close

Вход

Забыли?

вход по аккаунту

?

Particulate Optical Depth Retrievals over the Ocean from Collocated Lidar and Microwave Satellite Observations

код для вставкиСкачать
PARTICULATE OPTICAL DEPTH RETRIEVALS OVER THE OCEAN FROM
COLLOCATED LIDAR AND MICROWAVE SATELLITE OBSERVATIONS
by
Qiang Tang
A DISSERTATION
Submitted to the Faculty of the Stevens Institute of Technology
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Qiang Tang, Candidate
ADVISORY COMMITTEE
Knut Stamnes, Chairman
Date
Alan Blumberg, Chairman
Date
Yongxiang Hu
Date
Wei Li
Date
Ting Yu
Date
Julie Pullen
Date
Nickitas Georgas
Date
STEVENS INSTITUTE OF TECHNOLOGY
1 Castle Point Terrace
Hoboken, NJ 07030
2017
ProQuest Number: 10682263
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
ProQuest 10682263
Published by ProQuest LLC (2018 ). Copyright of the Dissertation is held by the Author.
All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
c
2017,
Qiang Tang. All rights reserved.
iii
PARTICULATE OPTICAL DEPTH RETRIEVALS OVER THE OCEAN FROM
COLLOCATED LIDAR AND MICROWAVE SATELLITE OBSERVATIONS
ABSTRACT
Retrievals of Aerosol/Cloud Optical Depth (AOD/COD) from backscatter measurements of the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument deployed on the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite rely on a single global mean extinction-to-backscatter
ratio, also known as the lidar ratio. However, the lidar ratio depends on the microphysical properties of the particulates. It has been shown that on a global basis
there is an uncertainty of about 20-30% in lidar ratios adopted in CALIOP retrievals,
which leads to large uncertainties in AOD/COD retrievals. In this study, we take
advantage of near simultaneous (almost “same” time and location) lidar (CALIOP)
and microwave (wind speed) observations to develop a new approach, that does not
rely on an assumed lidar ratio, to infer AOD/COD over the open ocean. Instead
the AOD/COD is inferred directly from backscatter measurements obtained from the
CALIOP lidar in conjunction with collocated sea surface wind speed data obtained
from the Advanced Microwave Scanning Radiometer (AMSR). This new method is
based on an ocean surface reflectance model relating wind-driven wave slope variances
to sea surface wind speeds. To properly apply this method, the impact of multiple
scattering between the ocean surface and the particulates should be taken into account. We take advantage of the 532 nm cross polarization feature of CALIOP and
introduce an empirical method based on the depolarization change at the ocean surface to correct for potential bias in ocean surface backscatter caused by whitecaps,
bubbles, foam, and multiple scattering. After the correction, AOD/COD values are
iv
derived with this method for individual CALIOP profiles obtained over the ocean.
From a comparison of AOD/COD retrievals obtained by our new method and by
other available retrieval methods, we found that our method can be used to improve
the standard CALIOP products and be applied to future space-borne active remote
sensor data.
Author: Qiang Tang
Advisor: Knut Stamnes
Date: November 27, 2017
Department: Physics, and Civil, Environmental and Ocean Engineering
Degree: Doctor of Philosophy
v
To my parents, for all their support and putting my through the best education.
To my wife Xiaoli and son Boyan, for their unending love and sacrifices.
vi
Acknowledgments
First of all, I would like to express my deepest gratitude to Prof. Knut Stamnes
for offering me the opportunity to work under his guidance and support. I am very
grateful for his encouragement and support throughout the years.
I would also like to thank Prof. Yongxiang Hu and Prof. Wei Li for their
continuous support, valuable insights and discussion. I would also like to express
thanks to my committee members, Prof. Alan Blumberg, Prof. Ting Yu, Prof. Julie
Pullen, and Prof. Nickitas Georgas for their inputs and discussion.
I would also like to express my gratitude to my wonderful colleagues. Many
thanks go to LLLabers - Dr. Yongzhen Fan, Dr. Nan Chen, Dr. Zhenyi Lin, Dr.
Lingling Fan, Dr. Min He, Dr. Snorre Stamnes, Dr. Denis Cohen, and Dr. Monica
Sikand, etc., for their helpful support and discussion.
Special thanks are extended to all individuals and organizations that provided
the data and tools used in this study, especially the CALIPSO and MODIS teams.
Finally, I would like to sincerely thank my parents Xiu Ping and Renxin Tang,
and my wife Xiaoli Guan for their generosity, love and affection.
vii
Table of Contents
Abstract
iii
Dedication
v
Acknowledgments
vi
List of Tables
x
List of Figures
xi
1 Introduction
1
1.1
Earth’s radiation budget, ice cloud/aerosol climate impacts
1
1.2
CALIPSO and CALIOP Introduction
8
1.3
Limitation of a Standard Elastic Lidar System like CALIOP
9
2 Cirrus Optical Depth (COD) over the Ocean obtained from Collocated CALIPSO and AMSR-E Observations
13
2.1
Research Background and Introduction
13
2.2
Analysis and Methodology
15
2.2.1
CALIOP Attenuated Backscatter Measurements
15
2.2.2
Ocean Surface Reflectance Model
18
2.2.3
The U -σ 2 Relationship and Pure Integrated Ocean Surface Backscat-
2.3
ter
20
2.2.4
Ice Cloud Optical Depth Retrievals
23
2.2.5
Lidar Ratio Retrievals from ocean surface wind
26
Results and Discussions
27
viii
2.3.1
Correction for backscatter due to “junk”
2.3.2
Comparison of Ice Cloud Optical Depths with CALIPSO L-2
2.3.3
2.4
27
Data
30
Adjustment of lidar ratios for CALIOP level 2 data
34
Conclusion
37
3 Column Aerosol Optical Depth Evaluation over Open Ocean with
collocated CALIPSO-MODIS-AERONET Measurements
38
3.1
Research Background and Introduction
38
3.2
Methodology
40
3.2.1
CALIOP aerosol algorithm and product
40
3.2.2
MODIS aerosol algorithms and products
45
3.2.3
Aerosol retrievals from CALIPSO-AMSR co-measurements
46
3.2.4
Multi-Layer Neural Network based on AccuRT
48
3.2.5
AERONET aerosol algorithm and product
49
3.3
3.4
Results and Discussions
50
3.3.1
Coastal Ocean Comparison
50
3.3.2
Open Ocean Comparison
54
3.3.3
Open Ocean Regional Seasonal Variations
58
3.3.4
AERONET Comparison
66
Conclusions
4 Summary and Conclusion
70
72
4.1
Conclusions
72
4.2
Future Directions
73
Bibliography
75
ix
Vita
93
x
List of Tables
1.1
Characteristics of CALIOP
9
xi
List of Figures
2.1
Relationship between sea surface wind speed and (i) slope variance of
wind-driven waves (right) and, (ii) ocean surface integrated backscatter
(left).
2.2
22
Upper panel: 2008 August daytime 1064 nm lidar ratios derived from
collocated CALIPSO and AMSR-E data. Note that the lidar ratio
remains fixed throughout the optical depth ranging from 0.5-3. Lower
panel: Zonal mean distribution of ice cloud lidar ratios. The mean
value is 31.7 sr, with a standard deviation of ±5.7 sr.
2.3
24
Upper panel: 2008 August daytime 532 nm lidar ratios derived from
collocated CALIPSO and AMSR-E data. Note that the lidar ratio
has little variation throughout the optical depth ranging from 0.5-3.
Lower panel: Zonal mean distribution of ice cloud lidar ratios. The
mean value is 32.3 sr, with a standard deviation of ±6.7 sr.
2.4
25
Relationship between ocean surface wind speed and CALIPSO 1064
nm attenuated backscatter. The yellow and green curves are based on
the empirical backscatter-wind relations proposed by Cox-Munk and
Wu. The underestimate of backscatter from the empirical relations is
due to contributions from“junk” in the water and sub-surface scattering. 28
2.5
By correcting CALIPSO observations for ”junk” backscatter, a good
match is obtained between the backscatter-wind relations and CALIPSO
measurements.
29
xii
2.6
Upper panel: Global distribution of cirrus optical depth at 1064 nm
derived from collocated CALIPSO - AMSR-E observations with corresponding cirrus optical depth from CALIPSO L-2 data. Lower panel:
Global distributions of cirrus optical depth at 1064 nm from both the
collocated CALIPSO - AMSR-E observations and CALIPSO level 2
data. Cirrus optical depths from our new approach generally have
very similar distribution pattern as the CALIPSO standard product,
which indicates that our approach is reliable.
2.7
31
Upper panel: Global distribution of cirrus optical depth at 532 nm
derived from collocated CALIPSO - AMSR-E observations with corresponding cirrus optical depth from CALIPSO L2 data. Lower panel:
Global distributions of cirrus optical depth at 532 nm from both the
collocated CALIPSO - AMSR-E observations and CALIPSO level 2
cloud data. Cirrus optical depths from our new approach generally
have very similar distribution pattern as the CALIPSO standard product, indicating that our approach is reliable.
2.8
32
Upper panel: August 2008 daytime comparison of ice cloud optical
depth (1064 nm on the left, 532 nm on the right) from CALIPSO L2
and CALIPSO–AMSR. Lower panel: Zonal mean distribution of the
optical depths from two sources.
2.9
33
August 2008 comparison of ice cloud optical depth for 1064nm from
revised CALIPSO L2 product and CALIPSO–AMSR. Lower panel:
3.1
Same comparison for 532 nm data.
36
Open ocean region selected for study.
44
xiii
3.2
Left Panel: MODIS RGB image with collocated CALIOP track labeled by the green dots on March 15th, 2007 over the Yellow sea area.
Right Panel: The corresponding CALIOP 532 nm total attenuated
backscatter image; the colorbar represents the strength of the attenuated backscatter.
3.3
50
Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset
for the corresponding collocated coastal case; Lower Panel: AOD retrieved from different algorithms: CALIOP (black dots), MODIS ocean
color(blue dots), MLNN (red dots) and OSRM (green dots).
3.4
51
Left Panel: MODIS RGB image with collocated CALIOP track labeled
by the green dots on March 1st, 2008 over Yellow sea area; Right Panel:
The corresponding CALIOP 532 nm total attenuated backscatter image, the colorbar represents the strength of the attenuated backscatter. 52
3.5
Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset
for the corresponding collocated coastal case. Lower Panel: AODs
retrieved from different algorithms: CALIOP (black dots), MODIS
ocean color (blue dots), MLNN (red dots) and OSRM (green dots).
3.6
53
Left Panel: MODIS RGB image with collocated CALIOP track marked
by the green dots on Janurary 1, 2016 over the Atlantic ocean close to
West Africa. Right Panel: The corresponding CALIOP 532 nm total
attenuated backscatter image; the colorbar represents the strength of
the attenuated backscatter.
3.7
54
Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset
for the corresponding collocated coastal case; Lower Panel: AOD retrieved from different algorithms: CALIOP (black dots), MODIS ocean
color(blue dots), MLNN (red dots) and OSRM (green dots).
55
xiv
3.8
Left Panel: MODIS RGB image with collocated CALIOP track labeled
by the green dots on January 19th, 2016 over Atlantic ocean close to
West Africa. Right Panel: The corresponding CALIOP 532 nm total
attenuated backscatter image; the colorbar represents the strength of
the attenuated backscatter.
3.9
56
Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset
for the corresponding collocated coastal case; Lower Panel: AOD retrieved from different algorithms: CALIOP (black dots), MODIS ocean
color(blue dots), MLNN (red dots) and OSRM (green dots).
57
3.10 Mean AOD distribution from different algorithms in the selected open
ocean area, January 2016.
59
3.11 Zonal mean of AODs and relative differences between each algorithm,
January 2016.
60
3.12 AOD probability distribution according to each algorithm for January
2016. The mean AODs are 0.12 (CALIPSO), 0.21 (MODIS), 0.23
(OSRM), and 0.23 (MLNN), respectively.
60
3.13 Mean AOD distribution from different algorithms in the selected open
ocean area, April 2016.
62
3.14 Zonal mean of AODs and relative differences between each algorithm,
April 2016.
3.15 AOD probability distribution according to each algorithm, April 2016.
63
63
3.16 Mean AOD distribution from different algorithms in the selected open
ocean area, July 2016.
64
3.17 Zonal mean of AODs and relative differences between each algorithm,
July 2016.
65
xv
3.18 AOD probability distribution according to each algorithm, July 2016l,
with mean AODs of 0.15, 0.2, 0.17, and 0.23, respectively.
65
3.19 Mean AOD distribution from different algorithms in the selected open
ocean area, October 2016.
66
3.20 Zonal mean of AODs and relative differences between each algorithm,
October 2016.
67
3.21 AOD probability distribution according to each algorithm, October
2016. The mean AODs are 0.11, 0.20, 0.20, and 0.19, respectively.
3.22 Ten ocean AERONET sites selected for study.
67
69
3.23 Comparisons of AODs derived by different algorithms with those from
AERONET. The slope of the linear regression of each comparison is
given in the subgraph’s title.
70
1
Chapter 1
Introduction
1.1
Earth’s radiation budget, ice cloud/aerosol climate impacts
Numerous publications show that the Earth is undergoing climate change (e.g., Intergovernmental Panel on Climate Change (IPCC) fourth and fifth assessment reports [40, 88]). Climate change has a vital impact on community, vegetation and
animal life, and the better preparation the community can get for the future climate,
the quicker and more effective mitigation strategies can be put in place. Climate
models are invaluable tools for predicting changes of the climate. Unfortunately, fully
representing the natural processes that drive climate is very difficult due to their complexity. The individual components in the climate system affect each other, and there
are numerous components and processes in the atmosphere, oceans, cryosphere (ice),
biosphere (vegetation), and lithosphere (the ground) that must be accounted for. We
could gain a valuable understanding of climate change through the Earth’s radiation
budget.
The Earth’s surface temperature is determined by the balance between incoming solar radiation and outgoing infrared radiation. If the Earth had no atmosphere,
the average temperature over the Earth surface, i.e., when the incoming solar energy
and the outgoing energy radiated from the Earth back to space are equal, would be
about -18◦ C. However, due to the natural greenhouse effect that takes place in the
atmosphere, the average temperature goes up to 15◦ C [39]. The greenhouse effect is
caused by so-called GreenHouse Gases (GHG), which are molecular species that absorb radiation emitted from the warmer surface and re-emit or re-distribute radiation
2
at colder temperatures. This process, traps the energy within the Earth system. The
most important greenhouse gases are water vapor (H2 O), followed by carbon dioxide
(CO2 ), methane (CH4 ), and ozone (O3 ) [39]. The radiation emitted upwards from
the altitude where there is not enough absorbing greenhouse gases to absorb the upward radiation, is lost to space. This mechanism has a warming effect on the Earth’s
surface because the atmospheric temperature decreases with height (negative lapse
rate) within the troposphere, where nearly all the greenhouse gases are located. The
amount of energy from the emission altitude corresponds to its temperature, which
is less than the surface, and therefore less energy is lost to space, compared to if the
surface radiation leaves the Earth system without being absorbed [26].
By the same mechanisms, other absorbers/emitters in the atmosphere, such as
clouds and aerosols, could also have a warming/cooling effect on climate. Ice cloud is a
colloid of ice particles dispersed in air, typical ice crystals that compose ice clouds are
hexagonal columns, hexagonal plates, dendritic crystals, and diamond dust. Depending on environmental temperature and humidity, ice crystals can develop from the
initial hexagonal prism into numerous symmetric or non-symmetric shapes. Aerosols
are small particles suspended in the atmosphere. They have natural sources such as
desert dust, sea salt, volcanic eruptions, and smoke from forest fires. They are also
produced from the burning of coal, oil, and other fossil fuels; manufacturing chemicals; and driving cars and trucks. Radiative Forcing (RF) is widely used to determine
the strength of the warming/cooling effect caused by GHG/clouds/aerosols. The RF
of GHG/clouds/aerosols is defined as the difference between incoming solar radiation and outgoing infrared radiation caused by the increased concentration of these
constituents. It describes the capacity of GHG/clouds/aerosols to affect that energy
balance, thereby contributing to climate change. Put more simply, RF expresses the
change in energy in the atmosphere due to GHG/clouds/aerosols. In this study, we
3
mainly focus on ice cloud and aerosols. For liquid clouds, our knowledge is much
better than for ice clouds and aerosols. We know that they have a strong impact
on the radiation budget because they strongly reflect incoming solar radiation, and
thereby cool the Earth system, and they are more uniform, and more similar from
cloud to cloud than ice clouds are, and hence easier to parameterize in models and
easier to measure [80].
