o f squares o f the appropriate terms in the equation
which minimizes E. If the formal uncertainty at a level is not significantly less than the
uncertainty o f the relative humidity constraint (bj), the measurements are not
contributing much information about a level (Wilheit, 1990).
Next, the convergence criterion C is then computed for the new humidity profile
and the process is repeated. Note that while the new humidity profiles are calculated
using both the observations and the statistical constraints, the convergence criterion
depends strictly on the agreement between retrieved and observed brightness
temperatures. The statistical constraints are used to guide the choice o f the next guess,
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
62
but in determining the best profile, we use the physical error model without any
statistics. If C<1, the retrieval is considered to match the observations to within the
noise level o f the instrument. C is stored for each iteration and the profile with the
lowest C is used. If C never falls below unity, the retrieval is reported as failing to
converge. The yield o f a retrieval method is defined as the percentage o f cases which
have a retrieved profile with C<1.
b. Retrieval heights
Because the level where a given sensitivity function peak occurs can vary from
one observation to another, it is difficult to know a priori for which levels the brightness
temperatures provide meaningful humidity information. This problem has been treated
by making the retrieval levels variable.
Wilheit and Al-Khalaf (1994) found that for each SSM/T-2 channel near 183
GHz, there is a nearly constant overburden (the integrated water vapor amount above a
given level) above the height at which the atmospheric temperature equals the observed
brightness temperature. This was the basis for an algorithm giving, for each o f those
channels, a height and a corresponding humidity.
The lowest retrieval level is set to 2 km below the lowest level from this single­
height, single-channel algorithm, with the restriction that it be between 1 and 2 km. The
upper bound o f 2 km is used to ensure that there are retrieval levels low in the
atmosphere where the TIWV sensitivity function (which is not used in the single­
channel, single-height algorithm) peaks, while the lower bound o f 1 km is used because
the algorithm can not resolve features below about 1 km anyway. The highest retrieval
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
63
level is set to 2 km above the highest level from the single-height, single-channel
algorithm, with the restriction that it does not exceed 11 km. A constant spacing o f 1 km
between levels is used, adjusting the height o f the highest retrieval level upward as much
as 1 km to achieve this spacing. This method results in from five to ten retrieval levels.
c. The first guess profile
Al-Khalaf (1995) used a global dataset o f radiosonde observations to provide a
climatological mean humidity profile for ocean, land, snow, and mixed backgrounds.
The mean profile for ocean is used as a basis for the first guess profile by initializing the
values o f R at each o f the levels. The information from the single-height single-channel
algorithm o f Wilheit and Al-Khalaf is then added for each o f the three channels in turn
by adding a Gaussian-shaped distribution to the RH profile, as follows:
R fg -n e » =
R f g - o ,J
+ ^
* _ /? ( /,* ) ] e x p
(hj-h*^ 2’
I 2.0 J
(A9),
where R ^ -0'11and Rj,g'new are the first guess humidities at level j before and after taking
the single-height single-channel algorithm into account, h* and R* the height and
relative humidity returned by the single-channel single-height algorithm, hj the height at
level j, R(h*) the relative humidity at h* interpolated from the Rj, and 2.0 is an arbitrary
scale factor. The effect o f this is to give a first guess profile which is has an RH close to
the values from the single-channel height algorithm at the appropriate levels, but which
approaches the climatological RH away from those levels.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
64
d. Clouds
If the new humidity profile has any supersaturated levels, cloud liquid water is
added to the model atmosphere at those levels. Because the effect o f cloud liquid water
is similar to that o f excess water vapor, this will cause a lower amount o f water vapor to
be retrieved. The relative humidity at each supersaturated level is further constrained by
changing the statistical constraint at that level to RH=100%±1%. This is realistic
because in the presence o f clouds, the relative humidity must be very close to saturation.
The addition o f cloud liquid water continues until the retrieved relative humidity at that
level is within the constraints.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
65
VITA
Clay Bruce Blankenship was bom in 1971 in Tuscumbia, Alabama. He received
a Bachelor o f Science in Physics from Auburn University in 1993. He received a Master
o f Science in M eteorology from Texas A&M University in 1997 and a Doctor o f
Philosophy in Atmospheric Sciences from Texas A&M in May 2000. He was married to
Andrea Leigh Morris in July 1997. Clay plans to study under a postdoctoral fellowship
at the Naval Research Laboratory in Monterey, California beginning in June 2000.
Correspondence may be addressed to a continuing email address,
snoweel@yahoo.com or via U.S. mail, c/o Bruce Blankenship, 190 Olive Drive,
Tuscumbia, AL 35674.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
2 731 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа