optical characterization of complex aerosol and cloud ... › ~ll360 › thesis_final.pdf ·...
TRANSCRIPT
Optical Characterization of Complex Aerosol and Cloud Particles:
Remote Sensing and Climatological Implications
LI LIU
Submitted in partial fulfillment of the
requirements for the degree
of Doctor of Philosophy
in the Graduate School of Arts and Sciences
COLUMBIA UNIVERSITY 2004
ABSTRACT
Optical Characterization of Complex Aerosol and Cloud Particles: Remote Sensing and Climatological Implications
Li Liu
Optical characterization of aerosol and cloud particles has been a challenge to
researchers involved in a wide range of disciplines including remote sensing and climate
studies. This thesis addresses several important atmospheric radiation problems involving
cloud and aerosol particles with complex structure. We solve these problems by (i)
extensively using state-of-the-art theoretical techniques to compute radiative properties of
nonspherical and composite atmospheric particulates; (ii) combining theoretical and high-
quality laboratory data for single scattering by irregular dust-like aerosols; (iii) applying
advanced retrieval algorithms to analyze satellite observations of tropospheric aerosols;
and (iv) validating satellite retrievals with high-quality ground-based data.
In Chapter 1, the superposition T-matrix method is used to compute electromagnetic
scattering by semi-external aerosol mixtures in the form of polydisperse, randomly
oriented two-particle clusters with touching components. The results are compared with
those for composition-equivalent external aerosol mixtures. It is concluded that
aggregation had a relatively weak effect on radiative properties of composite aerosols.
In Chapter 2, scattering and absorption characteristics of water cloud droplets
containing black carbon (BC) inclusions are calculated in the visible spectral range by a
combination of ray-tracing and Monte Carlo techniques. In addition, Lorenz-Mie
calculations are performed assuming that the same amount of BC particles are mixed
with water droplets externally. It is concluded that under normal conditions the effect of
BC inclusions on the radiative properties of cloud droplets is weak.
In Chapters 3 and 4, we compare and combine the results of laboratory
measurements of the Stokes scattering matrix for nonspherical quartz aerosols at a visible
wavelength in the scattering angle range 5°–173° and the results of Lorenz-Mie
computations for projected-area-equivalent spheres with the refractive index of quartz. A
synthetic normalized phase function is constructed and then used to analyze the potential
effect of particle nonsphericity on the results of retrievals of mineral tropospheric
aerosols based on radiance observations from Advanced Very High Resolution
Radiometer (AVHRR).
Chapter 5 presents the validation results of the aerosol optical thickness retrieved
from AVHRR channel 1 and 2 radiances. The satellite retrieved optical thickness is
compared with the accumulated historical ship-borne sun-photometer measurements.
Comparisons of single-scattering albedo and Ångström exponent values retrieved from
the AVHRR data and those measured in situ at Sable Island indicate that the currently
adopted value 0.003 can be a reasonable choice for the imaginary part of the aerosol
refractive index in global satellite retrievals.
In chapter 6, we analyze existing lidar observations of polar stratospheric clouds
(PSCs) and derive several constraints on PSC particle miscrophysical properties based on
extensive T-matrix computations of light scattering by polydispersions of randomly
oriented, rotationally symmetric nonspherical particles.
i
TABLE OF CONTENTS
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 1 Scattering and radiative properties of semi-external versus
external mixtures of different aerosol types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 2 The effect of black carbon on scattering and absorption of
solar radiation by cloud droplets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Optical properties of black carbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4 Ray-tracing/Monte Carlo model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5 Model computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.1 Internal mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.2 External mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.3 Numerical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
iv
vii
viii
1
16
16
18
21
29
30
31
32
33
33
34
35
37
40
ii
Chapter 3 Scattering matrix of quartz aerosols: comparison and
synthesis of laboratory and Lorenz-Mie results . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Measurements and Lorenz-Mie computations . . . . . . . . . . . . . . . . . . . .
3.3 Synthetic phase function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 4 Investigation of the effects of particle nonsphericity on aerosol
retrievals from AVHRR observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Effect of particle shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 5 Global validation of the operational two-channel AVHRR
retrieval product over the ocean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Validation of satellite aerosol optical thickness retrievals. . . . . . . . . . .
5.2.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.2 Ship data sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.3 Primary validation results . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 Validation of the aerosol single-scattering albedo . . . . . . . . . . . . . . . . .
5.4 Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 6 Constraints on PSC particle microphysics derived from lidar
observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
46
47
49
52
55
63
63
65
70
77
78
80
80
81
81
89
96
103
104
iii
6.2 T-matrix computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Observation data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.4 Analysis results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.5 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
105
108
110
118
iv
LIST OF FIGURES
1.1 (a) External, (b) semi-external, and (c) internal particle mixtures . . . . . . . . . . .
1.2 Phase function versus scattering angle for dust-sulfate semi-external (solid
curves) and external (dotted curves) mixtures . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 As in Fig. 1.2, but for sulfate-soot mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4 As in Fig. 1.2, but for scattering-matrix element ratios . . . . . . . . . . . . . . . . . . .
1.5 As in Fig. 1.3, but for scattering-matrix element ratios . . . . . . . . . . . . . . . . . . .
2.1 External (a) and internal (b) mixing of large cloud droplets and smaller
aerosol particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Single-scattering co-albedo ϖ−1 and asymmetry parameter g for mixtures
of cloud droplets and BC particles versus BC particle effective radius at a
wavelength of 0.55 µm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Single-scattering co-albedo ϖ−1 and asymmetry parameter g for mixtures
of cloud droplets and BC particles versus BC mass fraction . . . . . . . . . . . . . . .
2.4 Relative differences (in %) between the single-scattering co-albedo and
asymmetry parameter for external and internal mixtures of cloud droplets and
BC particles versus the BC mass fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1 Normalized distribution of the average area of the particle projection for
randomly oriented quartz aerosols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Laboratory data for nonspherical quartz aerosols and results of Lorenz-Mie
computations for projected-area-equivalent quartz spheres . . . . . . . . . . . . . . . .
3.3 The pattern of the differences between the Lorenz-Mie phase function for
spherical quartz particles (solid curve) and the phase function for
nonspherical quartz aerosols depends on the vertical position of the
experimental )(~1 Θa profile (dashed, dotted, and dot-dashed curves) . . . . . . . .
17
22
23
25
26
32
36
38
39
49
50
53
v
3.4 Synthetic and Lorenz-Mie phase functions for nonspherical quartz aerosols
and projected-area-equivalent quartz spheres, respectively . . . . . . . . . . . . . . .
4.1 Synthetic and Lorenz-Mie phase functions for nonspherical quartz aerosols
and projected-area-equivalent quartz spheres, respectively, used in the one-
channel retrieval algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Monthly averages of the ratio τN/τS and the respective scattering angle versus
longitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1 AVHRR retrieved aerosol optical thickness τSAT versus ship measurements
τSP at λ = 0.55 µm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 τSAT versus τSP for different AVHRR instrument (NOAA-7, 9, 11, 14) . . . . . .
5.3 Comparison of τSAT and τSP at λ = 0.55 µm for three increasing values of
diffuse component of surface reflection S = 0.002, 0.004, 0.005 and the
corresponding linear regression lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.4 Aerosol single-scattering albedo versus Ångström exponent for Re(m) = 1.5
and four increasing values of Im(m) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5 Monthly averages of the Ångström exponent and single-scattering albedo for
July 1999 derived from two-channel AVHRR data assuming a fixed aerosol
refractive index m = 1.5 + 0.003i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.6 The annual cycle of the aerosol single-scattering albedo measured in situ at
Sable Island . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7 The annual cycle of the aerosol optical thickness retrieved from channel–1
and –2 AVHRR data over Sable Island during the period November 1994–
December 1999 assuming that the imaginary part of the aerosol refractive
index is fixed at 0.001, 0.002, 0.003, and 0.005 . . . . . . . . . . . . . . . . . . . . . . . .
5.8 As in Fig. 5.7, but for the constrained Ångström exponent. . . . . . . . . . . . . . . .
6.1 Linear depolarization ratio )nm603(δ , backscatter color index α , and
56
67
69
85
87
88
91
92
93
94
95
vi
depolarization color index β versus effective equal-volume-sphere radius
effr for polydisperse, randomly oriented spheroids with a refractive index of
5.1=m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 As in Fig. 6.1, but for polydisperse, randomly oriented cylinders . . . . . . . . . . .
6.3 As in Fig. 6.1, but for the refractive index m = 1.308 typical of water ice at
visible wavelengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.4 As in Fig. 6.3, but for polydisperse, randomly oriented circular cylinders . . . .
6.5 The bars depict the respective ranges of the effective radius that reproduce the
values of δ , α , and β observed for type Ia PSCs, as shown in Figs. 6.1 and
6.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.6 The bars depict the respective ranges of the effective radius that reproduce the
values of δ and α observed for type Ib PSCs, as shown in Figs. 6.1 and 6.2 .
6.7 The bars depict the respective ranges of the effective radius that reproduce the
values of δ , α , and β observed for type II PSCs, as shown in Figs. 6.3 and
6.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
111
112
113
114
115
116
117
vii
LIST OF TABLES
1.1 Optical characteristics of dust-sulfate particle mixtures . . . . . . . . . . . . . . . . . .
1.2 Optical characteristics of sulfate-soot aerosol mixtures . . . . . . . . . . . . . . . . . . .
2.1 Black carbon concentration in cloud water . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Model parameter values used in this study . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1 Ship measurements of aerosol optical thickness in maritime areas used in this
study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Statistics of comparison of τSAT and τSP for three increasing values of diffuse
surface reflection S = 0.002, 0.004, 0.005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 Mean values of aerosol parameters retrieved from AVHRR data and
measured in situ at Sable Island . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1 Typical backscattering characteristics of PSCs observed by Browell et al. . . . .
6.2 Ranges of backscattering characteristics for different PSC types used in this
study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
24
32
35
82
89
96
108
109
viii
ACKNOWLEDGEMENTS
I would like to express my sincere gratitude to all those who made this thesis
possible. My deepest appreciation is to my thesis advisor Dr. Michael I. Mishchenko. I
thank him for kindly providing supervision throughout all my years in graduate school.
His ideas, insightful advice, support, and heart-warming encouragement have been
crucial to the successful completion of my thesis. I am thankful to my other advisory
committee members, Dr. James E. Hansen and Dr. Dick Ou, who also provided very
helpful guidance in my research work. I am indebted to Dr. Andrew A. Lacis. Like my
fourth mentor, he generously gave me advice on every aspect of my research. I
gratefully acknowledge Dr. William B. Rossow for his continuous encouragements,
suggestions, and teaching us climate science and science philosophy. I am also thankful
to Dr. Anthony Del Genio for being a strict but caring professor. I have benefited a lot
from his initial “push” at an early stage of my study here. I would like to thank Dr.
Larry Travis, Dr. Brian Cairns, Dr. Surabi Menon, and Dr. Makoto Sato for their
interest and comments on my research. I owe much to Dr. Yuanchong Zhang. He has
been the expert I fall back on whenever I have questions using ISCCP DX data.
This work was finished in collaboration with many other scientists. I would like to
thank Dr. Mishchenko for many of his source codes, including the T-matrix, and
Lorenz-Mie codes, and Dr. Igor V. Geogdzhayev for helping me to get familiar with the
one- and the two-channel retrieval algorithms using AVHRR observations. I gratefully
acknowledge Dr. Andreas Macke (University of Kiel, Germany) for providing us with
his Ray-tracing/Monte Carlo model. I would like to express my appreciation to Dr. Joop
W. Hovenier (Free University and University of Amsterdam, The Netherlands), Dr.
Hester Volten (University of Amsterdam and FOM-Institute for Atomic and Molecular
Physics, The Netherlands) and Dr. Olga Muñoz (Instituto de Astrofisica de Andalucia,
Spain) for providing us the results of laboratory measurements of the Stokes scattering
ix
matrix for mineral aerosols and the corresponding size distributions. Sincere gratitude
also goes to Dr. John A. Ogren at NOAA/CMDL for providing us the extensive in situ
aerosol data set collected at Sable Island, to Dr. Alexander Smirnov at NASA/GSFC,
Dr. Sergey M. Sakerin and Dr. Dmitry M. Kabanov of the Institute of Atmospheric
Optics, Russia, for providing us with historically accumulated ship-borne sun
photometer aerosol data for validation purposes.
I would like to thank Nadia T. Zakharova for her expertise in computer graphics,
Zoe Wai and Josefina Mora for helping to find papers that were not readily accessible,
Sabrina Hosein, Michael Shopsin, Angel Suarez, Raymond Encarnacion, and Kenneth
Bell for technical support. I also would like to extend my thanks to department
administrators Mia Leo and Missy Pinckert for being so devoted and kind to students.
I also would like to express thanks to all my fellow students at GISS – Jacek, Ting,
Kirstie, Mike, Jiping, Junye, Radley, Peng, Max, Reha, Johnny, Joy, Duane. They have
been a source of wisdom and have had much influence on me.
Last but not least, I am grateful to my family for their love, strength, and
continuous support.
1
INTRODUCTION
Aerosols are fine particles suspended in air. The impact of aerosols on the global
climate system through the direct and indirect radiative forcings is one of the major
uncertainties in present climate modeling (Hansen and Lacis, 1990; Charlson et al.,
1992; Lacis and Mishchenko, 1995; Hansen et al., 1998; Mishchenko et al., 2002).
Furthermore, aerosols play an important role in atmospheric chemistry and hence affect
the concentrations of other key-role atmospheric constituents, such as ozone. Therefore
knowledge of aerosol physical and chemical properties is critically important to climate
change and environmental studies.
Remote-sensing characterization of airborne particulates and their radiative effects
has been a challenging problem to atmospheric scientists due to the high spatial
inhomogeneity and heterogeneity and significant temporal variability of aerosols.
Furthermore, aerosol particles are often nonspherical, have a tendency to aggregate, and
can exist inside larger cloud droplets, thereby adding significantly to the complexity of
computing their scattering properties and radiative effects.
The objective of this thesis is to addresses several important atmospheric radiation
problems involving airborne particulates with complex morphology such as irregular
dust-like aerosols and polar stratospheric cloud particles, aerosol particle clusters, and
internal mixtures of aerosols and cloud droplets. These problems are solved by
(i) extensively using state-of-the-art theoretical techniques to compute radiative
properties of non spherical and composite scatterers;
(ii) combining theoretical and high-quality laboratory data for single scattering by
irregular dust-like aerosols;
(iii) applying advanced retrieval algorithms to analyze satellite observations of
tropospheric aerosols; and
2
(iv) validating satellite retrievals with high-quality ground-based data.
The thesis is divided into four parts with 6 chapters.
Effect of mixing
The uncertainty in current estimates of the aerosol radiative forcing (defined as an
externally imposed perturbation in the radiative energy budget of the Earth’s climate
system (IPCC, 2001)) remains large despite significant recent progress in theoretical
understanding, model simulations, and aerosol observations. There are three major areas
of uncertainty related to (i) the atmospheric burden and its anthropogenic component,
(ii) the aerosol optical parameters, and (iii) the method of translating the optical
parameters and burden into a radiative forcing (IPCC, 2001). The optical parameters
are uncertain because of poor knowledge of the aerosol size distribution, chemical
composition, state of mixing, method of mixing, degree of agglomeration, and shape.
The problem of the degree of mixing deserves special attention (IPCC, 2001) since
global modeling studies tend to assume that aerosols are externally mixed (different
aerosol particles are separated by distances much greater than their sizes and scatter
light independently of each other), which makes modeling the sources, atmospheric
transport, and radiative properties much simpler than in the case of internal mixtures
(one or several small aerosol particles are imbedded in a larger host particle). However,
many aerosol particles appear to be mixed internally (one or more particles are
imbedded in a larger host) or semi-externally (two or more aerosol particles are in
physical contact and form an aggregate). For example, Podzimek (1990) found that in a
polluted urban-marine environment haze elements can exist as insoluble, carbonaceous
particles on or in larger nonabsorbing droplets. Buseck et al. (2000) have found soot
inter-grown with sulfates in aerosol particles collected from marine particles over the
northern Atlantic, equatorial Pacific, Southern, and Indian oceans, and from polluted
3
continental regions in Europe and the United States. Significant aggregation of soot
and/or sulfate with mineral dust is common when dust from the Chinese interior mixes
with pollution plumes from major industrial cities.
Changing the mixing state and the degree of aggregation can alter aerosol optical
properties significantly. These changes may influence the results of remote sensing
studies of tropospheric aerosols and calculations of the direct aerosol forcing of climate
(e.g., Sato et al., 2003) and, therefore, must be accurately evaluated using theoretical
calculations and/or laboratory measurements. For example, uncertainties in the way
absorbing aerosols are mixed can introduce a range of a factor of two in the magnitude
of radiative forcing by black carbon (Haywood and Shine, 1995; Jacobson, 2000). The
results of Fuller et al. (1999) show that the radiative properties depend on a variety of
parameters that include particle size and shape, degree of crystallinity, type of
aggregation, density of aggregates, and type (internal or external) and extent of mixing
with non-absorbing materials such as sulfate.
Given the importance of the aerosol mixing problem, the first chapter of this thesis
will be devoted to a study of the scattering and radiative properties of semi-external
mixtures of different aerosol species, viz., sulfate, dust, and carbonaceous particles.
These three types of aerosols are believed to have a significant effect on the Earth’s
climate.
Secondly, it is well known that carbonaceous aerosols (internal or external mixtures
of organic carbon and black carbon), biomass burning aerosols in particular, are
efficient cloud condensation nuclei (e.g., Novakov and Penner, 1993; Novakov and
Corrigan, 1996). Once they are incorporated into cloud droplets, the enhanced
absorption by black carbon particles could potentially reduce the cloud albedo (Chýlek
et al., 1984), thereby causing a significant indirect forcing of climate (Charlson et al.,
1992). There have been few GCM studies evaluating the indirect forcing from
4
carbonaceous aerosols, and quantitative estimates of this effect are still highly
uncertain. Therefore, the goal of Chapter 2 is to investigate the effect of internally
mixed black carbon on scattering and absorption of solar radiation by cloud droplets
using an advanced ray-tracing/Monte Carlo model.
Effect of particle shape
The second major objective of this thesis is to study particle nonsphericity and its
effect on remote sensing of mineral tropospheric aerosols. The effect of particle shape
on scattering and absorption characteristics has been discussed in the literature for
several decades (e.g., van de Hulst, 1957; Mishchenko et al., 2000a; Mishchenko et al.,
2002a). It is commonly assumed that aerosol particles are spherical in computations of
radiative transfer because both absorption and scattering can be readily evaluated for
spherical particles by applying the Lorenz-Mie theory. Many studies indicate that this
assumption does not lead to significant errors in radiation flux computations (e.g., Lacis
and Mishchenko, 1995; Mishchenko et al., 1995; Mishchenko et al., 1997). However,
the assumption of sphericity can be problematic in satellite retrievals of aerosol optical
properties since the light scattering behavior of nonspherical particles may differ
significantly from that of spherical ones, especially in backscattering directions.
Global aerosol information can only be acquired using satellite passive and/or
active remote sensing (Karl, 1995). Most current satellite remote sensing of
tropospheric aerosols relies upon radiance measurements, such as the retrievals using
Advanced Very High Resolution Radiometer (AVHRR), Total Ozone Mapping
Spectrometer (TOMS), Moderate Resolution Imaging Spectroradiometer (MODIS),
Multiangle Imaging SpectroRadiometer (MISR), and Global Imager (GLI) data. As
discussed by Wang and Gordon (1994), the retrieval of aerosol optical thickness using
satellite reflectance measurements requires an aerosol model, namely, the specification
5
of the aerosols scattering phase function and single-scattering albedo. Currently the
majority of aerosol remote-sensing retrievals still rely on the conventional Lorenz-Mie
theory, which is strictly valid only for the spherical particle shape. However, the
climatically important dust-like and sea salt aerosols are apparently nonspherical. It is
therefore important to examine the applicability of using the Lorenz-Mie theory in
retrieval algorithms when such aerosol species are present. To address this problem
requires accurate information regarding scattering properties of aerosols. Theoretical
methods for simulating the scattering of light by various nonspherical shapes have
rapidly evolved over the last two decades (c.f., Mishchenko et al., 2000b). Numerous
studies (e.g., Nakajima et al., 1989; Mishchenko et al., 1995; Mishchenko et al., 1997;
Diner et al., 2001; Dubovik et al., 2002a; Dubovik et al., 2002b) indeed indicate the
need to account for particle non-sphericity in modeling the optical properties of dust-
like aerosols. Despite the significant progress, theoretical and numerical techniques are
still limited in their ability to model electromagnetic scattering by realistic
polydispersions of irregular particles. Therefore, laboratory measurement techniques
(Hovenier et al., 2000; Gustafson et al., 2000) remain an important source of
information on scattering properties of nonspherical aerosols.
In a recent paper, Volten et al. (2001) presented an extensive dataset which
includes the results of laboratory measurements in the visible of the Stokes scattering
matrix in a wide scattering angle range for several types of polydisperse, randomly
oriented mineral aerosols and accompanying size distribution data. A traditional
limitation of such laboratory measurements is the lack of data at very small and very
large scattering angle (in this case, from 0º to 5º and from 173º to 180º), which makes
the measurements less useful in those cases when the absolute phase function values for
the entire scattering-angle range are needed (Mishchenko et al., 1995; Krotkov et al.,
1999; Diner et al., 2001; Dubovik et al., 2002a; King et al., 1999). However, based on
6
the assumption that the diffraction forward-scattering peak is the same for spherical and
nonspherical projected-area-equivalent particles (Mishchenko et al., 1996; Mishchenko
et al., 1997), we have constructed a synthetic normalized phase function and used it in a
global one-channel retrieval algorithm to study the potential effect of nonsphericity on
the aerosol optical thickness retrieved from the AVHRR data. These results will be
described in Chapters 3 and 4.
Validation of AVHRR retrievals
The important role of tropospheric aerosols in forming the Earth’s climate has
motivated several dedicated research programs (Artaxo et al., 1998; Russell et al., 1999;
Raes and Bates, 2000), including the Global Aerosol Climatology Project (GACP)
(Mishchenko et al., 2002b). A major component of the GACP is a retrospective analysis
of the AVHRR radiance data set in order to infer the long-term global distribution of
aerosols, their properties, and seasonal and interannual variations. The papers by
Mishchenko et al. (1999, 2003) and Geogdzhayev et al. (2002) outlined the
methodology of inverting channel-1 and -2 AVHRR radiance data over the oceans,
described a detailed analysis of the sensitivity of monthly averages of retrieved aerosols
parameters to the assumptions made in different retrieval algorithms, and presented a
global aerosol climatology for the period extending from July 1983 to December 1999.
In order to retrieve aerosol optical thickness and Ångström exponent (the retrieved
aerosol parameters), several unavoidable assumptions must be made in the operational
two-channel retrieval algorithm (e.g., size distribution, real and imaginary parts of the
aerosol refractive index). These assumptions together with all the uncertainties due to
the instrument itself and the measurement procedure can generate errors in the final
retrieval products.
7
An important direction of our research has been validation of the satellite retrieval
results. This effort has involved comparisons and consistency checks with other
satellite, airborne, and ground-based datasets (e.g., Haywood et al., 2001; Smirnov et
al., 2002; Zhao et al., 2003) and models (e.g., Penner et al., 2002). This is not a simple
task because of the scale differences, collocation problems, and different approaches to
cloud screening. The ground-based networks such as the Aerosol Robotic Network
(AERONET) (Holben et al., 1998; Dubovik et al., 2000) and the Multi-Filter Rotating
Shadowband Radiometer (MFRSR) network (Alexandrov et al., 2001) provide
extensive coverage over land, but have only a few coastal sites, where the atmospheric
conditions and surface albedo may be significantly different from those in the open
ocean.
In a recent paper, Smirnov et al. (2002) summarized aerosol optical thickness
measurements in maritime and coastal areas. This comprehensive survey of ship-borne
measurements published over the last 30 years has proved to be a great asset for our
validation since much of the data were collected over undisturbed open ocean areas.
Furthermore, the availability of extensive in situ measurements of the single-scattering
albedo at Sable Island as reported by Delene and Ogren (2002) have enabled us to
examine the accuracy of the choice of the imaginary part of the refractive index in the
AVHRR retrieval algorithm, at least in the Atlantic Ocean area adjacent to that location.