Ice clouds (or Cirrus) cover about 50% of the Earth’s surface (see Chapter
2), and are very variable in their spatial distribution. Regionally, this number is
around 30% at mid-latitudes and between about 60% to 80% in the Tropics [27].
Overall, clouds are globally estimated to have a net cooling effect on the Earth’s
climate [30]. The climate impact of cloudiness mainly depends on the total loading
and the altitude of the clouds. Generally speaking, low- and middle-level clouds
tend to have a cooling effect, while upper-level clouds (Cirrus) may have a warming
effect [49]. Cirrus generally heat the upper atmosphere (by absorption) and cool the
surface (by reflecting solar radiation) at low latitudes and, inversely to an almost
equal degree, cool the upper atmosphere and warm the surface at high latitudes [60].
Cirrus are particularly important since they can induce either a net radiative heating
or cooling, depending on their microphysical characteristics.
One of the most important parameters used to describe cirrus is the cirrus
optical depth (COD) [9–13]. COD is proportional to the ice particle size and ice/water
path (IWP), which are quantities that largely determine the net radiative forcing of
clouds [107–109]. Cirrus, which can be characterized by COD, are one of the foci of
this dissertation since, although it is already known that they are important, we still
lack sufficient knowledge of the accuracy and magnitude of CODs.
Aerosols also play important roles in the radiation budget and have a great
impact on the environment and human life. Aerosols scatter and absorb solar and
4
infrared radiation in the atmosphere, leading to diminished visibility and climate
change. Moreover, aerosols alter the formation and precipitation efficiency of clouds
by acting as cloud condensation nuclei (CCN) and/or ice nuclei (IN) [40, 88, 93].
First, the direct radiative effect of aerosols on the climate is the mechanism by which
aerosols scatter and absorb solar and infrared radiation so that the radiative balance
of the Earth is altered. The effect may be either negative or positive, depending
on the aerosol optical properties (e.g., its single-scattering albedo and extinction
coefficient), which vary as a function of wavelength, humidity, aerosol concentration,
and the vertical distribution of the aerosols. Second, the indirect radiative effect is the
mechanism by which aerosols alter the formation of clouds by changing their radiative
properties, amounts, and lifetimes. The key parameter of this indirect effect is the
CCN effectiveness, which is a function of size, chemical composition, and mixing states
of the aerosols. Generally, if the water content in the air is fixed, the aerosol particle
number concentration increases as the cloud droplet number concentration increases.
This phenomenon clearly leads to a smaller cloud droplet size. Consequently, the
albedo and the lifetime of the cloud are altered. These effects are referred to as
the cloud albedo effect and the cloud lifetime effect respectively. Aerosols affect
the climate system by modifying the radiative balance and cloud properties. More
details and discussion of aerosol-induced climate change or their radiative forcing
are described in IPCC - 2007 report, which indicates that aerosols have the largest
uncertainty in radiative forcing due to their direct and indirect radiative forcing effects
mentioned above. We could also use aerosol optical depth (AOD) to characterize the
loading of aerosol in the atmosphere.
In response to the demand for better understanding of atmospheric ice clouds
or aerosols, measurements of ice cloud/aerosol distributions and properties have been
strenuously performed for years.
5
For ice clouds, progress has been made by in situ campaigns that have provided
valuable information on ice clouds from aircraft-borne instruments [4, 6, 7, 24, 33, 63].
However, such campaigns are very localized and few, and the measurements from
the clouds sampled at the in situ campaigns may not be comparable to clouds that
are formed at other geographical regions or formed by different mechanisms [e.g.,
publication from Cooper and Mcfarquhar [17,62]. To reduce the limitation of localized
measurements, global retrievals of ice clouds from passive infrared and visible satellite
measurements have been made since around 1980 [32,81,82]. There are many remote
sensing approaches, such as measuring terrestrial radiation from passive sensors at
microwave or infrared frequencies, or measuring solar reflection at visible or near
infrared frequencies (MODIS). More recently active instruments, such as cloud radar
and lidar are measuring cloud properties from satellites (CLOUDSAT/CALIOP).
For aerosols, ground-based measurements of aerosol properties, including number concentration, size distribution, chemical composition, and optical depth, are
performed either at specific sites or during field campaigns. These measurements
provide high-quality information. Ground-based remote sensing networks using sunphotometers (e.g., AERONET) and lidar instruments (e.g., MPLNET, EARLINET,
ADNET) around the world have been successfully deployed, developed, and maintained (cf. IPCC, 2007). The sun-photometer network, AERONET, consists of
approximately 150 sites that measure the column-averaged size distribution, singlescattering albedo, and Ångstrom exponent as well as the aerosol optical depth (AOD).
These sun-photometers are cost effective, and their measurements provide the whole
atmospheric column amounts of aerosols. However, the representativeness of the spatial or temporal distributions of these measurements is limited. In this sense, satellite
remote sensing is much more effective in worldwide coverage than ground-based and
aircraft measurements.
6
With the goal of obtaining better coverage on a global or synoptic scale, advanced satellite measurements have recently been conducted (IPCC, 2007, cf. Table
2.2 therein, [88]). For example, AOD satellite retrievals have been improved by the
addition of new-generation sensors such as MODIS, MISR, and POLDER [48], and
these aerosol retrieval products include both AOD and additional information, such
as aerosol fine-mode fraction and effective particle size. The aerosol products of the
Moderate Resolution Imaging Spectroradiometer (MODIS) are much superior to the
early-time satellite products of the 1970s and 1980s, such as the Advanced Very High
Resolution Radiometer (AVHRR) single/dual channel retrievals and the Total Ozone
Mapping Spectrometer (TOMS) ultraviolet-based retrievals. However, the MODIS
algorithm does not yield reliable retrievals over highly reflective surfaces (e.g., deserts,
snow/ice covered areas, or ocean glint), and its retrieval accuracy over land surfaces
is inferior to that over ocean surfaces. These are significant disadvantages in dust
aerosol research because dust source regions are almost invisible to MODIS. Multiangle Imaging SpectroRadiometer (MISR) retrievals have been performed using the
instrument’s multiple viewing capability to measure aerosol properties over ocean
and land surfaces, including highly reflective surfaces, such as deserts [61]. However, the MISR instrument has a much narrower swath width (360 km) than MODIS
(2330 km); thus, its observational coverage is very limited in comparison to that of
MODIS. Furthermore, despite the new developments in these satellite-borne instruments and their retrieval algorithms, large discrepancies exist between their aerosol
retrievals even over ocean regions where the retrieval accuracy is generally superior (e.g., Myhre et al., 2005 [65]). The satellite retrievals are often different from
ground-based sun-photometer observations as well. These discrepancies are due to
the different assumptions adopted in the cloud, aerosol, optical, and surface models
used in the retrieval algorithms.
7
For aerosol retrievals from passive instruments, some methods use simulations
of the top-of-atmosphere (TOA) radiation or path radiance to relate satellite-based
observations to aerosol properties by using theoretical models. These models are
usually based on measurements or established climatologies [16, 52]. In most cases
these algorithms make assumptions about the vertical distribution of aerosols and
surface reflectance, all of which have significant impact on the TOA radiation. MODIS
aerosol algorithms use multiwavelength radiances along with the scattering angle to
determine dust and non-dust aerosol types and the fine-mode fraction of the total
AOD [55, 78, 79, 105]. The MODIS collection-6 products are derived from four fine
modes and five coarse modes over the ocean. Recent improvements in the MODIS
retrievals, using the ”Dark Target” algorithm has enabled discrimination of dust
plumes from fine-mode pollution particles over land and ocean. The MODIS Dark
Target (DT) AOD algorithm makes use of the presence of a dark surface in two visible
channels, 0.47 and 0.66 µm, and the approximate transparency of the atmosphere
at a relatively long wavelength 2.12 µm to obtain an accurate estimation of the
atmosphere scattering [54, 55, 84, 101]. The dark target approach uses a set of ratios
and relationships between the 0.47, 0.67 and 2.1 µm channels to account for the
surface signal. This method works best over dark vegetated targets.
Meanwhile, lidars are active sensors that emit electromagnetic radiation by
themselves and, hence, can measure the distance from the instrument to the aerosol
particles. Lidar measurements are not affected by the conditions of the Earth surface, even if the measurement is performed from outer space. For aerosol retrievals
from active instruments, these algorithms compensate for some of the assumptions or
limitations of passive instruments. An example is the Cloud-Aerosol LIdar with Orthogonal Polarization (CALIOP) instrument aboard the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite [96, 98]. CALIOP is an
8
active sensor that makes range-resolved measurements of atmospheric constituents,
most importantly it can directly measure the vertical distribution and surface reflectance that are fundamental to most passive measurements. To determine aerosol
type, these algorithms use integrated attenuated backscatter measurements and volume depolarization ratio measurements, as well as surface type and layer altitude,
to determine aerosol type. However, the CALIOP aerosol extinction algorithms do
make a fundamental assumption (discussed in Chapter 3) for the extinction retrieval
of most aerosol layers.
Each satellite instrument implements a certain technique using a finite part
of the radiative spectrum. Having several retrieval datasets based on these different
instruments is an advantage, because they may be carefully combined to retrieve a
more accurate COD/AOD results. This fundamental approach is adopted in this
dissertation.
1.2
CALIPSO and CALIOP Introduction
The Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO)
satellite provides insightful view into the role that clouds and atmospheric aerosols
play in regulating Earth’s weather, climate, and air quality. CALIPSO combines an
active lidar instrument with passive infrared and visible imagers to probe the vertical
structure and properties of thin clouds and aerosols over the globe. The three coaligned nadir-viewing instruments on CALIPSO payload are
• the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP)
• the Imaging Infrared Radiometer (IIR)
• the Wide Field Camera (WFC)
9
CALIOP, which is the main instrument we focus, is a two-wavelength polarizationsensitive lidar that provides high-resolution vertical profiles of aerosols and clouds.
CALIOP utilizes three receiver channels: one measuring the 1064 nm backscatter
intensity and two channels measuring orthogonally polarized components of the 532
nm backscattered signal. Detail characteristics of CALIOP can be found from
Cloud-Aerosol LIDAR with Orthogonal Polarization (CALIOP)
Laser
Nd: YAG, diode-pumped, Q-switched, frequency doubled
Wavelength
532 nm, 1064 nm
Pulse energe
110 mJ/channel
Repetition rate
20.25 Hz
Receiver telescope
1.0 m diameter
Polariztion telescope
532 k and ⊥
Footprint/FOV
100 m / 130 µrad
Vertical resolution
30 - 60 m
Horizontal resolution
333 m
Table 1.1: Characteristics of CALIOP
1.3
Limitation of a Standard Elastic Lidar System like CALIOP
Consider a lidar system with a field of view so small that multiple scattering effects
can be neglected. The single-scatter lidar equation for such a system can be written
as
N (z) = η
P ∆t
A
βT (z)T 2 (z − zs )∆z
hν 4π(z − zs )2
(1.1)
where N (z) is the received signal photon count from a sample volume of thickness ∆z
at altitude z, η is the system efficiency, ∆t is the measurement integration period, P
is the average laser output power, hν is the photon energy, h is Planck’s constant, ν
is the optical frequency, A is the receiving telescope area, zs is the lidar altitude, and
βT (z) is the total volume backscatter coefficient [22, 23].
10
T 2 (z − zs ) is the two-way atmospheric transmission over the range z − zs given
by
Z
T (z − zs ) = exp −2
z
2
σT (l)dl
(1.2)
zs
where σT is the total atmospheric extinction coefficient.
A typical measurement from an elastic lidar is the attenuated backscatter coefficient, given by
βT0 (z)
2
Z
z
= βT (z)T (z − zs ) = βT (z) exp −2
0
σT (z )dz
0
(1.3)
zs
which is the product of the backscatter coefficient βT (z) and the two-way transmittance. This equation applies to the actual path of a laser beam in the atmosphere.
However, the more useful quantities are βT and σT , separately. With one measurement and two unknowns, some assumptions need to be adopted to solve this problem.
Both molecular and particulate (e.g. cloud and aerosol particles) components
contribute to the atmospheric extinction and backscatter coefficients:
σT (z) = σm (z) + σp (z) = Sm βm (z) + Sp βp (z)
(1.4)
βT (z) = βm (z) + βp (z)
(1.5)
where the subscripts m and p refer to molecular and particulate scattering, respectively, and S is the corresponding lidar ratio. Molecular scattering varies inversely to
the fourth power of the wavelength, whereas the wavelength dependence of particulate
(aerosols/clouds) scattering depends on the size distribution, shape, and refractive index of the particles.
The value of Sm can be calculated from the Rayleigh scattering phase function
of isotropic unpolarized light P R (π) = 43 [1 + cos2 (π)], and it is equal to
8π
3
sr. Here
11
a reference altitude zo is chosen to make sure the particulate scattering is negligible
[i.e., βp (zo ) ≈ 0]. The modeled backscatter signals at the reference altitude are used
to calibrate the measured backscatter.
For CALIPSO, the molecular backscatter coefficient profile is computed from
theory by using model atmosphere values for the temperature T (z) and pressure
P (z) obtained from the Global Modeling and Assimilation Office (GMAO; [90]). The
value of Sp , on the other hand, must be determined on a case-by-case basis and
depends on the aerosol composition, size distribution, and shape. These properties
depend primarily on the source of the aerosol and such factors as mixing, transport,
and in the case of hygroscopic aerosols hydration. The accuracy of the Sa value
used in lidar inversions depends on the correct identification of the aerosol type.
Although backscatter lidar measurements of CALIPSO are free from surface effects,
they do suffer from a need to assume an aerosol extinction-to-backscatter ratio Sa
(also referred to as the aerosol lidar ratio) to enable the calculation of extinction from
lidar backscatter signals. Though CALIOP has an algorithm to assign an aerosol
measurement to each aerosol layer, the lidar ratios for each type of aerosol are obtained
from a cluster analysis of AERONET measurements. Due to the sample size and
distribution of the cluster analysis, the assigned lidar ratios for aerosol types might
have big biases (more discussion in Chapter 3).
The main target of this study is to help reduce the large uncertainties related
to COD/AOD determinations from satellite measurements. These various instruments (MODIS, MISR, CALIOP, and so on) measure at different wavelengths, and
therefore obtain related, but different information about the atmosphere. Chapter 2
describes a comparison of COD retrievals from our new method with the standard
CALIOP level-2 cloud product. In Chapter 3, AOD retrievals from six different methods/algorithms are compared, and an assessment of CALIOP level-2 aerosol products
12
is made. Chapter 4 summarizes the benefits our new method, and concludes that the
methods developed in this dissertation can be used to improve the CALIOP operational products as well as future space-borne active remote sensing products.
13
Chapter 2
Cirrus Optical Depth (COD) over the Ocean obtained from Collocated
CALIPSO and AMSR-E Observations
2.1
Research Background and Introduction
Ice clouds (such as cirrus and contrails) are crucially important to radiative processes
and the heat balance of the Earth. They not only contribute to the distribution of
absorbed solar radiation, but also control the energy emitted to space by the climate
system by modulating the thermal emission. Previous studies have shown that ice
clouds affect the longwave radiation budget near the tropical tropopause [31,41], and
the net global radiation is especially sensitive to the optical depth of high clouds,
which mostly consists of ice crystals. Ice clouds are also suggested to have a warming
effect on the atmosphere [3, 102]. All these climate effects of ice clouds depend on
their vertical structure and optical properties. The launch of the Cloud-Aerosol Lidar
and Infrared Pathfinder Satellite Observations (CALIPSO) provided vertical profiles
of ice clouds, measured by the active Cloud Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument onboard CALIPSO [99]. These measurements represent a
significant step toward a better understanding of ice cloud temperature, as the temperature is directly linked to cirrus altitude. However, accurate derivation of optical
properties of semi-transparent ice clouds at visible wavelengths is still a significant
challenge.