These results will be described in Chapter 5.
Retrievals of PSC particle microphysics using the T-matrix method
The third part of the thesis deals with the application of the T-matrix method to
retrievals of polar stratospheric cloud (PSC) particle microphysics from lidar
observations. PSCs are important because heterogeneous reactions can occur on the
surfaces of the cloud particles. Condensation onto the clouds and subsequent reactions
8
remove nitrogen from the active gas phase (process known as “denoxification”). At the
same time, other reactions also liberate active chlorine and bromine. These compounds
then catalyze ozone destruction in the presence of sunlight, resulting in ozone depletion.
Therefore, an improved understanding of PSCs is essential to predictions of future
ozone depletion. A detailed review of PSC studies has been published by Toon et al.
(2000). The traditional classification of PSC types is based on lidar observations
(Browell et al., 1990). There are two main types of PSCs. Type Ia PSCs are
characterized by low backscattering but strong depolarization, whereas type Ib PSCs
exhibit the opposite behavior. Type II PSCs demonstrate both strong backscatter and
large depolarization ratios. Note that a non-zero depolarization ratio is indicative of the
presence of non-spherical particles (Mishchenko and Sassen, 1998).
Theoretical computations are required in order to interpret these measurements
quantitatively in terms of particle size and shape. As we have mentioned earlier, the past
two decades have seen much progress in both theoretical and numerical solutions of the
electromagnetic scattering problem for nonspherical particles. Although techniques
such as the finite difference time domain method (FDTDM; Yang and Liou, 2000; Sun
et al., 1999) and the discrete dipole approximation (DDA; Draine 2000) have no
restrictions on the particle shape, their relatively low efficiency and limited size
parameter range make their application to polydispersions of randomly oriented PSC
particles problematic. Therefore, we have based our analysis of PSC lidar observations
on the highly efficient T-matrix code (Mishchenko and Travis, 1998). The details of the
computations and the analysis results will be described in Chapter 6.
References
9
Alexandrov, M., A. Lacis, B. Carlson, and B. Cairns, 2001: MFRSR-based
climatologies of atmospheric aerosols, trace gases and water vapor. Proc. SPIE,
4168, 256-264.
Artaxo, P., P. Hobbs, Y. J. Kaufman, and Kirchhoff, editors, 1998: Smoke, clouds, and
radiation – Brazil. J. Geophys. Res., 103, 31,781-32,137.
Browell, E. V., C. F. Butler, S. Ismail, P. A. Robinette, A. F. Carter, N. S. Higdon, O.
B. Toon, M. R. Schoeberl, A. F. Tuck, 1990: Airborne lidar observations in the
wintertime Arctic stratosphere: polar stratospheric clouds. Geophys. Res. Lett., 17,
385-388.
Buseck, P. R., D. J. Jacob, M. Pósfai, J. Li, and J.R. Anderson, 2000: Minerals in the
air: An environmental perspective. Internat'l. Geol. Rev., 42, 577-593, 2000.
Charlson, R. J., S. E. Schwartz, J. M. Hales, R. D. Cess, J. A. Coakley, J. E. Hansen,
and D. J. Hoffman, 1992: Climate forcing by anthropogenic aerosols. Science, 255,
423-430.
Chýlek, P., V. Ramaswamy, and R. J. Cheng, 1984: Effect of graphitic carbon on the
albedo of clouds. J. Atmos. Sci., 41, 3076-3084.
Delene, D. J., and J. A. Ogren, 2002: Variability of aerosol optical properties at four
North American surface monitoring sites. J. Atmos. Sci., 59, 1135-1150.
Diner, D. J., W. A. Abdou, C. J. Bruegge, J. E. Conel, K. A. Crean, B. J. Gaitley, M. C.
Helmlinger, R. A. Kahn, J. V. Martonchik, S. H. Pilorz, and B. N. Holben, 2001:
MISR aerosol optical depth retrievals over southern Africa during the SAFARI-
2000 dry season campaign. Geophys. Res. Lett., 28, 3127-3130.
Dubovik, O., A. Smirnov, B. N. Holben, M. D. King, Y. J. Kaufman, T. F. Eck, and I.
Slutsker, 2000: Accuracy assessment of aerosol optical properties retrieved from
Aerosol Robotic Network (AERONET) Sun and sky radiance measurements. J.
Geophys. Res., 105, 9791-9806.
10
Dubovik, O., B. Holben, T. F. Eck, A. Smirnov, Y. J. Kaufman, M. D. King, D. Tanŕe,
and I. Slutsker, 2002: Variability of absorption and optical properties of key aerosol
types observed in worldwide locations. J. Atmos. Sci., 59, 590-608.
Dubovik, O., B. N. Holben, T. Lapyonok, A. Sinyuk, M. I. Mishchenko, P. Yang, and I.
Slutsker, 2002: Non-spherical aerosol retrieval method employing light scattering
by spheroids. Geophys. Res. Lett., 29(10)10.1029/2001GL014506.
Fuller, K. A., W. C. Malm, and S. M. Kreidenweis, 1999: Effects of mixing on
extinction by carbonaceous particles. J. Geophys. Res., 104, 15,941-15,954.
Geogdzhayev, I. V., M. I. Mishchenko, W. B. Rossow, B. Cairns, A. A. Lacis, 2002:
Global two-channel AVHRR retrievals of aerosol properties over the ocean for the
period of NOAA-9 observations and preliminary retrievals using NOAA-7 and
NOAA-11 data. J. Atmos. Sci., 59, 262-278.
Gustafson, B. Å. S, 2000: Microwave analog to light-scattering measurements. In:
Mishchenko, M. I., J. W. Hovenier, and L. D. Travis, editors. Light scattering by
nonspherical particles: theory, measurements, and applications. San Diego:
Academic press, p. 367-390.
Hansen, J. E., and A. A. Lacis, 1990: Sun and dust versus greenhouse gases: An
assessment of their relative roles in global climate change. Nature, 346, 713-719.
Hansen, J., Mki. Sato, A. Lacis, R. Ruedy, I. Tegen, and E. Matthews, 1998:
Perspective: Climate forcings in the industrial era. Proc. Natl. Acad. Sci., 95,
12,753-12,758.
Haywood, J. M., and K. P. Shine, 1995: The effect of anthropogenic sulphate and soot
aerosol on the clear sky planetary radiation budget. Geophys. Res. Lett., 22, 603-
606.
Haywood, J. M., P. N. Francis, I. Geogdzhayev, M. Mishchenko, and R. Frey, 2001:
Comparison of Saharan dust aerosol optical depth retrieved using aircraft mounted
11
pyranometers and 2-channel AVHRR algorithms. Geophys. Res. Lett., 28, 2393-
2396.
Holben, B. N., T. F. Eck, I. Slutsker, D. Tanŕe, et al., 1998: AERONET─A federated
instrument network and data archive for aerosol characterization. Remote Sens.
Environ., 66, 1-16.
Hovenier, J. W., 2000: Measuring scattering matrices of small particles at optical
wavelengths. In: Mishchenko, M. I., J. W. Hovenier, L. D. Travis, editors. Light
scattering by nonspherical particles: theory, measurements, and applications. San
Diego: Academic press, p. 355-365.
Intergovernmental Panel on Climate Change, Climate Change 2001: The Scientific
Basis, eds. Houghton, J. T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden,
X. Dai, K. Maskell, C. A. Johnson. Cambridge Univ. Press, Cambridge, U.K.
Jacobson, M. Z., 2000: A physically-based treatment of elemental carbon optics:
Implications for global direct forcing of aerosols. Geophys. Res. Lett., 27, 217-220.
Karl, T. R. (Ed.), 1995: Long-term climate monitoring by the global climate observing
system. Clim. Change, 31, 131-652.
King, M. D., Y. J. Kaufman, D. Tanŕe, and T. Nakajima, 1999: Remote sensing of
tropospheric aerosols from space: past, present, and future. Bull. Am. Meteorol.
Soc., 80, 2229-2259.
Krotkov, N. A., D. E. Flittner, A. J. Krueger, A. Kostinski, C. Riley, W. Rose, and O.
Torres, 1999: Effect of particle non-sphericity on satellite monitoring of drifting
volcanic ash clouds. JQSRT, 63, 613-630.
Lacis, A. A., and M. I. Mishchenko, 1995: Climate forcing, climate sensitivity, and
climate response: A radiative modeling perspective on atmospheric aerosols, in
Aerosol Forcing of Climate, edited by R. J. Charlson and J. Heinzenberg, pp. 11-
42, John Wiley, New York.
12
Mishchenko, M. I., A. A. Lacis, B. E. Carlson, and L. D. Travis, 1995: Nonsphericity of
dust-like tropospheric aerosols: Implications for aerosol remote sensing and climate
modeling. Geophys. Res. Lett., 22, 1077-1080.
Mishchenko, M. I., L. D. Travis, and A. Macke, 1996: Scattering of light by
polydisperse, randomly oriented, finite circular cylinders. Appl. Opt., 35, 4927-
4940.
Mishchenko, M. I., R. A. Kahn, and R. A. West, 1997: Modeling phase functions for
dustlike tropospheric aerosols using a shape mixture of randomly oriented
polydisperse spheroids. J. Geophys. Res., 102, 16,831-16,847.
Mishchenko, M.I., and K. Sassen, 1998: Depolarization of lidar returns by small ice
crystals: An application to contrails. Geophys. Res. Lett., 25, 309-312.
Mishchenko, M. I., and L. D. Travis, 1998: Capabilities and limitations of a current
Fortran implementation of the T-matrix method for randomly oriented, rotationally
symmetric scatters. JQSRT, 60, 309-324.
Mishchenko, M. I., I. V. Geogdzhayev, B. Cairns, W. B. Rossow, and A. A. Lacis,
1999: Aerosol retrievals over the ocean by use of channels 1 and 2 AVHRR data:
sensitivity analysis and preliminary results. Appl. Opt., 38, 7325-7341.
Mishchenko, M. I., J. W. Hovenier, and L. D. Travis, editors, 2000: Light scattering by
nonspherical particles: theory, measurements, and applications. San Diego:
Academic Press.
Mishchenko, M. I., W. J. Wiscombe, J. W. Hovenier, and L. D. Travis, 2000: Overview
of scattering by nonspherical particles. In: Mishchenko M. I., W. J. Hovenier, L. D.
Travis, editors. Light scattering by nonspherical particles: theory, measurements,
and applications. San Diego: Academic Press, p. 29-60.
Mishchenko, M. I., L. D. Travis, and A. A. Lacis, 2002: Scattering, absorption, and
emission of light by small particles. Cambridge: Cambridge University Press.
13
Mishchenko, M., J. Penner, and D. Anderson, 2002: Editorial: Global Aerosol
Climatology Project. J. Atmos. Sci., 59, 249.
Mishchenko, M. I., I. V. Geogdzhayev, L. Liu, J. A. Ogren, A. A. Lacis, W. B. Rossow,
J. W. Hovenier, H. Volten, and O. Muñoz, 2003: Aerosol retrievals from AVHRR
radiances: effects of particle nonsphericity and absorption and an updated long-
term global climatology of aerosol properties. JQSRT, 79-80, 953-972.
Nakajima, T., M. Tanaka, M. Yamano, et al., 1989: Aerosol optical characteristics in
the yellow sand events observed in May, 1982 at Nagasaki. II: Models. J. Meteorol.
Soc. Jpn., 67, 279-291.
Novakov, T., and J. E. Penner, 1993: Large contribution of organic aerosols to cloud-
condensation-nuclei concentrations. Nature, 365, 823-826.
Novakov, T., and C. E. Corrigan, 1996: Cloud condensation nucleus activity of the
organic component of biomass smoke particles. Geophys. Res. Lett., 23, 2141-
2144.
Penner, J. E., S. Y. Zhang, M. Chin, et al., 2002: A comparison of model- and satellite-
derived aerosol optical depth and reflectivity. J. Atmos. Sci., 59, 441-460.
Podzimek, J., 1990: Physical properties of coarse aerosol particles and haze elements in
a polluted urban-marine environment. J. Aerosol Sci., 21, 299-308.
Pilinis, C., and X. Li, 1998: Particle shape and internal inhomogeneity effects on the
optical properties of tropospheric aerosols of relevance to climate forcing. J.
Geophys. Res., 103, 3789-3800.
Raes, F., and T. Bates, editors, 2000: The second aerosol characterization experiment.
Tellus, 52B, 109-908.
Russell, P. B., P. V. Hobbs, and L. L. Stowe, editors, 1999: Tropospheric aerosol
radiative forcing observation experiment. J. Geophys. Res., 104, 2213-2319.
14
Sato, Mki., J. Hansen, D. Koch, A. Lacis, R. Ruedy, O. Dubovik, B. Holben, M. Chin,
and T. Novakov, 2003: Global atmospheric black carbon inferred from AERONET.
Proc. Natl. Acad. Sci., 100, 6319-6324, doi:10.1073/pnas.0731897100.
Smirnov, A., B. N. Holben, Y. J. Kaufman, O. Dubovik, T. F. Eck, I. Slutsker, C.
Pietras, and R. N. Halthore, 2002: Optical properties of atmospheric aerosol in
maritime environments. J. Atmos. Sci., 59, 501-523.
Sun, W., Q. Fu, and Z. Chen, 1999: Finite-difference time-domain solution of light
scattering by dielectric particles with a perfectly matched layer absorbing boundary
condition. Appl. Opt., 38, 3141-3151.
Toon, O. B., A. Tabazadeh, E. V. Browell, J. Jordan, 2000: Analysis of lidar
observations of Arctic polar stratospheric clouds during January 1989. J. Geophys.
Res., 105, 20,589-20,615.
van de Hulst, H. C., 1957: Light scattering by small particles, John Wiley, New York,
1957.
Volten, H., O. Muñoz, E. Rol, J. F. de Hann, W. Vassen, J. W. Hovenier, K. Muinonen,
and T. Nousiainen, 2001: Scattering matrices of mineral aerosol particles at 441.6
and 632.8 nm. J. Geophys. Res., 106, 17,375-17,401.
Wang, M., and H. R. Gordon, 1994: Estimating aerosol optical properties over the
oceans with the multiangle imaging spectroradiometer: some preliminary results.
Appl. Opt., 33, 4042-4057.
Yang, P., K. N. Liou, 2000: Finite difference time domain method for light scattering by
nonspherical and inhomogeneous particles. In M. I. Mishchenko, J. W. Hovenier,
L. D. Travis, editors. Light scattering by nonspherical particles: theory,
measurements, and applications. San Diego: Academic Press, San Diego, p.173-
221.
15
Zhao, T. X.-P., I. Laszlo, B. N. Holben, C. Pietras, and K. J. Voss, 2003: Validation of
two-channel VIRS retrievals of aerosol optical thickness over ocean and
quantitative evaluation of the impact from potential subpixel cloud contamination
and surface wind effect. J. Geophys. Res., 108 (D3), AAC 7.
16
Chapter 1
Scattering and radiative properties of semi-external versusexternal mixtures of different aerosol types1
Michael I. Mishchenkoa, Li Liua,b, Larry D. Travisa, Andrew A. Lacisa
aNASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025, USAbDepartment of Earth and Environmental Sciences, Columbia University, 2880 Broadway, New
York, NY 10025, USA
Abstract
The superposition T-matrix method is used to compute the scattering of unpolarized
light by semi-external aerosol mixtures in the form of polydisperse, randomly oriented
two-particle clusters with touching components. The results are compared with those for
composition-equivalent external aerosol mixtures, in which the components are widely
separated and scatter light in isolation from each other. It is concluded that aggregation
is likely to have a relatively weak effect on scattering and radiative properties of multi-
component tropospheric aerosols and can be replaced by the much simpler external-
mixture model in remote sensing studies and atmospheric radiation balance
computations.
Keywords: Aerosols; Scattering; Optical cross sections; Single-scattering albedo;
Asymmetry parameter; Scattering matrix; Remote sensing; Atmospheric radiation
1.1 Introduction
1 Submitted to Journal of Quantitative Spectroscopy & Radiative Transfer, 2003.
17
Different kinds of tropospheric aerosols (e.g., dust, carbonaceous, and sulfate
particles) can be suspended in air both in the form of external mixtures (different
aerosol particles are separated by distances much greater than their sizes and scatter
light independently of each other), semi-external mixtures (two or more aerosol
particles are in physical contact and form an aggregate), and internal mixtures (one or
several small aerosol particles are imbedded in a larger host particle) (see Fig. 1.1). For
example, Podzimek [1] found that in a polluted urban-marine environment haze
elements can exist as insoluble, carbonaceous particles on or in larger nonabsorbing
droplets. Significant aggregation of soot and/or sulfate with mineral dust is common
when dust from the Chinese interior mixes with pollution plumes from major industrial
cities. Since aerosol particles forming an aggregate are in the near field zone of each
(a)
(b)
(c)
Figure 1.1: (a) External, (b) semi-external, and (c) internal particle mixtures.
18
other, the scattering and radiative properties of semi-external aerosol mixtures can differ
from those of composition-equivalent external mixtures. These potential differences
may influence the results of remote sensing studies of tropospheric aerosols and
calculations of the direct aerosol forcing of climate (see, e.g., [2] and references therein)
and, therefore, must be accurately evaluated using theoretical calculations and/or
laboratory measurements [3–5].
In this paper we calculate the optical cross sections and the elements of the Stokes
scattering matrix of semi-external two-component mixtures of carbonaceous, dust, and
sulfate aerosols – the three types of aerosols which are believed to have a significant
effect on the earth’s climate – and compare them with those of equivalent external
mixtures computed with the standard Lorenz-Mie theory [6,7]. This comparison is used
to derive conclusions about the likely effect of aggregation on the scattering and
radiative properties of multi-component aerosols.
1.2 Computations
Important single-scattering characteristics of randomly oriented particles with a
plane of symmetry are the ensemble-averaged scattering, Csca, and extinction, Cext, cross
sections and the elements of the normalized scattering matrix )(ΘF , where Θ is the
scattering angle [7]. In the standard {I, Q, U, V} representation of polarization, the
scattering matrix has the well-known block-diagonal form [6,7],
−
=
)()(00
)()(00
00)()(
00)()(
)(
42
23
21
11
ΘΘΘΘ
ΘΘΘΘ
Θ
ab
ba
ab
ba
F , (1.1)
so that only eight elements of )(ΘF are nonzero and only six of them are independent.
19
The (1, 1) element of the scattering matrix )(1 Θa is traditionally called the phase
function and satisfies the normalization condition
∫ =π
ΘΘΘ0
1 1)(sind21
a . (1.2)
Additional useful quantities are the ensemble-averaged absorption cross section, Cabs =
Cext – Csca, the single-scattering albedo, ϖ = Csca/Cext, and the asymmetry parameter
defined as
∫=π
ΘΘΘΘ0
1 cos)(sind21
ag . (1.3)
The computations reported here were performed at a visible wavelength of λ =
0.628 µm assuming that each aerosol mode is represented by the power law size
distribution of the following type [7,8]:
≤≤
−=−
otherwise.,0
,,2
)(21
32
12
2
22
21 rrrr
rr
rr
rn (1.4)
The effective radius reff and effective variance effv of the size distribution are defined
as
2eff )(d
1 2
1
rrrnrG
rr
rπ∫= , (1.5)
22eff2
effeff )()(d
1 2
1
rrrrnrrG
r
rπ−= ∫v , (1.6)
where
2)(d2
1
rrnrGr
rπ∫= (1.7)
is the average area of the particle geometrical projection [7,8]. Dust aerosols are usually
20
characterized by a broad range of effective radii from about 0.2 µm to greater than 5
µm. The most common dust mode occurring at a large distance from the source is
around 1 µm in radius. The effective radius of sulfate particles ranges from less than 0.1
to larger than 0.5 µm and depends on relative humidity. Soot aerosols are usually fine
particles with effective radii around 0.1 µm.
In our computations, we have used two types of aerosol particle mixture. Mixture 1
is composed of dust-like particles with an effective radius of 1 µm and an equal number
of sulfate particles. The effective radius of the sulfate component was set at the
following four representative values: 0.1, 0.2, 0.5, and 1 µm. Mixture 2 consists of equal
numbers of sulfate particles with an effective radius of 0.5 µm and soot aerosols. The
effective radius of soot particles was set at 0.05, 0.1, 0.2, and 0.5µm. We used the
following values of the relative refractive index for the three different aerosol species:
1.53 + 0.008i for dust, 1.44 for sulfates, and 1.75 + 0.435i for soot [9]. The effective
variance for all aerosol types was fixed at 0.2, thereby representing a moderately wide
size distribution. Although solid aerosol particles should be presumed to have
nonspherical shapes, our main interest here is in evaluating the potential effect of
aggregation on light scattering. Therefore, for the sake of simplicity, we have assumed
that all three aerosol species consist of spherical particles.
We calculated the scattering of unpolarized incident light by polydisperse,
randomly oriented two-sphere clusters with touching components and compared the
results with the results of Lorenz-Mie calculations assuming that the same component
particles were mixed externally and acted as independent scatterers. In our
computations for two-sphere clusters, we used a highly efficient code developed by
Mishchenko and Mackowski [10] and publicly available at
http://www.giss.nasa.gov/~crmim. This code is based on the T-matrix solution of
Maxwell’s equations and precisely computes the scattering of light by randomly
21
oriented two-sphere clusters with sizes comparable to and larger than the wavelength.
The efficiency of the methods is the result of combining the power of the superposition
approach in treating light scattering by composite particles [11,12] and the analyticity of
the T-matrix formulation in application to randomly oriented nonspherical scatterers
[13]. The main idea of the method is to employ the superposition approach to calculate
the T matrix of a bisphere in the natural coordinate system with the z-axis connecting
the centers of the component spheres and then to use this T matrix in an analytical
procedure to directly compute the orientation-averaged optical cross sections and the
elements of the scattering matrix for randomly oriented bispheres. The analytical
averaging over orientations makes this approach much faster than that based on the
standard numerical averaging and, thus, suitable for computations for realistic size
distributions. The Lorenz-Mie code used is described in [7] and is also available at
http://www.giss.nasa.gov/~crmim. The numerical data are summarized in Figs. 1.2–1.5
and Tables 1.1 and 1.2 and are discussed in the following session.
1.3 Discussion
Figures 1.2 and 1.3 illustrate the differences between the phase functions for semi-
external (solid curves) and external (dotted curves) aerosol mixtures. One can see that
the relative phase-function differences are rather small and, on average, do not exceed
20%. A notable exception is the exact forward-scattering direction, where the
interference effects result in a significant enhancement of the aggregate phase functions
[14]. The phase-function differences are especially small when the effective radius of
one component is much smaller than that of the other, in which cases the thin solid and
dotted curves are almost (or even completely) indistinguishable. This obviously
happens because the contribution of the smaller component to the total scattered signal
becomes negligibly small.
22
The differences in the other scattering-matrix elements are also surprisingly small
(Figs. 1.4 and 1.5). The only clear indications of the fact that two-sphere aggregates are
nonspherical particles are the deviation of the ratio )()( 12 ΘΘ aa from unity and the
deviation of the ratios )180()180( 13 °° aa and )180()180( 14 °° aa from –1 (cf. [7,14]).
Tables 1.1 and 1.2 demonstrate that the differences in the integral photometric
characteristics (the ensemble averaged extinction, scattering, and absorption cross
sections, the single-scattering albedo, and the asymmetry parameter) between the semi-
external and the composition-equivalent external mixtures do not exceed 15% and are
0 60 120 180Scattering angle (deg)
0.1
1
10
100
1000
a1
reff (µm)
0.1
1.0
Figure 1.2: Phase function versus scattering angle for dust-sulfate semi-external (solid curves)and external (dotted curves) mixtures. The effective radius of dust particles is fixed at 1 µm.The thin and thick curves, respectively, show the results for the smallest (reff = 0.1 µm) and thelargest (reff = 1 µm) sulfate aerosol mode considered in this paper.
23
often much smaller, particularly for Cabs, ϖ , and g. Owing to mutual shadowing, the
scattering cross sections of the semi-externally mixed aerosols are always smaller than
those of their externally mixed counterparts, whereas the absorption cross section hardly
changes.
The small differences in the single-scattering albedo and the asymmetry parameter
in combination with the small phase function and polarization differences suggest that
aggregation is likely to have a relatively weak effect on remote sensing retrievals of the
0 60 120 180
Scattering angle (deg)
0.1
1
10
100
a1
reff (µm)
0.05
0.5
Figure 1.3: As in Fig. 1.2, but for sulfate-soot mixtures. The effective radius of sulfateparticles is fixed at 0.5 µm. The thin and thick curves, respectively, show the results for thesmallest (reff = 0.05) and the largest (0.5 µm) soot aerosol mode considered in this paper.
24
aerosol physical characteristics and computations of the aerosol direct radiative forcing.
This is especially true when one of the aerosol components is much larger than the other
and, thereby, dominates the total optical characteristics of the mixture. It is also quite
obvious that the semi-external mixing does not cause as pronounced an enhancement of
the black carbon absorption as that caused by internal mixing [3]. As a consequence, it
appears to be rather safe to use the much simpler external mixture model in radiative
transfer computations irrespective of the actual form of mixing.
Table 1.1: Optical characteristics of dust-sulfate particle mixturesa
0.1 � 0.2 � � 0.5 � 1reff (µm)
SEb E � SE E � � SE E � SE E
Cext (µm2) 4.540 4.542 4.578 4.635 �� 5.319 5.794 � 8.309 9.361Csca (µm2) 3.952 3.955 3.988 4.048
��
4.727 5.207� �
7.716 8.774Cabs (µm2) 0.588 0.587 0.590 0.587
� �
0.592 0.587� �
0.593 0.5870.871 0.871 0.871 0.873
� �
0.889 0.899� �
0.929 0.937g 0.712 0.712 0.711 0.712 � � 0.716 0.715 � 0.726 0.720
a The effective radius of dust particles is fixed at 1 µm.b “SE” denotes semi-external mixing and “E” denotes external mixing.
Table 1.2: Optical characteristics of sulfate-soot aerosol mixtures a
0.05 0.1 0.2 0.5reff (µm)
SEb E SE E SE E SE E
Cext (µm2) 1.259 1.259 1.272 1.282 1.367 1.426 2.176 2.423Csca (µm2) 1.256 1.257 1.255 1.266 1.269 1.332 1.563 1.822Cabs (µm2) 0.003 0.002 0.017 0.016 0.098 0.094 0.613 0.601
0.998 0.998 0.986 0.988 0.928 0.934 0.718 0.752g 0.722 0.721 0.726 0.720 0.737 0.717 0.780 0.751
a The effective radius of sulfate particles is 0.5 µm.b “SE” denotes semi-external mixing and “E” denotes external mixing.
The analysis of this paper is limited to simple two-particle aggregates. Since more
ϖ
ϖ
25
complex aerosol aggregates can also be encountered in the atmosphere, we plan to
extend this study by performing calculations using the code described in [15] and
applicable to randomly oriented clusters with three or more components.
0
25
50
75
100
a 2 /a
1 (%
)
−100
−50
0
50
100
a 3 /a
1 (%
)
−100
−50
0
50
100
a 4 /a
1 (%
)
0 60 120 180Scattering angle (deg)
−100
−50
0
50
100
−b1 /a
1 (%
)
0 60 120 180Scattering angle (deg)
−100
−50
0
50
100
b 2 /a
1 (%
)
reff (µm)
0.1
1.0
Figure 1.4: As in Fig. 1.2, but for scattering-matrix element ratios.
26
Acknowledgement
This research was supported by the NASA Radiation Sciences Program managed
by Donald Anderson.
0
25
50
75
100
a 2 /a
1 (%
)
−100
−50
0
50
100
a 3 /a
1 (%
)
−100
−50
0
50
100
a 4 /a
1 (%
)
0 60 120 180Scattering angle (deg)
−100
−50
0
50
100
−b1 /a
1 (%
)
0 60 120 180Scattering angle (deg)
−100
−50
0
50
100
b 2 /a
1 (%
)
reff (µm)
0.05
0.5
Figure 1.5: As in Fig. 1.3, but for scattering-matrix element ratios.
27
References
[1] Podzimek J. Physical properties of coarse aerosol particles and haze elements in a
polluted urban-marine environment. J Aerosol Sci 1990;21:299–308.
[2] Sato M, Hansen J, Koch D, Lacis A, Ruedy R, Dubovik O, Holben B, Chin M,
Novakov T. Global atmospheric black carbon inferred from AERONET. Proc
Natl Acad Sci, in press.
[3] Chýlek P, Videen G, Ngo D, Pinnick RG, Klett JD. Effect of black carbon on the
optical properties and climate forcing of sulfate aerosols. J Geophys Res
1995;100:16325–32.
[4] Fuller KA. Scattering and absorption cross sections of compounded spheres. II.
Calculations for external aggregation. J Opt Soc Am A 1995;12:881–92.
[5] Fuller KA, Malm WC, Kreidenweis SM. Effects of mixing on extinction by
carbonaceous particles. J Geophys Res 1999;104:15941–54.
[6] van de Hulst HC. Light scattering by small particles. New York: Wiley, 1957.
[7] Mishchenko MI, Travis LD, Lacis AA. Scattering, absorption, and emission of
light by small particles. Cambridge: Cambridge University Press, 2002.
[8] Hansen JE, Travis LD. Light scattering in planetary atmospheres. Space Sci Rev
1974;16:527–610.
[9] d’Almeida GA, Koepke P, Shettle EP. Atmospheric aerosols. Hampton, VA:
Deepak, 1991.
[10] Mishchenko MI, Mackowski DW. Light scattering by randomly oriented
bispheres. Opt Lett 1994;15:1604–6.
28
[11] Fuller KA. Optical resonances and two-sphere systems. Appl Opt 1991;30:4716-
31.
[12] Mackowski DW. Calculation of total cross sections of multiple-sphere clusters. J
Opt Soc Am A 1994;11:2851–61.
[13] Mishchenko MI. Light scattering by randomly oriented axially symmetric
particles. J Opt Soc Am A 1991;8:871–82.
[14] Mishchenko MI, Mackowski DW, Travis LD. Scattering of light by bispheres
with touching and separated components. Appl Opt 1995;34:4589–99.
[15] Mackowski DW, Mishchenko MI. Calculation of the T matrix and the scattering
matrix for ensembles of spheres. J Opt Soc Am A 1996;13:2266–78.
29
Chapter 2
The effect of black carbon on scattering and absorption ofsolar radiation by cloud droplets1
Li Liua,b, Michael I. Mishchenkoa, Surabi Menona,b, Andreas Mackec,Andrew A. Lacisa
aNASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025, USAbDepartment of Earth and Environmental Sciences, Columbia University, 2880 Broadway, New
York, NY 10025, USAcInstitute for Oceanography, University of Kiel, Duesternbrooker Weg 20, D-24105 Kiel,
Germany
Abstract
Scattering and absorption characteristics of water cloud droplets containing black
carbon (BC) inclusions are calculated at a visible wavelength of 0.55 µm by a
combination of ray-tracing and Monte Carlo techniques. In addition, Lorenz-Mie
calculations are performed assuming that the same amount of BC particles are mixed
with water droplets externally. The results show that it is unlikely under normal
conditions that BC aerosols can modify scattering and absorption properties of cloud
droplets in any significant way except for geographical locations very close to major
sources of BC. The differences in the single-scattering co-albedo and asymmetry
parameter between BC-fraction-equivalent internal and external mixtures are negligibly
small for normal black carbon loadings, which makes possible the use of the much
simpler external mixing model in radiative transfer modeling irrespective of the actual
form of mixing. For a fixed amount of BC internally mixed with cloud droplets, the
absorption is maximal when the effective radius of the BC inclusions is about 0.05–0.06
1 Published in Journal of Quantitative Spectroscopy & Radiative Transfer, 74, 195-204, 2002.
30
µm.
Keywords: Black carbon; Cloud water droplets; Internal and external mixing; Single-
scattering co-albedo; Asymmetry parameter
2.1 Introduction
Black carbon (BC) has long been recognized as an important atmospheric pollutant
[1]. It plays a significant role in the absorption of solar radiation by atmospheric
aerosols and possibly also by clouds. Enhanced absorption by black carbon particles
imbedded in water droplets could potentially reduce the cloud albedo [2], thereby
causing a significant indirect forcing of climate [3].
The effect of BC impurities on the absorption of solar radiation by cloud water
droplets was considered by Danielson et al. [4] using an idealized model with an inner
sphere of an absorbing aerosol particle surrounded by a concentric shell of pure water.
Chýlek et al. [2] calculated the spectral cloud albedo using an effective medium
approximation, which substitutes a heterogeneous internal water-carbon mixture by a
fictitious homogeneous material characterized by an effective refractive index.
However, the applicability of various effective medium approximations to water
droplets containing relatively large soot inclusions remains somewhat uncertain and
requires further theoretical and experimental research [5]. Recently, the exact solution
for electromagnetic scattering by a host sphere containing one or several non-concentric
spherical inclusions has become available (e.g., [6–10]). However, the practical
implementations of this solution are still limited in terms of the maximal size parameter
of the host and the number and size of inclusions and become very time-consuming
when applied to realistic cloud water droplets with multiple randomly positioned
inclusions. Therefore, in this paper we address the problem of scattering and absorption
31
of solar radiation by cloud droplets containing BC inclusions using the ray-
tracing/Monte Carlo approach developed by Macke et al. [11–13]. This approximate
technique assumes that the size of the host particle is much larger than the wavelength
of the incident radiation and that the inclusions are randomly and sparsely distributed,
and we expect that these conditions are adequately satisfied by an average cloud water
droplet (radius ~10 µm) at visible wavelengths. The results thus obtained are compared
with those calculated with the standard Lorenz-Mie formulation and assuming that the
same amount of BC particles are mixed with water droplets externally. This
comparison is used to derive conclusions about the specific effects of internal mixing on
radiative properties of cloud droplets contaminated with soot.
2.2 Optical properties of black carbon
An important consequence of the presence of BC in the atmosphere is increased
absorption of solar radiation [14]. The magnitude of absorption depends on the BC
refractive index (especially its imaginary part) and the size, shape, and porosity of BC
particles. It also depends on whether the BC particles are mixed with cloud droplets
internally or externally and is a function of the average size of the cloud droplets and
the exact location of BC inclusions within the droplets. The external and internal
mixture models are shown schematically in Fig. 2.1. Since there is no obvious reason to
assume that BC inclusions should have a preferential location inside water droplets, the
scattering and absorption properties of contaminated water droplets should be averaged
over a random distribution of BC particle locations.
The imaginary part of the BC complex refractive index depends on the original
composition of the material that was burned and on the burning process itself.
Consequently, there are no universal well-defined optical constants of BC [15].
Following the suggestion by d’Almeida et al. [16], we have adopted for this study the
32
refractive index 1.75 + 0.44i. This refractive index was also used in recent publications
by Chýlek et al. [14,15].
(a)
(b)
Figure 2.1: External (a) and internal (b) mixing of large cloud droplets and smaller aerosolparticles.
Table 2.1: Black carbon concentration in cloud water
ReferencesMean BC
)kgµg( 1−Range of BC
)kgµg( 1− LocationDegree ofinternalmixing (%)
Twohy et al. [29] 23–79 Southern California
Chýlek et al. [23] 40 10–61Nova Scotia
(Canada)9
Kou [30] 16 8–41Nova Scotia
(Canada)6
Bahrmann and Saxena [31] 74.2 20.7–196.9 North Carolina 13
Hallberg et al. [34] 6
2.3 Observations
In order to evaluate the effect of BC on the radiation balance of the Earth’s
atmosphere, one needs global information on the distribution of BC throughout the
33
atmosphere [17,18]. Although there have been several measurements of the BC
concentration in cloud water droplets, the observational data are incomplete, and no
clear global or regional picture can be deduced. Rather than rely on a definite set of
local observations, Chýlek et al. [15] estimated the lower and upper bounds on the black
carbon mixing ratio (by mass) in cloud water for stratus type clouds to be 9104.2 −× and
6108 −× . A summary of BC concentration measurements is given in Table 2.1.
2.4 Ray-tracing/Monte Carlo model
The ray-tracing/Monte Carlo technique is a simple and efficient hybrid method
combining ray optics and Monte Carlo radiative transfer concepts. This method permits
the treatment of light scattering and absorption by arbitrarily shaped host particles
containing small, randomly positioned spherical and nonspherical inclusions and is
valid for host particles that are large compared to the wavelength of the incident
radiation. The ray-tracing program takes care of individual reflection and refraction
events at the outer boundary of the host particle, while the Monte Carlo routine
simulates the process of multiple internal scattering by the inclusions. A detailed
description of this model is provided in [11–13]. The respective computer code is
publicly available at the following URL http://www.ifm.uni-kiel.de/fb/fb1/me/research/
Projekte/RemSens/SourceCodes/codes.html.
2.5 Model computations
Since BC is found in cloud droplets in measurable amounts, it has been of interest
from the climatic standpoint to determine whether the typical BC concentrations could
cause significant excess absorption of solar radiation relative to absorption by cloud
droplets alone. This absorption would be caused by both increased single-scattering co-
34
albedo ϖ−1 and increased asymmetry parameter g of the cloud droplet/BC mixture,
where ϖ is the single-scattering albedo. To find the upper bound on the absorption
effect of BC, we consider the maximal plausible BC mixing ratio in cloud water
6108 −× (by mass) as estimated by Chýlek et al. [15]. The specific density of BC is
largely unknown and depends on the actual burning process. In this paper, we have
adopted a low BC specific density of 3cmg1 −=ρ consistent with our desire to estimate
an upper limit on the BC absorption effect. The refractive index of water at a visible
wavelength of 0.55 µm and the water specific density are taken to be i10233.1 9−×+
and 1 3cmg − , respectively. Below we will consider separately the cases of internal and
external mixing of BC and cloud water.
2.5.1 Internal mixing
First, we consider a spherical 10 µm-radius water droplet containing a 6108 −×
fraction (by mass) of polydisperse, randomly distributed, spherical BC particles. The
size distribution of the small BC inclusions was determined on only a very few
occasions (e.g., [19–22]). Most of BC particles are in the submicron size range with a
typical mode radius between 0.03 and 0.06 µm [23]. Instead of specifying a fixed
effective radius effr of the BC inclusions, we vary it from 0.01 to 0.22 µm, with effr =
0.22 µm corresponding to one inclusion per host particle under the condition that such a
composite particle contains the 6108 −× BC fraction (by mass), to see how the effective
radius of these inclusions affects the total scattering and absorption properties of cloud
droplets. We assume that the size distribution of the BC particles is given by the
standard gamma distribution [24]:
−×= −
ab
rrrn bb expconst)( )31( , (2.1)
35
where effra = and effvb = . In this study, the effective variance effv is fixed at 0.1,
representing a moderately wide size distribution. The model variable values used in this
study are summarized in Table 2.2. The scattering and absorption properties of BC-
contaminated cloud droplets at the wavelength λ = 0.55 µm (corresponding to the
maximum in the spectral distribution of the solar radiation) have been calculated by a
combination of ray-tracing and Monte Carlo techniques as mentioned above. The
single-scattering properties of BC particles have been computed assuming the spherical
particle shape and using the Lorenz-Mie code described by Mishchenko et al. [25] and
available at http://www.giss.nasa.gov/~crmim.
Table 2.2: Model parameter values used in this study a
)cmg( 3−ρ Refractive index effr (µm) veffMixingratio bymass BC W BC W BC W BC W
Relative BCRefractiveindex b
6108 −× 1 1 1.75+0.44i 910233.1 −×+ i0.01–0.22
10 0.1 – i331.0316.1 +a BC = black carbon; W = water.b The BC refractive index is divided by that of water at 0.55µm.
2.5.2 External mixing
BC particles not only can act as cloud condensation nuclei and be found inside
cloud droplets but can also exist outside the droplet as interstitial aerosols. The optical
properties of externally mixed cloud droplets and BC aerosols can be well represented
by the traditional Lorenz-Mie theory provided that the cloud and aerosol particles are
widely separated [26]. We have performed the Lorenz-Mie computation assuming the
same mass fraction of BC. Assuming independent scattering, the total single-scattering
albedo ϖ and asymmetry parameter g of the mixture are given by [27]
36
0.00 0.05 0.10 0.15 0.20 0.25Effective Radius (µm)
0.0002
0.0003
0.0004
0.0005
Sin
gle-
Sca
tterin
g C
o-A
lbed
o
Internal Mixing
External Mixing
0.00 0.05 0.10 0.15 0.20 0.25Effective Radius (µm)
0.8832
0.8833
0.8834
0.8835
0.8836
0.8837
0.8838
Asy
mm
etry
Par
amet
er
Figure 2.2: Single-scattering co-albedo ϖ−1 and asymmetry parameter g for mixtures ofcloud droplets and BC particles versus BC particle effective radius at a wavelength of 0.55 µm.The effective variance of the BC particle size distribution is fixed at 0.1, and the BC massfraction is fixed at 6108 −× . The solid curves show the results for the internal mixture, whereasthe dashed curves represent the external mixture. The dash-dotted curve depicts the asymmetryparameter of pure 10 µm-radius water droplets.
BCext,Wext,
BCsca,Wsca,
NCC
NCC
++
=ϖ , (2.2)
BCsca,Wsca,
BCsca,BCWsca,W
NCC
CNgCgg
++
= , (2.3)
where scaC and extC are the scattering and extinction cross sections per particle and N
37
is the number of BC particles per cloud droplet assuming that the BC fraction by mass
is fixed at 6108 −× . The subscripts W and BC correspond to water droplets and BC
aerosols, respectively.
2.5.3. Numerical results
Fig. 2.2 shows the single-scattering co-albedo ϖ−1 and asymmetry parameter g
of water droplets internally and externally mixed with BC aerosols at the wavelength
0.55 µm as a function of the BC particle effective radius effr computed for the BC mass
fraction 6108 −× . It is clear that internal mixing enhances absorption compared to
external mixing, which has already been pointed out in previous studies (see, e.g.,
[2,29,28]). The absorption is maximized at 05.0eff ≈r µm for internal mixing
( 4106.41 −×≈ϖ− ) and at 08.0eff ≈r µm for external mixing 4104.41 −×≈ϖ− . Since
the imbedded BC particles decrease the ray-tracing part of the total phase function of
heterogeneous droplets and thus increase the fractional contribution of the diffraction
part, the total asymmetry parameter of BC-contaminated water droplets increases
relative to that of pure water droplets [11].
Although we have adopted the upper limit on the BC mass fraction equal to
6108 −× [15], the measured BC fractions in water cloud droplets are usually two orders
of magnitude smaller [19–22]. To demonstrate the effect of varying BC amount, Fig.
2.3 depicts the single-scattering co-albedo and asymmetry parameter as a function of
the BC mass fraction at the same wavelength 0.55 µm, with effr and effv of BC
particles fixed at 0.05 µm and 0.1, respectively. We have chosen the value effr = 0.05
µm because it maximizes the absorption effect of internally mixed BC particles and
because this value appears to be quite realistic according to the measurement results
reported in [23,32,33].
38
0.001 0.01 0.1 1 10BC Mass Fraction in Units of 8 x10−6
0
0.001
0.002
0.003
0.004
0.005
Sin
gle-
Sca
tterin
g C
o-A
lbed
o
Internal Mixing
External Mixing
0.001 0.01 0.1 1 10BC Mass Fraction in Units of 8 x10−6
0.8825
0.8830
0.8835
0.8840
0.8845
Asy
mm
etry
Par
amet
er
Figure 2.3: Single-scattering co-albedo ϖ−1 and asymmetry parameter g for mixtures ofcloud droplets and BC particles versus BC mass fraction. The BC particle effective radius is0.05 µm and the effective variance is 0.1. The solid curves show the results for the internalmixture, whereas the dashed curves represent the external mixture. The dash-dotted curvedepicts the asymmetry parameter of pure 10 µm-radius water droplets.
Fig. 2.3 shows that absorption increases almost linearly with increasing BC mass
fraction. However, it is also obvious that the traditionally measured amounts of BC
cannot cause significant indirect forcing by strongly increasing cloud absorption.
39
0.001 0.01 0.1 1 10BC Mass Fraction in Units of 8 x10−6
−40
−20
0
20
40
Relative Difference in Single-Scattering Co-Albedo (%)
0.001 0.01 0.1 1 10BC Mass Fraction in Units of 8 x10−6
−0.15
−0.1
−0.05
0
0.05
Relative Difference in Asymmetry Parameter (%)
Figure 2.4: Relative differences (in %) between the single-scattering co-albedo and asymmetryparameter for external and internal mixtures of cloud droplets and BC particles versus the BCmass fraction. The BC particle effective radius and effective variance are 0.05 µm and 0.1,respectively.
The absolute difference in the single-scattering co-albedo and asymmetry
parameter results between the cases of internal and external mixing is negligible when
the BC mass fraction is less than 7108 −× . The latter value is still an order of magnitude
larger than those measured for the majority of water clouds [23,29-31]. Fig. 2.4 shows
the relative external/internal differences (in percent) in ϖ−1 and g as a function of the
BC amount at the wavelength 0.55 µm. The values of effr and effv of the BC particles
40
are 0.05 µm and 0.1, respectively. The relative differences in the asymmetry parameter
are very small, less than 0.13% in absolute value even when the BC mass fraction is as
high as 5108 −× . The relative differences in the single-scattering co-albedo are about –
13% when the BC mass fraction is greater than 7104.2 −× and reach –28% when the BC
fraction is 88 108104.2 −− ×−× (thus representing a very clean atmosphere). Taking into
account that the majority of BC particles remain outside cloud droplets ([23,29-31,34])
and that the differences in ϖ−1 and g between the internal and external mixtures are
very small, we conclude that irrespective of the actual form of mixing, one can always
use the much simpler external mixing scheme in radiative transfer modeling with great
confidence.
2.6 Discussion
Despite the use of a different approach to compute the optical properties of BC-
contaminated cloud droplets, our conclusions are in remarkable agreement with those
derived by Chýlek et al. [2] and Twohy el al. [29]. Chýlek et al. [2] found that a mass
fraction of internally mixed graphitic carbon of about 6107 −× is required to increase the
single-scattering co-albedo of droplets forming thick stratus clouds from 710− (pure
water) to 310− . Twohy et al. [19] concluded that BC mass concentrations in excess of
20,000 µg per 1 kg of cloud water are necessary to reduce the albedo of a cloud with an
optical thickness of 30 by 0.03. The observed BC mass concentrations [23,29–31] are
usually too low to reduce the cloud albedo in any significant way.
There are a number of uncertainties about the BC optical constants and their
variability with type of BC. As a result, the measured refractive indices vary
appreciably. The absorption by a small black carbon particle can be shape dependent
and may be enhanced by porosity. In this study, we have assumed that BC particles are
41
randomly distributed inside water droplets, whereas a preferential location of BC
impurities may cause an enhanced absorption. Absorption also depends on the
(variable) size of cloud droplets. Although disregarding these uncertainties may result
in a biased quantitative estimate of the effect of BC particles on cloud droplet optical
properties, it is unlikely to affect our conclusions in a significant way because the
observed BC concentrations are so small.
Acknowledgements
This research was funded by the NASA Global Aerosol Climatology Project
managed by Donald Anderson.
42
References
[1] Penner JE, Novakov T. Carbonaceous particles in the atmosphere: A historical
perspective to the fifth international conference on carbonaceous particles in the
atmosphere. J Geophys Res 1996;101:19373–8.
[2] Chýlek P, Ramaswamy V, Cheng RJ. Effect of graphitic carbon on the albedo of
clouds. J Atmos Sci 1984;41:3076–84.
[3] Charlson RJ, Schwartz SE, Hales JM, Cess RD, Coakley Jr. JA, Hansen JE,
Hofmann DJ. Climate forcing by anthropogenic aerosols. Science 1992;255:423–
30.
[4] Danielson RE, Moore DR, van de Hulst HC. The transfer of visible radiation
through clouds. J Atmos Sci 1969;26:1078–87.
[5] Chýlek P, Videen G, Geldart DJW, Dobbie JS, Tso HCW. Effective medium
approximations for heterogeneous particles. In: Mishchenko MI, Hovenier JW,
Travis LD, editors. Light scattering by nonspherical particles: theory, measurements,
and applications. Academic Press, San Diego, 2000, p. 273–308.
[6] Fikioris JG, Uzunoglu NK. Scattering from an eccentrically stratified dielectric
sphere. J Opt Soc Am 1979;69:1359–66.
[7] Borghese F, Denti P, Saija R. Optical properties of spheres containing a spherical
eccentric inclusion. J Opt Soc Am A 1992;9:1327–35.
[8] Fuller KA. Scattering and absorption cross sections of compounded spheres. I.