Ice cloud optical properties, such as extinction coefficient and optical depth,
can be derived from either passive or active measurements. Passive sensors, such
as the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Aqua
14
and Terra satellites, retrieve height-integrated cloud optical properties using shortwave and infrared radiances [71]. However, the non-sphericity of ice crystals, cloud
multilayer structure, and the tendency for passive retrievals to be dominated by local radiation properties near the cloud top, make the retrieval of COD particularly
challenging. The accuracy of the retrieval depends on the diversity of crystal sizes
and shapes, and is very sensitive to ice crystal micro-physical assumptions [107]. On
the other hand, some active sensors, such as radars, underestimate particle size and
are not very sensitive to small particles, implying that they may miss some semitransparent ice clouds [19]. Other active sensors, such as elastic lidars, whether they
are ground-based or space-borne [2,15,38,97], have difficulties because lidar retrievals
rely on an assumed extinction-to-backscatter ratio (aka the lidar ratio) and correction
for multiple scattering in order to derive the optical depth from attenuated backscatter profiles.
In order to reduce the errors in lidar retrievals, attempts were made to benefit from the synergy provided by coincident observations from two different instruments [34, 66, 92]. For example, cirrus return signals may be used to calibrate lidar
measurements [74], and accurate aerosol and cloud optical properties may be determined by use of collocated CloudSat cloud profiling radar (CPR) and CALIPSO lidar
measurements [42,43]. Reagan [75] developed a new algorithm to improve spaceborne
lidar observations using lidar return signals from ground/sea reflections. Aerosol optical depths can be retrieved using this method from CALIPSO lidar ocean surface
returns and an appropropriate ocean surface reflectance model based on ocean surface
wind speed [94].
In this study, we take advantage of synergetic A-train constellation observations (i.e., for CALIPSO and AQUA, a temporal separation of 60 to 75 s and a perfect
spatial collocation at the ocean surface level), using the ocean surface backscatter sig-
15
nal from CALIOP, and the ocean surface wind speed measured from the AMSR-E
instrument onboard AQUA, to develop a new approach of deriving accurate COD
values.
2.2
2.2.1
Analysis and Methodology
CALIOP Attenuated Backscatter Measurements
For an elastic backscatter lidar system such as CALIOP, the lidar equation [22, 23]
gives a relation between the received signal and the atmospheric backscatter, and the
solution of the lidar quation can be used to retrieve profiles of particulate backscatter
and extinction as follows:
β 0 (r) = β(r)T 2 (r) ≈
P (r)r2
,
CE
(2.1)
where β 0 (r) (m−1 sr−1 ) is the original attenuated backscatter coefficient, which is the
main product of elastic lidar. P (r) is the lidar signal received from a scattering volume
at range r. The calibration factor C includes the amplifier gain, the transmitterreceiver overlap function, and losses in the transmitting and receiving optics. E
is average laser output power. The volume backscatter coefficient β(r) at range
r, can typically be split into two terms: β(r) = βM (r) + βP (r) with contributions
from molecules (subscript M ) and particulates (subscript P , including aerosols, water
droplets and ice particles). The two-way atmospheric transmittance T 2 (r) between
2
the lidar and the scattering volume can be written as the product TO2 3 (r)·TM
(r)·TP2 (r).
The three parts of this product are
16
1. the two-way transmittance due to absorption by ozone
TO2 3 (r)
Z
r
0
= exp −2
αO3 (r )dr
0
= exp [−2τO3 ] ;
(2.2)
0
where αO3 (r0 ) and τO3 are the ozone absorption coefficient and optical depth;
2. the two-way transmittance due to molecular scattering
2
TM
(r)
Z
r
0
0
= exp −2
σM (r )dr
0
Z r
0
0
= exp −2SM
βM (r )dr = exp [−2τM ] ;
0
where σM (r0 ), SM , and τM are the molecular scattering coefficient, the molecular
extinction-to-backscatter ratio (aka the lidar ratio), and the molecular scattering optical depth;
3. the two-way transmittance due to attenuation by particulate matter
TP2 (r)
Z r
0
0
= exp −2η
σP (r )dr
0
Z r
0
0
= exp −2ηSP
βP (r )dr = exp [−2ητP ] ;
(2.3)
0
where η, σP (r0 ), SP , and τP are the multiple scattering factor, the particulate
extinction coefficient, the particulate lidar ratio, and the particulate optical
depth, respectively.
The attenuated backscatter coefficients β 0 (r) are relatively easy to obtain from
the raw lidar signal after calibration. For CALIOP, the nighttime signal at 532 nm is
calibrated by normalizing the observed signal to the predicted molecular signal using
a scattering model from The Global Modeling and Assimilation Office (GMAO) in
17
the region between 30 and 34 km. The daytime 532 nm calibration is obtained by
interpolating the calibration constant between the adjacent nighttime data. The
1064 nm calibration constant is determined by comparing the 1064 nm signals to
the 532 nm signals from some properly selected target, such as high cirrus clouds.
The calibration uncertainty is mainly due to underestimating the existence of tiny
aerosols in the stratosphere [98]. However, retrieval of the more useful unattenuated
particulate backscatter coefficient requires more information, such as the particulate
lidar ratio, SP . For CALIOP, the aerosol models and the particulate lidar ratios
are based on a clustering analysis of AERONET data [69], and recent assessments
show that on a global basis there are about 20-30% uncertainty ranges for CALIPSO
modeled lidar ratios [59, 67, 86], which could induce roughly 20-30% underestimation
of optical depth.
In this paper, we mainly focus on the backscatter signal from the open ocean
surface. In order to reduce the error in the determination of the lidar signal peak
at the ocean surface due to sampling and sensor transient response, we consider the
0
(sr−1 ) (performed 3 bins above
integrated attenuated ocean surface backscatter γatt
and 1 bin below the ocean surface, each bin has vertical resolution of 30 m), defined
as
0
γatt
Z
top
=
β 0 (r)dr.
(2.4)
base
0
0
Here γatt
is due to (1) γocean
= γocean T 2 (r), the attenuated pure ocean surface
backscatter, which is Fresnel backscatter from the wind-roughened ocean surface with
two-way atmospheric attenuation, without any contamination from “junk” (such as
0
whitecaps, bubbles, foams etc.); and (2) γother
= γother T 2 (r), the “junk” backscatter
due to the ocean subsurface, whitecaps and bubbles [14, 37] after attenuation. The
contribution due to backscatter in part (2) can be effectively estimated from CALIOP
18
depolarization measurements [37]. Thus, we have
0
0
0
γatt
= γocean
+ γother
0
= γocean exp [−2 (τO3 + τM + ητP )] + γother
.
2.2.2
(2.5)
Ocean Surface Reflectance Model
An ocean surface reflectance model can be used to interpret lidar observations [18,37].
The integrated sea surface backscatter, γocean (sr−1 ), for a nadir pointing system can
be expressed as [37]:
γocean =
tan2 θ
ρ
exp(−
)
4πσ 2 cos4 θ
2σ 2
(2.6)
where ρ is the Fresnel reflectance, θ is the slope of wind-driven waves, and σ 2 is the
variance of the slope distribution of ocean surface waves.
CALIPSO was pointed at 0.3◦ prior to November 28, 2007 to avoid specular
reflections from calm waters and horizontally oriented ice crystals. The off-nadir angle
was switched to 3◦ afterward to reduce the maximum values of integrated attenuated
backscatter measured at both wavelengths, and to increase the minimum lidar ratios
retrieved for strongly scattering ice clouds [99]. To make sure the specular reflection
of the signal backscattered from the sea surface is received by a space-based lidar, the
slope of the waves must be equal to the lidar incidence angle, θL . Hence, the pure
ocean surface backscatter for CALIOP is obtained by:
γocean =
ρ
tan2 θL
exp(−
)
4πσ 2 cos4 θL
2σ 2
(2.7)
where ρ = 0.0209 for sea water at 532 nm and ρ = 0.0193 at 1064 nm for small angles
of incidence.
19
Equation (2.7) provides a direct relation between the unattenuated sea surface
backscatter and the variance of the slope distribution. From this equation we could
get pure backscatter without attenuation from a wind-roughened ocean surface. Ideally the pure unattenuated backscatter from the ocean surface will not change the
polarization status.
However, at high wind speeds (U > 9 m/s), the wind-driven waves start to
break and form bubbles, whitecaps, and foam, which leads to polarized backscatter
signals. Hence, ocean surface backscatter will be contaminated by such “junk” as
well as the ocean sub-surface backscatter.
Another potential bias is due to multiple scattering that may lead to higher
ocean surface backscatter and lower retrieved cirrus cloud optical depth. This method
is accurate if the surface backscatter is from the laser beam that interacts with ocean
surface only, ignoring possible interactions with ice clouds. In another words, the best
situation is that the surface backscatter has no multiply scattered contribution from
ice clouds. However, due to surface roughness, there is a chance that the laser light is
scattered by both ice clouds and the ocean surface before it is detected by the receiver.
How severe can this multiple scattering problem be? This is a fundamental question
to both this method and to newer lidar concepts such as upcoming space based
lidar missions (e.g., High Spectral Resolution Lidar (HSRL) onboard the EarthCARE
satellite, to be launched in August 2019; and the Differential Absorption Lidar (DIAL)
onboard the Merlin satellite, to be launched in December 2019). The potential biases
due to multiple scattering may increase with ice cloud and aerosol loadings.
To address this problem, we should take advantage of the cross-polarization
feature of CALIOP and use the perpendicularly polarized signal of the ocean surface
to help correct for it, since the multiply scattered ocean surface signals are also
polarized while the singly scattered signal from the ocean surface is not. So, it
20
is possible to reduce/eliminate the potential biases due to multiple scattering by
removing the multiple scattering contribution from the ocean surface backscatter.
Based on Monte-Carlo simulations, the contribution due to multiple scattering is
roughly four times the perpendicular component of the ocean surface signal.
The overall contribution from “junk” and multiple scattering can be assessed
from real time CALIOP data and Monte-Carlo simulations by using a lidar depolarization ratio of 15% [37]. At small wind speeds, the lidar depolarization ratio is close
to 0. The sub-surface backscatter can also be approximated by the depolarization
technique. The total correction can be expressed as
0
γother
≈
0
γocean,⊥
0
0
+ γocean,⊥
= 7.67γocean,⊥
0.15
(2.8)
0
where γocean,⊥
, is the measured attenuated perpendicular ocean surface backscatter for
the CALIOP instrument. The correlation between AMSR-E wind speed and CALIOP
lidar backscatter is almost doubled after this whitecap and sub-surface correction [37].
2.2.3
The U -σ 2 Relationship and Pure Integrated Ocean Surface Backscatter
The relationship between the wind speed U and the variance of the slope distribution σ 2 has been the subject of many studies based on different measurements
[18, 37, 100]. In 1954 Cox and Munk [18] first introduced a linear U -σ 2 relation
σ 2 = 0.003 + 0.00512U based on measurements of the bidirectional sea surface reflectance patterns of reflected sunlight. Wu [100] revised the linear relation to two
log-linear relations using laboratory measurements. Hu et al. [37] refined the U -σ 2
relation into three segmented functions based on comparison between CALIPSO lidar sea surface backscatter and collocated AMSR-E wind speed measurements. The
21
refined U -σ 2 relation based on CALIPSO–AMSR-E observations is given by [37]

√


0.0146
U
U < 7m/s



σ2 =
0.003 + 0.00512U
7m/s ≤ U < 13.3m/s




 0.138 log U − 0.084
U ≥ 13.3m/s.
10
(2.9)
From Eqs. (2.7) and (2.9), we can directly link the integrated sea surface backscatter
γocean with the sea surface wind speed U .
In this study, we take advantage of global sea surface wind speed measurements
to obtain pure sea surface backscatter directly (“pure” stands for the backscatter not
contaminated by contributions from bubbles, whitecaps and foam). Sea surface wind
speeds have been widely observed from in-situ platforms, such as ships [8], and spacebased instruments, such as AMSR-E [95]. The AMSR-E wind speed product with a
spatial resolution of about 20 km agrees well with other wind speed measurements
[20, 106].
Here we use the AMSR-E wind speed product to derive the pure integrated sea
surface backscatter. Figure 2.1 shows the relationship between ocean surface wind
speed and, 1) ocean surface integrated backscatter; 2) slope variance of wind-driven
waves. This figure explains the basic physics of the reflectance model. At small wind
speeds, the slope of wind-driven waves is small and does not vary too much. The waves
are “flat” and the corresponding backscatter is large. As the wind speed increases,
the waves become steeper, the slope varies in a large range, and the corresponding
backscatter becomes small.
22
Figure 2.1: Relationship between sea surface wind speed and (i) slope variance of
wind-driven waves (right) and, (ii) ocean surface integrated backscatter (left).
23
2.2.4
Ice Cloud Optical Depth Retrievals
From Eq. (2.5), by using CALIOP lidar profiles for very clear atmospheric conditions
(no water clouds, no aerosols etc.), one can obtain the ice cloud optical depth as
0
0
1
γatt − γother
1
− ln
− τM − τO3
τP =
η
2
γocean
(2.10)
where γocean can be derived through Eq. (2.7) and (2.9) by using sea surface wind speed
measurements. The multiple scattering factor, η, introduced by Platt [72, 73] is a
convenient parameter to correct the apparent two-way transmittance for contribution
from multiple scattering. Multiple scattering effects in ice clouds are significant. In
the CALIOP algorithm, η is assumed to be constant throughout the layer based on
this approach, identical within and below the cloud, and equal to 0.6. In the singlescattering limit, η is equal to 1 [97, 98]. Another approach proposed by Josset et
al. [43] yields a multiple scattering factor of 0.61 ± 0.15, which is essentially identical
to the value used operationally. Ozone and molecular optical depths, τO3 and τM , can
be obtained from meteorological analyses produced by NASA’s GMAO.
The ice cloud optical depth derived from this approach is a direct measurement, obtained without invoking any assumption about aerosol and cloud physical
properties, such as the lidar ratio.
24
Figure 2.2: Upper panel: 2008 August daytime 1064 nm lidar ratios derived from
collocated CALIPSO and AMSR-E data. Note that the lidar ratio remains fixed
throughout the optical depth ranging from 0.5-3. Lower panel: Zonal mean distribution of ice cloud lidar ratios. The mean value is 31.7 sr, with a standard deviation of
±5.7 sr.
25
Figure 2.3: Upper panel: 2008 August daytime 532 nm lidar ratios derived from collocated CALIPSO and AMSR-E data. Note that the lidar ratio has little variation
throughout the optical depth ranging from 0.5-3. Lower panel: Zonal mean distribution of ice cloud lidar ratios. The mean value is 32.3 sr, with a standard deviation of
±6.7 sr.
26
2.2.5
Lidar Ratio Retrievals from ocean surface wind
For a given group of particles, the lidar ratio SP (sr) is defined as the ratio of the extinction coefficient σ (m−1 ) to the backscatter coefficient β (m−1 sr−1 ). For CALIPSO,
the particulate lidar ratio is based on cluster analysis of ground-based AERONET
data. Hence, CALIPSO uses a fixed lidar ratio for different aerosol or cloud types.
For atmospheric conditions in which molecular scattering is negligible compared to
scattering by large particles, T 2 (r) ≈ TP2 (r), we can find another way to obtain the
lidar ratio directly from the basic lidar equation [23].
From Eq. (2.3), when differentiating TP2 (r) and combining the result with
Eq. (2.1), we obtain a first-order differential equation
dTP2 (r)
= −2ηSp β 0 (r).
dr
(2.11)
For a range r∗ , if the two-way transmittance TP2 (r∗ ) is known, and the particulate
lidar ratio SP can be easily derived by solving the above equation to obtain
SP =
where γr0 ∗ =
R r∗
0
1 − TP2 (r∗ )
2ηγr0 ∗
(2.12)
β 0 (r)dr is the column-integrated attenuated backscatter.
In this study, we can obtain SP if T 2 (r∗ ) can be accurately estimated from
Eq. (2.5) by using the ocean surface as a target. The retrieved lidar ratio can be used
instead of the approximate value used in the CALIPSO extinction algorithm. It can
also be applied to any other elastic lidar system.
Figures 2.2 and 2.3 show the feasibility of deriving ice cloud lidar ratios from
ocean surface wind speeds. At both wavelengths, the lidar ratios are rather stable
throughout the optical depth of ice clouds. At small optical depth τ close to 0.5,
27
the ice clouds are very thin and the retrieved lidar ratios have stable values with
small variations, which are mainly due to temperature and latitude dependence [25].
The zonal mean lidar ratios at the two wavelengths are given in the lower panels of
Figs. 2.2 and 2.3. The hump of the 1064 nm lidar ratios around 15◦ S - 20◦ S may
indicate “warmer” ice clouds found in those areas.