Theory for external aggregation. J Opt Soc Am A 1994;11:3251–60.
[9] Videen G, Ngo D, Chýlek P. Effective-medium predictions of absorption by
graphic carbon in water droplets. Opt Lett 1994;19:1675–7.
[10] Fuller KA, Mackowski DW. Electromagnetic scattering by compounded spherical
particles. In: Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by
nonspherical particles: theory, measurements, and applications. Academic Press, San
43
Diego, 2000, p. 225–72.
[11] Macke A, Mishchenko MI, Cairns B. The influence of inclusions on light
scattering by large ice particles. J Geophys Res 1996;101:23311–6.
[12] Mishchenko MI, Macke A. Asymmetry parameters of the phase function for isolated
and densely packed spherical particles with multiple internal inclusions in the
geometric optics limit. JQSRT 1997;57:767–94.
[13] Macke A. Monte Carlo calculations of light scattering by large particles with
multiple internal inclusions. In: Mishchenko MI, Hovenier JW, Travis LD, editors.
Light scattering by nonspherical particles: theory, measurements, and applications.
San Diego: Academic Press, 2000, p. 309–22.
[14] Chýlek P, Videen G, Ngo D, Pinnick RG, Klett JD. Effect of black carbon on the
optical properties and climate forcing of sulfate aerosols. J Geophys Res
1995;100:16325–32.
[15] Chýlek P, Lesins GB, Videen G, Wong JGD, Pinnick RG, Ngo D, Klett JD. Black
carbon and absorption of solar radiation by clouds. J Geophys Res
1996;101:23365–71.
[16] D’Almeida GA, Koepke P, Shettle EP. Atmospheric Aerosols: Global Climatology
and Radiative Characteristics. Hampton, Va.: A. Deepak, 1991.
[17] Penner JE, Eddleman H, Novakov T. Towards the development of a global
inventory for black carbon emission. Atmos Environ A 1993;27:1277–95.
[18] Chýlek P, Wong J. Effect of absorbing aerosols on global radiation budget.
Geophys Res Lett 1995;22:929–31.
[19] Heintzenberg J. Size-segregated measurements of particulate elemental carbon and
aerosol light absorption at remote Arctic locations. Atmos Environ 1982;16:2461–
9.
[20] Pueschel RF, Blake DF, Snetsinger KG, Hansen ADA, Verma S, Kato K. Black
44
carbon (soot) aerosol in the lower stratosphere and upper troposphere. Geophys
Res Lett 1992;19:1659–62.
[21] Parungo F, Nagamoto C, Zhou MY, Hanson ADA, Harris J. Aeolian transport of
aerosol black carbon from China to the ocean. Atmos Environ 1994;28:3251–60.
[22] Blake DF, Kato K. Latitudinal distribution of black carbon soot in the upper
troposphere and lower stratopshere. J Geophys Res 1995;100:7195–202.
[23] Chýlek P, Banic CM, Johnson B, Damiano PA, Isaac GA, Leaitch WR, Liu PSK,
Boudala FS, Winter B, Ngo D. Black carbon: atmospheric concentrations and
cloud water content measurements over southern Nova Scotia. J Geophys Res
1996;101:29105–10.
[24] Hansen JE, Travis LD. Light scattering in planetary atmosphere. Space Sci Rev
1974;16: 527–610.
[25] Mishchenko MI, Dlugach JM, Yanovitskij EG, Zakharova NT. Bidirectional
reflectance of flat, optically thick particulate layers: an efficient radiative transfer
solution and applications to snow and soil surfaces. JQSRT 1999;63:409–32.
[26] Mishchenko MI, Mackowski DW, Travis LD. Scattering of light by bispheres with
touching and separated components. Appl Opt 1995;34:4589–99.
[27] Mishchenko MI, Travis LD, Lacis AA. Scattering, absorption, and emission of light
by small particles. Cambridge University Press, Cambridge, 2002.
[28] Ackerman TP, Toon OB. Absorption of visible radiation in atmosphere containing
mixtures of absorbing and nonabsorbing particles. Appl Opt 1981;20:3661–8.
[29] Twohy CH, Clarke AD, Warren SG, Radke LF, Charlson RJ. Light-absorbing
material extracted from cloud droplets and its effect on cloud albedo. J Geophys
Res 1989;94:8623–31.
[30] Kou L. Black carbon: atmospheric measurements and radiative effect. Ph.D thesis.
Halifax (Canada): Dalhousie University.
45
[31] Bahrmann CP, Saxena VK. Influence of air mass history on black carbon
concentrations and regional climate forcing in southeastern United States. J
Geophys Res 1998;103: 23153–61.
[32] Stockham JD, Fenton DL, Johnson RH. Turbine engine particulate emission
characterization. Report ADA073198. [Available from NTIS, Springfield, VA
22161.]
[33] Borghesi A, Bussoletti E, Colangeli L, Minafra A, Rubini F. The absorption
efficiency of submicron amorphous carbon particles between 2.5 and 40µm.
Infrared Phys 1983;23:85–91.
[34] Hallberg A, Ogren JA, Noone KJ, Heintzenberg J, Berner A, Solly I, Kruisz C,
Reischl G, Fuzzi S, Facchini MC, Hansson H-C, Wiedensohler A, Svenningsson
IB. Phase partitioning for different aerosol species in fog. Tellus B 1992;44:545–
55.
46
Chapter 3
Scattering matrix of quartz aerosols: comparison andsynthesis of laboratory and Lorenz-Mie results1
Li Liu,a,b Michael I. Mishchenko,b,* Joop W. Hovenier,c,d Hester Volten,d,e
Olga Muñozf
aDepartment of Earth and Environmental Sciences, Columbia University, 2880 Broadway,
New York, NY 10025, USAbNASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025, USAcDepartment of Physics and Astronomy, Free University, De Boelelaan 1081, 1081 HV
Amsterdam, The NetherlandsdAstronomical Institute “Anton Pannekoek,” University of Amsterdam, Kruislaan 403,
1098 SJ Amsterdam, The NetherlandseFOM-Institute for Atomic and Molecular Physics, Kruislaan 407, 1098 SJ Amsterdam,
The NetherlandsfInstituto de Astrofisica de Andalucia, P.O. Box 3004, Granada 18080, Spain
Abstract
This paper compares and combines the results of laboratory measurements of the
Stokes scattering matrix for nonspherical quartz aerosols at a visible wavelength in the
scattering angle range 5°–173° and the results of Lorenz-Mie computations for
projected-area-equivalent spheres with the refractive index of quartz. A synthetic
normalized phase function is constructed based on the laboratory data and the
assumption that the diffraction forward-scattering peak is the same for spherical and
nonspherical projected-area-equivalent particles. The experimental scattering matrix for
1 Published in Journal of Quantitative Spectroscopy & Radiative Transfer, 79-80, 911-920,2003.
47
the nonspherical quartz particles is poorly represented by the Lorenz-Mie results for
most scattering angles. However, the asymmetry parameters for the synthetic phase
function and for the equivalent spherical particles are similar.
Keywords: Electromagnetic scattering; Nonspherical particles; Polarization
3.1 Introduction
Particle nonsphericity has been shown to be an important factor that must be
carefully addressed in optical characterization of mineral atmospheric aerosols [1–4].
Despite the significant recent progress [5–9], theoretical and numerical techniques are
still limited in their ability to simulate electromagnetic scattering by realistic
polydispersions of irregular particles. Therefore, laboratory measurement techniques
[10,11] remain an important source of information on scattering properties of
nonspherical aerosols.
In a recent paper, Volten et al. [12] presented an extensive dataset which includes
the results of laboratory measurements in the visible of the Stokes scattering matrix in a
wide scattering angle range for several types of polydisperse, randomly oriented mineral
aerosols and accompanying size distribution data. A traditional limitation of such
laboratory measurements is the lack of data at very small and very large scattering
angles (in this case, from 0° to 5° and from 173° to 180°), which precludes the
determination of the absolute angular dependence of the phase function )(1 Θa by using
the standard normalization condition,
1)(sind21
10
=∫ ΘΘΘπ
a , (3.1)
where Θ is the scattering angle. As a consequence, Volten et al. plotted the relative
quantity )30()()(~111 °= aaa ΘΘ rather than )(1 Θa , which makes their measurements
48
less useful in those cases when the absolute phase function values are needed [1–4,13].
The main purpose of this paper is to explore what insight and knowledge can be
gained by comparing and synthesizing an experimental and a theoretical scattering
matrix in the visible part of the spectrum. A unique opportunity to do so was provided
by a sample of randomly oriented quartz particles, because (i) its scattering matrix at a
wavelength of 441.6 nm was measured in the laboratory over a wide range of scattering
angles [12]; (ii) independently measured size distribution data are available for this
sample [12]; and (iii) independently measured values of the refractive index of quartz at
visible wavelengths are also available [14]. Consequently, in this case the experimental
scattering matrix can be contrasted with the most physically relevant theoretical Lorenz-
Mie scattering matrix, namely the one that is valid for projected-area-equivalent spheres
with the same refractive index.
The organization of this paper is as follows. First, we parallel the laboratory study
of Volten et al. [12] for the quartz particle sample by performing theoretical Lorenz-Mie
computations for projected-area-equivalent quartz spheres. Second, we construct a
synthetic phase function for the quartz particle sample using the relative angular profile
of the phase function measured by Volten et al., assuming that the forward-scattering
diffraction peak is independent of the particle shape and depends only on the
distribution of surface-equivalent-sphere radii, and using the normalization condition of
Eq. (3.1). The synthetic phase function is then used to compute the corresponding value
of the asymmetry parameter,
ΘΘΘΘΘπ
cos)(sind21
cos 10
a∫= . (3.2)
Finally, we briefly compare all elements of the scattering matrix and the asymmetry
parameter for the quartz particle sample and for projected-area-equivalent quartz
spheres.
49
3.2 Measurements and Lorenz-Mie computations
Volten et al. [12] used a laser particle sizer to measure the normalized projected
area distribution )(log rS of the quartz particle sample, where )d(log)(log rrS is the
fraction of the total projected area of the sample contributed by particles with radii in
the size range from rlog to )d(loglog rr + . The equivalent-sphere radius r of a
nonspherical particle is defined as the radius of a sphere that has a projected area equal
to the average projected area of the nonspherical particle in random orientation. This
distribution was presented in tabular form by Volten [15] and is shown in Fig. 3.1. Note
that equal areas under the curves correspond to equal contributions to the total projected
area. The effective radius effr and effective variance effv of this broad size distribution
are 2.3 µm and 2.4, respectively, where [16]
2eff )(d
1 max
min
rrrnrG
rr
rπ= ∫ , (3.3)
−1 0 1 2 logr
0
0.2
0.4
0.6
0.8
1
S(lo
gr)
Figure 3.1: Normalized distribution of the average area of the particle projection for randomlyoriented quartz aerosols. Here r is expressed in micrometers.
50
22eff2
effeff )()(d
1 max
min
rrrrnrrG
r
rπ−= ∫v , (3.4)
0.01
0.1
1
10
100
Lorenz-Mie results
Measurement results
Pha
se fu
nctio
n
−0.4
−0.2
0
0.2
−b1 /a
1
0
0.5
1
a 2 /a
1
−1
−0.5
0
0.5
1
a 3 /a
1
0 60 120 180Scattering angle (deg)
−0.6
−0.4
−0.2
0
0.2
b 2 /a
1
0 60 120 180Scattering angle (deg)
−1
−0.5
0
0.5
1
a 4 /a
1
Figure 3.2: Laboratory data for nonspherical quartz aerosols and results of Lorenz-Miecomputations for projected-area-equivalent quartz spheres. Experimental errors are shown byvertical error bars.
51
and
2)(dmax
min
rrnrGr
rπ= ∫ (3.5)
is the average area of the particle geometrical projection. Here rrn d)( is the fraction of
projected-area-equivalent spheres with radii between r and r + dr.
The open circles in Fig. 3.2 show the experimentally determined elements of the
scattering matrix versus scattering angle at a wavelength of 441.6 nm. The
measurements were performed at 5° intervals for scattering angles, Θ , in the range
from 5° to 170° and at 1° intervals for Θ from 170° to 173°. The experimental phase
function is normalized by its value at Θ = 30° and is plotted on a logarithmic scale. The
other elements are shown relative to the phase function. Within measurement errors,
the scattering matrix has the standard block-diagonal form,
− )()(00
)()(00
00)()(
00)()(
42
23
21
11
ΘΘΘΘ
ΘΘΘΘ
ab
ba
ab
ba
, (3.6)
thereby indicating that during the measurement, the quartz particles suspended in the air
jet were randomly oriented and formed a macroscopically isotropic and mirror-
symmetric scattering medium [17].
For comparison, the solid curves in Fig. 3.2 show the results for projected-area-
equivalent spheres calculated with the Lorenz-Mie code described in [18] and available
on the Internet at http://www.giss.nasa.gov/~crmim. For these computations, we
employed the number size distribution )(rn derived from the measured projected area
distribution shown in Fig. 3.1. Furthermore, we used the real part of the refractive index
1.559 as typical of quartz at wavelengths close to 440 nm [14]. Since quartz is
essentially nonabsorbing at visible wavelengths, the imaginary part of the refractive
52
index was set to zero. The theoretical Lorenz-Mie phase function is normalized
according to Eq. (3.1). Note that Volten et al. [12] use the time factor )iexp( tω rather
than the factor )iexp( tω− adopted in [18], which causes a sign difference in the
numerical values of the ratio )()( 12 ΘΘ ab . Therefore, the sign of the experimentally
measured ratio )()( 12 ΘΘ ab in Fig. 3.2 is opposite to that in [12].
3.3 Synthetic phase function
As we have already mentioned, Volten et al. [12] measured the relative phase
function )(~1 Θa rather than the actual phase function. Therefore, although the laboratory
data give the relative angular profile of the phase function in the scattering angle
interval from 5° to 173°, the exact vertical position of the experimental curve in (Θ , a1)
coordinates remains uncertain. The dashed and dot-dashed curves in Fig. 3.3 show two
extreme vertical positions of the experimental curve intended to match the phase
function values for nonspherical and projected-area-equivalent spherical quartz particles
at side- and backscattering angles, respectively. It is seen that in both cases spherical-
nonspherical phase function differences at other angles are very large and can exceed a
factor of 10. Placing the experimental curve in an intermediate position minimizes the
differences at side- and backscattering angles on average (dotted curve), but they can
still exceed a factor of 3. It is thus clear that irrespective of the actual vertical position
of the experimental curve, spherical-nonspherical phase function differences remain
very large at specific scattering angles.
In order to get a better idea of a plausible vertical position of the phase function for
nonspherical quartz particles, we did the following. It is known that the phase function
at small scattering angles for particles greater than a wavelength is mostly determined
by Fraunhofer diffraction and is largely the same for spherical and projected-area-
53
equivalent nonspherical particles with moderate aspect ratios and regular shapes (e.g.,
[19,20]). Therefore, we used the results of Lorenz-Mie computations for projected-area-
equivalent quartz spheres in the scattering angle interval from 0° to 5° and shifted the
experimental )(~1 Θa -curve in the vertical direction until its value at Θ = 5° matched the
Lorenz-Mie result. Finally, the experimental phase function was extrapolated from Θ =
173° to Θ = 180° using cubic splines.
We then evaluated the left-hand-side of Eq. (3.1) in order to check whether this
synthetic phase function satisfied the normalization condition. The result was 0.848
rather than the expected value unity. Since the contribution of the interval from Θ =
0 60 120 180Scattering angle (deg)
0.01
0.1
1
10
100
Pha
se fu
nctio
n
Figure 3.3: The pattern of the differences between the Lorenz-Mie phase function for sphericalquartz particles (solid curve) and the phase function for nonspherical quartz aerosols depends onthe vertical position of the experimental )(~
1 Θa profile (dashed, dotted, and dot-dashed curves).
54
173° to 180° to the integral was only 0.005, it is obvious that the use of spline
extrapolation could not explain the large discrepancy between the computed and
expected integral values.
There are several potential contributors to this discrepancy, including the
following.
• Experimental errors. These are always a potential source of complications.
However, in this case the errors (indicated by error bars in Fig. 3.2) appear to be too
small to be a likely explanation of the above discrepancy.
• The possible inaccuracy of the underlying assumption, made on the basis of
computations for regular nonspherical shapes, that the phase functions for projected-
area-equivalent spherical and nonspherical particles are the same in the forward-
scattering direction. One should not exclude the possibility that this assumption may
not be sufficiently precise for the irregular (including complicated surface structure)
quartz aerosols under consideration, especially at scattering angles as large as 5º.
• Inaccuracies in the measured size distribution of the quartz aerosols, especially for
the smallest particles. For example, we have found that truncating the measured size
distribution by leaving out all particles with equivalent-sphere radii smaller than
0.31 µm and then renormalizing the resulting size distribution improves the
normalization of the synthetic phase function significantly. Therefore, we should not
exclude the possibility that the size distribution of the quartz particles during the
scattering matrix measurements contained fewer small particles than was deduced
from the separate measurements with the particle sizer.
• Multiple scattering effects in the laboratory measurements. However, these appear
to be very unlikely, since the shape of the curves for )(~1 Θa did not show any
significant difference when the amount of scattering mineral aerosol particles was
doubled (see also [21]).
55
• Some constructive interference of light singly scattered by particles in the forward
direction [22] may have contributed to the intensity measured at small scattering
angles.
Given these uncertainties, we have decided to use, as a tentative fix, the following
simple procedure. Since it is likely that the most “vulnerable” quantity is the phase
function at the smallest scattering angle, we kept changing the experimental )5(~1 °a
value in very small increments and repeated the process of compiling the synthetic
phase function until it satisfied the normalization condition of Eq. (3.1) to better than
0.001. The result is shown in Fig. 3.4 by the dotted curve and is contrasted with the
Lorenz-Mie phase function for projected-area-equivalent quartz spheres depicted by the
solid curve. The respective asymmetry parameters evaluated using Eq. (3.2) are 0.669
and 0.698.
It is often convenient to represent a phase function by expanding it in Legendre
polynomials )(cosΘnP [17]:
∑=
Θ=Θmax
011 ).(cos)(
n
nn
n Pa α (3.7)
The expansion coefficients for the synthetic phase function were computed by
evaluating numerically the integral in the formula
∫ ΘΘΘΘ+=π
α0 11 )(cos)(sin)
2
1( n
n Padn (3.8)
and are available from the corresponding author upon request.
3.4 Discussion and conclusions
Figure 3.4 represents the main result of this paper. It closely resembles Fig. 6(a) of
Jaggard et al. [23] depicting experimental and theoretical Lorenz-Mie results for Raft
56
River soil dust. In agreement with the results of Jaggard et al. and previous theoretical
studies of light scattering by polydisperse, randomly oriented spheroids and circular
cylinders [19,20], Fig. 3.4 reveals the following three distinct regions:
nonsphere < sphere from Θ ~15°–20° to Θ ~65°;
nonsphere >> sphere from Θ ~65° to Θ ~150°;
nonsphere << sphere from Θ ~150° to Θ = 180°.
The asymmetry parameter for the nonspherical quartz particles, as determined from the
synthetic phase function, is smaller than that for the projected-area-equivalent quartz
spheres, but not by much, which also agrees well with theory [19,20].
0 60 120 180Scattering angle (deg)
0.1
1
10
100
1000
Pha
se fu
nctio
nLorenz-Mie phase function
Synthetic phase function
Figure 3.4: Synthetic and Lorenz-Mie phase functions for nonspherical quartz aerosols andprojected-area-equivalent quartz spheres, respectively.
57
The differences between the Lorenz-Mie and the synthetic phase function are quite
significant: they can exceed a factor of two at side-scattering angles and are even
greater in the backscattering direction. Large differences between the experimental and
Lorenz-Mie results also occur for all other scattering matrix elements (see Fig. 3.2) and
follow the general pattern discussed by Mishchenko et al. [24]. Specifically, the degree
of linear polarization for unpolarized incident light, )()( 11 ΘΘ ab− , tends to be positive
at side-scattering angles for the nonspherical particles, but shows a broad negative region
at side-scattering angles and a narrow positive feature at Θ ~ 165º caused by the primary
rainbow for the spherical aerosols. Whereas 1)()( 12 ≡ΘΘ aa for spherically symmetric
scatterers, the )()( 12 ΘΘ aa curve for the nonspherical quartz aerosols significantly
deviates from unity and exhibits strong backscattering depolarization. Similarly,
)()()()( 1413 ΘΘΘΘ aaaa ≡ for spherically symmetric particles, whereas )()( 14 ΘΘ aa
for the nonspherical quartz aerosols tends to be significantly greater than )()( 13 ΘΘ aa for
most angles, especially at backscattering directions. Furthermore, the ratios )()( 12 ΘΘ ab
for the nonspherical and spherical quartz aerosols show significant differences at scattering
angles in the range °<<° 170120 Θ . Thus our results reinforce previous indications that
for most scattering angles, the phase function and the other elements of the scattering
matrix for nonspherical aerosols are inadequately represented by Lorenz-Mie results
computed for the same size distribution and refractive index and caution against the use
of the latter in optical characterization of nonspherical particles.
The idea of compiling a synthetic phase function using a combination of
experimental and theoretical Lorenz-Mie results appears to be attractive because of its
simplicity and may be a useful practical tool in cases when experimental data are not
available in the entire scattering angle interval from 0º to 180º. Somewhat similar
procedures were described by Hill et al. [25] and Moreno et al. [26]. We have
58
mentioned, however, that there are several issues that may make the application of this
procedure less straightforward than one would like it to be. Our current research is
focused on addressing the potential complexifying factors one-by-one.
Finally, we note that Mishchenko et al. [27] used the synthetic phase function to
analyze the potential effect of nonsphericity on the results of retrievals of mineral
tropospheric aerosols based on radiance observations from earth-orbiting satellites.
Acknowledgements
It is a pleasure to thank J. F. de Haan and W. Vassen for many fruitful discussions
and two anonymous reviewers for constructive comments. This research was supported
by the NASA Radiation Sciences Program managed by Donald Anderson.
59
References
[1] Mishchenko MI, Lacis AA, Carlson BE, and Travis LD. Nonsphericity of dust-like
tropospheric aerosols: implications for aerosol remote sensing and climate
modeling. Geophys Res Lett 1995;22:1077–80.
[2] Krotkov NA, Flittner DE, Krueger AJ, Kostinski A, Riley C, Rose W, Torres O.
Effect of particle non-sphericity on satellite monitoring of drifting volcanic ash
clouds. J Quant Spectrosc Radiat Transfer 1999;63:613–30.
[3] Diner DJ, Abdou WA, Bruegge CJ, Conel JE, Crean JA, Gaitley BJ, Helmlinger
MC, Kahn RA, Martonchik JV, Pilorz SH. MISR aerosol optical depth retrievals
over southern Africa during the SAFARI-2000 dry season campaign. Geophys Res
Lett 2001;28:3127–30.
[4] Dubovik O, Holben B, Eck TF, Smirnov A, Kaufman YJ, King MD, Tanŕe D,
Slutsker I. Variability of absorption and optical properties of key aerosol types
observed in worldwide locations. J Atmos Sci 2002;59:590-608.
[5] Draine BT. The discrete dipole approximation for light scattering by irregular targets.
In: Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by
nonspherical particles: theory, measurements, and applications. San Diego:
Academic Press, 2000, p. 131–45.
[6] Lumme K. Scattering properties of interplanetary dust particles. In: Mishchenko MI,
Hovenier JW, Travis LD, editors. Light scattering by nonspherical particles: theory,
measurements, and applications. San Diego: Academic Press, 2000, p. 555–83.
[7] Mishchenko MI, Travis LD, Macke A. T-matrix method and its applications. In:
Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by nonspherical
particles: theory, measurements, and applications. San Diego: Academic Press, 2000,
p. 147–72.
[8] Yang P, Liou KN. Finite difference time domain method for light scattering by
60
nonspherical and inhomogeneous particles. In: Mishchenko MI, Hovenier JW, Travis
LD, editors. Light scattering by nonspherical particles: theory, measurements, and
applications. San Diego: Academic Press, 2000, p. 173–221.
[9] Muinonen K. Light scattering by stochastically shaped particles. In: Mishchenko MI,
Hovenier JW, Travis LD, editors. Light scattering by nonspherical particles: theory,
measurements, and applications. San Diego: Academic Press, 2000, p. 323–52.
[10] Hovenier JW. Measuring scattering matrices of small particles at optical
wavelengths. In: Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering
by nonspherical particles: theory, measurements, and applications. San Diego:
Academic Press, 2000, p. 355–65.
[11] Gustafson BÅS. Microwave analog to light-scattering measurements. In:
Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by nonspherical
particles: theory, measurements, and applications. San Diego: Academic Press, 2000,
p. 367–90.
[12] Volten H, Muñoz O, Rol E, de Haan JF, Vassen W, Hovenier JW, Muinonen K,
Nousiainen T. Scattering matrices of mineral aerosol particles at 441.6 nm and 632.8
nm. J Geophys Res 2001;106:17 375–401.
[13] King MD, Kaufman YJ, Tanré D, Nakajima T. Remote sensing of tropospheric
aerosols from space: past, present, and future. Bull Am Meteorol Soc 1999;80:2229–
59.
[14] Billings BH, Frederikse HPR, Bleil DF, Lindsay RB, Cook RK, Marion JB,
Crosswhite HM, Zemansky MW, editors. American Institute of Physics handbook.
New York: McGraw-Hill, 1972, p. 6–27.
[15] Volten H. Light scattering by small planetary particles: an experimental study. Ph.D.
dissertation, Free University, Amsterdam, 2001.
[16] Hansen JE, Travis LD. Light scattering in planetary atmospheres. Space Sci Rev
61
1974;16: 527–610.
[17] Mishchenko MI, Hovenier JW, Travis LD. Concepts, terms, notation. In:
Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by nonspherical
particles: theory, measurements, and applications. San Diego: Academic Press, 2000,
p. 3–27.
[18] Mishchenko MI, Dlugach JM, Yanovitskij EG, Zakharova NT. Bidirectional
reflectance of flat, optically thick particulate layers: an efficient radiative transfer
solution and applications to snow and soil surfaces. J Quant Spectrosc Radiat
Transfer 1999;63:409–32.
[19] Mishchenko MI, Travis LD, Macke A. Scattering of light by polydisperse, randomly
oriented, finite circular cylinders. Appl Opt 1996;35:4927–40.
[20] Mishchenko MI, Travis LD, Kahn RA, West RA. Modeling phase functions for
dustlike tropospheric aerosols using a shape mixture of randomly oriented
polydisperse spheroids. J Geophys Res 1997;102:16 831–47.
[21] Hovenier JW, Volten H, Muñoz O, van der Zande WJ, Waters LBFM. Laboratory
studies of scattering matrices for randomly oriented particles. Potentials problems
and perspectives. JQSRT 2003, this issue.
[22] Mishchenko MI, Mackowski DW, Travis LD. Scattering of light by bispheres with
touching and separated components. Appl Opt 1995;34:4589–99.
[23] Jaggard DL, Hill C, Shorthill RW, Stuart D, Glantz M, Rosswog F, Taggart B,
Hammond S. Light scattering from particles of regular and irregular shape. Atmos
Environ 1981;15:2511–19.
[24] Mishchenko MI, Wiscombe WJ, Hovenier JW, Travis LD. Overview of scattering
by nonspherical particles. In: Mishchenko MI, Hovenier JW, Travis LD, editors.
Light scattering by nonspherical particles: theory, measurements, and applications.
San Diego: Academic Press, 2000, p. 29–60.
62
[25] Hill SC, Hill AC, Barber PW. Light scattering by size/shape distributions of soil
particles and spheroids. Appl Opt 1984;23:1025–31.
[26] Moreno F, Muñoz O, Lopez-Moreno JJ, Molina A, Ortiz JL. A Monte Carlo code
to compute energy fluxes in cometary nuclei. Icarus 2002;156:474-84.
[27] Mishchenko MI, Geogdzhayev IV, Liu L, Ogren JA, Lacis AA, Rossow WB,
Hovenier JW, Volten H, Muñoz O. Aerosol retrievals from AVHRR radiances:
effects of particle nonsphericity and absorption and an updated long-term global
climatology of aerosol properties. JQSRT 2003, this issue.
63
Chapter 4
Investigation of the effects of particle nonsphericity onaerosol retrievals from AVHRR observations1
Abstract
In this paper, we use the synthetic phase function [1] to analyze the potential effect
of particle nonsphericity on the results of retrievals of mineral tropospheric aerosols
from Advanced Very High-Resolution Radiometer (AVHRR) radiance observations.
Our results reconfirm the previously reached conclusion that the nonsphericity of dusk-
like and dry sea salt aerosols can lead to very large errors in the retrieved optical
thickness if one mistakenly applies the scattering model for spherical particles.
Keywords: Atmospheric aerosols; Nonspherical particles; Remote sensing
4.1 Introduction
The important role of tropospheric aerosols in forming the Earth’s climate is now
well recognized [2–4] and has motivated several dedicated research programs [5–7],
including the Global Aerosol Climatology Project (GACP) established in 1998 as part
of the National Aeronautics and Space Administration’s Radiation Sciences Program
and the World Climate Research Programme’s Global Energy and Water Cycle
Experiment [8]. A major component of the GACP is a retrospective analysis of the
Advanced Very High Resolution Radiometer (AVHRR) radiance data set in order to
1 Published in Journal of Quantitative Spectroscopy & Radiative Transfer, 79-80, 953-972,2003.
64
infer the long-term global distribution of aerosols, their properties, and seasonal and
interannual variations.
In recent publications [9,10], Mishchenko et al. and Geogdzhayev et al. described
an advanced aerosol retrieval algorithm based on using channel–1 and –2 AVHRR data
over the oceans and applied it to the ISCCP DX radiance dataset [11]. Specifically, the
algorithm retrieves the aerosol optical thickness τ and Ångström exponent A for each
pixel by minimizing the difference between two radiances measured in the 0.65 and
0.85µm channels at the specific illumination and observation angles determined by the
satellite orbit on the one hand and the radiances computed theoretically for a realistic
atmosphere–ocean model on the other hand. The Ångström exponent is defined as
1)(ln
)]([ln
λλλλ
=
−=d
CdA ext , (4.1)
where λ1 = 0.65µm is the nominal wavelength of the AVHRR channel 1 and Cext is the
ensemble-averaged extinction cross-section per particle. With only two pieces of data
per pixel available, one can retrieve only the two model parameters and must assign
fixed global values to the remaining parameters describing the complex atmosphere–
ocean system, thereby introducing potential biases in the aerosol product. We have
performed an extensive study of the expected accuracy of the algorithm and its
sensitivity to various a priori assumptions and used it in the development of a
preliminary global climatology of the aerosol optical thickness and size for the period
extending from July 1983 to August 1994.
It is well known that the climatically-important dust-like aerosols have
nonspherical shapes, whereas the operational GACP algorithm is based on Lorenz-Mie
computations and assumes the spherical particle shape irrespective of aerosol type. It is,
therefore, important to examine how nonsphericity can affect the results of AVHRR
aerosol retrievals.
65
4.2 Effect of particle shape
Desert dust can dominate the aerosol population over large Atlantic Ocean areas off
the north-west coast of Africa, the Persian Gulf region, and large areas of the Pacific
Ocean off the coast of China. Sea salt particles are believed to be the dominant aerosol
species at high southern latitudes. It has been demonstrated recently that particle
nonsphericity is an important factor that must be carefully addressed in optical
characterization of mineral aerosols such as dust-like and dry sea salt particles [12–15].
Previous analyses of this issue were based either on semi-empirical approximate
approaches [16,17] or on theoretical computations of the phase function for simplified
model shapes [13–15,18-21]. However, theoretical and numerical techniques are still
limited in their ability to simulate electromagnetic scattering by realistic polydispersions
of irregular particles. Therefore, laboratory measurement techniques [22,23] remain an
important source of information on scattering properties of natural nonspherical
aerosols.
Recently, Volten et al. [24,25] presented an extensive dataset which includes the
results of laboratory measurements for several types of polydisperse, randomly oriented
mineral aerosols at 441.6 and 632.8 nm wavelengths. A limitation of these
measurements is the lack of data at very small and very large scattering angles (from 0°
to 5° and from 173° to 180°), which precludes the determination of the absolute angular
dependence of the phase function )(1 Θa by using the standard normalization condition,
∫ =ΘΘΘπ
0 1 1)(sin2
1ad , (4.2)
where Θ is the scattering angle. As a consequence, Volten et al. plotted the relative
quantity )30()()(~111
oaaa Θ=Θ , rather than )(1 Θa , which limits the applicability of
their results in satellite remote sensing. However, Liu et al. [1] used the relative angular
profile of the phase function measured by Volten et al. [24,25] for a quartz particle
66
sample at 441.6 nm to construct a synthetic phase function on the entire interval
]180,0[ oo∈Θ by assuming that the forward-scattering diffraction peak is independent of
the particle shape and depends only on the distribution of surface-equivalent-sphere
radii and the wavelength, and then using the normalization condition (4.2). The quartz
phase function was selected because the refractive index of quartz is rather well known
in the visible spectral range. Furthermore, the experimental angular profile of the phase
function for the quartz sample appeared to be only weakly dependent on the wavelength
and was similar to those of the other mineral aerosol samples studied by Volten et al.
irrespective of their exact size distribution and chemical composition.
Since the synthetic phase function is available only for one average particle size
relative to the wavelength, it cannot be used in the operational two-channel algorithm in
order to retrieve simultaneously the aerosol optical thickness and size. However, it can
be used in a one-channel algorithm analogous to that used by Stowe et al. [26], wherein
the average particle size (and, thus, the phase function) is assumed to be fixed globally
and the only retrieved parameter is τ . Although this approach does not generate a
product similar to that based on the two-channel algorithm [9,10], it allows us to study,
at least semi-quantitatively, the potential effect of nonsphericity on the aerosol optical
thickness retrieved from the AVHRR data. This follows, indeed, from the theoretical
observation that the spherical–nonspherical differences in the phase function should be
very similar for the two close AVHRR channels [19].
Volten et al. [24,25] defined the equivalent-sphere radius r of a nonspherical
particle as the radius of a sphere that has a projected area equal to the average projected
area of the nonspherical particle in random orientation. The effective radius reff and
effective variance veff [27] of the quartz particle sample studied in [24,25] are 2.3 µm
and 2.4, respectively. The laboratory measurements were performed at a wavelength of
441.6 nm. Therefore, the effective radius of a phase-function-equivalent size
67
distribution at the AVHRR channel–1 wavelength 650 nm is 3.39µm, thereby
representing the coarse mode of dust-like aerosols. The extrapolation of the results
obtained at 441.6 nm to 650 nm is based on the assumption that the refractive index of
quartz remains nearly constant in this spectral interval, which is indeed the case.
Fig. 4.1 depicts the synthetic phase function as well as the phase function for
projected-area-equivalent quartz spheres calculated using the Lorenz-Mie code
described in [28] and available on-line at http://www.giss.nasa.gov/~crmim. In close
agreement with previous results of Jaggard al. [29] for Raft River soil dust and the
results of theoretical studies of light scattering by polydisperse, randomly oriented
0 60 120 180Scattering angle (deg)
0.1
1
10
100
1000
Pha
se fu
nctio
n
Lorenz-Mie phase function
Synthetic phase function
Figure 4.1: Synthetic and Lorenz-Mie phase functions for nonspherical quartz aerosols andprojected-area-equivalent quartz spheres, respectively, used in the one-channel retrievalalgorithm (see text).
68
spheroids and circular cylinders [19,30], Fig. 4.1 exhibits the following three
conspicuous regions:
nonsphere < sphere from Θ ~ 15°–20° to Θ ~ 65°,
nonsphere >> sphere from Θ ~ 65° to Θ ~ 150°, (4.3)
nonsphere << sphere from Θ ~ 150° to Θ = 180°.
The differences between the Lorenz-Mie and the synthetic phase function are quite
significant at side-scattering angles, where they can exceed a factor of two, and are even
greater at backscattering angles.
We have used both phase functions in the one-channel retrieval algorithm and
applied the latter to AVHRR data collected in 1987 over a region often dominated by
Sahara dust aerosols and extending from Central Meridian to 60°W and from 7°S to
18°Ν. We then calculated the monthly average of the ratio of the aerosol optical
thickness τN retrieved with the phase function of the nonspherical quartz particles to the
optical thickness τS retrieved with the phase function computed for the projected-area-
equivalent spherical quartz aerosols at λ = 0.65 µm. This ratio is plotted in Fig. 4.2 as a
function of longitude for 1°-wide horizontal belts along with the corresponding monthly
averages of the scattering angle.
Fig. 4.2 is an excellent illustration of the relationships summarized by Eq. (4.3) and
shows that τN/τS > 1 for Θ > 150° and τN/τS < 1 for Θ < 150°. Owing to the specific
NOAA-9 spacecraft orbit, the scattering angle for this area is, for the most part, greater
than 150° during the four months studied. As a consequence, the algorithm based on the
Lorenz-Mie phase function tends to generate significantly smaller optical thicknesses
than that based on the phase function representative of nonspherical aerosols. The ratio
τN/τS reaches values exceeding 3.5 in April, but stays closer to unity in July.
69
Accordingly, the monthly averages of this ratio over the entire area studied are 1.545 for
January, 1.975 for April, 1.053 for July, and 1.974 for October.
0
1
2
3
4
τ(no
nsph
eric
al)/
τ(sp
heric
al) Jan 87
0
1
2
3
4
Apr 87
120o
150o
180o
Sca
tterin
g an
gle
120o
150o
180o
0
1
2
3
4
τ(no
nsph
eric
al)/
τ(sp
heric
al) 7oS−6oS
2oS−1oS
3oN−4oN
8oN−9oN
13oN−14oN
17oN−18oN
Jul 87
0
1
2
3
4
Oct 87
60oW 50oW 40oW 30oW 20oW 10oW 0o120o
150o
180o
Sca
tterin
g an
gle
Longitude60o 50oW 40oW 30oW 20oW 10oW 0o
120o
150o
180o
Longitude
Figure 4.2: Monthly averages of the ratio τN/τS and the respective scattering angle versuslongitude.
70
Our results obviously reinforce the previously reached conclusion [18] that the
nonsphericity of mineral particles can have a profound effect on the reflected intensity
and must be explicitly accounted for in aerosol retrievals based on satellite radiance
data. Unfortunately, the AVHRR data by themselves provide no means of identifying
particle type and shape. This is also true for any instrument taking reflectance data at
only one scattering geometry per pixel, such as the MODerate resolution Imaging
Spectrometer (MODIS) [31,32]. Although small values of the Ångström exponent can
be indicative of the presence of large mineral particles, this test cannot be expected to
distinguish between dry (nonspherical) and wet (spherical) sea salt particles. It thus
appears to be difficult, if not impossible, to develop a simple and reliable procedure
which improves the AVHRR retrieval algorithm by introducing the necessary
corrections when the particles happen to be nonspherical. However, this may be feasible
with multi-angle instruments such as the Multiangle Imaging Spectro-Radiometer
(MISR) [13,33], the POLarization and Directionality of Earth Reflectances (POLDER)
instrument [34,35], and the Earth Observing Scanning Photopolarimeter [36], especially
when polarization of the reflected light is also measured.
4.3 Conclusions
To summarize, we conclude the use of the phase function typical of irregular
mineral aerosols in the retrieval algorithm based on channel–1 AVHRR radiances
indicates that the nonsphericity of dust-like and dry sea salt aerosols can lead to very
large errors in the retrieved optical thickness if one mistakenly applies look-up tables
based on Lorenz-Mie computations. The errors change with season and geographical
location and cannot be corrected using AVHRR data alone since the latter do not
provide a reliable indication of aerosol type.
71
Acknowledgements
This research was supported by the NASA Radiation Sciences Program managed
by Donald Anderson.
72
References
[1] Liu L, Mishchenko MI, Hovenier JW, Volten H, Muñoz O. Scattering matrix of
quartz aerosols: comparison and synthesis of laboratory and Lorenz-Mie results.
JQSRT 2003;79-80:911-20.
[2] Penner JE, Charlson RJ, Hales JM, Laulainen NS, Leifer R, Novakov T, Ogren J,
Radke LF, Schwartz SE, Travis L. Quantifying and minimizing uncertainty of
climate forcing by anthropogenic aerosols. Bull Am Meteorol Soc 1994;75:375–
400.
[3] Hansen JE, Sato M, Lacis A, Ruedy R, Tegen I, Matthews E. Perspective: climate
forcings in the industrial era. Proc Natl Acad Sci 1998;95:12753–58.
[4] Kaufman YJ, Tanré D, Gordon HR, Nakajima T, editors. Passive remote sensing
of tropospheric aerosol and atmospheric corrections of the aerosol effect. J
Geophys Res 1997;102:16 815–17 217.
[5] Artaxo P, Hobbs P, Kaufman YJ, Kirchhoff, editors. Smoke, clouds, and
radiation–Brazil. J Geophys Res 1998;103:31 781–32 137.
[6] Russell PB, Hobbs PV, Stowe LL, editors. Tropospheric aerosol radiative forcing
observational experiment, J Geophys Res 1999;104:2213–2319.
[7] Raes F, Bates T, editors. The second aerosol characterization experiment. Tellus
2000;52B:109–908.
[8] Mishchenko M, Penner J, Anderson D, editors. Global aerosol climatology project.
J Atmos Sci 2002;59:249–783.
[9] Mishchenko MI, Geogdzhayev IV, Cairns B, Rossow WB, Lacis AA. Aerosol
retrievalsover the ocean by use of channels 1 and 2 AVHRR data: sensitivity
analysis and preliminary results. Appl Opt 1999;38:7325–41.
[10] Geogdzhayev IV, Mishchenko MI, Rossow WB, Cairns B, Lacis AA. Global two-
channel AVHRR retrievals of aerosol properties over the ocean for the period of
73
NOAA-9 observations and preliminary retrievals using NOAA–7 and NOAA–11
data. J Atmos Sci 2002;59:262–78.
[11] Rossow WB, Schi+er RA. Advances in understanding clouds from ISCCP. Bull
Am Meteorol Soc 1999;80:2261–87.
[12] Nakajima T, Tanaka M, Yamano M, et al. Aerosol optical characteristics in the
yellow sand events observed in May, 1982 at Nagasaki. II: Models. J Meteorol Soc
Japan 1989;67:279–91.
[13] Diner DJ, Abdou WA, Bruegge CJ, Conel JE, Crean KA, Gaitley BJ, Helmlinger
MC, Kahn RA, Martonchik JV, Pilorz SH, Holben BN. MISR aerosol optical
depth retrievals over southern Africa during the SAFARI-2000 dry season
campaign. Geophys Res Lett 2001;28:3127–30.
[14] Dubovik O, Holben B, Eck TF, Smirnov A, Kaufman YJ, King MD, Tanré D,
Slutsker I. Variability of absorption and optical properties of key aerosol types
observed in worldwide locations. J Atmos Sci 2002;59:590–608.
[15] Dubovik O, Holben BN, Lapyonok T, Sinyuk A, Mishchenko MI, Yang P,
Slutsker I. Non-spherical aerosol retrieval method employing light scattering by
spheroids. Geophys Res Lett 2002 (in press).
[16] Pollack JB, Cuzzi JN. Scattering by nonspherical particles of size comparable to a
wavelength: a new semi-empirical theory and its application to tropospheric
aerosols. J Atmos Res 1980;37:868–81.
[17] von Hoyningen-Huene W, Posse P. Nonsphericity of aerosol particles and their
contribution to radiative forcing. JQSRT 1997;57:651–68.
[18] Mishchenko MI, Lacis AA, Carlson BE, Travis LD. Nonsphericity of dust-like
tropospheric aerosols: implications for aerosol remote sensing and climate
modeling. Geophys Res Lett 1995;22:1077–80.
74
[19] Mishchenko MI, Travis LD, Kahn RA, West RA. Modeling phase functions for
dustlike tropospheric aerosols using a shape mixture of randomly oriented
polydisperse spheroids. J Geophys Res 1997;102:16831–47.
[20] Krotkov NA, Flittner DE, Krueger AJ, Kostinski A, Riley C, Rose W, Torres O.
Effect of particle non-sphericity on satellite monitoring of drifting volcanic ash
clouds. J Quant Spectrosc Radiat Transfer 1999;63:613–30.
[21] Yang P, Liou KN, Mishchenko MI, Gao B-C. Efficient finite-difference time-
domain scheme for light scattering by dielectric particles: application to aerosols.
Appl Opt 2000;39:3727–37.
[22] Hovenier JW. Measuring scattering matrices of small particles at optical
wavelengths. In: Mishchenko MI, Hovenier JW, Travis LD, editors. Light
scattering by nonspherical particles: theory, measurements, and applications. San
Diego: Academic Press, 2000, p. 355–65.
[23] Gustafson BÅS. Microwave analog to light-scattering measurements. In:
Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by
nonspherical particles: theory, measurements, and applications. San Diego:
Academic Press, 2000, p. 367–90.
[24] Volten H, Muñoz O, Rol E, de Haan JF, Vassen W, Hovenier JW, Muinonen K,
Nousiainen T. Scattering matrices of mineral aerosol particles at 441.6 nm and
632.8 nm. J Geophys Res 2001;106:17375–401.
[25] Volten H. Light scattering by small planetary particles: an experimental study.
PhD dissertation, Free University, Amsterdam, 2001.
[26] Stowe LL, Ignatov AM, Singh RR. Development, validation, and potential
enhancements to the second-generation operational aerosol product at the National
Environmental Satellite, Data, and Information Service of the National Oceanic
and Atmospheric Administration. J Geophys Res 1997;102:16923–34.
75
[27] Hansen JE, Travis LD. Light scattering in planetary atmospheres. Space Sci Rev
1974;16:527– 610.
[28] Mishchenko MI, Travis LD, Lacis AA. Scattering, absorption, and emission of
light by small particles. Cambridge: Cambridge University Press, 2002.
[29] Jaggard DL, Hill C, Shorthill RW, Stuart D, Glantz M, Rosswog F, Taggart B,
Hammond S. Light scattering from particles of regular and irregular shape. Atmos
Environ 1981;15:2511– 19.
[30] Mishchenko MI, Travis LD, Macke A. Scattering of light by polydisperse,
randomly oriented, finite circular cylinders. Appl Opt 1996;35:4927– 40.
[31] King MD, Kaufman YJ, Menzel WP, Tanré D. Remote sensing of cloud, aerosol,
and water vapor properties from the Moderate Resolution Imaging Spectrometer
(MODIS). IEEE Trans Geosci Remote Sens 1992;30:2– 27.
[32] Tanré D, Kaufman YJ, Herman M, Mattoo S. Remote sensing of aerosol properties
over oceans using the MODIS/EOS spectral radiances. J Geophys Res
1997;102:16971– 88.
[33] Kahn R, Banerjee P, McDonald D, Diner DJ. Sensitivity of multiangle imaging to
aerosol optical depth and to pure-particle size distribution and composition over
ocean. J Geophys Res 1998;103:32195– 213.
[34] Goloub P, Tanré D, Deuzé JL, Herman M, Marchand A, Bréon F-M. Validation of
the first algorithm applied for deriving the aerosol properties over the ocean using
the POLDER/ADEOS measurements. IEEE Trans Geosci Remote Sens
1999;37:1586– 96.
[35] Deuzé JL, Goloub P, Herman M, Marchand A, Perry G, Susana S, Tanré D.
Estimate of the aerosol properties over the ocean with POLDER. J Geophys Res
2000;105:15329– 46.
76
[36] Chowdhary J, Cairns B, Mishchenko M, Travis L. Retrieval of aerosol properties
over the ocean using multispectral and multiangle photopolarimetric
measurements from the Research Scanning Polarimeter. Geophys Res Lett
2001;28:243– 6.
77
Chapter 5
Global validation of the operational two-channel AVHRR retrieval product over the ocean1
Abstract
In this chapter we present validation results for the aerosol optical thickness τSAT,
the Ångström exponent A, and the single-scattering albedo ϖ retrievals. These retrievals
are part of the global long-term aerosol climatology derived by applying a two-channel
algorithm to Advanced Very High Resolution Radiometer (AVHRR) data. The satellite
retrieved optical thicknesses are compared with ship-borne sun-photometer results, τSP.