The mean derived lidar ratio derived from our method is 31.7 ± 5.7 sr for 1064
nm, 32.3±6.7 sr for 532 nm. For CALIPSO, the ice cloud lidar ratio used in the level2 algorithm is 25 sr. The 20% difference in the lidar ratio would cause a significant
bias in the extinction and optical depth retrievals. Based on simultaneous CALIPSO
and CloudSat observations, Josset et al. [43] found that the lidar ratio used in the
CALIPSO operational products is underestimated by 25%, and that the lidar-derived
optical depth is underestimated by about 30%. The lidar ratio for semitransparent
cirrus was suggested by Josset et al. to be rather stable over the ocean (33 ± 5 sr)
with slight variations depending on temperature and latitude, in agreement with our
results.
2.3
2.3.1
Results and Discussions
Correction for backscatter due to “junk”
Figures 2.4 and 2.5 demonstrate the importance of correcting for backscatter due to
“junk” in the water and sub-surface scattering.
0
As the wind speed increases, the observed ocean surface backscatter γatt
from
CALIPSO starts to be dominated by the backscatter from “junk”. The pure ocean
surface backscatter estimated from osean surface wind speed does not include the
influence from whitecaps, bubbles, and foam. After correcting for “junk” backscatter using CALIPSO’s polarization measurements, a good match is obtained between
28
Figure 2.4: Relationship between ocean surface wind speed and CALIPSO 1064 nm
attenuated backscatter. The yellow and green curves are based on the empirical
backscatter-wind relations proposed by Cox-Munk and Wu. The underestimate of
backscatter from the empirical relations is due to contributions from“junk” in the
water and sub-surface scattering.
29
Figure 2.5: By correcting CALIPSO observations for ”junk” backscatter, a good
match is obtained between the backscatter-wind relations and CALIPSO measurements.
30
the backscatter from the ocean surface reflectance model and CALIPSO measurements. By comparing the estimated backscatter with measurements, the column
optical depth can be assessed. In this paper, we constrain our study to focus solely
on the “aerosol-free” sky with single-layer non-water clouds detected by CALIPSO.
CALIPSO’s level-2 aerosol/cloud layer product, combined with the vertical feature
mask product, are used to ensure that the selected profiles satisfy our purpose.
2.3.2
Comparison of Ice Cloud Optical Depths with CALIPSO L-2 Data
Figures 2.6 and 2.7 show the global distribution of ice cloud optical depth at 1064
and 532 nm from our method and CALIPSO level 2 data. As can be seen, both data
sets have very similar distribution patterns, which ensures our approach is reliable on
a global scale. While the patterns of optical depth at other regions agree well, the
optical depth derived from CALIPSO underestimated ice cloud optical depth between
30◦ S and 65◦ S. This phenomena exists at both wavelength, and it can be explained
by the use of a fixed (inaccurate) lidar ratio in the standard CALIPSO product.
Figure 2.8 compares the single-layer ice cloud optical depths obtained from
the CALIPSO-AMSR algorithm derived in this paper, and from the CALIPSO level
2 ice cloud data at 1064 nm and 532 nm). The colors represent the frequencies of the
optical depth. The single-layer ice clouds are constrained by neglecting water clouds
from the level 2 data. The wind speed from AMSR used in the retrieval is selected
to be in the range [3, 9] m/s to ensure the quality of the new method of deriving ice
cloud optical depth.
Please reduce the size of Figs. 2.6 and 2.7 to allow room for the
caption!!
It can be seen from Fig. 2.8 that the optical depths from our new approach follow
the trend of the CALIOP L-2 products. However CALIPSO level 2 data tend to un-
31
Figure 2.6: Upper panel: Global distribution of cirrus optical depth at 1064 nm
derived from collocated CALIPSO - AMSR-E observations with corresponding cirrus
optical depth from CALIPSO L-2 data. Lower panel: Global distributions of cirrus
optical depth at 1064 nm from both the collocated CALIPSO - AMSR-E observations
and CALIPSO level 2 data. Cirrus optical depths from our new approach generally
have very similar distribution pattern as the CALIPSO standard product, which
indicates that our approach is reliable.
32
Figure 2.7: Upper panel: Global distribution of cirrus optical depth at 532 nm derived
from collocated CALIPSO - AMSR-E observations with corresponding cirrus optical
depth from CALIPSO L2 data. Lower panel: Global distributions of cirrus optical
depth at 532 nm from both the collocated CALIPSO - AMSR-E observations and
CALIPSO level 2 cloud data. Cirrus optical depths from our new approach generally
have very similar distribution pattern as the CALIPSO standard product, indicating
that our approach is reliable.
33
Figure 2.8: Upper panel: August 2008 daytime comparison of ice cloud optical depth
(1064 nm on the left, 532 nm on the right) from CALIPSO L2 and CALIPSO–AMSR.
Lower panel: Zonal mean distribution of the optical depths from two sources.
34
derestimate the cirrus optical depth for τ < 0.3 (which implies that very thin cirrus
clouds, or semitransparent ice clouds are neglected or underestimated by the standard
method) and τ > 1.0 (which means that the standard method may have issues with
the cirrus lidar ratios used in the retrieval algorithm). The optical depth from the
standard method is about 20% lower than that from our new method. Other investigators, using different techniques/data, found similar underestimates in CALIPSO
L-2 data [5,43]. The main reasons for the differences are: 1) The ice cloud lidar ratios
adopted in the CALIPSO operational optical depth retrieval algorithm (Sice ≈ 25±10
sr) are not very accurate, implying that uncertainties in the lidar ratio, due to natural variability and misclassification of cloud type, propagate non-linearly into the
estimates of cloud layer optical depth. 2) The “standard” CALIOP cloud retrieval
algorithm screens out some cloud layers containing horizontally oriented ice crystals
that produced anomalously high specular backscatter from the near nadir-pointing
CALIOP beam, implying that reliable extinction estimates cannot be retrieved in
these cases. [5] 3) The CALIOP daytime calibration accuracy may still have issues
due to the presence of small amounts of aerosols in the atmosphere. 4) Potential bias
in the AMSR-E ocean surface wind speeds exists.
2.3.3
Adjustment of lidar ratios for CALIOP level 2 data
We have found that if the ice cloud lidar ratio used in CALIPSO level-2 data were to
be changed from 22 to 32, the difference between the CALIPSO L-2 product and our
new method will be greatly reduced, as can be seen from Fig. 2.9. In another words,
there is considerable room for improvement in the uncertainty of the lidar ratio in
the CALIPSO retrieval algorithm. Even though there are still some minor differences
between the two methods, we could infer a better linear correlation between these
two products, especially at nighttime (upper panel of Fig. 2.8).
35
36
Figure 2.9: August 2008 comparison of ice cloud optical depth for 1064nm from
revised CALIPSO L2 product and CALIPSO–AMSR. Lower panel: Same comparison
for 532 nm data.
37
2.4
Conclusion
In this study we introduced an empirical method that makes use of the perpendicularly polarized signal of the ocean surface to remove the impact of multiple scattering on the retrieved optical depth. This approach may apply to HSRL which uses
molecular backscatter as a target. Since both the ocean surface and the molecular
backscatter do not lead to depolarization, part of the perpendicular component of the
ocean surface signal comes from multiple scattering which causes biases.
We have developed a simple but reliable method to retrieve the ice cloud optical depth from collocated ocean surface wind and lidar backscatter observations.
A comparison of optical depths derived by this method with those obtained from
CALIPSO standard products, shows that the patterns of optical depths roughly agree
with CALIPSO level 2 data on a global scale. The ice cloud optical depths lie mostly
in the range between 0 and 3, and the results from our method are about 18% higher
than CALIPSO level 2 data.
The results in this paper will allow improvements in ice cloud discrimination
as well as enhancements of CALIPSO extinction profile retrievals and uncertainty
estimates.
38
Chapter 3
Column Aerosol Optical Depth Evaluation over Open Ocean with
collocated CALIPSO-MODIS-AERONET Measurements
3.1
Research Background and Introduction
Aerosols have a significant impact on air quality, weather, and climate, and they can
also significantly influence the radiation budget of the Earth system by interacting
with clouds to modify the incoming solar radiation or outgoing long-wave radiation
in different ways, such as their direct and indirect radiative forcing [85]. Evaluating
these impacts and influences requires an accurate observational characteristics of large
scale of temporal and spatial distributions of aerosols. There are several ways that
could provide high-confident aerosol measurements over the globe. First and the most
vital one is the AErosol RObotic NETwork (AERONET), a global network of groundbased sun photometers. AERONET measures routine aerosol optical properties at
fixed locations, such as aerosol optical depth (AOD), at high temporal and spectral
resolutions to better understand aerosol distributions in the atmosphere. However,
AERONET measurements are highly location related, or limited in spatial coverage.
Knowledge of aerosol characteristics over large spatial scales is also important in order
to determine their impact on the radiation budget and hence climate. This spatial
limitation can be addressed by satellite remote sensing, which provides systematic
global real time AOD observations.
The capability of passive/active satellite measurements provides an unique opportunity to enhance the understanding of the importance of aerosol impacts globally.
AODs from passive sensors such as the Moderate resolution Imaging Spectroradiome-
39
ter (MODIS) [54, 55, 78] and Multiangle Imaging Spectroradiometer (MISR) [45, 46],
can be used to discover the strong observational constraints for the aerosol direct
effect on solar radiation at the top-of-atmosphere (TOA) [77]. This capability has
also been used to enhance the surface monitoring networks for air quality forecasts
and to provide observation-based estimates of the long-range transport of aerosols
[1, 47, 83]. Though passive sensors have little information on the vertical distribution of aerosols [45, 70], they can still provide valuable assessments of total column
aerosol amount in cloud-free scenes. However, the AODs from passive sensors depend
on the aerosol models being adopted in the retrieval and cloud/aerosol classification
algorithms [58, 91].
For active sensors, such as The Cloud-Aerosol Lidar with Orthognal Polarization (CALIOP) [96,98,99] and High Spectral Resolution Lidar (HSRL) [28,29,64], the
most vital improvement is that they can provide the range-solved vertical structure of
aerosol and cloud distributions. This information is very important in order to properly determine aerosol/cloud radiative effects. CALIPSO is the first active-technique
satellite that provides a unique opportunity to profile aerosol/cloud vertical distributions globally and assess model simulations of these distributions on global and
annual scales. However, the column or layer AOD from CALIPSO may be less accurate than that retrieved by passive sensors due to the limitations of standard elastic
backscatter lidar algorithms [44, 59]. For example, AODs retrieved from CALIOP
using standard methods are biased due to the use of assumed lidar ratios. It has
been proven that the standard CALIPSO agorithm underestimates AODs by about
20-30% due to use of too small lidar ratios for some aerosol types [42, 43]. The objectives of this study are to develop an alternative AOD retrieval method based on
combined satellite measurements, and to analyze regional and seasonal variations of
total column AODs from six different aerosol products (including two new retrieval
40
methods) and examine differences in AODs obtained by these methods over coastal
and open ocean areas.
The rest of this chapter is organized as follows. Section 3.2.1 briefly introduces
CALIOP lidar measurements, Section 3.2.2 MODIS measurements, and their major
uncertainties. Section 3.2.3 describes AODs derived from the ocean surface reflectance
method and Section 3.2.4 from multi-layer Neural Network techniques. In Section 3.3,
results are presented and discussed including regional variations of aerosol optical
depths on a seasonal basis (Section 3.3.3) through comparisons of CALIPSO and
MODIS measurements with AODs from the other two methods. Concluding remarks
are provided in Section 3.4.
3.2
3.2.1
Methodology
CALIOP aerosol algorithm and product
The CALIPSO mission was launched on April 28, 2006 with an equator-crossing time
of about 1:30 P.M. and 1:30 A.M., and a 16-day repeating cycle. The primary instrument onboard the CALIPSO satellite is the Cloud-Aerosol Lidar with Orthogonal
Polarization (CALIOP), a two-wavelength, polarization sensitive lidar [97]. Starting
from June 13, 2006, CALIOP collects almost continuously high-resolution (333 m
in the horizontal and 30 m in the vertical in low and middle troposphere) profiles
of the attenuated backscatter by aerosols and clouds at 532 nm and 1064 nm wavelengths along with polarized backscatter in 532 nm channel. Spatial averaging over
large scales (from 5 km averaging to 80 km averaging) is usually done to improve
signal-to-noise-ratio (SNR) for reliable aerosol detection and retrievals [98, 99].
The detectable atmospheric features are first classified into different groups
(aerosol or cloud) using the CALIOP cloud-aerosol discrimination (CAD) algorithm
41
[56]. The confidence level for the aerosol-cloud classification is determined by a CAD
score given by the CALIOP CAD algorithm, which ranges from -100 to 0 for aerosols
and +100 to 0 for clouds. A larger absolute value of the CAD score indicates higher
confidence of the feature classification. The CALIOP team also developed a scene
classification algorithm to divide the aerosol layer into six types, which are smoke,
polluted continental, polluted dust, dust, clean continental, and clean marine, with
respective extinction-to-backscatter ratio or lidar ratio (S) of 70, 70, 65, 40, 35, and
20 sr at 532 nm [68]. These fixed values of the lidar ratio are based on cluster
analysis of AERONET ground-based site measurements. For each layer detected in
the CALIOP backscatter data, a set of feature classification flags are used to report
feature type (e.g., cloud vs. aerosol vs. stratospheric layer) and subtype. With certain
ranges of the lidar ratio for each layer of aerosols, extinction coefficients and optical
depths can be estimated from CALIOP backscatter measurements using an extinction
retrieval algorithm. Details of CALIOP aerosol detection, classification and retrieval
algorithms can be found in the CALIOP Algorithm Theoretical Basis Document
(ATBD) and some updates provided in recent publications [56, 68, 98, 99, 103].
Several validation efforts have shown that CALIOP is very accurate at determining aerosol/cloud vertical distributions [50, 51]. Comparisons of simultaneous
CALIOP and ground-based lidar over Korea show that the top and base of cloud
and aerosol layers are generally in agreement within 0.10 km and the aerosol extinction profiles are generally in agreement within 30% in cloud-free, and nighttime,
semi-transparent cirrus cloud conditions [50]. Comparisons with the High Spectral
Resolution Lidar (HSRL) measurements from two U.S. field campaigns show that
the CALIOP and HSRL are in good agreement on the vertical profile of aerosols,
while the average extinction determined by CALIOP is biased by 20-50% compared
to HSRL [68].
42
However, recent publications show that the lidar ratios being used by the
CALIOP algorithm are biased by 20-30 % on a global basis. Hence, uncertainty
in the determination of the lidar ratio S becomes a major factor contributing to the
uncertainty in aerosol extinction and optical depth obtained from CALIOP aerosol retrievals [43]. Lidar ratios typically vary about 30% within a given aerosol type [68,99].
Some types of aerosols have higher variability. Misclassification of aerosol type contributes additional uncertainty. At small AOD, the AOD fractional uncertainty can
be approximately estimated as the fractional uncertainty of lidar ratio S. However,
with an increase of AOD, the fractional uncertainty of AOD increases rapidly. For
example, a fractional uncertainty of 30% for lidar ratio S would result in an AOD
fractional uncertainty of 50% for AOD around 0.5 and nearly 100% for AOD around
1.0 [68]. An AOD uncertainty implies an uncertainty in the retrieved profile. For
weakly attenuating layers, the shape of the profile is fairly representative, although
the magnitude is biased either high or low. For denser layers, retrieval errors tend to
accumulate toward the base of the layer, but again the shape of the upper portions
of the layer is credible.
Another significant source of uncertainty in the retrieved AOD occurs when
the base of the aerosol layer is incorrectly identified (most commonly above the true
layer base). This happens when the lidar signal is completely attenuated in layers
with optical depths greater than 3.0, but can also happen in attenuating aerosol when
the detection algorithm incorrectly identifies clear air while still inside the layer. In
either case, the AOD is biased low but the error affects only the lower part of the
profile. Additional retrieval errors arise from the use of an incorrect Sa value in the
inversion of the lidar single-scattering equation as a result of the wrong choice of
aerosol model. If the type is incorrectly chosen, a wrong lidar ratio will be used in
the extinction retrieval leading to errors in the CALIOP AOD. Multiple scattering
43
effects, which are not taken into account in CALIOP AOD retrievals, can also be a
source of CALIOP AOD error in the presence of dense dust [57].
Retrieval uncertainties propagate through the chain of algorithms used for creating each successive data product thus eventually impacting the accuracy of the
column AOD [99, 104]. Errors in the CALIOP AOD arise from inaccurate or incomplete detection and misclassification of aerosol/cloud layers and retrieval errors.