The comparison of the spatial and temporal statistics of the satellite retrievals and the
ship measurements shows a strong correlation. A systematic overestimation in τSAT by
about 11% relative to τSP for average aerosol loadings was found for the initial satellite
dataset. Increasing the diffuse component of the surface reflectance from 0.002 to 0.004
produces a better match, with the ensemble average at 0.55 µm differing by only about
3.6% from the ship truth and having a small offset of 0.03. Comparisons of single-
scattering albedo and Ångström exponent values retrieved from the AVHRR data and
those measured in situ at Sable Island indicate that the currently adopted value 0.003
can be a reasonable choice for the imaginary part of the aerosol refractive index in the
global satellite retrievals.
1 Part of this work has been published in Journal of Quantitative Spectroscopy & Radiative Transfer, 79-80, 953-972, 2003; the rest is in preparation for submission.
78
5.1 Introduction
The Advanced Very High Resolution (AVHRR) radiance dataset with its two-
decade record and near global coverage is a unique potential source of information
about atmospheric aerosols. Mishchenko et al. (1999) and Geogdzhayev et al. (2002)
developed an advanced retrieval algorithm that utilizes AVHRR channels 1 and 2 to
retrieve aerosol optical thickness and Ångström exponent over the ocean and conducted
comprehensive sensitivity studies. By using the algorithm, a long-term climatology has
been created, which combines the data from several AVHRR instruments and currently
spans the period from 1983 to 2001.
Because of the ill-posed character of the inverse scattering problem, various a priori
assumptions must be made in order to retrieve aerosol properties from remotely sensed
data. Any variability in the assumed parameters leads to uncertainty in the satellite
retrievals. Therefore validation of the retrieved results is critical and has been an
important direction in our research. This chapter has two major objectives. The first one
is to validate the satellite-retrieved aerosol optical thickness. This task poses significant
problems since sun-photometer aerosol measurements over ocean are scarce. A number
of publications describe inter-comparisons of various satellite, ground-based and model
aerosol data. Kinne et al. (2001) and Penner et al. (2002) performed a comparison of
monthly statistics of aerosol satellite retrievals and model results with the AERONET-
derived (Holben et al. 1998) statistics. Kinne et al. found that the aerosol optical
thickness derived using the original algorithm described by Mishchenko et al. (1999)
for the period of NOAA-9 observations are systematically higher compared to the
averages from several coastal AERONET sites. Penner et al. used a more recent aerosol
product and found a significantly better agreement. A study by Myhre et al. (2002) uses
the most recent two-channel AVHRR retrievals and shows a much better agreement
than the paper by Kinne et al.
79
The discrepancies between the ground-based measurements and the satellite
retrievals may be associated with space and time co-location, different approaches to
cloud screening, sensor calibration error, improper assumption about ocean surface
reflectance, inconsistency between the aerosol microphysical model (such as aerosol
size distribution and refractive index) used in the retrieval algorithm and that in the real
world. In addition, higher surface albedos in the coastal regions compared to the open
ocean values assumed in the retrieval algorithm and desert particle nonsphericity may
also have contributed to the discrepancies.
In a recent paper, Smirnov et al. (2002) summarized aerosol optical thickness τ
measurements in maritime and coastal areas. The comprehensive survey of ship-borne
measurements published over the last 30 years is a great asset for our validation since
the ship data cover the same period of time as many of the AVHRR retrievals. The data
have been collected over open ocean areas, and thus we do not have to consider the
possible effects of coastline and shallow water on the satellite retrievals (e.g., Zhao et al
2003). This will be the subject of the following section.
Second, the inherent limitations of a retrieval algorithm based on utilizing only two
pieces of data per pixel (Mishchenko et al. 1999 and Geogdzhayev et al. 2002) force
one to adopt a spatially and temporally fixed value of the imaginary part of the aerosol
refractive index. The current version of the algorithm (Geogdzhayev et al. 2002) uses
the value 0.003, which is smaller than the value 0.005 used in the initial version
(Mishchenko et al. 1999), the reason being that this decrease may help to achieve a
better balance between the nonabsorbing sea salt aerosols and the absorbing
anthropogenic and dust aerosols on a global scale. Although the effect of this change on
the retrieved optical thickness and Ångström exponent was found to be relatively small,
one can expect a more significant effect on the retrieved single-scattering albedo. The
latter is not a formal operational product, but follows implicitly from the retrieved
80
Ångström exponent and the assumed shape of the aerosol size distribution and the real
and imaginary parts of the aerosol refractive index. The availability of extensive in situ
measurements of the single-scattering albedo at Sable Island as reported by Delene and
Ogren (2002) enable us to examine the accuracy of our choice of the imaginary part of
the refractive index, at least in the Atlantic Ocean area adjacent to that location. This
will be done in Section 5.3.
5.2 Validation of satellite aerosol optical thickness retrievals
5.2.1 Methodology
The space-time collocation between satellite and sun photometer observation is
an important part of the aerosol validation process (Ignatov et al. 1995; Zhao et al.
2002). Many validations performed for spaceborne and airborne aerosol retrievals can
be found in the literature (e.g., Ignatov et al. 1995; Stowe et al. 1997; Nakajima and
Higurashi 1997; Tanŕe et al. 1997; Goloub et al. 1999; Zhao et al. 2002; Remer et al.
2002; Ichoku et al. 2002), each with different validation concepts and procedures. Since
a two-channel algorithm can retrieve only two aerosol parameters and must rely on
globally fixed values of all other model parameters, and because the retrieval accuracy
can be plagued by factors such as imperfect cloud screening and calibration
uncertainties, it appears more appropriate to talk about the “calibration” of the
algorithm in terms of minimizing the difference between the actual and the retrieved
global and regional long-term averages of the aerosol properties.
Considering that ship aerosol data are relatively scarce, that each research cruise
did not last longer than a few months, and that the route of the expedition varied, it is
not realistic to perform a monthly statistics comparison. We can however calculate the
aerosol statistics over a certain geographical location and within a certain range of time
81
provided that there are space-time collocated ship measurements. In our validation, we
take into account all the satellite retrieved data points over each location considered.
5.2.2 Ship data sets
Judging from a recent review paper (Smirnov et al., 2002), there is a total of 77
known published results of aerosol optical thickness measurements in maritime and
coastal areas from 1967 to 2001. Not all the data summarized by Smirnov et al. can be
used to validate the results of the satellite observations. First, 33 records were taken
before July 1983 when our satellite climatology begins. Even for those aerosol data
collected after July 1983, still sometimes measurement accuracy was unknown
(Smirnov et al. 2002). Table 5.1 summarizes the ship measurements used in this study.
The instruments, the method, and the errors of aerosol optical thickness determination
were described in the earlier papers (c.f. Smirnov et al. 1995; Kabanov and Sakerin
1997; Moulin et al. 1997; Smirnov et al. 2000; Sakerin and Kabanov 2002 etc.). It was
estimated that the errors in these measured aerosol optical thickness were not exceeding
0.02 in the visible range compared to the 0.03-0.05 error of reconstructing aerosol
optical thickness from the satellite (Rao et al. 1989; Ignatov et al. 1995; Tanѓe et al.
1997). Thus the accuracy of the sun photometer optical thickness is accurate enough to
use as a ground truth standard for validation of our retrieval product.
5.2.3 Primary validation results
Results of the comparison of the aerosol optical thickness at wavelength λ = 0.55
µm retrieved from the AVHRR data τSAT and ship-borne sun photometer measured τSP
for 58 aerosol statistics are presented in Figure 5.1. Each data point is numbered
according to the corresponding dataset in Table 5.1. The figure indeed shows the
aerosol optical thickness spatial distribution pattern described by Smirnov et al. (2002).
The atmosphere over the Pacific Ocean is more transparent compared to that of the
82
Table 5:1: Ship measurements of aerosol optical thickness in maritime areas used in this study
Reference Datea Area τ ± στb N/H/Dc
Volgin et al. (1988) Mediterranean Sea 1. 08/13/86-08/28/86 33N-40N, 21E-27E 0.20 ± 0.07 –/25/11 2. 09/06/86-09/15/86 34N-40N, 12E-29E 0.17 ± 0.05 –/21/9 Pacific Ocean 3. 11/01/85-11/04/85 10N-20N, 145E-158E 0.05 ± 0.02 –/7/4 4. 11/05/85-11/13/85 8S-9N, 159E-175E 0.08 ± 0.02 –/5/4 5. 11/16/85-12/01/85 21S-13S, 166W-172E 0.06 ± 0.01 –/24/15 6. 12/24/85-12/29/85 10N-14N, 124E-140E 0.07 ± 0.01 –/6/4 Shifrin et al. (1989) Indian Ocean 7. 11/17/83-11/18/83 11.6N-12.4N, 45.8E-53.8E 0.14 ± 0.04 –/2/2 Sakerin et al. (1991) & Korotaev et al. (1993) Atlantic Ocean 8. 09/01/89-09/10/89 36.27N-59.45N, 9.77W-2.85E 0.20 ± 0.13 29/–/6 9. 09/22/89-10/03/89 33.34N-41.96N, 68.96W-48.60W 0.08 ± 0.07 –/–/7 10. 10/04/89-10/11/89 17.02N-28.24N, 68.81W-43.32W 0.08 ± 0.03 –/–/8 11. 10/12/89-10/26/89 6.23N-15.19N, 38.03W-16.94W 0.31 ± 0.19 –/–/9 12. 11/17/89-12/04/89 2.02N-16.42N, 21.57W-17.01W 0.30 ± 0.10 82/–/14 13. 12/05/89-12/14/89 20.96N-31.28N, 19.40W-12.37W 0.06 ± 0.03 –/–/7 Mediterranean Sea 14. 12/15/89-12/20/89 35.18N-40.22N, 6.85W-26.56E 0.07 ± 0.05 –/–/6 Sakerin et al. (1993) Atlantic Ocean 15. 07/04/91-08/03/91 24.99N-32.97N, 30.38W-16.92W 0.24 ± 0.17 125/–/2816. 08/04/91-08/09/91 33.36N-39.76N, 64.33W-34.39W 0.14 ± 0.03 –/–/6 17. 08/18/91-09/08/91 38.28N-40.30N, 68.93W-61.24W 0.23 ± 0.27 –/–/20 18. 09/19/91-09/26/91 38.32N-41.02N, 70.22W-16.30W 0.13 ± 0.08 –/–/6 19. 09/09/91-09/18/91 36.76N-39.29N, 76.55W-73.10W 0.28 ± 0.20 –/–/9 Mediterranean Sea 20. 09/29/91-09/30/91 37.23N-37.25N, 5.75E-10.48E 0.38 –/–/2 Atlantic Ocean 21. 05/01/94-05/22/94 22.82N-28.14N, 16.94W-15.21W 0.18 ± 0.16 –/–/19 Villevalde et al. (1994) Pacific Ocean 22. 12/17/88-12/27/88 39.4N-40.0N, 170.0E-170.0E 0.13 ± 0.0 –/2/1 23. 12/30/88-12/27/88 25.0N-30.0N, 154.7E-180.0E 0.094 ± 0.031 –/8/6
83
Table 5.1 (continued)
Reference Date Area τ ± στ N/H/D
24. 01/13/89-01/18/89 10.8N-22.4N, 133.8E-156.2E 0.136 ± 0.037 –/12/6 25. 01/23/89-02/03/89 1.6S-18.4N, 174.9E-180.0E 0.147 ± 0.044 –/15/9 26. 02/11/89-02/21/89 40.0S-38.5S, 157.6E-179.9E 0.127 ± 0.015 –/3/3 Indian Ocean 27. 3/13/1989-03/16/89 23.0S-9.1S, 90.0E-90.0E 0.075 ± 0.021 –/2/2 Sakerin et al. (1995) Atlantic Ocean 28. 03/01/95-03/07/95 3.70N-14.70N, 20.47W-13.09W 0.41 ± 0.10 –/–/6 29. 03/08/95-04/04/95 1.65S-0.29N, 10.92W-9.64W 0.14 ± 0.08 –/–/19 30. 04/09/95-04/13/95 20.53N-36.40N, 17.90W-12.46W 0.27 ± 0.11 –/–/4 31. 04/14/95-04/21/95 40.21N-50.16N, 11.07W-1.23W 0.13 ± 0.07 –/–/7 Smirnov et al. (1995a) Mediterranean Sea 32. 12/14/89-12/21/89 37N-40N, 1W-13E 0.04 ± 0.02 –/4/3 33. 01/25/90-01/27/90 36N-38N, 3W-4E 0.06 ± 0.03 –/4/3 34. 08/25/91-08/30/91 36N-38N, 2E-25E 0.23 ± 0.12 –/6/4 Black Sea 35. 08/19/91-08/24/91 41N-44N, 28E-38E 0.33 ± 0.06 –/6/5 Atlantic Ocean 36. 09/02/91-09/23/91 37N-48N, 64W-24W 0.14 ± 0.07 –/8/8 37. 12/29/89-01/19/90 23N-27N, 28W-18W 0.07 ± 0.03 –/20/12 38. 10/03/91-10/18/91 21N-28N, 60W-20W 0.24 ± 0.04 –/20/12 39. 01/21/90-01/24/90 28N-35N, 15W-9W 0.08 ± 0.02 –/4/3 Gibraltar Area 40. 10/26/91-10/27/91 35N-36N, 13W-9W 0.21 ± 0.05 –/3/2 Smirnov et al. (1995b) Baltic Sea 41. 05/16/84-05/18/84 56.1N-56.1N, 11.8E-19.1E 0.461 ± 0.006 –/2/2 Atlantic Ocean 42. 05/20/84-06/06/84 60.5N-65.0N, 4.8W-2.7W 0.200 ± 0.078 –/4/4 Baltic Sea 43. 07/06/84-07/07/84 59.0N-59.3N, 21.1E-23.5E 0.087 ± 0.008 –/2/2 Atlantic Ocean 44. 05/23/88-05/26/88 62.6N-64.0N, 3.3W-4.7E 0.155 ± 0.040 –/5/3 45. 06/19/88-07/03/88 67.1N-68.8N, 2.0W-0.4W 0.110 ± 0.071 –/2/2 46. 05/14/89-05/26/89 58.0N-67.3N, 6.2W-7.0E 0.098 ± 0.017 –/6/5 Kabanov et al. (1997)
84
Table 5.1 continued
Reference Date Area τ ± στ N/H/D
47. 08/25/96-08/26/96 44.03N-44.03N, 63.55W-63.55W 0.13 ± 0.11 –/–/2 48. 08/27/96-09/13/96 29.15N-41.24N, 58.24W-21.44W 0.08 ± 0.04 –/–/18 49. 09/14/96-09/15/96 44.16N-46.70N, 15.61W-10.32W 0.09 ± 0.02 –/–/2 English Channel 50. 09/16/96-09/17/97 49.55N-51.83N, 3.70W-2.76E 0.100 ± 0.021 –/–/3 Moorthy et al. (1997) Indian Ocean 51. 01/07/96-01/22/96 5S-8.83N, 60E-69E 0.186 ± 0.10d –/15/13 Moulin et al. (1997) Atlantic Ocean 52. 09/14/91-09/29/91 15.50N-27.70N, 31.15W-17.90W 0.46 ± 0.31 –/–/13 Kuśmierczyk-Michulec (1999) Baltic Sea 53. 07/02/97-07/15/97 54N-58.5N, 11E-21E 0.205 ± 0.094 145/–/12 Smirnov et al. (2000) Atlantic Ocean 54. 07/08/96-07/08/96 34.93N-37.30N, 69.95W-67.34W 0.29 ± 0.07 24/–/1 55. 07/18/96-07/18/96 32.37N-32.78N, 64.87W-64.58W 0.06 ± 0.01 8/–/1 56. 07/26/96-07/26/96 36.74N-38.17N, 70.89W-68.48W 0.16 ± 0.04 12/–/1 57. 07/27/96-07/27/96 35.82N-38.24N, 71.09W-68.30W 0.25 ± 0.13 28/–/1 58. 07/08/96-07/27/96 32.37N-40.76N, 74.01W-64.57W 0.19 ± 0.12 130/–/7 a Read xx/xx/xx as mm/dd/yy. b τ and στ are the mean value and the standard deviation of aerosol optical thickness at a wavelength 0.55 µm. c N: number of sets of data; H: number of half day averages; D: number of observation days. d The measurements were taken at a wavelength of 0.50 µm.
Atlantic Ocean, inland seas and coastal zones. There is generally a good agreement
between the satellite retrievals and the ship data. One can see that over the Pacific
Ocean the retrieved τSAT values are higher relative to the ship-measured τSP. Aerosol
optical thickness over the Atlantic Ocean shows large variability, where the systematic
bias of τSAT relative to τSP , is somewhat less obvious compared to that over the Pacific
85
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5Measured τ(0.55 µm)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5A
VH
RR
ret
rieve
d τ(
0.55
µm
)
1
2
3
4
5
6 7 8
9
10
11
12
13
14
15
16
17 18
19
20
21
28
29
30
31
32
33
34
35
36
37
38
39
40
47
48
49
50
51
52
53
54
55
56
57
58
22
23
24
25
26
27
41
42
43
44
45
46
Mediterranean Sea
Pacific Ocean
Indian Ocean
Atlantic Ocean
Black Sea
Gibraltar Area
English Channel
Baltic Sea
Figure 5.1: AVHRR retrieved aerosol optical thickness τSAT versus ship measurements τSP at λ = 0.55 µm. Different sea areas are represented by different colors. The number near each data point corresponds to the aerosol data shown in Table 5.1.
86
Ocean although again we see a slight overestimation in τSAT. Since there is very little
aerosol data for the other areas except perhaps Mediterranean Sea, the comparison
results are less conclusive. Performing a linear regression analysis yields the following
relation between τSAT and τSP: τSAT = 0.047+0.836 τSP, with a high correlation coefficient
R of 0.90 and a standard error σ of 0.04.
However the ensemble average <τSAT > is 0.188, which is 11.2% higher than <τSP >.
This is more than the 3%-5% accuracy estimated in previous studies (Rao et al. 1989;
Ignatov et al. 1995; Tanŕe et al. 1997). The systematic errors of the retrieval algorithm
for low (τSP = 0), average (τSP = <τSAT >), and high (τSP = 1) aerosol loadings are found to
be 0.047, 0.019, and –0.117 respectively. This is comparable to the values reported by
Zhao et al. (2003) after the cloud and wind effects are minimized in their retrievals.
These values tell us that τSAT tends to be overestimated in low τ and underestimated in
high aerosol optical thickness.
The non-unity slope in the regression may be associated with a regional bias:
incorrect assumptions in the aerosol model of the retrieval algorithm for the sites of the
ship measurements. Due to the high aerosol inhomogeniety, this problem cannot be
easily solved by current retrieval algorithm which uses a global uniform aerosol model.
A possible improvement is to adopt regional aerosol models which take into account,
e.g., the stronger absorptivity of soot and dust-like particles (e.g., Fuller et al. 1999) and
the nonsphericity of mineral aerosols (e.g., Liou and Takano 1994; Mishchenko et al.
1997). This is a great challenge and will be our future research subject. A nonzero
intercept may be associated with sensor calibration error, improper assumption of ocean
surface reflection, and subpixel cloud contamination (Zhao et al. 2002).
In order to improve the agreement between the satellite retrievals and the ship data
we can try to reduce the positive intercept. An overestimation of τSAT at low τ can result
from either an underestimation of ocean surface reflection and/or an error in calibration
87
of satellite radiances, namely an error in the offset (deep space count) value. In order to
distinguish between the two possible causes we repeated the comparisons using data
from a single AVHRR instrument (NOAA-7, 9, 11, 14) each with its own calibrations
and the results are shown in Figure 5.2. No obvious satellite-dependent discrepancy was
found. We are thus inclined to consider the diffuse subsurface reflectance as a cause of
the overestimation.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5Measured τ(0.55 µm)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
AV
HR
R r
etrie
ved
τ(0.
55 µ
m)
NOAA 7
NOAA 9
NOAA 11
NOAA 14
Figure 5.2: τSAT versus τSP for different AVHRR instrument (NOAA-7, 9, 11, 14).
88
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5Measured τ(0.55 µm)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
AV
HR
R r
etrie
ved
τ(0.
55 µ
m)
S=0.002
S=0.004
S=0.005
y=0.047+0.836xR=0.899 σ=0.042
y=0.032+0.844xR=0.899 σ=0.043
y=0.021+0.850xR=0.895 σ=0.044
Figure 5.3: Comparison of τSAT and τSP at λ = 0.55 µm for three increasing values of diffuse component of surface reflection S = 0.002, 0.004, 0.005 and the corresponding linear regression lines. The dotted line depicts a 1:1 relationship.
Figure 5.3 shows τSAT versus τSP at λ = 0.55 µm and the corresponding linear
regression statistics using three increasing values of diffuse component of surface
reflectance S. Dotted line represents a 1:1 relationship. The detailed statistics are
summarized in table 5.2. Note our current two-channel retrieval algorithm uses the
value of S equals to 0.002. Although choosing S = 0.005 tends to minimize the offset,
the satellite retrievals are extremely sensitive to the selection of the surface reflectance
89
Table 5.2. Statistics of comparison of τSAT and τSP for three increasing values of diffuse surface reflection S = 0.002, 0.004, 0.005.
Intercept Slope Mean Systematic Errors (∆τ) Random Error S
a b R
τSAT τSP Minimum Mean Maximum σ 0.002 0.047 0.836 0.899 0.188 0.169 0.047 0.019 -0.117 0.042 0.004 0.032 0.844 0.899 0.175 0.169 0.032 0.006 -0.124 0.043 0.005 0.021 0.850 0.895 0.165 0.169 0.021 -0.004 -0.129 0.044
Systematic error is defined as ∆τ = τSAT – τSP (or ∆τ = a + b τSP – τSP), for minimum, mean and maximum ∆τ, τSP = 0, mean of τSP and 1 correspondingly. R is correlation coefficient between τSAT and τSP. Parameters a, b, and σ are intercept, slope, and standard deviation, respectively, of the linear regression line.
model in low τ. One cannot fail to notice that one retrieval falls to zero if S = 0.005 is
adopted in the retrieval algorithm and an underestimation of τSAT relative to τSP. Overall,
S = 0.004 is a good choice in terms of minimizing the difference between the actual and
the retrieved aerosol optical thickness, with ensemble average of τSAT differs that of τSP
only by 3.6%, within the 0.03-0.05 range estimated by previous papers (Rao et al. 1989;
Ignatov et al. 1995; Tanŕe et al. 1997). Linear regression of τSAT versus τSP produces a
good match τSAT = 0.032+0.844 τSP with a high correlation coefficient of 0.9 and a small
standard error of 0.04. This is encouraging considering that AVHRR was not specially
designed for aerosol retrievals and the sensor is less advanced and well calibrated
compared to that of the MODerate Resolution Imaging Spectrometer (MODIS) and the
Multiangle Imaging SpectroRadiometer (MISR) and considering that all the available
satellite retrievals were included in the comparisons.
90
5.3 Validation of the aerosol single-scattering albedo
Although the aerosol single-scattering albedo ϖ is not included explicitly in the
operational product generated by the two-channel retrieval algorithm (Mishchenko et al.
1999 and Geogdzhayev et al. 2002), it can be determined from the implicit relationship
between ϖ and the Ångström exponent provided that the aerosol refractive index m is
fixed. Fig. 5.4 shows this relationship computed using the Lorenz-Mie theory for Re(m)
= 1.5 and four increasing values of Im(m). As in (Mishchenko et al. 1999 and
Geogdzhayev et al. 2002), we use the simple power law distribution of aerosol radii
>
≤<
≤
=−
,,0
,,)(
,),(
),(
2
211
1
rr
rrrrrC
rrC
rnα
α
α
α (5.1)
with r1 = 0.1 µm, r2 = 10 µm, and ]5,5.2[∈α , where the normalization constant C(α) is
chosen such that
∫∞
=0
.1),( αrndr (5.2)
Obviously, the ϖ(A) dependence is always monotonic, and ϖ always decreases with
particle size. Furthermore, ϖ always decreases with increasing Im(m ).
The current version of the two-channel algorithm uses the value Im(m) = 0.003
(Geogdzhayev et al. 2002). Fig. 5.5 illustrates the single-scattering albedo for July 1999
(lower panel) determined from the retrieved constrained Ångström exponent (upper
panel). The latter means that only the Ångström exponent values falling in the interval
(0.05,1.72) are retained (Geogdzhayev et al. 2002). It is seen that ϖ varies from 0.86 to
0.98 and is indeed smaller in areas dominated by larger particles. Unfortunately, this
implicit relationship can be unphysical in that it does not allow large but weakly
absorbing particles (e.g., sea salt aerosols) to have greater single-scattering albedos than
91
0 0.5 1 1.5 2Ångström exponent
0.7
0.8
0.9
1
Sin
gle−
scat
terin
g al
bedo
at λ
= 0
.55
µm
m = 1.5 + 0.001i
m = 1.5 + 0.002i
m = 1.5 + 0.003i
m = 1.5 + 0.005i
Figure 5.4: Aerosol single-scattering albedo versus Ångström exponent for Re(m) = 1.5 and four increasing values of Im(m).
smaller but strongly absorbing particles (e.g., soot aerosols). However, the choice of a
constant Im(m) can still be optimized in such a way that it provides a realistic global
average value of the single-scattering albedo. Although the latter is not known at
present, the extensive in situ measurements of ϖ at Sable Island (43.933°N, 60.007°W)
reported by Delene and Ogren (2002) can be used to validate our choice of Im(m), at
least for that region of the Atlantic Ocean.
The measurements by Delene and Ogren cover the period from 23 November 1994
to 15 April 2000 and are summarized in Fig. 5.6. For comparison, Fig. 5.6 also depicts
the results of AVHRR retrievals averaged over a 100×100 km square centered at Sable
Island for the period from November 1994 to December 1999. In this study, we have
92
90oS
60oS
30oS
EQ
30oN
60oN
90oN
DL 120oW 60oW CM 60oE 120oE DL
ÅNGSTRÖM EXPONENT(a)
0
0.5
1
1.5
2
90oS
60oS
30oS
EQ
30oN
60oN
90oN
DL 120oW 60oW CM 60oE 120oE DL
SINGLE−SCATTERING ALBEDO(b)
0.86
0.89
0.92
0.95
0.98
Figure 5.5: Monthly averages of the Ångström exponent and single-scattering albedo for July 1999 derived from two-channel AVHRR data assuming a fixed aerosol refractive index m = 1.5 + 0.003i.
93
JAN MAR MAY JUL SEP NOV AVER Month of year
0.85
0.9
0.95
1
Sin
gle−
scat
terin
g al
bedo
at λ
= 0
.55
µmIm(m)=0.001 Im(m)=0.002 Im(m)=0.003 Im(m)=0.005
Figure 5.6: The annual cycle of the aerosol single-scattering albedo measured in situ at Sable Island (Delene and Ogren 2002). The whiskers denote the 5 and 95 percentiles, the bottom and top of the box denote the 25 and 75 percentiles, and the horizontal line within the box denotes the median value. The statistics are based on the hourly averaged data for all valid measurements obtained during the period 11/23/1994–04/15/2000. The horizontal axis shows the month of year, with the last tick mark representing the statistics for the entire study period. The circle, triangle, plus, and diamond signs represent the average monthly ϖ values retrieved from channel–1 and –2 AVHRR data during the period 11/1994–12/1999 assuming that the imaginary part of the aerosol refractive index is fixed at 0.001, 0.002, 0.003, and 0.005, respectively. The AVHRR results for March, April, and December were not computed because of the insufficient number of cloud-free pixels during these months.
94
Monthly average τ for 94/11−99/12
JAN MAR MAY JUL SEP NOV AVER Month of year
0
0.2
0.4
0.6A
eros
ol o
ptic
al th
ickn
ess
at λ
= 0
.55
µm
Im(m)=0.001 Im(m)=0.002 Im(m)=0.003 Im(m)=0.005
Figure 5.7: The annual cycle of the aerosol optical thickness retrieved from channel–1 and –2 AVHRR data over Sable Island during the period November 1994–December 1999 assuming that the imaginary part of the aerosol refractive index is fixed at 0.001, 0.002, 0.003, and 0.005. The horizontal axis shows the month of year, with the last tick mark representing the average results for the entire study period. The AVHRR results for December were not computed because of the insufficient number of cloud-free pixels.
95
Constrained monthly average Ångström exponent for 94/11−99/12
JAN MAR MAY JUL SEP NOV AVER Month of year
0
0.3
0.6
0.9
1.2
1.5Å
ngst
röm
exp
onen
t
Im(m)=0.001 Im(m)=0.002 Im(m)=0.003 Im(m)=0.005
Figure 5.8: As in Fig. 5.7, but for the constrained Ångström exponent. The AVHRR results for March, April, and December were not computed because of the insufficient number of cloud-free pixels during these months.
compared our AVHRR-retrieved single-scattering albedo ϖ with the ground-based
results and investigate its seasonal variability using several discrete values of the
imaginary part of the aerosol refractive index Im( m ) in order to accommodate the
strong variability of aerosol absorption. It is seen that the in situ ϖ results may be best
reproduced by an Im(m) value between 0.002 and 0.003. Figs. 5.7 and 5.8 show a rather
weak dependence of the retrieved aerosol optical thickness and Ångström exponent on
Im(m) for Im(m) in the range [0.002, 0.003] and suggests that the current choice
96
Im(m)=0.003 is quite consistent, at least for locations in the vicinity of Sable Island.
This conclusion is only reinforced by the inspection of Table 5.3, which suggests that an
Im(m) value close to 0.0025 is needed to reproduce the long-term mean in situ value of
the single-scattering albedo and a value close to 0.0035 could reproduce the long-term
mean in situ value of the Ångström exponent (Delene and Ogren 2002).
Table 5.3: Mean values of aerosol parameters retrieved from AVHRR data and measured in situ at Sable Island.
Im(m) ϖ τ A Retrieved values for 11/1994-12/1999 0.001 0.978 0.160 0.710 0.002 0.962 0.164 0.735 0.003 0.949 0.172 0.758 0.005 0.928 0.184 0.786 In situ values for 11/1994-04/2000 (Delene and Ogren 2002) 0.956 0.770
5.4 Discussion and conclusions
We have used ship-borne aerosol data with known accuracy estimates to validate
our global two-channel AVHRR satellite retrievals. The results of comparison statistics
for three increasing values of diffuse surface reflectance are summarized in Table 5.2.
Adjusting the diffuse component of surface reflection S from 0.002 value in the
operational algorithm to 0.004 reduces the positive offset of 0.047 to 0.032. The
positive bias in AVHRR retrieved τSAT compared to ship τSP at mean aerosol optical
thickness was reduced from 0.019 (11.2 % relative to τSP) to 0.006 (3.6% relative to τSP).
We also used AERONET sun photometer observations to evaluate our satellite
retrievals and obtained similar results to the study by Zhao et al. (2003)
97
The small systematic error at low τ and the random error may be attributed to
calibration errors, radiometric noise, and measurement instability. These sensors-related
error sources are anticipated to have less effect on the aerosol retrievals from more
advanced instruments such as MODIS and MISR. Our further research will focus on the
comparisons of AVHRR-derived aerosol properties over oceans with these new satellite
data products.
The proportional systematic error at large τ is mainly the result of improper
assumptions on the aerosol microphysical model in the retrieval algorithm. For
example, Zhao et al. (2002) have showed that the proportional systematic error at large
aerosol optical thickness can be reduced by increasing the imaginary part of the aerosol
refractive index Im(m). However one can argue that good agreement with a limited
ground-based or in situ dataset does not guarantee the global applicability of the
satellite-retrieved product. Our comparison indicated that the currently adopted value
Im(m)=0.003 can be a reasonable choice for the imaginary part of the aerosol refractive
index in the global AVHRR retrievals by comparing single-scattering albedo and
Ångström exponent values generated by the operational two-channel algorithm and
those measure in situ at Sable Island (Delene and Ogren 2002).
98
References
Delene, D. J., and J. A. Ogren, 2002: Variability of aerosol optical properties at four
North American surface monitoring sites. J. Atmos. Sci., 59, 1135-1150.
Fuller, K. A., W. C. Malm, and S. M. Kreidenweis, 1999: Effects of mixing on
extinction by carbonaceous particles. J. Geophys. Res., 104, 15 941-15 954.
Geogdzhayev, I. V., M. I. Mishchenko, W. B. Rossow, B. Cairns, and A. A. Lacis,
2002: Global two-channel AVHRR retrievals of aerosol properties over the ocean
for the period of NOAA-9 observations and preliminary retrievals using NOAA-7
and NOAA-11 data. J. Atmos. Sci., 59, 262-278.
Goloub, P., D. Tanŕe, J. L. Deuzé, M. Herman, A. Marchand, and F. M. Breon, 1999:
Validation of the first algorithm applied for deriving the aerosol properties over the
ocean using the POLDER/ADEOS measurements. IEEE Trans. Geosci. Remote
Sens., 37, 1586-1596.
Holben, B. N., T. F. Eck, I. Slutsker, D. Tanŕe, J. P. Buis, A. Setzer, E. Vermote, J. A.
Reagan, Y. J. Kaufman, T. Nakajima, F. Lavenu, I. Jankowiak, and A. Smirnov,
1998: AERONET ─ A federated instrument network and data archive for aerosol
characterization. Rem. Sens. Environ., 66, 1-16.
Ichoku, C., D. A. Chu, S. Mattoo, Y. J. Kaufman, L. A. Remer, D. Tanŕe, I. Slutsker,
and B. N. Holben, 2002: A spatio-temporal approach for global validation and
analysis of MODIS aerosol products. Geophys. Res. Lett., 29 (12), MOD-1.
Ignatov, A. M., L. L. Stowe, S. M. Sakerin, and G. K. Korotaev, 1995: Validation of the
NOAA/NESDIS satellite aerosol product over the North Atlantic in 1989. J.
Geophys. Res., 100, 5123-5132.
Kabanov, D. M., and S. M. Sakerin, 1997: Results of investigations of the aerosol
optical thickness and the water vapor column density in the atmosphere of central
Atlantic. Atmos. Oceanic Opt., 10, 913-918.
99
Kabanov, D. M., and S. M. Sakerin, 1997: About method of atmospheric aerosol optical
thickness determination in near IR-spectral range. Atmos. Oceanic Opt., 10, 540-
545.
Kinne, S., B. Holben, T. Eck, A. Smirnov, O. Dubovik, I. Slutsker, D. Tanŕe, G.
Zibozdi, U. Lohmann, S. Ghan, R. Easter, M. Chin, P. Ginoux, T. Takemura, I.
Tegen, D. Koch, R. Kahn, E. Vermote, L. Stowe, O. Torres, M. Mishchenko, I.
Geogdzhayev, and A. Higurashi, 2001: How well do aerosol retrievals from
satellites and representation in global circulation models match ground-based
AERONET aerosol statistics? In Remote sensing and climate modeling: Synergies
and Limitations (M. Beniston and M. M. Verstraete, Eds.), Kluwer Academic
Publishers, Dordrecht, pp. 103-158.
Korotaev, G. K., S. M. Sakerin, A. M. Ignatov, L. L. Stowe, and E. P. McClain, 1993:
Sun-Photometer observations of aerosol optical thickness over the North Atlantic
from a Soviet Research vessel for validation of satellite measurements. J. Atmos.
Oceanic Technol., 10, 725-735.
Kuśmierczyk-Michulec J., O. Krüger, and R. Marks, 1999: Aerosol influence on the
sea-viewing wide-field-of-view sensor bands: Extinction measurements in a marine
summer atmosphere over the Baltic Sea. J. Geophys. Res. 104, 14,293-14,307.
Liou, K. N., and Y. Takano, 1994: Light scattering by nonspherical particles: Remote
sensing and climatic applications. Atmos. Res., 31, 271-298.
Mishchenko, M. I., L. D. Travis, R. A. Kahn, and R. A. West, 1997: Modeling phase
functions for dustlike tropospheric aerosols using a shape mixture of randomly
oriented polydisperse spheroids. J. Geophys. Res., 102, 16 831-16 847.
Mishchenko, M. I., I. V. Geogdzhayev, B. Cairns, W. B. Rossow, and A. A. Lacis,
1999: Aerosol retrievals over the ocean by use of channels 1 and 2 AVHRR data:
sensitivity analysis and preliminary results. Appl. Opt., 38, 7325-7341.
100
Mishchenko, M. I., I. V. Geogdzhayev, L. Liu, J. A. Ogren, A. A. Lacis, W. B. Rossow,
J. W. Hovenier, H. Volten, and O. Muñoz, 2003: Aerosol retrievals from AVHRR
radiances: effects of particle nonsphericity and absorption and an updated long-
term global climatology of aerosol properties. J. Quant. Spectrosc. Radiat. Transfer
(in press).
Moorthy, K. K., S. K. Satheesh, and B. V. K. Murthy, 1997: Investigations of marine
aerosols over the tropical Indian Ocean. J. Geophys. Res. 102, 18,827-18,842.
Moulin, C., F. Dulac, C. E. Lambert, P. Chazette, I. Jankowiak, B. Chatenet, and F.
Lavenu, 1997: Long-term daily monitoring of Saharan dust load over ocean using
Meteosat ISCCP-B2 data. 2. Accuracy of the method and validation using Sun
photometer measurements. J. Geophys. Res., 102, 16,959-16,969.
Myhre, G., F. Stordal, M. Johnsrud, A. Ignatov, M. I. Mishchenko, I. V. Geogdzhayev,
D. Tanŕe, P. Goloub, T. Nakajima, A. Higurashi, O. Torres, and B. N. Holben,
2002: Intercomparison of satellite retrieved aerosol optical depth over ocean, J.
Geophys. Res. (submitted).
Nakajima, T., and A. Higurashi, 1997: AVHRR remote sensing of aerosol optical
properties in the Persian Gulf region, summer 1991. J. Geophys. Res., 102, 16 935-
16 946.
Rao, C. R. N., L. L. Stowe, and El Pl McClain, 1989: Remote sensing of aerosols over
the oceans using AVHRR data: Theory, practice and applications. Int. J. Remote
Sens., 10, 743-749.
Remer, L. A., D. Tanŕe, Y. J. Kaufman, C. Ichoku, S. Mattoo, R. Levy, D. A. Chu, B.
N. Holben, O. Dubovik, A. Smirnov, J. V. Martins, R-R. Li, and Z. Ahmad, 2002:
Validation of MODIS aerosol retrieval over ocean. Geophy. Res. Lett., 29(12),
MOD-3.
101
Sakerin, S. M., S. V. Afonin, T. A. Eremina, A. M. Ignatov, and D. M. Kabanov, 1991:
General characteristics and statistical parameters of spectral transmission of the
atmosphere in some regions of the Atlantic. Atm. Opt., 4, 504-510.
Sakerin, S. M., I. L. Dergileva, A. M. Ignatov, and D. M. Kabanov, 1993: Enhancement
of the turbidity of the atmosphere over the Atlantic in the post Mt. Pinatubo
eruption period. Atmos. Oceanic Opt., 6, 711-714.
Sakerin, S. M., D. M. Kabanov, and V. V. Pol’kin, 1995: Optical investigations of the
atmosphere during the 35th mission of the research vessel AKADEMIK MSTISLAV
KELDYSH. Atmos. Oceanic. Opt., 8, 981-988.
Sakerin, S. M., and D. M. Kabanov, 2002: Spatial inhomogeneities and the spectral
behavior of atmospheric aerosol optical depth over the Atlantic ocean. J. Atmos.
Sci., 59, 484-500.
Shifrin, K. S., V. M. Volgin, B. N. Volkov, O. A. Ershov, and A. V. Smirnov, 1989:
Optical thickness of aerosols over the sea. Sov. J. Remote Sen., 5, 591-605.
Smirnov, A., O. Yershov, Y. Villevalde, 1995a: Measurement of aerosol optical depth
in the Atlantic Ocean and Mediterranean Sea. Proc. SPIE, 2582, 203-214.
Smirnov, A., Y. Villevalde, N. T. O’Neill, A. Royer, and A. Tarussov, 1995b: Aerosol
optical depth over the oceans: Analysis in terms of synoptic air mass types. J.
Geophys. Res. 100, 16 639-16 650.
Smirnov A., B. N. Holben, O. Dubovik, N. T. O’Neill, L. A. Remer, T. F. Eck, I.
Slutsker, and D. Savoie. Measurement of atmospheric optical parameters on U.S.
Atlantic coast sites, ships, and Bermuda during TARFOX, 2000: J. Geophys. Res.
105, 9887-9901.
Smirnov, A., B. N. Holben, Y. J. Kaufman, O. Dubovik, T. F. Eck, I. Slutsker, C.
Pietras, and R. N. Halthore, 2002: Optical properties of atmospheric aerosol in
maritime environments. J. Atmos. Sci., 59, 501-523.
102
Stowe, L. L., A. M. Ignatov, and R. R. Singh, 1997: Development, Validation, and
potential enhancements to the second-generation operational aerosol product at the
National Environmental Satellite, Data, and Information Service of the National
Oceanic and Atmospheric Administration. J. Geophys. Res., 102, 16 923-16 934.
Tanŕe, D., Y. J. Kaufman, M. Herman, and S. Mattoo, 1997: Remote sensing of aerosol
properties over oceans using the MODIS/EOS spectral radiances. J. Geophys. Res.,
102, 16 971-16 988.
Villevalde, Y. V., A. V. Smirnov, N. T. O’Neill, S. P. Smyshlyaev, and V. V.
Yakovlev, 1994: Measurement of aerosol optical depth in the Pacific Ocean and the
North Atlantic. J. Geophys. Res. 99, 20 983-20 988.
Volgin, V. M., O. A. Yershov, A. V. Smirnov, and K. S. Shifrin, 1988: Optical depth of
aerosol in typical sea areas. Izv. Acad. Sci. USSR Atmos. Oceanic Phys., 24, 772-
777.
Zhao, T. X.-P., L. L. Stowe, A. Smirnov, D. Crosby, J. Sapper, and C. R. McClain,
2002: Development of a global validation package for satellite oceanic aerosol
retrieval based on AERONET sun/sky radiometer observations and its application
to NOAA/NESDIS operational aerosol retrievals. J. Atmos. Sci., 59, 294-312.
Zhao, T. X.-P., I. Laszlo, B. N. Holben, C. Pietras, and K. J. Voss, 2003: Validation of
two-channel VIRS retrievals of aerosol optical thickness over ocean and
quantitative evaluation of the impact from potential subpixel cloud contamination
and surface wind effect. J Geophys. Res., 108 (D3), AAC 7.
103
Chapter 6
Constraints on PSC particle microphysics derived fromlidar observations1
Li Liua, Michael I. Mishchenkob
aDepartment of Earth and Environmental Sciences, Columbia University, and NASA GISS, 2880Broadway, New York, NY 10025, USA
bGoddard Institute for Space Studies, 2880 Broadway, New York, NY 10025, USA
Abstract
Based on extensive T-matrix computations of light scattering by polydispersions of
randomly oriented, rotationally symmetric nonspherical particles, we analyze existing
lidar observations of polar stratospheric clouds (PSCs) and derive several constraints
on PSC particle miscrophysical properties. We show that sharp-edged nonspherical
particles (finite circular cylinders) exhibit less variability of lidar backscattering
characteristics with particle size and aspect ratio than particles with smooth surfaces
(spheroids). For PSC particles significantly smaller than the wavelength, the
backscatter color index α and the depolarization color index β are essentially shape-
independent. Observations for type Ia PSCs can be reproduced by spheroids with aspect
ratios larger than 1.2, oblate cylinders with diameter-to-length ratios greater than 1.6,
and prolate cylinders with length-to-diameter ratios greater than 1.4. The effective
equal-volume-sphere radius for type Ia PSCs is about m8.0 µ or larger. Type Ib PSCs
are likely to be composed of spheres or nearly spherical particles with effective radii
smaller than 0.8 µm. Observations for type II PSCs are consistent with large ice crystals
1 Published in Journal of Quantitative Spectroscopy & Radiative Transfer, 70, 817-831, 2001
104
(effective radius greater than 1 µm) modeled as cylinders or prolate spheroids.
Keywords: Scattering, Depolarization, Polar stratospheric clouds, Remote sensing,
Nonspherical particles
6.1 Introduction
Although polar stratospheric clouds (PSCs) are unlikely to cause a significant direct
radiative forcing of climate, their critical role in chemical ozone depletion is now well
recognized (e.g., [1] and references therein). A detailed review of PSC studies has
recently been published by Toon et al. [2]. The traditional classification of PSC types is
based on lidar observations [3]. Type Ia PSCs are characterized by low backscattering
but strong depolarization, whereas type Ib PSCs exhibit the opposite behavior. Type II
PSCs demonstrate both strong backscatter and large depolarization ratios. Type Ib
clouds are believed to consist of droplets of supercooled ternary solutions (STSs) of
water, nitric acid, and sulfuric acid [4], whereas type Ia PSCs are thought to form by
condensation of nitric acid tri- or dihydrate (NAT or NAD). Type II PSCs are thought
to consist of water ice crystals. Although more recent observations indicate the
occurrence of other types of PSCs (e.g., [5–9]), the traditional classification covers the
majority of PSC observations (e.g., [10]) and is, therefore, the focus of this paper.
Our aim is to derive certain constraints on PSC particle microphysics based solely
on remote-sensing lidar observations. Although these constraints may not always
provide a definitive identification of the shape, size, and composition of PSC particles,
they can significantly narrow down the plausible range of particle microphysical
parameters and can be useful in analyses of in situ physical and chemical measurements
as well as in modeling PSC particle formation and evolution.
Analyzing lidar observations of PSCs is a challenging problem because many PSCs
105
are likely to consist of nonspherical solid particles with sizes comparable to the
wavelength of the lidar light (at least for lidars operating in the visible and near-infrared
spectral ranges). Theoretical computations of lidar backscatter and depolarization by
wavelength-sized nonspherical particles may not rely on the geometrical optics
approximation [11,12] but must be based on directly solving the Maxwell equations
using an exact numerical technique [13]. Although techniques such as the finite
difference time domain method (FDTDM; [14,15]) and the discrete dipole
approximation (DDA; [16]) have no restrictions on the particle shape, their relatively
low efficiency and limited size parameter range make their application to
polydispersions of randomly oriented PSC particles problematic. Therefore, we base
our analysis on the T-matrix code [17] specifically designed for polydisperse, randomly
oriented, rotationally symmetric particles such as spheroids and finite circular cylinders.
The code takes advantage of the analytical procedure for averaging scattering and
absorption characteristics over the uniform distribution of particle orientations [18,19],
uses the matrix inversion scheme based on a special form of the LU-factorization
method [20], and extends the range of particle size parameters by employing extended-
precision floating-point FORTRAN variables [21]. Although the code is limited to
rotationally symmetric particle shapes, it is much faster and is applicable to
significantly larger size parameters than FDTDM and DDA. Furthermore, by changing
the aspect ratio of spheroids and the length-to-diameter ratio of circular cylinders one
can model a wide variety of oblate and prolate shapes with either smooth or sharp-
edged surfaces.
6.2 T-matrix computations
Assuming that the laser light is linearly polarized, we define the linear
backscattering depolarization ratio )(λδ at the lidar wavelength λ as the ratio of the
106
aerosol backscatter returns in the perpendicular, )(λβ ⊥a , and parallel, )(// λβa , planes
relative to the emitted polarization plane [22-24]:
)()(
)(// λβ
λβ=λδ ⊥
a
a (6.1)
The backscatter ratio )(λR is defined as the ratio of the total molecular, )(λβm , and
aerosol, )(λβa , backscatter returns to the molecular backscatter return [3]:
)()()(
)(λβ
λβ+λβ=λm
amR . (6.2)
Similarly, the aerosol backscatter ratio is given by
1)()()(
)( −λ=λβλβ=λ RR
m
aa . (6.3)
The color ratio ),( 12 λλS is defined as the ratio of the aerosol backscatter ratios at
wavelengths 2λ and 1λ :
)()(
),(1
212 λ
λ=λλa
a
R
RS . (6.4)
The wavelength dependence of the aerosol backscatter is often characterized by the
backscatter color index
)ln(),(ln
412
12
λλλλ−=α S
. (6.5)
Analogously, the wavelength dependence of the depolarization ratio is described by the
depolarization color index
)ln(])()(ln[
12
12
λλλδλδ−=β . (6.6)
It is thus assumed that the aerosol backscatter aβ is proportional to α−λ (given the 4−λ
dependence of the molecular backscatter) and the depolarization ratio δ is proportional
to β−λ . A positive α (β ) indicates that the aerosol backscatter (depolarization ratio)
107
decreases with increasing wavelength [25]. A large α indicates the abundance of
particles with radii smaller than the lidar wavelengths, whereas a small α indicates the
predominance of large particles [2,6]. The depolarization ratio vanishes for molecular
scattering and spherical particles. A nonzero δ usually indicates the presence of
nonspherical particles provided that the contribution of multiple scattering to the
detected signal is negligible [23,24,26]. (Note, however, that wavelength-sized
needlelike and platelike particles can generate depolarization ratios close to zero [27]).