Both of these are influenced by signal-to-noise ratios (SNRs), which are lower during
daytime. CALIOP is not able to detect features with attenuated backscatter signal
below 2 ∼ 4 × 10−4 km−1 sr−1 in the troposphere [98]. Hence, the minimum detectable
extinction coefficient is 1 ∼ 2 × 10−2 km−1 for a lidar ratio around 50. Although the
CAD algorithm works well in a majority of cases examined, several specific layer
types are prone to misclassification. Above or close to source regions, heavy dust or
smoke might be misclassified as clouds. Dust transported to the upper troposphere
and tropopause may be misclassified as clouds, which biases the AOD low. Optically
thin clouds in the polar regions may be misclassified as aerosol [56].
The major data used in this study are CALIOP Level 1 standard product and
Level 2 Version 4.10 standard aerosol layer product from January, April, July, and
October 2016. The product provides the column AOD, top and base of the aerosol
layer, and layer integrated properties such as attenuated backscatter (IAB), lidar
ratio (S), volume depolarization ratio (VDR), and CAD score, among others.
When comparing CALIOP and AERONET data, differences can result if the
atmospheric column at the coincident station is not homogenous. Omar et al. [68]
found that the median relative AOD difference between the two measurements is
25%, and thatCALIPSO finds clouds in more than 45% of the coincident atmospheric
columns classified by AERONET as clear. In this sense, since the analysis focuses
on the CALIOP observations in cloud-free conditions, the cloud-free condition of a
44
profile is determmined by CALIOP cloud information and CAD scores, including all
the columns that are completely cloud-free. The cloud information comes from the
CALIOP 5-km cloud layer product.
Figure 3.1: Open ocean region selected for study.
Two steps of data screenings are applied to attain good quality CALIOP data.
The first step is to exclude cloud-contaminated profiles and aerosol layers that have
low CAD scores in the selected open ocean region. We include profiles with CAD
scores between 50 and 00, which ensures that the profiles we select have high confidence. The blue box in Fig. 3.1 shows the region of interest. In this region a dust
belt extends through different seasons roughly from 15◦ to 25◦ N and from 20◦ W to
40◦ W). The region above 25◦ N can be considered as an open ocean area. Over the
dust region, the occurrence of dense dust or smoke would yield attenuated backscatter
and color ratios that are more likely to overlap with cloud histograms, resulting in a
lower level of confidence of cloud-aerosol discrimination [56]. The other screening step
45
is to exclude CALIOP profiles where the retrieval algorithm has to adjust the initially
selected lidar ratio. Such adjustment usually occurs for complex features with high
AOD that are vertically adjacent to or embedded in clouds or other features [67, 68].
In such cases, the retrieved AOD and extinction profiles are generally not accurate
and the associated uncertainty cannot be reasonably assessed [98, 99, 103].
3.2.2
MODIS aerosol algorithms and products
There are two main AOD products from MODIS. One is obtained by using the NASA
SeaDAS software provided by the MODIS ocean color group. The MODIS-Aqua
satellite Collection 6 L1B data were processed with SeaDAS using the NIR algorithm
and the NIR/SWIR algorithm. Whitecaps and sunglint were corrected for using the
standard method in SeaDAS. Polarization effects of the MODIS sensor were also implicitly corrected for (Meister et al., 2005). MODIS data are selected from a 3×3
box centered at the locations of the collocated CALIOP measurements to be processed. AOD values at 412, 443, 488, 531, 547, 667, and 869 nm are provided by the
SeaDAS algorithm, and AOD values at 532 nm are obtained by interpolation. Since
this product provides AODs over the ocean, we mark it as “AOD-MOD-OCEAN”. A
main concerns with this product is that during the SeaDAS processing, dust aerosol
models are not considered and the Atmospheric Correction (AC) algorithm does not
work well in coastal water areas. Hence, the outcome is expected to be questionable
in areas with dust loadings over coastal water.
The other AOD product is based on a Dark Target (DT) algorithm developed
by the MODIS Atmosphere group. The detailed retrieval principle of the MODIS
DT algorithm can be found in recent publications [54, 55, 79]. The DT AOD product
at 3 km spatial scale is obtained from spectral reflectances using a look-up-table and
inversion method based on the ratio of visible and shortwave infrared reflectances
46
similar to that of the 10-km product [77, 78]. MODIS AOD products with high resolutions (3 km) are expected to address aerosol gradients and pollution sources missed
at 10 km. The quality of MODIS aerosol retrievals generally depends on the accuracy
of the surface reflectance and the aerosol model, and over- or underestimation under
clear and polluted conditions are normally caused by an error in these two factors. In
this study, MYD04 Collection 6 DT aerosol products were obtained over the selected
open ocean region, and only the highest-quality-flag (QF=3) AOD observations were
selected for analysis.
Validation studies that examine CALIOP estimates of AOD typically reveal a
low bias in the CALIOP data relative to other measurements. Kittaka et al. [53] and
Redemann et al. [76] used different strategies to compare CALIOP total column AODs
to spatially and temporally collocated AODs retrieved from MODIS measurements.
Both studies found that the CALIOP daytime AOD estimates are lower than the
corresponding MODIS values, and both suggested to filter the data to obtain the
most reliable AOD values from CALIOP data.
In this work, we extend CALIOP AOD validation efforts by comparing 532
nm column AODs obtained from the CALIPSO version 4 data product to MODIS as
well as ground-based sunphotometer measurements obtained at multiple AERONET
sites around the globe.
3.2.3
Aerosol retrievals from CALIPSO-AMSR co-measurements
In Chapter 2 we established an ocean surface reflectance model. If the variance σ 2
of wave slope distribution is known, the 180◦ ocean surface backscatter γocean can be
expressed as [37]
γocean
tan2 θ
ρ
exp −
.
=
4πσ 2 cos4 θ
2σ 2
(3.1)
47
Cox and Munk [18] first introduced an empirical linear relationship between the slope
variance σ 2 and the ocean surface wind speed U , described as σ 2 = 0.003 + 0.00512U .
Based on collocated CALIOP-AMSR measurements, Hu et al. [37] revised this relationship to an empirical fractional equation

√


0.0146
U
U < 7 m/s



σ2 =
0.003 + 0.00512U
7m/s ≤ U < 13.3 m/s




 0.138 log U − 0.084
U ≥ 13.3 m/s.
10
(3.2)
From the above equation, if the sea surface wind U speed is given, the theoretical ocean surface backscatter γocean can be estimated from the wind speed U .
The attenuated ocean surface backscatter measurement from CALIOP is a product
of ocean surface backscatter and two-way atmospheric transmittance
0
γocean
= (γocean + γother )T 2 (z) = (γocean + γother ) exp (−2τcolumn ) .
(3.3)
Note in the above equation, the ocean surface backscatter is split into two parts:
the theoretical ocean surface backscatter γocean and backscatter γother due to “other”
scatterers in the air-water interface, such as whitecaps, bubbles, foam, ocean subsurface backscatter, and multiple-scattering in the atmosphere above the ocean surface. These “junk” contributions to the backscatter, γother , can be effectively assessed
using a depolarization technique [37]. Details about how to remove “junk” backscatter using the depolarization ratio over ocean surface were provided in Chapter 2.
Once γother has been accurately removed, the column optical depth can be obtained
from Eq. (3.3):
τcolumn
1
= ln
2
γocean + γother
0
γocean
.
(3.4)
48
In this study, since we focus on cloud-free situations for which the column
aerosol optical depth τaerosol is obtained by
τaerosol = τcolumn − τM − τO3
(3.5)
where τM and τO3 are the optical depths of atmospheric molecules and ozone, respectively. Accurate values of τM and τO3 can be obtained using meteorological analyses
available from NASA’s GMAO.
The advantage of this method is that no lidar ratio assumption is needed to
retrieve τaerosol from this method. It only depends on AMSR-E ocean surface wind
speed measurements, which have been shown to have high accuracy.
3.2.4
Multi-Layer Neural Network based on AccuRT
To address atmospheric correction issues related to MODIS ocean products, Fan
et al. [21] introduced a new approach that is based on neural network techniques.
A radiative transfer model for the coupled atmosphere-ocean system (AccuRT, see
Chapter 10 of Stamnes et al., 2017 [89]) was used to compute the Rayleigh-corrected
TOA radiance Lrc . Information about AOD and water-leaving radiance Lw is embedded in the Lrc signal. The radiative transfer simulations show that there is a spectral
similarity between the Rayleigh-corrected radiance Lrc and the water-leaving radiance
Lw . Based on this spectral similarity, a Multi-Layer Neural Network (MLNN) can be
trained to find a relation between Lrc and Lw . Also, there is significant correlation
between the spectral Lrc and the spectral AOD, implying that it should be possible to
use a MLNN method to retrieve spectral AOD values from spectral Lrc data. A comparison with AERONET measurements shows that the MLNN algorithm significantly
improves retrieval of normalized Lw in blue bands and yields minor improvements in
49
green and red bands. AODs in coastal regions (turbid water, heavy aerosol loadings)
derived by the MLNN method also improved. However, since the MLNN method
uses the same aerosol models as in SeaDAS, it may also have problems in retrieving
AODs over dust areas.
3.2.5
AERONET aerosol algorithm and product
AERONET AOD level-2 version 2 products are adopted in this study. AERONET
consists of more than 180 sun and sky scanning radiometers located at surface sites
throughout the world [36]. Unlike CALIPSO, which estimates AOD based on an
inferred (or assumed) lidar ratio, aerosol type and composition, the AERONET sunphotometers directly measure AOD at seven wavelengths (approximately 0.340, 0.380,
0.440, 0.500, 0.675, 0.870, and 1.02 µm) with an estimated uncertainty of 0.010 [35].
Unfortunately, AERONET does not provide aerosol optical depths at the CALIOP
visible wavelength (0.532 nm), so we use a 2nd order variation of the Ångstrom relation to interpolate between all available AERONET wavelengths [87]:
ln τA (λ) = a0 + a1 ln λ + a2 (ln λ)2 .
(3.6)
That is, the aerosol optical depths τA (λ) provided by AERONET are used to determine the coefficients a0 , a1 , a2 via a 2nd-order polynomial regression. These coefficients then can be used to accurately determine the aerosol optical depth at any
wavelength in the visible region. In this study, we selected ten AERONET sites over
the ocean globally to assess AOD retrieved from the other algorithms.
50
3.3
Results and Discussions
In this study, we did 3 comparisons for different purposes. First we compared AOD
retrieved from CALIOP standard algorithm with that from MODIS ocean color product, MLNN algorithm and our method, namely as Ocean Surface Reflectance Method
(OSRM) at the coastal region. Then we picked an open ocean area marked in Fig. 3.1
to conduct comparisons of AODs obtained from several different algorithms for different seasons. Lastly, AODs from 10 AERONET sites were used as a reference to
assess biases in the AOD retrieval algorithms being compared.
3.3.1
Coastal Ocean Comparison
Figure 3.2: Left Panel: MODIS RGB image with collocated CALIOP track labeled
by the green dots on March 15th, 2007 over the Yellow sea area. Right Panel: The
corresponding CALIOP 532 nm total attenuated backscatter image; the colorbar
represents the strength of the attenuated backscatter.
In this coastal ocean case, as shown in Figs. 3.2 and 3.3, we can see that a
strong aerosol event is detected from the CALIOP total attenuated backscatter profile
image, and two dense aerosol layers can be found between 34◦ N and 37◦ N (from
CALIOP vertical feature mask dataset). There is a small amount of cloud/lowconfidence aerosol contamination at the edge of the aerosol layers. The MODIS
51
Figure 3.3: Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset for
the corresponding collocated coastal case; Lower Panel: AOD retrieved from different
algorithms: CALIOP (black dots), MODIS ocean color(blue dots), MLNN (red dots)
and OSRM (green dots).
52
ocean color AOD product missed the peak around 35.7◦ N, which may due to an overadjustment of the reflectance. The AOD from CALIOP level-2 data is very low; the
main reason being the poor aerosol lidar ratios adopted by the CALIOP extinctionretrieval algorithm. The MLNN algorithm [21] fixed potential atmospheric correction
issues in MODIS ocean color product. The AODs retrieved by our new OSRM method
matches perfectly with those obtained by the MLNN method. Even though the R2 coefficient of MLNN-OSRM is improved by only a modest amount compared with
MLNN-CALIOP (from 0.87 to 0.96), the relative differences decrease from over 100%
to less than 10%, which is a big improvement.
Figure 3.4: Left Panel: MODIS RGB image with collocated CALIOP track labeled by
the green dots on March 1st, 2008 over Yellow sea area; Right Panel: The corresponding CALIOP 532 nm total attenuated backscatter image, the colorbar represents the
strength of the attenuated backscatter.
Figures 3.4 and 3.5 show another case when multi-layer aerosols were detected
by CALIOP. Although this case is more targeted on the open ocean, the retrieved
AODs from OSRM and MLNN match well with each other, with little oscillations
in our OSRM method, which might be due to oscillations of the sea surface wind
speed. The AODs available in the CALIPSO Level-2 data are much smaller, and the
CALIPSO algorithm tends to derive fixed-value AODs in some regions (e.g. between
latitudes 33◦ N and 34◦ N), which may be due to the application of several averaging
53
Figure 3.5: Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset
for the corresponding collocated coastal case. Lower Panel: AODs retrieved from
different algorithms: CALIOP (black dots), MODIS ocean color (blue dots), MLNN
(red dots) and OSRM (green dots).
54
processes. The R2 -coefficients between different methods jump from 0.65 to 0.88, and
the relative differences decrease from 96% to 10.2%.
3.3.2
Open Ocean Comparison
Figure 3.6: Left Panel: MODIS RGB image with collocated CALIOP track marked
by the green dots on Janurary 1, 2016 over the Atlantic ocean close to West Africa.
Right Panel: The corresponding CALIOP 532 nm total attenuated backscatter image;
the colorbar represents the strength of the attenuated backscatter.
For the open ocean case shown in Figs. 3.6 and 3.7, the AODs from our method
match well with those obtained by the MLNN method, in the absence of cloudcontaminations. We can see that between 15.5◦ N and 23◦ N latitudes, the AODs
obtained by our OSRM method, the MODIS and the MLNN methods have very
similar trends, while the AODs obtained by the standard CALIOP method are systematically lower than the others. Between latitudes 23◦ N and 27◦ N, the trend is a
little messy due to the existence of tiny broken clouds. However, the R2 -coefficients
increase from 0.0085 for MLNN-CALIPSO to 0.2577 for MLNN-OSRM. The AODs
from CALIPSO level-2 data is 20-30% smaller than those from the other methods,
which is mainly due to the underestimation of the lidar raio as well as some aerosols
55
Figure 3.7: Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset for
the corresponding collocated coastal case; Lower Panel: AOD retrieved from different
algorithms: CALIOP (black dots), MODIS ocean color(blue dots), MLNN (red dots)
and OSRM (green dots).
56
being misclassified into the wrong types by the CALIOP aerosol extinction algorithm.
From this case we can conclude that AOD retrievals from the MODIS ocean color
product and the MLNN method have some issues when the aerosol layers are contaminated or mixed by tiny clouds, or aerosols close to the edge of clouds will be
misinterpreted by these two algorithms. Since the MLNN method currently makes
use of the SeaDAS cloud screening algorithm, the SeaDAS and MLNN algorithms are
expected to behave similarly when the cloud screening is inadequate.
Figure 3.8: Left Panel: MODIS RGB image with collocated CALIOP track labeled by
the green dots on January 19th, 2016 over Atlantic ocean close to West Africa. Right
Panel: The corresponding CALIOP 532 nm total attenuated backscatter image; the
colorbar represents the strength of the attenuated backscatter.
In this case, CALIPSO detected an intense aerosol layer between 20◦ N and
24◦ N. Looking further into the information provided by the CALIOP Vertical Feature
Mask data, we found that the aerosol type of this layer is classified as dust, which
is transported from the desert in West Africa. The remainder of the aerosol layer
between 25◦ N and 30◦ N is classified as marine aerosol. We note that the AODs from
our OSRM method agree well with those obtained by the MLNN method, while both
57
Figure 3.9: Upper Panel: Vertical feature mask from CALIOP level-2 VFM dataset for
the corresponding collocated coastal case; Lower Panel: AOD retrieved from different
algorithms: CALIOP (black dots), MODIS ocean color(blue dots), MLNN (red dots)
and OSRM (green dots).