The lidar scattering properties of particles depend upon the lidar wavelengths as
well as on the particle size distribution, shape, and refractive index. We model the
shape of nonspherical PSC particles using smooth spheroids with varying aspect ratios
and sharp-edged circular cylinders with varying length-to-diameter ratios. Previous
analyses of PSC observations assumed different analytical representations of the
particle size distribution, including gamma and log normal distributions. However, it
has been demonstrated by Hansen and Travis [28] that many size distributions can be
well represented by just two parameters, the effective radius effr and effective variance
effv . Specifically, they have shown that different size distributions but with the same
effr and effv can be expected to have similar scattering properties. In this study, we
have adopted a simple power law distribution given by
>≤≤
≤=
,for0
,for)(
,for
)(
2
213
1
1
rr
rrrrrC
rrC
rn (6.7)
where C is a normalization constant and r is the radius of the volume-equivalent
sphere. The parameters 1r and 2r are chosen such that the effective variance is fixed at
1.0 , representing a moderately wide distribution. An important advantage of the power
law distribution is that for the same effv , it has a (much) smaller value of the maximal
108
radius 2r than optically equivalent gamma and log normal distributions, thereby
significantly accelerating theoretical computations [29].
Most of our theoretical computations pertain to two typical lidar wavelengths of
603 nm and 1064 nm and assume a spectrally independent refractive index.
Measurements by Deshler et al. [30] showed that the refractive index of PSC particles at
visible and near-infrared wavelengths was close to 47.1 ± 01.0 in the lower
nondepolarizing layer of the cloud and 1.52–1.56 ± 04.0 in the upper depolarizing
layer. These values are in reasonable agreement with those reported by Middlebrook et
al. [31] and Berland et al. [32]. A typical value of the refractive index of water ice at
visible and near-infrared wavelengths is 1.308 [33]. Based on this evidence, we chose
the refractive indices 1.5 and 1.308 as typical of type I and II PSC particles,
respectively. We thus used the same refractive index for type Ia and Ib PSCs, the
reason being that their refractive indices appear to be close and that their chemical
composition remains to be somewhat uncertain. The computations have been
performed for the range 0 – 3 µm of effective radii for prolate and oblate spheroids with
aspect ratios from 1 to 2, prolate cylinders with length-to-diameter ratios from 1 to 2,
and oblate cylinders with diameter-to-length ratios from 1 to 2.
6.3 Observational data
Table 6.1 summarizes the results of lidar observations of PSCs at nm6031 =λ and
Table 6.1: Typical backscattering characteristics of PSCs observed by Browell et al. [3].
R (%)δCloud type
nm6031 =λ nm10642 =λα
nm6031 =λ nm10642 =λβ
Ia 1.2 – 1.5 2 – 5 0.4 30 – 50 30 – 50 ~0Ib 3 – 8 5 – 20 2 – 3 0.5 – 2.5 <4 No dataII >10 >20 <0.8 >10 >10 ~0
109
nm10642 =λ performed by Browell et al. [3] during the NASA/NOAA Airborne
Arctic Stratospheric Expedition in the winter of 1988–1989. Larsen et al. [34]
performed extensive balloon-borne backscatter sonde observations of type I PSCs
during the winters of 1989, 1990, 1995, and 1996 and concluded that the color ratio
)nm480,nm940(S was greater than 10 for type Ia PSCs and was in the range 5–8 for
type Ib PSCs. The range of the color ratio )nm490,nm940(S observed by Vömel et
al. [35] for type II PSCs was 11–15.
The backscatter ratio defined by Eq. (6.2) depends on the PSC particle number
concentration. Since the latter cannot be directly retrieved from lidar measurements and
is a priori unknown, we have not used the measurements of the backscatter ratio in our
analysis. Unlike R, the backscatter color index, the depolarization ratio, and the
depolarization color index are independent of the particle number concentration and can
be directly used to infer particle microphysical characteristics. We have converted the
color ratio values measured by Larsen et al. [34] and Vömel et al. [35] into the
respective backscatter color index values and combined the latter with the
measurements by Browell et al. [3]. Table 6.2 summarizes the ranges of the lidar
quantities α, δ, and β used in our analysis. Note that instead of using a constant β value
equal to zero for type Ia and II PSCs, we allowed it to vary within a narrow range [–0.4,
0.4], which seems to be a reasonable assumption given the natural variability of PSC
particles. The same approach was used to define the plausible range of variability of the
backscatter color index for type 1a PSCs.
Table 6.2: Ranges of backscattering characteristics for different PSC types used in this study.
Cloud Type α (%)nm)603(δ βIa 0 – 0.8 30 – 50 –0.4 – 0.4Ib 1 – 3 0.5 – 2.5 No dataII –0.2 – 0.8 >10 –0.4 – 0.4
110
6.4 Analysis results
Figures 6.1–6.4 summarize the results of the T-matrix computations. In general,
the curves of all backscattering characteristics versus effective equivalent-sphere radius
are (much) less aspect-ratio dependent for cylinders than for spheroids. This is not
surprising since spheroids with an aspect ratio of one are perfect spheres, whereas
cylinders with a length-to-diameter ratio of one are already distinctly nonspherical
particles. There is a rapid increase of the depolarization ratio with increasing effective
radius from 0 to about 0.5 µm. Maximal δ values for most nonspherical particles are
observed at effective equal-volume-sphere radii between 0.5 µm and 1.5 µm. The most
notable exception are nearly spherical spheroids, whose depolarization ratio increases
with particle size rather monotonically. It is obvious that there is no definitive
relationship between the magnitude of depolarization and the degree of particle
asphericity [36]. For example, prolate spheroids with as small an aspect ratio as 1.05
(Fig. 6.1) produce depolarization ratios exceeding 65%. The maximal depolarization
value (~70%) is produced by prolate ice spheroid with an aspect ratio of 1.2 (Fig. 6.3).
The fact that this maximal value is caused by wavelength-sized particles indicates that
multiple internal reflections in very large particles as discussed by Liou and Lahore [37]
are not the only mechanism generating strong depolarization and not necessarily the
mechanism producing maximal depolarization values [22].
In the large-particle limit, both α and β are expected to tend to zero for
nonabsorbing scatterers. This trend is more visible for cylinders than for spheroids.
However, both color indices exhibit a significant degree of variability in the range of
sizes studied for cylinders as well as for spheroids. In the Rayleigh limit, α must tend
to 4 and β must tend to zero. This theoretical behavior is indeed well reproduced by
our T-matrix computations.
111
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70
Line
ar D
epol
ariz
atio
n R
atio
δ(%
)Prolate Spheroids
0
10
20
30
40
50
60
70
ε21.81.61.41.21.11.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Obate Spheroids
ε21.81.61.41.21.11.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4
Bac
ksca
tter
Col
or In
dex
α
Prolate Spheroids
ε21.81.61.41.21.11.051
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4Oblate Spheroids
ε21.81.61.41.21.11.051
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8
Dep
olar
izat
ion
Col
or In
dex
β
Prolate Spheroids
-2
0
2
4
6
8
ε21.81.61.41.21.11.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8Oblate Spheroids
-2
0
2
4
6
8
ε21.81.61.41.21.11.05
Figure 6.1: Linear depolarization ratio )nm603(δ , backscatter color index α , and
depolarization color index β versus effective equal-volume-sphere radius effr for polydisperse,
randomly oriented spheroids with a refractive index of 5.1=m . ε is the ratio of the largest tothe smallest semi-axes of a spheroid. The light and dark shaded areas show the observed rangesof these parameters for type Ia and Ib PSCs, respectively.
112
Figures 6.5–6.7 show the ranges of the effective equivalent-sphere radius that
reproduce the observed values of the respective backscattering characteristics for
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Li
near
Dep
olar
izat
ion
Rat
io δ
(%)
Prolate Cylinders
L/D21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Obate Cylinders
D/L21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4
Bac
ksca
tter
Col
or In
dex
α
Prolate Cylinders
L/D21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4Oblate Cylinders
D/L21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8
Dep
olar
izat
ion
Col
or In
dex
β
Prolate Cylinders
-2
0
2
4
6
8
L/D21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8Oblate Cylinders
-2
0
2
4
6
8
D/L21.81.61.41.21
Figure 6.2: As in Fig. 6.1, but for polydisperse, randomly oriented cylinders. The shapes ofprolate and oblate cylinders are specified by length-to-diameter and diameter-to-length ratios,respectively.
113
different particle shapes and refractive indices. To represent a plausible PSC particle, a
combination of model size, shape, and refractive index must simultaneously reproduce
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Li
near
Dep
olar
izat
ion
Rat
io δ
(%)
Prolate Spheroids
0
10
20
30
40
50
60
70
ε21.81.61.41.21.11.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Obate Spheroids
ε21.81.61.41.21.11.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4
Bac
ksca
tter
Col
or In
dex
α
Prolate Spheroids
ε21.81.61.41.21.11.051
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4Oblate Spheroids
ε21.81.61.41.21.11.051
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8
Dep
olar
izat
ion
Col
or In
dex
β
Prolate Spheroids
-2
0
2
4
6
8
ε21.81.61.41.21.11.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8Oblate Spheroids
-2
0
2
4
6
8
ε21.81.61.41.21.11.05
Figure 6.3: As in Fig. 6.1, but for the refractive index m = 1.308 typical of water ice at visiblewavelengths. The shaded areas show the respective ranges of the backscattering characteristicsobserved for type II PSCs.
114
all observed lidar characteristics.
Figure 6.5 shows that cylinders with diameter-to-length and length-to-diameter
ratios larger than about 1.5 and effective equivalent-sphere radii larger than about 0.8
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Li
near
Dep
olar
izat
ion
Rat
io δ
(%)
Prolate Cylinders
L/D21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.50
10
20
30
40
50
60
70Obate Cylinders
D/L21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4
Bac
ksca
tter
Col
or In
dex
α
Prolate Cylinders
L/D21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-2
-1
0
1
2
3
4Oblate Cylinders
D/L21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8
Dep
olar
izat
ion
Col
or In
dex
β
Prolate Cylinders
-2
0
2
4
6
8
L/D21.81.61.41.21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Effective Equal-Volume-Sphere Radius ( µm )
-2
0
2
4
6
8Oblate Cylinders
-2
0
2
4
6
8
D/L21.81.61.41.21
Figure 6.4: As in Fig. 6.3, but for polydisperse, randomly oriented circular cylinders.
115
0 1 2 3 4
Effective Equal-Volume-Sphere Radius (µm)
Prolate Spheroids
ε=1
ε=1.05
ε=1.1
ε=1.2
ε=1.4
ε=1.6
ε=1.8
ε=2
0 1 2 3 4
Oblate Spheroids
Type Ia PSCs
ε=1
ε=1.05
ε=1.1
ε=1.2
ε=1.4
ε=1.6
ε=1.8
ε=2
0 1 2 3 4
Effective Equal-Volume-Sphere Radius (µm)
Prolate Cylinders
L/D=1
L/D=1.2
L/D=1.4
L/D=1.6
L/D=1.8
L/D=2
δ
β
0 1 2 3 4
Oblate Cylinders
D/L=1
D/L=1.2
D/L=1.4
D/L=1.6
D/L=1.8
D/L=2
α
Figure 6.5: The bars depict the respective ranges of the effective radius that reproduce thevalues of δ , α , and β observed for type Ia PSCs, as shown in Figs. 6.1 and 6.2.
116
Figure 6.6: The bars depict the respective ranges of the effective radius that reproduce thevalues of δ and α observed for type Ib PSCs, as shown in Figs. 6.1 and 6.2.
0 1 2 3 4
Effective Equal-Volume-Sphere Radius (µm)
Prolate Spheroids
ε=1
ε=1.05
ε=1.1
ε=1.2
ε=1.4
ε=1.6
ε=1.8
ε=2
0 1 2 3 4
Oblate Spheroids
Type Ib PSCs
ε=1
ε=1.05
ε=1.1
ε=1.2
ε=1.4
ε=1.6
ε=1.8
ε=2
0 1 2 3 4
Effective Equal-Volume-Sphere Radius (µm)
Prolate Cylinders
L/D=1
L/D=1.2
L/D=1.4
L/D=1.6
L/D=1.8
L/D=2
δ
0 1 2 3 4
Oblate Cylinders
D/L=1
D/L=1.2
D/L=1.4
D/L=1.6
D/L=1.8
D/L=2
α
117
Figure 6.7: The bars depict the respective ranges of the effective radius that reproduce thevalues of δ , α , and β observed for type II PSCs, as shown in Figs. 6.3 and 6.4.
0 1 2 3 4
Effective Equal-Volume-Sphere Radius (µm)
Prolate Spheroids
ε=1
ε=1.05
ε=1.1
ε=1.2
ε=1.4
ε=1.6
ε=1.8
ε=2
0 1 2 3 4
Oblate Spheroids
Type II PSCs
ε=1
ε=1.05
ε=1.1
ε=1.2
ε=1.4
ε=1.6
ε=1.8
ε=2
0 1 2 3 4
Effective Equal-Volume-Sphere Radius (µm)
Prolate Cylinders
L/D=1
L/D=1.2
L/D=1.4
L/D=1.6
L/D=1.8
L/D=2
δ
β
0 1.0 2.0 3.0 4.0
Oblate Cylinders
D/L=1
D/L=1.2
D/L=1.4
D/L=1.6
D/L=1.8
D/L=2
α
118
µm are the likely model representatives of type Ia PSC particles. Prolate and oblate
spheroids with aspect ratios 1.2 and larger are also acceptable solutions, although the
ranges of the effective radius that reproduce all three lidar observables may be narrower
than those for the cylinders and may be too shape-dependent to be realistic.
Figure 6.6 suggests that type Ib PSC particles are likely to have effective radii less
that 0.8 µm and are best represented by spheres or spheroids with very small aspect
ratios. Spheroids with aspect ratios larger than 1.1 and cylinders may also qualify, but
must have radii smaller than a few tenths of a micrometer. Note, however, that
choosing the narrower range [2, 3] of α values observed by Browell et al. [3] would
constrain type Ib PSC particles to spheres or nearly spherical spheroids with effective
radii close to 0.5 mµ (cf. [2]).
According to Fig. 6.7, type II PSC particles are well represented by ice cylinders
and prolate spheroids ( 2.1≥ε ) with effective equivalent-sphere radii exceeding 1 µm.
Although oblate ice spheroids with 1.1=ε , 1.2, and 1.6 are also potential candidates,
the respective effective radius ranges appear to be too narrow to be realistic.
6.5 Concluding remarks
We have used the current advanced version of the T-matrix method to perform
massive computations of backscattering lidar characteristics for polydisperse, randomly
oriented spheroids and circular cylinders and analyzed the existing lidar measurements
of three predominant types of PSCs. Our analysis is an extension of that previously
published by Toon et al. [2] and determines the likely ranges of particle physical
parameters that reproduce the existing lidar data. We have not discussed the plausibility
of our results from the standpoint of the physics and chemistry of PSC formation and
evolution and hope that such a discussion will be the subject of further research.
An obvious limitation of our analysis is the use of simple, rotationally symmetric
119
shapes and a restricted range of particle sizes. Although our model shapes and the
range of effective radii from 0 to 3 µm may be relevant to many real PSCs, further
effort is obviously warranted in order to include particles with larger sizes (e.g., [38])
and more irregular shapes (e.g., [14]). Another desirable extension would be an
analysis of less frequently encountered types of PSCs potentially composed of a mixture
of different particle species [2].
Acknowledgement
This research was funded by the NASA Global Aerosol Climatology Project
managed by Donald Anderson. We thank K.H. Fricke for valuable comments on a
preliminary version of this paper.
120
References
[1] Carslaw KS, Wirth M, Tsias A, Luo BP, Dörnbrack A, Leutbecher M, Volkert H,
Renger W, Bacmeister JT, Reimer E, Peter Th. Increased stratospheric ozone
depletion due to mountain-induced atmospheric waves. Nature 1998; 391: 675–
678.
[2] Toon OB, Tabazadeh A, Browell EV, Jordan J. Analysis of lidar observations of
Arctic polar stratospheric clouds during January 1989. J Geophys Res 2000; 105:
20589–20615.
[3] Browell EV, Butler CF, Ismail S, Robinette PA, Carter AF, Higdon NS, Toon OB,
Schoeberl MR, Tuck AF. Airborne lidar observations in the wintertime Arctic
stratosphere: polar stratospheric clouds. Geophys Res Lett 1990; 17: 385–388.
[4] Carslaw KS, Luo BP, Clegg SL, Peter Th, Brimblecombe P, Crutzen PJ.
Stratospheric aerosol growth and HNO3 gas phase depletion from coupled HNO3
and water uptake by liquid particles. Geophys Res Lett 1994; 21: 2479–2482.
[5] Stefanutti L, Morandi M, Del Guasta M, Godin S, David C. Unusual PSCs
observed by LIDAR in Antarctica. Geophys Res Lett 1995; 22: 2377–2380.
[6] Shibata T, Iwasaka Y, Fujiwara M, Hayashi M, Nagatani M, Shiraishi K, Adachi
H, Sakai T, Susumu K, Nakura Y. Polar stratospheric clouds observed by lidar
over Spitsbergen in the winter of 1994/1995: liquid particles and vertical
“sandwich” structure. J Geophys Res 1997; 102: 10829–10840.
[7] Rosen JM, Kjome NT, Larsen N, Knudsen BM, Kyrö E, Kivi R, Karhu J, Neuber
R, Beninga I. Polar stratospheric cloud threshold temperatures in the 1995-1996
arctic vortex. J Geophysics Res 1997; 102: 28195–28202.
[8] Tsias A, Wirth M, Carslaw KS, Biele J, Mehrtens H, Reichardt J, Wedekind C,
121
Weiß V, Renger W, Neuber R, von Zahn U, Stein B, Santacesaria V, Stefanutti L,
Fierli F, Bacmeister J, Peter T. Aircraft lidar observations of an enhanced type Ia
polar stratospheric cloud during APE-POLECAT. J Geophys Res 1999; 104:
23961–23969.
[9] Reichardt J, Tsias A, Behrendt A. Optical properties of PSC Ia-enhanced at UV
and visible wavelengths: model and observations. Geophys Res Lett 2000; 27:
201–204.
[10] David C, Bekki S, Godin S, Mégie G, Chipperfield MP. Polar stratospheric clouds
climatology over Dumont d’Urville between 1989 and 1993 and the influence of
volcanic aerosols on their formation. J Geophys Res 1998; 103: 22163–22180.
[11] Macke A, Mishchenko MI, Muinonen K, Carlson BE. Scattering of light by large
nonspherical particles: Ray tracing approximation versus T-matrix method. Opt Lett
1995; 20: 1934–1936.
[12] Liou KN, Takano Y, Yang P. Light scattering and radiative transfer in ice crystal
clouds: applications to climate research. In Mishchenko MI, Hovenier JW, Travis
LD, editors. Light scattering by nonspherical particles: theory, measurements, and
applications. San Diego: Academic Press, San Diego, 2000, p. 417–449.
[13] Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by nonspherical
particles: theory, measurements, and applications. San Diego: Academic Press,
2000.
[14] Yang P, Liou KN. Finite difference time domain method for light scattering by
nonspherical and inhomogeneous particles. In Mishchenko MI, Hovenier JW, Travis
LD, editors. Light scattering by nonspherical particles: theory, measurements, and
applications. San Diego: Academic Press, San Diego, 2000, p. 173–221.
[15] Sun W, Fu Q, Chen Z. Finite-difference time-domain solution of light scattering by
122
dielectric particles with a perfectly matched layer absorbing boundary condition.
Appl Opt 1999; 38: 3141–3151.
[16] Draine BT. The discrete dipole approximation for light scattering by irregular targets.
In Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by
nonspherical particles: theory, measurements, and applications. San Diego:
Academic Press, San Diego, 2000, p. 131–145.
[17] Mishchenko MI, Travis LD. Capabilities and limitations of a current Fortran
implementation of the T-matrix method for randomly oriented, rotationally
symmetric scatterers. J Quant Spectrosc Radiat Transfer, 1998; 60: 309–324.
[18] Mishchenko MI. Light scattering by randomly oriented axially symmetric particles. J
Opt Soc Am A 1991; 8: 871–882. [Errata: 9: 497 (1992).]
[19] Mishchenko MI. Light scattering by size-shape distributions of randomly oriented
axially symmetric particles of a size comparable to a wavelength. Appl Opt 1993; 32:
4652–4666.
[20] Wielaard DJ, Mishchenko MI, Macke A, Carlson BE. Improved T-matrix
computations for large, nonabsorbing and weakly absorbing nonspherical particles
and comparison with geometrical-optics approximation. Appl Opt 1997; 36: 4305–
4313.
[21] Mishchenko MI, Travis LD. T-matrix computations of light scattering by large
spheroidal particles. Opt Commun 1994; 109: 16–21.
[22] Mishchenko MI, Hovenier JW. Depolarization of light backscattered by randomly
oriented nonspherical particles. Opt Lett 1995; 20: 1356–1358.
[23] Gobbi GP. Polarization lidar returns from aerosols and thin clouds: a framework for
the analysis. Appl Opt 1998; 37: 5505–5508.
123
[24] Sassen K. Lidar backscatter depolarization technique for cloud and aerosol research.
In Mishchenko MI, Hovenier JW, Travis LD, editors. Light scattering by
nonspherical particles: theory, measurements, and applications. San Diego:
Academic Press, San Diego, 2000, p. 393–416.
[25] Toon OB, Browell EV, Kinne S, Jordan J. An analysis of lidar observations of
polar stratospheric clouds. Geophys Res Lett 1990; 17 393–396.
[26] Sassen K. The polarization lidar technique for cloud research: a review and current
assessment. Bull Am Meteorol Soc 1991; 72: 1848–1866.
[27] Zakharova NT, Mishchenko MI. Scattering properties of needlelike and platelike ice
spheroids with moderate size parameters. Appl Opt 2000; 39: 5052–5057.
[28] Hansen JE, Travis LD. Light scattering in planetary atmosphere. Space Sci Rev
1974; 16: 527–610.
[29] Mishchenko MI and Travis LD. Light scattering by polydispersions of randomly
oriented spheroids with sizes comparable to wavelengths of observation. Appl Opt
1994; 33: 7206–7225.
[30] Deshler T, Nardi B, Adriani A, Cairo F, Hansen G, Fierli F, Hauchecorne A,
Pulvirenti L. Determining the index of refraction of polar stratospheric clouds
above Andoya (69ºN) by combining size-resolved concentration and optical
scattering measurements. J Geophys Res 2000; 105: 3943–3953.
[31] Middlebrook AM, Berland BS, George SM, Tolbert MA, Toon OB. Real refractive
indices of infrared-characterized nitric-acid/ice films: implications for optical
measurements of polar stratospheric clouds. J Geophys Res 1994; 99: 25655–25666.
[32] Berland BS, Haynes DR, Foster KL, Tolbert MA, George SM, Toon OB.
Refractive indices of amorphous and crystalline NHO3/H2O films representative of
124
polar stratospheric clouds. J Phys Chem 1994; 98: 4358–4364.
[33] Warren SG. Optical constants of ice from the ultraviolet to the microwave. Appl
Opt 1984; 23: 1206–1225.
[34] Larsen N, Knudsen BM, Rosen JM, Kjome NT, Neuber R, Kyrö E. Temperature
histories in liquid and solid polar stratospheric cloud formation. J Geophys Res
1997; 102; 23505–23517.
[35] Vömel H, Rummukainen M, Kivi R, Karhu J, Turunen T, Kyrö E, Rosen J, Kjome
N, Oltmans S. Dehydration and sedimentation of ice particles in the Arctic
stratospheric vortex. Geophys Res Lett 1997; 24: 795–798.
[36] Mishchenko MI, Sassen K. Depolarization of lidar returns by small ice crystals: an
application to contrails. Geophys Res Lett 1998; 25: 309–312.
[37] Liou KN, Lahore H. Laser sensing of cloud composition: a backscatter
depolarization technique. J Appl Meteorol 1974; 13: 257–263.
[38] Goodman J, Verma S, Pueschel RF, Hamill P, Ferry GV, Webster D. New
evidence of size and composition of polar stratospheric cloud particles. Geophys
Res Lett 1997; 24: 615–618.