58
the CALIPSO level-2 data and the MODIS ocean color data underestimate the AODs
by about 25% between 20◦ N and 24◦ N. The lidar raio for dust-type aerosols in the
CALIOP algorithm is 40 sr, but it has been shown that a more reliable/trustable
dust lidar ratio is around 60 sr. This uncertainty in the lidar ratio causes CALIOP to
underestimate AODs for dust aerosols. There is no dust aerosol model in the MODIS
ocean color aerosol retrieval algorithm. Thus the MODIS ocean color algorithm will
misclassify this dust aerosol layer into a different type and thereby underestimate the
optical depth.
3.3.3
Open Ocean Regional Seasonal Variations
We selected four months (January, April, July and October 2016) of collocated
CALIOP-MODIS data to study the seasonal variation of the aerosol distribution
in an open ocean area between 15◦ N and 45◦ N and between -60◦ W and -20◦ W. The
reason for choosing this area is that it covers and open ocean region partially affected
by dust. The AODs across the dust belt extending roughly from 15◦ to 25◦ N and
from 20◦ W to 40◦ W, vary through different seasons, while the region above 25◦ N can
be considered as a dust-free open ocean area.
First, looking at the January 2016 collocated data shown in Figs. 3.10-3.12,
we see that AODs obtained by the OSRM, MLNN, and MODIS methods have very
similar geolocation distribution, whereas the standard CALIOP retrieval missed most
of the aerosols. In January, the surface temperature is low compared to other seasons,
and there are few dust events occurring over the ocean surface; dust transport from
the desert in Africa to the ocean can be neglected in January. The underestimated
lidar ratios of marine aerosol used in the standard CALIOP algorithms leads to an
underestimation of the retrieved AODs.
From the zonal means of AODs shown in Figs. 3.11, we could also find the
59
big differences between CALIPSO level-2 data and other methods. By applying our
method to CALIPSO-AMSR-2 measurements, the relative differences in AODs can
be shrink from over 100% to less than 20%.
Fig. 3.12 shows the probability distributions of the AODs from these algorithms. We could see that AODs from MODIS ocean color, MLNN and OSRM
have similar PDFs, with peak value (mode value) around 0.1, but the AODs from
CALIPSO have a very different distribution, with peak value shifted to around
0.04 0.05, which also tells that CALIPSO level-2 aerosol products underestimate
AODs due to the biased lidar ratios been used, and under some circumstances CALIPSO
may classify aerosol types wrong. The mean AODs derived from CALIOP, MODIS
ocean color, MLNN, and OSRM methods are 0.12, 0.21, 0.23, and 0.23, respectively.
Figure 3.10: Mean AOD distribution from different algorithms in the selected open
ocean area, January 2016.
60
Figure 3.11: Zonal mean of AODs and relative differences between each algorithm,
January 2016.
Figure 3.12: AOD probability distribution according to each algorithm for January
2016. The mean AODs are 0.12 (CALIPSO), 0.21 (MODIS), 0.23 (OSRM), and 0.23
(MLNN), respectively.
61
For AODs in April 2016 (Figs. 3.13-3.15), the distributions above 25◦ N obtained from the MLNN, MODIS, and OSRM methods are close to each other, whereas
CALIOP still cannot gather accurate AODs over this dust-free open ocean area. The
open ocean area below 25◦ N is affected by dust events. Starting from April, more
frequent dust events occur due to transport from West Africa to the adjacent ocean
surface. AODs obtained by the OSRM method correctly reflect this phenomenon,
while the other three algorithms do not, either due to a misclassification of dust
aerosols into other types (MODIS and MLNN), or an underestimation of the dust
lidar ratio (CALIOP). From the zonal mean of the AODs we get the same result.
The relative difference in AODs decreases to 20% above 25◦ N.
From Fig. 3.15 we can see that AODs from the MODIS ocean color, MLNN, and
OSRM methods have very similar PDFs, with a peak value (mode value) around 0.1
(MLNN and MODIS) and 0.15 (OSRM), while the peak PDF value of AODs derived
from CALIPSO is shifted to 0.04 0.05. The mean AODs derived from CALIOP,
MODIS ocean color, MLNN, and OSRM are 0.12, 0.19, 0.18, and 0.21, respectively.
In July 2016 (Figs. 3.16-3.18), as the surface temperature increases, dust
aerosols transported from West Africa to the ocean surface cover the area below 25◦ N.
From the distributions of AODs from CALIPSO and OSRM we can clearly see the
heavy dust loadings. However, AODs retrieved from MODIS and MLNN algorithms
are underestimated, because these two algorithms likely will misclassify dust aerosols
in this area into different types of aerosols or clouds. Though the standard CALIOP
algorithm can derive the approximate dust distribution, the AODs are still biased low
due to the low dust lidar ratio. Above 30◦ N, AODs from CALIOP and OSRM seem
to have a discrepancy compared with those derived from the MODIS and MLNN
methods. This disrepancy may be due to the strong and frequent atmosphere-ocean
interactions occurring in this open ocean area in July, such as precipitation and water
62
Figure 3.13: Mean AOD distribution from different algorithms in the selected open
ocean area, April 2016.
63
Figure 3.14: Zonal mean of AODs and relative differences between each algorithm,
April 2016.
Figure 3.15: AOD probability distribution according to each algorithm, April 2016.
64
evaporation over the ocean surface.
From Fig. 3.18 we see that AODs obtained from the MODIS ocean color and
MLNN methods have very similar PDFs, with a peak value (mode value) of 0.12.
The AODs derived from OSRM method has a wider PDF with more large values.
The peak value of the PDF of AODs derived from the CALIPSO method is shifted
to 0.05. The mean AODs derived from CALIOP, MODIS ocean color, MLNN, and
OSRM methods are 0.15, 0.20, 0.17, and 0.23, respectively.
Figure 3.16: Mean AOD distribution from different algorithms in the selected open
ocean area, July 2016.
The last comparison is for the month of October 2016 (Figs. 3.19-3.21). The
influence of dust events below 25◦ N is reduced as the surface temperature drops in
October. From the distributions of AODs derived from the MODIS, MLNN, and
OSRM methods, we can clearly notice the impact of decreased dust transport. AODs
65
Figure 3.17: Zonal mean of AODs and relative differences between each algorithm,
July 2016.
Figure 3.18: AOD probability distribution according to each algorithm, July 2016l,
with mean AODs of 0.15, 0.2, 0.17, and 0.23, respectively.
66
derived from CALIOP in October is similar to that in January. AODs zonal mean
also indicate the same result: the MODIS, MLNN, and OSRM methods obtain very
similar distributions at different latitudes. AODs derived from the MODIS ocean
color, MLNN, and OSRM methods have very similar PDFs, with a peak value (mode
value) around 0.09. The CALIOP AOD PDF peak is shifted to 0.05. The mean
AODs derived from the CALIOP, MODIS ocean color, MLNN, and OSRM methods
are 0.11, 0.20, 0.20, and 0.19, respectively.
Figure 3.19: Mean AOD distribution from different algorithms in the selected open
ocean area, October 2016.
3.3.4
AERONET Comparison
To better assess AODs derived from different algorithms, we would like to introduce a
reference to be used for comparison. The aerosol models being used in both CALIOP
67
Figure 3.20: Zonal mean of AODs and relative differences between each algorithm,
October 2016.
Figure 3.21: AOD probability distribution according to each algorithm, October 2016.
The mean AODs are 0.11, 0.20, 0.20, and 0.19, respectively.
68
and MODIS are somewhat based on the AERONET measurements, and AERONET
sunphotometers can directly measure AODs without any assumptions. Thus, we selected 10 AERONET sites in ocean areas and tried to find collocated CALIOP and
MODIS AOD measurements suitable for making comparisons. Figure 3.22 gives the
AERONET locations used for this purpose. Allowing relatively large spatial differences between CALIOP and AERONET observations being compared will generate a
larger data set at the expense of losing representativeness. We chose a maximum spatial separation of 80 km between the CALIOP footprint and AERONET sites. The
reason for choosing 80 km is two-fold: (1) this choice maintains enough coincidence
points, and (2) the CALIPSO horizontal resolutions are 5, 20, and 80 km. Hence, our
80 km criterion corresponds to the coarsest resolution of the CALIPSO data.
Of all collocations, only a few sites had temporally coincident measurements
(within 30 minutes of the CALIPSO overpass). In this study, a total of 316 coincidences of CALIPSO overpasses occurred within 80 km of currently active or inactive
AERONET sites within 1 hour across the globe.
By comparing the collocated AOD measurements from CALIOP, MODIS ocean
color group, MODIS atmospheric group, and the MLNN algorithm with AODs from
AERONET (Fig. 3.23), we see that CALIOP AODs show the largest discrepancy with
AERONET AODs, the slope of linear fitting is 0.65, implying that AODs derived from
CALIOP are underestimated by about 35%. In general, this bias is primarily due
to the underestimated aerosol lidar ratios adopted in the CALIOP aerosol retrieval
algorithm. The linear slopes for the OSRM-AERONET pair, the MLNN-AERONET
pair, and the MODIS-OCEAN-AERONET pair are 0.94, 0.96 and 0.96, respectively.
AODs from our OSRM method is biased by about 5-6% compared to the reference
AERONET value. This bias is similar to that of the MODIS ocean color product and
the MLNN product. The linear slope of the MODIS-ATMOSPHERE-AERONET
69
pair is 1.04, implying that the MODIS atmosphere product overestimates the AODs
over the open ocean by about 4%. These results show that our OSRM algorithm
provides good AOD retrievals that can replace the current CALIOP product.
Figure 3.22: Ten ocean AERONET sites selected for study.
70
Figure 3.23: Comparisons of AODs derived by different algorithms with those from
AERONET. The slope of the linear regression of each comparison is given in the
subgraph’s title.
3.4
Conclusions
In this study we used the ocean surface reflectance method (OSRM) to retrieve aerosol
optical depths (AODs). AODs in coastal regions, open ocean regions, and dustaffected open ocean regions have been estimated by using our OSRM approach in
conjunction with AMSR-E ocean surface wind speed measurements and CALIOP
backscatter measurements. AODs derived from the new method were compared
with those from the MODIS ocean color product, a Multi-Layer Neural Network
(MLNN) method, the MODIS atmosphere product, and AERONET sunphotometers
measurements. It has been shown that the current CALIOP method systematically
underestimates dust-type aerosols and polluted marine type aerosols because of the
uncertainties in the aerosol lidar ratios adopted in the CALIOP algorithm. in coastal
regions, our OSRM method can improve the AOD retrievals compare to the current
CALIOP method by 20-50%. In open ocean regions, AODs derived by our method is
71
also much more reliable (closer to the AERONET reference AODs) than those derived
from the current CALIOP algorithm.
72
Chapter 4
Summary and Conclusion
4.1
Conclusions
This PhD dissertation consist of three parts. In Part I, radative forcings (climate
impacts) of clouds and aerosols are introduced. Ice clouds (cirrus) and aerosols represent two large sources of uncertainty in the Earth’s radiative energy budget. To
improve representation of ice clouds and aerosols in climate models, accurate cirrus
optical depths (CODs) and aerosol optical depths (AODs) is needed. Though current
passive or active remote sensing instruments can provide an estimate of COD/AOD
based on several different retrieval algorithms, these algorithms all rely on assumptions and limitations that lead to uncertainties. The assumptions made in standard
elastic lidar retrieval methods have been discussed in Part I.
In Part II, to address issues with current remote sensing COD lidar retrieval
algorithms, a new method is developed that takes advantage of synergetic satellite
measurements to derive accurate COD values over the ocean. This new method is
based on an ocean surface reflectance model (OSRM) that relates wind-driven wave
slope variances to ocean surface backscatter. The slope variance σ 2 is related to
the ocean surface wind speed U . A nonlinear σ 2 − U relation, derived from collocated CALIPSO-AMSR-E measurements over the ocean, is adopted in this study.
To correctly apply this method, the “contamination” of the desired backscatter signal due bubbles, whitecaps, foam, ocean sub-surface backscatter, and multiple scattering must be taken into account. CALIOP is a polarization sensitive lidar. A
depolarization technique can be used to effectively isolate the “pure” ocean surface
73
backscatter signal from the effects of the ‘’contaminants”, which, in contrast to the
desired (non-depolarized) “pure” backscatter signal, will change the polarization of
the backscattered signal. The CODs derived from this method is much more reliable
than those derived from the current CALIOP operational product. CODs derived
from our OSRM method are about 18% higher than those produced by the standard
CALIOP product. Cirrus lidar ratios can also be derived from this method.
Part III primarily focuses on AOD retrievals using our new OSRM method.
The new AODs are then compared to AODs available in the standard CALIOP level-2
aerosol product, the MODIS ocean color product, the MODIS atmosphere product, a
Multi-Layer Neural Network based retrieval, and AERONET measurements. These
comparisons show that AODs derived from our OSRM method in coastal regions
and open ocean areas are much more reliable than those derived from the standard
CALIOP method.
4.2
Future Directions
It was shown in Part II that the ice cloud lidar ratios are relatively independent of
ice cloud optical depth, implying that for different COD values, the ice clouds tend
to have the same lidar ratio. Hence, the lidar ratio may depend on other factors such
as ice crystal structure, orientation, and temperature. The cirrus lidar ratio derived
in Part II may depend on cirrus temperature and other parameters. Hence, it might
be interesting to examine the temperature-dependence of cirrus lidar ratios.
Another interesting topic is the multiple scattering effect of ice clouds. CALIOP
adopts a multiple scattering factor of 0.6 to correct for multiple scattering of the laser
beam in ice clouds. Is this fixed-valued multiple scattering factor accurate or not?
Evidence to answer this question is currently lacking. However, our OSRM method
74
could provide a clue to estimate this multiple scattering factor, and determine if and
how it correlates with other cirrus parameters, such as cirrus temperature.
75
Bibliography
[1] Jassim Al-Saadi, James Szykman, R. Bradley Pierce, Chieko Kittaka, Doreen
Neil, D. Allen Chu, Lorraine Remer, Liam Gumley, Elaine Prins, Lewis Weinstock, Clinton MacDonald, Richard Wayland, Fred Dimmick, and Jack Fishman. Improving national air quality forecasts with satellite aerosol observations.
Bulletin of the American Meteorological Society, 86(9):1249–1261, 2005.
[2] A. Ansmann, M. Riebesell, U. Wandinger, C. Weitkamp, E. Voss, W. Lahmann,
and W. Michaelis. Combined raman elastic-backscatter lidar for vertical profiling of moisture, aerosol extinction, backscatter, and lidar ratio. Applied Physics
B, 55(1):18–28, Jul 1992.
[3] David Atlas, Sergey Y Matrosov, Andrew J Heymsfield, Ming-Dah Chou, and
David B Wolff. Radar and radiation properties of ice clouds. Journal of Applied
Meteorology, 34(11):2329–2345, 1995.
[4] Richard T. Austin, Andrew J. Heymsfield, and Graeme L. Stephens. Retrieval
of ice cloud microphysical parameters using the cloudsat millimeter-wave radar
and temperature. Journal of Geophysical Research: Atmospheres, 114(D8):n/a–
n/a, 2009. D00A23.
[5] Melody Avery, David Winker, Andrew Heymsfield, Mark Vaughan, Stuart
Young, Yongxiang Hu, and Charles Trepte. Cloud ice water content retrieved
from the caliop space-based lidar. Geophysical Research Letters, 39(5), 2012.
76
[6] Anthony J. Baran. From the single-scattering properties of ice crystals to climate prediction: A way forward. Atmospheric Research, 112(Supplement C):45
– 69, 2012.
[7] Anthony J. Baran, Paul J. Connolly, A. J. Heymsfield, and A. Bansemer. Using in situ estimates of ice water content, volume extinction coefficient, and
the total solar optical depth obtained during the tropical active campaign to
test an ensemble model of cirrus ice crystals. Quarterly Journal of the Royal
Meteorological Society, 137(654):199–218, 2011.
[8] Mark A Bourassa, Rosario Romero, Shawn R Smith, and James J OBrien. A
new fsu winds climatology. Journal of climate, 18(17):3686–3698, 2005.
[9] S. A. Buehler, E. Defer, F. Evans, S. Eliasson, J. Mendrok, P. Eriksson, C. Lee,
C. Jiménez, C. Prigent, S. Crewell, Y. Kasai, R. Bennartz, and A. J. Gasiewski.
Observing ice clouds in the submillimeter spectral range: the cloudice mission proposal for esa’s earth explorer 8. Atmospheric Measurement Techniques,
5(7):1529–1549, 2012.
[10] S. A. Buehler, C. Jimnez, K. F. Evans, P. Eriksson, B. Rydberg, A. J. Heymsfield, C. J. Stubenrauch, U. Lohmann, C. Emde, V. O. John, T. R. Sreerekha,
and C. P. Davis. A concept for a satellite mission to measure cloud ice water path, ice particle size, and cloud altitude. Quarterly Journal of the Royal
Meteorological Society, 133(S2):109–128, 2007.
[11] S. A. Buehler and V. O. John. A simple method to relate microwave radiances
to upper tropospheric humidity. Journal of Geophysical Research: Atmospheres,
110(D2):n/a–n/a, 2005. D02110.
77
[12] S. A. Buehler, M. Kuvatov, T. R. Sreerekha, V. O. John, B. Rydberg, P. Eriksson, and J. Notholt. A cloud filtering method for microwave upper tropospheric
humidity measurements. Atmospheric Chemistry and Physics, 7(21):5531–5542,
2007.
[13] S. A. Buehler, S. Östman, C. Melsheimer, G. Holl, S. Eliasson, V. O. John,
T. Blumenstock, F. Hase, G. Elgered, U. Raffalski, T. Nasuno, M. Satoh,
M. Milz, and J. Mendrok. A multi-instrument comparison of integrated water vapour measurements at a high latitude site. Atmospheric Chemistry and
Physics, 12(22):10925–10943, 2012.
[14] Jack L Bufton, Frank E Hoge, and Robert N Swift. Airborne measurements
of laser backscatter from the ocean surface. Applied optics, 22(17):2603–2618,
1983.
[15] Wei-Nai Chen, Chih-Wei Chiang, and Jan-Bai Nee. Lidar ratio and depolarization ratio for cirrus clouds. Appl. Opt., 41(30):6470–6476, Oct 2002.
[16] Mian Chin, Paul Ginoux, Stefan Kinne, Omar Torres, Brent N. Holben,
Bryan N. Duncan, Randall V. Martin, Jennifer A. Logan, Akiko Higurashi,
and Teruyuki Nakajima. Tropospheric aerosol optical thickness from the gocart model and comparisons with satellite and sun photometer measurements.
Journal of the Atmospheric Sciences, 59(3):461–483, 2002.
[17] Steven J Cooper and Timothy J Garrett. Identification of small ice cloud particles using passive radiometric observations. Journal of Applied Meteorology
and Climatology, 49(11):2334–2347, 2010.
78
[18] Charles Cox and Walter Munk. Measurement of the roughness of the sea surface
from photographs of the suns glitter. Journal of the Optical Society of America,
44(11):838–850, November 1954.
[19] Julien Delano and Robin J. Hogan. Combined cloudsat-calipso-modis retrievals
of the properties of ice clouds. Journal of Geophysical Research: Atmospheres,
115(D4):n/a–n/a, 2010. D00H29.
[20] Naoto Ebuchi, Hans C Graber, and Michael J Caruso. Evaluation of wind
vectors observed by quikscat/seawinds using ocean buoy data. Journal of Atmospheric and Oceanic Technology, 19(12):2049–2062, 2002.
[21] Yongzhen Fan, Wei Li, Kenneth J. Voss, Charles K. Gatebe, and Knut Stamnes.
Neural network method to correct bidirectional effects in water-leaving radiance.
Appl. Opt., 55(1):10–21, Jan 2016.
[22] Frederick G Fernald. Analysis of atmospheric lidar observations- some comments. Applied optics, 23(5):652–653, March 1984.
[23] Frederick G Fernald, Benjamin M Herman, and John A Reagan. Determination of aerosol height distributions by lidar. Journal of Applied meteorology,
11(3):482–489, April 1972.
[24] W. Frey, S. Borrmann, D. Kunkel, R. Weigel, M. de Reus, H. Schlager,
A. Roiger, C. Voigt, P. Hoor, J. Curtius, M. Krämer, C. Schiller, C. M. Volk,
C. D. Homan, F. Fierli, G. Di Donfrancesco, A. Ulanovsky, F. Ravegnani, N. M.
Sitnikov, S. Viciani, F. D’Amato, G. N. Shur, G. V. Belyaev, K. S. Law, and
F. Cairo. In situ measurements of tropical cloud properties in the west african
monsoon: upper tropospheric ice clouds, mesoscale convective system outflow,
79
and subvisual cirrus. Atmospheric Chemistry and Physics, 11(12):5569–5590,
2011.
[25] A. Garnier, J. Pelon, M. A. Vaughan, D. M. Winker, C. R. Trepte, and
P. Dubuisson. Lidar multiple scattering factors inferred from calipso lidar and
iir retrievals of semi-transparent cirrus cloud optical depths over oceans. Atmospheric Measurement Techniques, 8(7):2759–2774, 2015.
[26] A. Gettelman, J. E. Harries, and P. W. Mote. Sparc assessment of upper tropospheric and stratospheric water vapour, chapter distribution and variability of
water vapour in the upper troposphere and lower stratosphere. SPARC Report,
(2):197–263, December 2000.
[27] A. Guignard, C. J. Stubenrauch, A. J. Baran, and R. Armante. Bulk microphysical properties of semi-transparent cirrus from airs: a six year global climatology and statistical analysis in synergy with geometrical profiling data from
cloudsat-calipso. Atmospheric Chemistry and Physics, 12(1):503–525, 2012.
[28] John W. Hair, Loren M. Caldwell, David A. Krueger, and Chiao-Yao She. Highspectral-resolution lidar with iodine-vapor filters: Measurement of atmosphericstate and aerosol profiles. Appl. Opt., 40(30):5280–5294, Oct 2001.
[29] Johnathan W. Hair, Chris A. Hostetler, Anthony L. Cook, David B. Harper,
Richard A. Ferrare, Terry L. Mack, Wayne Welch, Luis Ramos Izquierdo, and
Floyd E. Hovis. Airborne high spectral resolution lidar for profiling aerosol
optical properties. Appl. Opt., 47(36):6734–6752, Dec 2008.
[30] Dennis L. Hartmann and David Doelling. On the net radiative effectiveness of
clouds. Journal of Geophysical Research: Atmospheres, 96(D1):869–891, 1991.
80
[31] Dennis L Hartmann, James R Holton, and Qiang Fu. The heat balance of the
tropical tropopause, cirrus, and stratospheric dehydration. Geophys. Res. Lett,
28(10):1969–1972, May 2001.
[32] Andrew K. Heidinger and Michael J. Pavolonis. Gazing at cirrus clouds for
25 years through a split window. part i: Methodology. Journal of Applied
Meteorology and Climatology, 48(6):1100–1116, 2009.
[33] Andrew J. Heymsfield and Greg M. McFarquhar. High albedos of cirrus in the
tropical pacific warm pool: Microphysical interpretations from cepex and from
kwajalein, marshall islands. Journal of the Atmospheric Sciences, 53(17):2424–
2451, 1996.
[34] Robin J. Hogan. Fast approximate calculation of multiply scattered lidar returns. Appl. Opt., 45(23):5984–5992, Aug 2006.
[35] BN Holben, D Tanre, A Smirnov, TF Eck, I Slutsker, N Abuhassan, WW Newcomb, JS Schafer, B Chatenet, F Lavenu, et al. An emerging ground-based
aerosol climatology: Aerosol optical depth from aeronet. Journal of Geophysical Research: Atmospheres, 106(D11):12067–12097, 2001.
[36] Brent N Holben, TF Eck, I Slutsker, D Tanre, JP Buis, A Setzer, E Vermote,
JA Reagan, YJ Kaufman, T Nakajima, et al. Aeroneta federated instrument
network and data archive for aerosol characterization. Remote sensing of environment, 66(1):1–16, 1998.
[37] Yongxiang Hu, K Stamnes, M Vaughan, Jacques Pelon, C Weimer, D Wu,
M Cisewski, W Sun, P Yang, B Lin, et al. Sea surface wind speed estimation
from space-based lidar measurements. Atmospheric Chemistry and Physics,
13(8):3593–3601, July 2008.
81
[38] Yongxiang Hu, Mark Vaughan, Zhaoyan Liu, Bing Lin, Ping Yang, David Flittner, Bill Hunt, Ralph Kuehn, Jianping Huang, Dong Wu, Sharon Rodier,
Kathy Powell, Charles Trepte, and David Winker. The depolarization - attenuated backscatter relation: Calipso lidar measurements vs. theory. Opt.
Express, 15(9):5327–5332, Apr 2007.
[39] IPCC. Environmental Physics. John Whiley Sons LTD, 1999.
[40] IPCC. Summary for Policymakers, book section SPM, page 130. Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA, 2013.
[41] Eric J Jensen, Owen B Toon, Stephanie A Vay, Joëlle Ovarlez, Randy May,
TP Bui, Cynthia Twohy, Bruce W Gandrud, Rudolf F Pueschel, and Ulrich
Schumann. Prevalence of ice-supersaturated regions in the upper troposphere:
Implications for optically thin ice cloud formation. Journal of Geophysical Research: Atmospheres, 106(D15):17253–17266, August 2001.
[42] D. Josset, J. Pelon, A. Protat, and C. Flamant. New approach to determine
aerosol optical depth from combined calipso and cloudsat ocean surface echoes.
Geophysical Research Letters, 35(10):n/a–n/a, 2008. L10805.
[43] Damien Josset, Jacques Pelon, Anne Garnier, Yongxiang Hu, Mark Vaughan,
Peng-Wang Zhai, Ralph Kuehn, and Pat Lucker. Cirrus optical depth and lidar
ratio retrieval from combined calipso-cloudsat observations using ocean surface
echo. Journal of Geophysical Research: Atmospheres, 117(D5):n/a–n/a, 2012.
D05207.
[44] Damien Josset, Raymond Rogers, Jacques Pelon, Yongxiang Hu, Zhaoyan Liu,
Ali Omar, and Peng-Wang Zhai. Calipso lidar ratio retrieval over the ocean.
Opt. Express, 19(19):18696–18706, Sep 2011.
82
[45] Ralph A. Kahn, Barbara J. Gaitley, John V. Martonchik, David J. Diner, Kathleen A. Crean, and Brent Holben. Multiangle imaging spectroradiometer (misr)
global aerosol optical depth validation based on 2 years of coincident aerosol
robotic network (aeronet) observations. Journal of Geophysical Research: Atmospheres, 110(D10):n/a–n/a, 2005. D10S04.
[46] Ralph A. Kahn, W.-H. Li, Catherine Moroney, David J. Diner, John V. Martonchik, and Evan Fishbein. Aerosol source plume physical characteristics
from space-based multiangle imaging. Journal of Geophysical Research: Atmospheres, 112(D11):n/a–n/a, 2007. D11205.
[47] Y. J. Kaufman, I. Koren, L. A. Remer, D. Tanr, P. Ginoux, and S. Fan. Dust
transport and deposition observed from the terra-moderate resolution imaging spectroradiometer (modis) spacecraft over the atlantic ocean. Journal of
Geophysical Research: Atmospheres, 110(D10):n/a–n/a, 2005. D10S12.
[48] Yoram J Kaufman, Didier Tanré, and Olivier Boucher. A satellite view of
aerosols in the climate system. Nature, 419(6903):215–223, 2002.
[49] Vitaly I Khvorostyanov and K Sassen. Microphysical processes in cirrus and
their impact on radiation. Cirrus, edited by DK Lynch, K. Sassen, D. Starr,
and G. Stephens, pages 397–432, 2002.
[50] S.-W. Kim, S. Berthier, J.-C. Raut, P. Chazette, F. Dulac, and S.-C. Yoon.
Validation of aerosol and cloud layer structures from the space-borne lidar caliop
using a ground-based lidar in seoul, korea. Atmospheric Chemistry and Physics,
8(13):3705–3720, 2008.
[51] Sang-Woo Kim, Soon-Chang Yoon, Jiyoung Kim, and Seung-Yeon Kim. Seasonal and monthly variations of columnar aerosol optical properties over east
83
asia determined from multi-year modis, lidar, and aeronet sun/sky radiometer
measurements. Atmospheric Environment, 41(8):1634–1651, 2007.
[52] S. Kinne, M. Schulz, C. Textor, S. Guibert, Y. Balkanski, S. E. Bauer,
T. Berntsen, T. F. Berglen, O. Boucher, M. Chin, W. Collins, F. Dentener, T. Diehl, R. Easter, J. Feichter, D. Fillmore, S. Ghan, P. Ginoux,
S. Gong, A. Grini, J. Hendricks, M. Herzog, L. Horowitz, I. Isaksen, T. Iversen,
A. Kirkevåg, S. Kloster, D. Koch, J. E. Kristjansson, M. Krol, A. Lauer, J. F.
Lamarque, G. Lesins, X. Liu, U. Lohmann, V. Montanaro, G. Myhre, J. Penner,
G. Pitari, S. Reddy, O. Seland, P. Stier, T. Takemura, and X. Tie. An aerocom
initial assessment optical properties in aerosol component modules of global
models. Atmospheric Chemistry and Physics, 6(7):1815–1834, 2006.
[53] C. Kittaka, D. M. Winker, M. A. Vaughan, A. Omar, and L. A. Remer. Intercomparison of column aerosol optical depths from calipso and modis-aqua.
Atmospheric Measurement Techniques, 4(2):131–141, 2011.
[54] R. C. Levy, S. Mattoo, L. A. Munchak, L. A. Remer, A. M. Sayer, F. Patadia,
and N. C. Hsu. The collection 6 modis aerosol products over land and ocean.
Atmospheric Measurement Techniques, 6(11):2989–3034, 2013.
[55] Robert C. Levy, Lorraine A. Remer, Shana Mattoo, Eric F. Vermote, and
Yoram J. Kaufman. Second-generation operational algorithm: Retrieval of
aerosol properties over land from inversion of moderate resolution imaging spectroradiometer spectral reflectance. Journal of Geophysical Research: Atmospheres, 112(D13):n/a–n/a, 2007. D13211.
[56] Zhaoyan Liu, Mark Vaughan, David Winker, Chieko Kittaka, Brian Getzewich,
Ralph Kuehn, Ali Omar, Kathleen Powell, Charles Trepte, and Chris Hostetler.
84
The calipso lidar cloud and aerosol discrimination: Version 2 algorithm and
initial assessment of performance. Journal of Atmospheric and Oceanic Technology, 26(7):1198–1213, 2009.
[57] Zhaoyan Liu, David Winker, Ali Omar, Mark Vaughan, Charles Trepte, Yong
Hu, Kathleen Powell, Wenbo Sun, and Bing Lin. Effective lidar ratios of dense
dust layers over north africa derived from the caliop measurements. Journal of
Quantitative Spectroscopy and Radiative Transfer, 112(2):204–213, 2011.
[58] U. LOHMANN, W. R. LEAITCH, L. BARRIE, K. LAW, Y. YI,
D. BERGMANN, C. BRIDGEMAN, M. CHIN, J. CHRISTENSEN,
R. EASTER, J. FEICHTER, A. JEUKEN, E. KJELLSTRM, D. KOCH,
C. LAND, P. RASCH, and G.-J. ROELOFS. Vertical distributions of sulfur
species simulated by large scale atmospheric models in cosam: Comparison
with observations. Tellus B, 53(5):646–672, 2001.
[59] F. J. S. Lopes, E. Landulfo, and M. A. Vaughan. Evaluating calipso’s 532
nm lidar ratio selection algorithm using aeronet sun photometers in brazil.
Atmospheric Measurement Techniques, 6(11):3281–3299, 2013.
[60] David K Lynch, Kenneth Sassen, David O’C Starr, and Graeme Stephens.
Cirrus. Oxford University Press, 2002.
[61] J. V. Martonchik, D. J. Diner, R. Kahn, B. Gaitley, and B. N. Holben. Comparison of misr and aeronet aerosol optical depths over desert sites. Geophysical
Research Letters, 31(16):n/a–n/a, 2004. L16102.
[62] Greg M McFarquhar, Darrel Baumgardner, Aaron Bansemer, Steven J Abel,
Jonathan Crosier, Jeff French, Phil Rosenberg, Alexei Korolev, Alfons Schwarzoenboeck, Delphine Leroy, et al. Processing of ice cloud in situ data collected
85
by bulk water, scattering, and imaging probes: Fundamentals, uncertainties,
and efforts toward consistency. Meteorological Monographs, 58:11–1, 2017.
[63] Greg M. McFarquhar, Junshik Um, Matt Freer, Darrel Baumgardner, Gregory L. Kok, and Gerald Mace. Importance of small ice crystals to cirrus properties: Observations from the tropical warm pool international cloud experiment
(twp-ice). Geophysical Research Letters, 34(13):n/a–n/a, 2007. L13803.
[64] Christopher J. McPherson, John A. Reagan, Joel Schafer, David Giles, Rich
Ferrare, John Hair, and Chris Hostetler. Aeronet, airborne hsrl, and calipso
aerosol retrievals compared and combined: A case study. Journal of Geophysical
Research: Atmospheres, 115(D4):n/a–n/a, 2010. D00H21.
[65] G. Myhre, F. Stordal, M. Johnsrud, D. J. Diner, I. V. Geogdzhayev, J. M.
Haywood, B. N. Holben, T. Holzer-Popp, A. Ignatov, R. A. Kahn, Y. J. Kaufman, N. Loeb, J. V. Martonchik, M. I. Mishchenko, N. R. Nalli, L. A. Remer,
M. Schroedter-Homscheidt, D. Tanré, O. Torres, and M. Wang. Intercomparison of satellite retrieved aerosol optical depth over ocean during the period september 1997 to december 2000. Atmospheric Chemistry and Physics,
5(6):1697–1719, 2005.
[66] Hajime Okamoto, Suginori Iwasaki, Motoaki Yasui, Hiroaki Horie, Hiroshi
Kuroiwa, and Hiroshi Kumagai. An algorithm for retrieval of cloud microphysics using 95-ghz cloud radar and lidar. Journal of Geophysical Research:
Atmospheres, 108(D7):n/a–n/a, 2003. 4226.
[67] A. H. Omar, D. M. Winker, J. L. Tackett, D. M. Giles, J. Kar, Z. Liu, M. A.
Vaughan, K. A. Powell, and C. R. Trepte. Caliop and aeronet aerosol opti-
86
cal depth comparisons: One size fits none. Journal of Geophysical Research:
Atmospheres, 118(10):4748–4766, 2013.
[68] Ali H. Omar, David M. Winker, Mark A. Vaughan, Yongxiang Hu, Charles R.
Trepte, Richard A. Ferrare, Kam-Pui Lee, Chris A. Hostetler, Chieko Kittaka,
Raymond R. Rogers, Ralph E. Kuehn, and Zhaoyan Liu. The calipso automated
aerosol classification and lidar ratio selection algorithm. Journal of Atmospheric
and Oceanic Technology, 26(10):1994–2014, 2009.
[69] Ali H. Omar, Jae-Gwang Won, David M. Winker, Soon-Chang Yoon, Oleg
Dubovik, and M. Patrick McCormick. Development of global aerosol models
using cluster analysis of aerosol robotic network (aeronet) measurements. Journal of Geophysical Research: Atmospheres, 110(D10):n/a–n/a, 2005. D10S14.
[70] C. Pierangelo, A. Chédin, S. Heilliette, N. Jacquinet-Husson, and R. Armante.
Dust altitude and infrared optical depth from airs. Atmospheric Chemistry and
Physics, 4(7):1813–1822, 2004.
[71] S. Platnick, M. D. King, S. A. Ackerman, W. P. Menzel, B. A. Baum, J. C.
Riedi, and R. A. Frey. The modis cloud products: algorithms and examples from
terra. IEEE Transactions on Geoscience and Remote Sensing, 41(2):459–473,
Feb 2003.
[72] C. M. R. Platt. Lidar and radiometric observations of cirrus clouds. Journal of
the Atmospheric Sciences, 30(6):1191–1204, 1973.
[73] C. M. R. Platt, S. A. Young, R. T. Austin, G. R. Patterson, D. L. Mitchell,
and S. D. Miller. Lirad observations of tropical cirrus clouds in mctex. part i:
Optical properties and detection of small particles in cold cirrus. Journal of the
Atmospheric Sciences, 59(22):3145–3162, 2002.
87
[74] J. A. Reagan, X. Wang, and M. T. Osborn. Spaceborne lidar calibration from
cirrus and molecular backscatter returns. IEEE Transactions on Geoscience
and Remote Sensing, 40(10):2285–2290, Oct 2002.
[75] J.A. Reagan and D.A. Zielinskie. Spaceborne lidar remote sensing techniques
aided by surface returns. Optical Engineering; (United States), 30:1, Jan 1991.
[76] J. Redemann, M. A. Vaughan, Q. Zhang, Y. Shinozuka, P. B. Russell, J. M.
Livingston, M. Kacenelenbogen, and L. A. Remer. The comparison of modisaqua (c5) and caliop (v2 and v3) aerosol optical depth. Atmospheric Chemistry
and Physics, 12(6):3025–3043, 2012.
[77] L. A. Remer and Y. J. Kaufman. Aerosol direct radiative effect at the top of
the atmosphere over cloud free ocean derived from four years of modis data.
Atmospheric Chemistry and Physics, 6(1):237–253, 2006.
[78] L. A. Remer, Y. J. Kaufman, D. Tanr, S. Mattoo, D. A. Chu, J. V. Martins,
R.-R. Li, C. Ichoku, R. C. Levy, R. G. Kleidman, T. F. Eck, E. Vermote, and
B. N. Holben. The modis aerosol algorithm, products, and validation. Journal
of the Atmospheric Sciences, 62(4):947–973, 2005.
[79] Lorraine A. Remer, Richard G. Kleidman, Robert C. Levy, Yoram J. Kaufman,
Didier Tanr, Shana Mattoo, J. Vanderlei Martins, Charles Ichoku, Ilan Koren,
Hongbin Yu, and Brent N. Holben. Global aerosol climatology from the modis
satellite sensors. Journal of Geophysical Research: Atmospheres, 113(D14):n/a–
n/a, 2008. D14S07.
[80] R. R. Rogers and M. K. Yau. Technical report. McGill University, 1976.
88
[81] William B. Rossow and Robert A. Schiffer. Isccp cloud data products. Bulletin
of the American Meteorological Society, 72(1):2–20, 1991.
[82] William B. Rossow and Robert A. Schiffer. Advances in understanding clouds
from isccp. Bulletin of the American Meteorological Society, 80(11):2261–2287,
1999.
[83] Y. Rudich, Y. J. Kaufman, U. Dayan, Hongbin Yu, and R. G. Kleidman. Estimation of transboundary transport of pollution aerosols by remote sensing
in the eastern mediterranean. Journal of Geophysical Research: Atmospheres,
113(D14):n/a–n/a, 2008. D14S13.
[84] A. M. Sayer, L. A. Munchak, N. C. Hsu, R. C. Levy, C. Bettenhausen, and
M.-J. Jeong. Modis collection 6 aerosol products: Comparison between aqua’s
e-deep blue, dark target, and merged data sets, and usage recommendations.
Journal of Geophysical Research: Atmospheres, 119(24):13,965–13,989, 2014.
2014JD022453.
[85] M. Schulz, C. Textor, S. Kinne, Y. Balkanski, S. Bauer, T. Berntsen, T. Berglen,
O. Boucher, F. Dentener, S. Guibert, I. S. A. Isaksen, T. Iversen, D. Koch,
A. Kirkevåg, X. Liu, V. Montanaro, G. Myhre, J. E. Penner, G. Pitari, S. Reddy,
Ø. Seland, P. Stier, and T. Takemura. Radiative forcing by aerosols as derived from the aerocom present-day and pre-industrial simulations. Atmospheric
Chemistry and Physics, 6(12):5225–5246, 2006.
[86] G. L. Schuster, M. Vaughan, D. MacDonnell, W. Su, D. Winker, O. Dubovik,
T. Lapyonok, and C. Trepte. Comparison of calipso aerosol optical depth retrievals to aeronet measurements, and a climatology for the lidar ratio of dust.
Atmospheric Chemistry and Physics, 12(16):7431–7452, 2012.
89
[87] Gregory L Schuster, Oleg Dubovik, and Brent N Holben. Angstrom exponent
and bimodal aerosol size distributions. Journal of Geophysical Research: Atmospheres, 111(D7), 2006.
[88] Susan Solomon, Dahe Qin, Martin Manning, Zhenlin Chen, Merlinda Marquis,
Kristen B Averyt, M Tignor, Henry L Miller, et al. Contribution of working group i to the fourth assessment report of the intergovernmental panel on
climate change, 2007, 2007.
[89] K. Stamnes, G. E. Thomas, and J. J. Stamnes. Radiative transfer in the atmosphere and ocean. Cambridge University Press, 2 edition, 2017.
[90] Max J Suarez, Arlindo daSilva, Dick Dee, Stephen Bloom, Michael Bosilovich,
Steven Pawson, Siegfried Schubert, Man-Li Wu, Meta Sienkiewicz, and Ivanka
Stajner. Documentation and validation of the goddard earth observing system
(geos) data assimilation system, version 4. 2005.
[91] C. Textor, M. Schulz, S. Guibert, S. Kinne, Y. Balkanski, S. Bauer, T. Berntsen,
T. Berglen, O. Boucher, M. Chin, F. Dentener, T. Diehl, R. Easter, H. Feichter,
D. Fillmore, S. Ghan, P. Ginoux, S. Gong, A. Grini, J. Hendricks, L. Horowitz,
P. Huang, I. Isaksen, I. Iversen, S. Kloster, D. Koch, A. Kirkevåg, J. E. Kristjansson, M. Krol, A. Lauer, J. F. Lamarque, X. Liu, V. Montanaro, G. Myhre,
J. Penner, G. Pitari, S. Reddy, Ø. Seland, P. Stier, T. Takemura, and X. Tie.
Analysis and quantification of the diversities of aerosol life cycles within aerocom. Atmospheric Chemistry and Physics, 6(7):1777–1813, 2006.
[92] Claire Tinel, Jacques Testud, Jacques Pelon, Robin J. Hogan, Alain Protat,
Julien Delano, and Dominique Bouniol. The retrieval of ice-cloud properties
90
from cloud radar and lidar synergy. Journal of Applied Meteorology, 44(6):860–
875, 2005.
[93] Sean Twomey. Atmospheric aerosols. 1977.
[94] Srikanth L. Venkata and John A. Reagan. Aerosol retrievals from calipso lidar
ocean surface returns. Remote Sensing, 8(12), 2016.
[95] Frank J Wentz and Thomas Meissner. Amsr ocean algorithm theoretical basis
document, version 2. Remote Sensing Systems, Santa Rosa, CA, 2000.
[96] David M. Winker, William H. Hunt, and Matthew J. McGill. Initial performance assessment of caliop. Geophysical Research Letters, 34(19):n/a–n/a,
2007. L19803.
[97] David M Winker, Jacques R Pelon, and M Patrick McCormick. The calipso
mission: Spaceborne lidar for observation of aerosols and clouds. Proc. SPIE,
4893(1):1–11, March 2003.
[98] David M Winker, Mark A Vaughan, Ali Omar, Yongxiang Hu, Kathleen A
Powell, Zhaoyan Liu, William H Hunt, and Stuart A Young. Overview of the
calipso mission and caliop data processing algorithms. Journal of Atmospheric
and Oceanic Technology, 26(11):2310–2323, 2009.
[99] DM Winker, Jacques Pelon, JA Coakley Jr, SA Ackerman, RJ Charlson, PR Colarco, Pierre Flamant, Q Fu, RM Hoff, C Kittaka, et al. The calipso mission: A
global 3d view of aerosols and clouds. Bulletin of the American Meteorological
Society, 91(9):1211–1229, 2010.
91
[100] Jin Wu. Mean square slopes of the wind-disturbed water surface, their magnitude, directionality, and composition. Radio Science, 25(1):37–48, February
1990.
[101] Y. Wu, M. de Graaf, and M. Menenti. Improved modis dark target aerosol
optical depth algorithm over land: angular effect correction. Atmospheric Measurement Techniques, 9(11):5575–5589, 2016.
[102] Qiong Yang, Qiang Fu, and Yongxiang Hu. Radiative impacts of clouds in
the tropical tropopause layer. Journal of Geophysical Research: Atmospheres,
115(D4):n/a–n/a, 2010. D00H12.
[103] Stuart A. Young and Mark A. Vaughan. The retrieval of profiles of particulate extinction from cloud-aerosol lidar infrared pathfinder satellite observations
(calipso) data: Algorithm description. Journal of Atmospheric and Oceanic
Technology, 26(6):1105–1119, 2009.
[104] Hongbin Yu, R. E. Dickinson, M. Chin, Y. J. Kaufman, B. N. Holben, I. V.
Geogdzhayev, and M. I. Mishchenko. Annual cycle of global distributions of
aerosol optical depth from integration of modis retrievals and gocart model
simulations. Journal of Geophysical Research: Atmospheres, 108(D3):n/a–n/a,
2003. 4128.
[105] Hongbin Yu, Lorraine A. Remer, Mian Chin, Huisheng Bian, Richard G. Kleidman, and Thomas Diehl. A satellite-based assessment of transpacific transport of pollution aerosol.
Journal of Geophysical Research: Atmospheres,
113(D14):n/a–n/a, 2008. D14S12.
92
[106] Huai-Min Zhang, John J Bates, and Richard W Reynolds. Assessment of composite global sampling: Sea surface wind speed. Geophysical Research Letters,
33(17), 2006.
[107] Z. Zhang, P. Yang, G. Kattawar, J. Riedi, L. C. Labonnote, B. A. Baum,
S. Platnick, and H.-L. Huang. Influence of ice particle model on satellite ice
cloud retrieval: lessons learned from modis and polder cloud product comparison. Atmospheric Chemistry and Physics, 9(18):7115–7129, 2009.
[108] Zhibo Zhang, Steven Platnick, Ping Yang, Andrew K. Heidinger, and Jennifer M. Comstock. Effects of ice particle size vertical inhomogeneity on the
passive remote sensing of ice clouds. Journal of Geophysical Research: Atmospheres, 115(D17):n/a–n/a, 2010. D17203.
[109] Limin Zhao and Fuzhong Weng. Retrieval of ice cloud parameters using the
advanced microwave sounding unit. Journal of Applied Meteorology, 41(4):384–
395, 2002.
93
Vita
Qiang Tang
Address
80 Buff Road, Tenafly, NJ 07670
Place of birth
Lanzhou, Gansu, China
Date of birth
October 20, 1985
Education
Stevens Institute of Technology, Hoboken, NJ
Doctoral Candidate in Physics
expected date of graduation, November 2017
Stevens Institute of Technology, Hoboken, NJ
Master in Physics
Graduated in May 2012
China Agricultural University, Beijing, China
Bachelor in Applied Meteorology
Graduated in June 2006
Professional
Experience
Teaching Assistant
PEP-112: Electricity and Magnetism
PEP-201: Modern Physics
PEP-221: Physics Lab I for Scientists
Publications
Q. Tang, Y. Hu, W. Li, J. Huang, & K. Stamnes (2017).
Cirrus Optical Depth over the Ocean obtained from
Collocated CALIPSO and AMSR-E Observations.
Applied Optics, Submitted
Q. Tang, Y. Fan, W. Li, Y. Hu, & K. Stamnes (2017).
Column Aerosol Optical Depth Evaluation over Open Ocean
with collocated CALIPSO-MODIS-AERONET Measurements.
Applied Optics, In Process
Huang, J., Fu, Q., Su, J., Tang, Q., Minnis, P., Hu, Y.,
Yi, Y., & Zhao, Q. (2009).
Taklimakan dust aerosol radiative heating derived from
CALIPSO observations using the Fu-Liou radiation model
with CERES constraints.
Atmos. Chem. Phys., 9, 4011-4021, doi:10.5194/acp-9-4011-2009
94
Chen, Y., Mao, X., Huang, J., Zhang, H., Tang, Q.,
Pan, H. & Wang, C. (2009).
Vertical distribution characteristics of aerosol during a
long-distance transport of heavy dust pollution.
China Environmental Science, 29(5), pp.449-454.
Huang, J., P. Minnis, Y. Yi, Q. Tang, X. Wang, Y. Hu, Z. Liu,
K. Ayers, C. Trepte, & D. Winker. (2007).
Summer dust aerosols detected from CALIPSO over the
Tibetan Plateau .
Geophys. Res. Lett., 34, L18805, doi:10.1029/2007GL029938
Документ
Категория
Без категории
Просмотров
0
Размер файла
4 183 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа