AUGUST 1998 NOTES AND CORRESPONDENCE 1043

The Extinction-to-Backscatter Ratio of Tropospheric Aerosol: A Numerical Study

JÖRG ACKERMANN
Meteorologisches Institut, Universität München, Munich, Germany
12 May 1997 and 7 October 1997

ABSTRACT
An adequate estimation of the aerosol extinction-to-backscatter ratio S is important for solving the under-
determined single scattering lidar equation and for investigating the climate impact of aerosols. In this study,
the extinction-to-backscatter ratios for the Nd:YAG wavelengths are calculated for continental, maritime, and
desert aerosols; the corresponding aerosol components are varied within the expected natural variabilities of the
particle number mixing ratios.
For continental aerosol, S increases with the relative humidity f from 40 to 80 sr. For maritime aerosol, the
extinction-to-backscatter ratios lie between 15 and 30 sr for 355 and 532 nm and between 25 and 50 sr for 1064
nm. The desert aerosol exhibits a weak dependence of S on f and ranges between 42 and 48 sr for 355 nm and
between 17 and 25 sr for 532 and 1064 nm. For practical applications, the calculated values of S are fitted by
a power series expansion with respect to their dependence on f.

1. Introduction particles are height dependent. Thus, for the measure-

ments with an upward-looking monostatic lidar system,
Backscatter lidar measurements are considered useful
due to the change of the relative humidity along the
for the derivation of atmospheric extinction profiles.
measurement path, the lidar ratio may differ consider-
However, the backscatter lidar equation is underdeter-
ably from an assumed constant range-independent val-
mined due to its dependence on the two unknowns,
ue. It is one of the objectives of this study to quantify
backscatter coefficient and extinction coefficient (e.g.,
this variability for different aerosol types by numerical
Klett 1981, 1985; Hughes et al. 1985; Sasano et al. 1985;
Bissonnette 1986; Kovalev 1993). This dependence ne- simulations.
cessitates the prescription of the ratio between the aero- Since the extinction-to-backscatter ratio is also im-
sol extinction coefficient and the aerosol backscatter co- portant in estimating the climate impact of aerosols,
efficient, that is, the lidar ratio. there are several studies of an experimental determi-
Since the lidar ratio connects two optical quantities nation of the lidar ratio. The results are manifold: using
that both depend on the wavelength of the incident light, the slope method, Salemink et al. (1984) found a linear
the refractive index, and the aerosol size distribution, increase of the lidar ratio from about 25 to 70 sr between
its behavior is not obvious: in real atmospheres, aerosol 40% and 80% relative humidity near the surface. They
is a mixture of different components and hence different reported no significant difference between 532 and 1064
refractive indices. Each of these components has a dis- nm. Combining lidar, sun photometer, and optical par-
tinct size distribution, which, assuming it is lognormally ticle counter measurements, Takamura et al. (1994) de-
distributed, is defined by its median radius and the width rived values between 20 and 70 sr for 532 nm. Waggoner
of the distribution. Therefore, the combination of dif- et al. (1972) retrieved an aerosol scatter-to-backscatter
ferent aerosol components with different chemical prop- ratio of about 84 sr for relative humidities less than 75%
erties and size distributions is manifold. using nephelometer data at 680 nm and ruby lidar data
In addition, some aerosol components change their at 694 nm. From ruby lidar slant path measurements,
size distributions and refractive indices due to variations Spinhirne et al. (1980) deduced an average extinction-
of the relative humidity. Since the relative humidity in- to-backscatter ratio of 19.5 sr for the PBL near Tucson,
creases often with altitude within the planetary bound- Arizona. The most actual data are published by Rosen
ary layer (PBL), the physical–chemical properties of the et al. (1997): from a combination of nephelometer and
backscatter sonde measurements a mean extinction-to-
backscatter ratio of 41.6 sr at 490 nm and 32.2 sr at
700 nm is calculated for near-surface aerosols over the
Corresponding author address: Dr. Jörg Ackermann, MIM - Me-
teorologisches Inst., Universität München, Theresienstr. 37, 80333
southwestern United States.
München, Germany. In case of elastic backscattering, range-dependent val-
E-mail: [email protected] ues of the lidar ratio are known only theoretically, when

q 1998 American Meteorological Society

1044 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 15

the physical–chemical properties and the size distribu- bining Eqs. (1) and (2), the lognormal distribution can
tion of the aerosol particles along the lidar line are be expressed by
known. However, the fulfillment of these requirements
would make the lidar measurements unnecessary. Thus,
the lidar ratio must be estimated with respect to the
measurement site and to any available a priori infor-
n i(r) 5
Ntot m i
(2p)1/2 lns i r
exp
2 ln 2s i[
2ln 2 (r/r m,i )

The following considerations refer to Ntot 5 1 cm23 .

.
] (3)

mation about the actual state of the atmosphere.

The a priori information about the actual tempera- For a given wavelength l the optical quantities of
ture–pressure relationship, and hence the Rayleigh back- each component depend on the radius r and the refrac-
scattering and extinction properties of the atmosphere, tive index m i . Due to the influence of relative humidity,
is often determined from nearby routine measurements the hygroscopic aerosol components change their re-
(e.g., radiosonde ascents). In these cases one also has fractive index (Hänel 1976):
information about the vertical humidity profile, which
1 2
3
for the majority of atmospheric conditions is one of the r0, i
m i 5 m w 1 (m 0, i 2 m w ) . (4)
most important prerequisites for the lidar ratio estima- r m, i
tion. In Eq. (4), m w denotes the refractive index of water, and
From the geographical location of the measurement m 0, i is the refractive index of the dry particle of com-
site, it is possible to perform a crude preselection of the ponent i with the median radius r 0,i . The exponent in-
aerosol type: over continents, mainly three components dicates that Eq. (4) describes a volume-weighted av-
occur (water soluble, insoluble, and soot particles); over erage of refractive indices of water and the considered
oceans, the aerosol consists of sea salt, water soluble; component.
in coastal regions there are additional soot particles; and By means of Mie theory, the extinction efficiency
over deserts, minerals in combination with water soluble QExt,i (r, m i , l ) and the backscatter efficiency QBack,i (r,
particles are present. Hence, the variability of aerosols m i , l ) can be calculated. This procedure is repeated for
results mainly from the fact that for one aerosol type, each aerosol component. Accordingly, the resulting lidar
the mixing ratios of the distinct components are un- ratio S is given by
known. On the basis of this assumption, this study is a

OE Q
`
systematic and extensive numerical investigation about M

the lidar ratio of tropospheric aerosols. The physical Ext, i (r, m i , l)pr 2 n i (r) dr
i51

OE Q
0
basis and the input data are presented in sections 2 and S5 , (5)
`
3, respectively. The results for the different aerosol types M

are discussed separately in sections 4 to 6, whereby the Back,i (r, m i , l)pr n i (r) dr
2

i51 0
main emphasis is put on the calculation of the lidar ratio
for varying aerosol components and different relative where the numerator expresses the extinction contri-
humidities. bution a i and the denominator the backscatter contri-
bution b i , that is,
2. Physical basis
In the following considerations, it is assumed that the
EO ` M

i51
a i dr

EO
0
aerosol is an external mixture of internally mixed com- S5 ` M
. (6)
ponents. Each aerosol component (index i) is lognor- b i dr
mally distributed with respect to the particle radius r: 0 i51

n i (r) 5
(2p)
1/2
Ni
lns i r [
exp 2
2 ln 2s i ]
ln 2 (r/r m,i )
(1)
Figure 1 illustrates an example for the calculation of
the lidar ratio. A continental aerosol, which consists of
the three components water soluble, soot, and insoluble,
is chosen (Fig. 1a). The relative humidity, which influ-
where N i is the number density of the component, and
r m,i and s i denote the median radius and the standard ences the microphysical properties of the water soluble
deviation of n i (r), respectively. component, is assumed to be 90%, and the wavelength
The number mixing ratio m i (i.e., the normalized num- of the incident light is 532 nm.
ber particle concentration) is defined as With knowledge of the refractive indices, the extinc-
tion efficiencies and backscatter efficiencies are calcu-
Ni Ni lated. These quantities are weighted with the size dis-
mi 5 5
ON
M . (2)
Ntot tribution of the corresponding aerosol component. To
i51
i obtain the extinction and backscattering for a distinct
particle radius, the contributions of the three aerosol
Here, M is the number of aerosol components and Ntot components are summed. The resulting curves are plot-
is the total number of particles per unit volume. Com- ted in Fig. 1b. From the maxima of the curves note that,
AUGUST 1998 NOTES AND CORRESPONDENCE 1045

input parameters are summarized in Tables 1 and 2.

Briefly, the chemical compositions of the components
are as follows: the water soluble particles consist mainly
of nitrate and sulfate and have no distinct sources (Hess
et al. 1998). The chemical composition of insoluble par-
ticles is manifold since they originate from different
soils or plants or man-made activities. Soot is generated
in all kinds of combustion and is supposed to be gra-
phitic carbon. Sea salt consists mainly of sodium chlo-
ride, but also of other kinds of salt, that occur in sea-
water. The mineral component is a mixture of quartz
and clay. The chainlike character of soot particles is not
accounted for. Instead, one soot particle is considered
as one primary particle of such a chain. Compared to
the representation of a soot particle as a cluster, the
number concentration of the particles is enlarged, and
the mode radius of the corresponding size distribution
is very small.
The internal variabilities of the distinct components
make clear that the refractive indices given in Table 1
FIG. 1. Determination of the lidar ratio for a three-component aero-
sol. (a) The size distributions. (b) Backscatter and extinction are
do not refer to one distinct chemical composition but
summed for the distinct components. rather to a mixture of different substances. Consequent-
ly, external mixing of the components must be seen as
external mixing of internally mixed components.
the main contribution to both backscattering and ex- The Mie code of Bohren and Huffman (1983) is ap-
tinction at 532 nm comes from particles with radii be- plied to the aerosol data. The possible nonsphericity of
tween 0.1 and 0.6 mm. The resulting lidar ratio is the the particles is not respected. This simplification can
ratio of the integrals of the extinction curve and the cause serious errors for all nonhygroscopic components,
backscatter curve shown in Fig 1b. especially when the particles are large compared to the
The procedure described above is repeated for each wavelength of the incident light.
combination of aerosol components discussed in the fol- It is expected that the measurements of continental
lowing sections. aerosols are most common. Thus, the calculated lidar
ratios for the continental aerosol will be discussed in
more detail than the results for the maritime aerosol and
3. Input data the desert aerosol.
As input for the Mie calculations, aerosol climatology
data from Hess et al. (1998), based on prior climatol- 4. Continental aerosol
ogies of d’Almeida et al. (1991) and Shettle and Fenn
(1979), are used. These climatologies include for each The continental aerosol type consists of the three
aerosol component the refractive index m (m 5 h 2 components: insoluble, soot, and water soluble. The me-
ik), the median radii r m for different relative humidities dian radius of the latter component grows with increas-
f, and the standard deviation s of the corresponding ing relative humidity (see Table 2).
lognormal distribution. The wavelengths considered in For given optical properties and a distinct relative
this study are 1064 nm (near infrared), 532 nm (visible), humidity, the variability of the lidar ratio is caused by
and 355 nm (near UV), which are the fundamental, dou- different number mixing ratios. Consequently, it is nec-
bled, and tripled wavelengths of an Nd:YAG laser. essary to consider a wide range of possible combinations
For the aerosol components used in this study the of number mixing ratios with the boundary condition

TABLE 1. The real parts (h) and imaginary parts (k) of the refractive indices for the aerosol components used in this study. The
subscripts denote the three considered wavelengths in nanometers.

Component h355 k355 h532 k532 h1064 k1064

Water soluble 1.530 5.00 3 1023
1.530 5.64 3 1023
1.520 1.64 3 1022
Insoluble 1.530 8.00 3 1023 1.530 8.00 3 1023 1.510 8.00 3 1023
Soot 1.750 4.64 3 1021 1.750 4.46 3 1021 1.760 4.43 3 1021
Mineral 1.530 1.66 3 1022 1.530 6.33 3 1023 1.530 4.30 3 1023
Sea salt 1.510 3.22 3 1028 1.500 1.12 3 1028 1.470 1.95 3 1024
Water 1.343 6.00 3 1029 1.333 1.61 3 1029 1.326 1.39 3 1025
1046 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 15

TABLE 2. The parameters of the lognormal distributions for different relative humidities of the aerosol components used in this study. The
unity of rm is in micrometers (Hess et al. 1998).

Component rm (0%) rm (50%) rm (70%) rm (80%) rm (90%) rm (95%) rm (98%) rm (99%) s

Water soluble 0.0212 0.0262 0.0285 0.0306 0.0348 0.0399 0.0476 0.0534 2.239
Insoluble 0.4710 0.4710 0.4710 0.4710 0.4710 0.4710 0.4710 0.4710 2.512
Soot 0.0118 0.0118 0.0118 0.0118 0.0118 0.0118 0.0118 0.0118 2.000
Mineral (nuc.) 0.0700 0.0700 0.0700 0.0700 0.0700 0.0700 0.0700 0.0700 1.950
Mineral (acc.) 0.3900 0.3900 0.3900 0.3900 0.3900 0.3900 0.3900 0.3900 2.000
Mineral (coa.) 1.9000 1.9000 1.9000 1.9000 1.9000 1.9000 1.9000 1.9000 2.150
Sea salt (acc.) 0.2090 0.3360 0.3780 0.4160 0.4970 0.6050 0.8010 0.9950 2.030
Sea salt (coa.) 1.7500 2.8200 3.1700 3.4900 4.1800 5.1100 6.8400 8.5900 2.030

that the sum of these mixing ratios equals 1. In the continental areas. The high soot content of 0.9 is of the
following, the mixing ratios for water soluble, insoluble, same order of magnitude as the value of 0.823 proposed
and soot particles are denoted m1 , m 2 , and m 3 , respec- by Hess et al. (1998) for the aerosol type ‘‘urban.’’
tively. To respect the influence of the relative humidity f on
The mixing ratio m1 is varied between 0.1 and 1 in the water soluble component and hence on the lidar
steps of 0.1. Accordingly, since m 2 is about four orders ratio, the calculations are performed for relative hu-
less, m 3 has values between 0.9 and 0. To account for midities of 0%, 50%, 70%, 80%, 90%, 95%, 98%, and
the influence of the insoluble component, m 2 is varied 99%, respectively.
between 6 3 1026 and 6 3 1025 in steps of 6 3 1026 . Figure 2 shows the calculated mean lidar ratios Ŝ and
Compared to the composition of the standard aerosol the deviations from Ŝ for relative humidities of 0%,
types proposed by Hess et al. (1998), it is expected that 50%, 90%, and 99%, and for the wavelengths 355, 532,
the choice of these 100 different combinations covers and 1064 nm, respectively. These nine contour plots
the high variability of the aerosol that can occur over enable us to investigate in detail the influence of the
three aerosol components on the values of S.
For all three wavelengths, the variabilities of S have
their maxima at f 5 0%. In particular, a high soot con-
tent can cause large positive deviations of the lidar ratio
from its mean value. This result is due to the large
imaginary part of the refractive index and hence the
extinction of these particles. For 1064 nm, this effect
can be compensated or even exceeded by a high amount
of insoluble particles (see Fig. 2 at the lower left).
The contributions of the three different components
to the lidar ratio can be estimated from the slopes of
the isolines in Fig. 2. When the absolute of the slopes
is near zero, the influence of the mixing ratios of soot
and water soluble particles is small, whereas when the
absolute of the slopes grows to infinity, the insoluble
component has hardly any influence on S. Thus, the
mixing ratio of the insoluble particles has for all relative
humidities a small influence on the lidar ratio for 355
nm. In contrast, the change of the insoluble component
plays the dominant role for the variability of the lidar
ratio at 1064 nm.
The standard deviation s of the lidar ratio is calculated
for each wavelength and relative humidity. The result
is shown in Fig. 3. The error bars of length 2s at a
definite relative humidity must be regarded as the vari-
ability of the lidar ratio driven by the assumption made
on the variability of the distinct values of m1 , m 2 , and
FIG. 2. Deviations of the calculated lidar ratios from the mean m3.
values of Ŝ for 355, 532, and 1064 nm and relative humidities of 0%, Since the water soluble component is the only com-
50%, and 90%. The dotted, dashed, and full lines indicate negative,
zero, and positive deviations from Ŝ, respectively, and the contour
ponent whose properties are affected by the relative hu-
interval is 1 sr. The mixing ratios of the four standard aerosol types midity, all variations of Ŝ and s with respect to their
proposed by Hess et al. (1998) are shown by open symbols. values for f 5 0% are considered as variations of the
AUGUST 1998 NOTES AND CORRESPONDENCE 1047

FIG. 3. Dependence of the lidar ratios on the relative humidity for

FIG. 4. Same as Fig. 3 but for maritime aerosol.
different wavelengths for continental aerosol. The curves are the re-
sults of power series expansions, the symbols indicate the calculated
values of Ŝ, and the error bars are calculated on the assumptions made
about the variabilities of the distinct components.
water soluble and the sea salt components grow with
increasing relative humidity (see Table 2).
water soluble component. The change of Ŝ and s due
In this study, the water soluble component varies be-
to increasing f has two causes: first, a decrease of the
tween 0.3 and 1 in steps of 0.1, the sea salt accumulation
real and the imaginary part of the refractive index and
mode ranges between 2 3 1023 and 2 3 1022 (step
second, a shift of the lognormal distribution to larger
width: 1.8 3 1023 ), and the sea salt coarse mode varies
radii due to the growth of the particles (Fitzgerald 1984).
between 3 3 1027 and 3 3 1026 (step width: 2.7 3
Since in most cases the extinction increases, in general
1027 ). For every combination, the mixing ratio of the
the latter effect is dominant.
soot component is chosen to sum up all four mixing
For relative humidities up to 95%, there are only small
ratios to 1. This process yields maximum soot contents
differences between Ŝ for 355 nm and Ŝ for 532 nm.
of nearly 0.7, a value that is somewhat higher than that
For larger relative humidities, the lidar ratio increases
at 532 nm and decreases at 355 nm. This decrease is of 0.576, which was proposed by Hess et al. (1998) for
caused by the strong increase of the backscatter coef- the aerosol type ‘‘maritime polluted.’’ Since Hess et al.
ficient at 355 nm due to the growth of the water soluble (1998) note the strong variability of the soot content
particles. For f , 95%, the decrease of extinction with over coastal areas, it seems appropriate to respect such
increasing wavelength is the governing effect that caus- high soot contents in a systematic analysis as well.
es the lidar ratio for 1064 nm to be smaller than for 355 All calculations are repeated for eight different values
nm and for 532 nm. The high variability of Ŝ with f of the relative humidity (see section 4), and the results
suggests that the physical–chemical properties of the for the mean lidar ratios Ŝ and their standard deviations
water soluble particles dominate over the influence of s are plotted in Fig. 4. The largest values of Ŝ, which
soot and insoluble particles. have remarkable small standard deviations, are calcu-
The calculated uncertainties for the different wave- lated for 1064 nm. For the lidar ratio of maritime aero-
lengths decrease with increasing relative humidity: for sol, the accumulation mode of the sea salt particles is
355 nm, the uncertainties range from 61% to 614%, the dominant component. This is caused by the median
for 532 nm from 61% to 69%, and for 1064 nm from radii of the accumulation sea salt component, which,
65% to 611%. depending on the relative humidity, ranges between 209
and 995 nm, that is, the order of magnitude of the Nd:
YAG wavelengths.
5. Maritime aerosol Since air masses of the PBL over oceans usually have
The maritime aerosol occurs in the PBL over oceans relative humidities of more than 50%, it can be stated
and over nearby coastal regions. It is composed of four that the lidar ratio of maritime aerosol decreases with
components: water soluble, soot, and the accumulation relative humidity for most atmospheric conditions. This
and the coarse mode of sea salt. The particles of the decrease is a function of wavelength: for 355 nm, the
1048 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 15

cumulation mode (median radius: 390 nm). According-

ly, the influence of the water soluble component is rather
small: only for relative humidities of more than 90% is
the lidar ratio considerably changed. Such atmospheric
conditions are unlikely to occur over desert areas. Thus,
the lidar ratios can be assumed to be independent of f
for the majority of cases.
For all values of f, the sensitivity of Ŝ due to changes
of the desert aerosol composition is most pronounced
for 532 nm. This is evident from the uncertainties of Ŝ:
for 355 nm, the uncertainty ranges between 62% and
66%; for 532 nm, between 610% and 616%; and for
1064 nm, between 66% and 610%.

7. Empirical formula for Ŝ( f )

The mean lidar ratios Ŝ were discussed above for
continental, maritime, and desert aerosol for eight dif-
ferent values of the relative humidity f. As mentioned
in section 1, the information about the vertical profile
of f can be deduced from radiosonde data. For practical
FIG. 5. Same as Fig. 3 but for desert aerosol.
applications, it is required that the corresponding mean
values of the lidar ratio can be derived immediately from
the vertical profile of f. On the basis of the results of
variations of Ŝ with f are nearly negligible, and for 1064 sections 4 through 6, for each wavelength and each
nm, the variability can be as high as 25 sr. aerosol type, a power series expansion is performed for
The calculated uncertainties of Ŝ for maritime aerosol Ŝ( f ):
have less dependence on f than for the continental aero-
sol type: for 355 nm, the uncertainties range between
Oaf
J

614% and 618%; for 532 nm, they lie between 68% Ŝ( f ) ø j
j21
, (7)
j51
and 613%; and for 1064 nm, the corresponding values
are 61% and 65%. with a j the polynomial coefficients and J the total num-
ber of the coefficients. The value of J is found by min-
imizing a least squares merit function (Press et al. 1992)
6. Desert aerosol
and varies between 5 and 10 depending on the shape
The desert aerosol originates from desert areas and of the original function. The resulting coefficients are
mainly consists of the water soluble component and the listed in Table 3.
nucleation, accumulation, and coarse particle mode of To compare the polynomial fit with the input data,
minerals. Since the mineral components do not change the fitted curves and the computed values of Ŝ( f ) are
their physical–chemical properties with relative humid- plotted in Figs. 3–5 for continental, maritime, and desert
ity, all changes of the lidar ratio of desert aerosol with aerosol, respectively. In all cases, the curves lie within
the relative humidity result from the water soluble com- the error bars of the computed lidar ratios, indicating
ponent. that the power series expansions are mathematically suf-
The calculation of the lidar ratios is performed in the ficiently accurate.
same manner as described in section 4. For the com- A comparison between these results and the outcomes
putations, the water soluble component varies between of the experimental studies mentioned in section 1 is
0.6 and 0.9 (step width: 0.1), the mineral accumulation rather difficult because of the different wavelengths used
mode ranges from 1 3 1022 to 4 3 1022 (step width: and in most cases the lack of information about the
1 3 1022 ), and the mineral coarse mode lies between variability of the considered aerosols with relative hu-
5 3 1025 and 8 3 1025 (step width: 1 3 1025 ). The midity.
mixing ratio of the mineral nucleation mode is chosen For continental aerosols, Salemink et al. (1984) report
to sum up the total mixing ratio to 1. a stronger dependence of S on f, but their absolute val-
The calculated lidar ratios and their standard devia- ues (25–70 sr) do not disagree with the values deter-
tions are summarized in Fig. 5. The values of Ŝ for 355 mined in this study. The data of Takamura et al. (1994)
nm are about twice the values for 532 and 1064 nm. do not conflict with the results of this study, even if
Similar to the dominance of the accumulation mode of they determined low values of about 20 sr, too. The
sea salt for the lidar ratio of maritime aerosol, the lidar values of Waggoner et al. (1972) seem to be very high,
ratio of desert aerosol is governed by the mineral ac- even when a large soot content is assumed. Low values
AUGUST 1998 NOTES AND CORRESPONDENCE 1049

ranging from about 21 to 32 sr are found by an earlier

1022
1023
1025
1027
101
numerical study of McCormick et al. (1968), which ne-

1064 nm
1.706 3
22.101 3
1.263 3
22.411 3
1.480 3
glected the wavelength dependence and the imaginary

0.
0.
0.
0.
0.
part of the refractive index (i.e., absorption) and as-
sumed a Junge size distribution.
By a minimum search technique, Sasano and Browell
(1989) determined the lidar ratio of maritime aerosol
1022
1023
1025
1027
101 from multiple wavelength lidar observations. For 300
532 nm
Desert

1.993 3
24.398 3
2.865 3
25.518 3
3.430 3
nm, they reported values ranging from 5.1 to 19.4 sr,

0.
0.
0.
0.
0.
which are somewhat lower than the values given in this
study for 355 nm. However, from the wavelength de-
pendence of S shown in Fig. 4, it seems that calculated
values for 300 nm would lie close to the values reported
1023
1023
1025
1027
101

by Sasano and Browell (1989). Their maximum value

TABLE 3. Coefficients for the power series expansions of the lidar ratio for different aerosol types.

355 nm

of 30 sr for 600 nm also agrees with the results of this

4.191 3
29.523 3
2.131 3
24.507 3
3.014 3
0.
0.
0.
0.
0.
study.
A comparison of the results for the desert aerosol with
the outcomes of Spinhirne et al. (1980) provides a good
agreement. Their retrieved value of 19.5 sr at 694 nm
10212
10214
10216

fits well with the values given in Fig. 5 for dry particles.
1021
1022
1024
1026
1027
1029
101

The measurements of Rosen et al. (1997) were taken

1064 nm
3
3
3
3
3
3
3
3
3
3

at rural deserts, nearby mountains, and polluted areas

3.387
5.038
21.675
5.497
23.393
21.035
1.086
2.185
21.397
22.345

under fair weather conditions after sunset. Thus, as a

first-order approximation, their average values of 41.6
sr at 490 nm and of 32.2 sr at 700 nm can be regarded
as a mean value of the lidar ratio for the continental
1.277 3 1021
5.191 3 1023
21.037 3 1024
4.364 3 1027

and the desert aerosol. On the assumption of f 5 50%,

1.884 3 101
Maritime
532 nm

Eq. (7) provides 51.9, 38.6, and 31.0 sr for 355, 532,
0.
0.
0.
0.
0.

and 1064 nm, respectively. Evidently, both the wave-

length dependence and the absolute values are in close
agreement with the data of Rosen et al. (1997). The
same is true on the assumption of slightly higher or
1022
1023
1025
1027
101

lower relative humidities than the arbitrary chosen value

355 nm
1.646 3
8.312 3
3.890 3
28.203 3
4.225 3

of 50%.
0.
0.
0.
0.
0.

Thus, it is expected that the empirical formulation of

Ŝ( f ) via Eq. (7) with the coefficients listed in Table 3
will be a useful tool for an estimation of the lidar ratio
for the Nd:YAG wavelengths. Nevertheless, it must be
10211

emphasized that other assumptions on the soot prop-

1022
1023
1025
1026
1027
1029
101
1064 nm

erties, on the shape of the mineral particles, on external

4.093 3
7.091 3
2.006 3
1.767 3
26.958 3
2.242 3
22.708 3
1.149 3

versus internal mixtures of the aerosol components, and

0.
0.

on the growth relations of the size distributions may

affect the outcome of the lidar ratio calculations.
1021
1023
1024
1026
1028

8. Concluding remarks
101
Continental
532 nm
4.531 3
2.628 3
23.085 3
1.334 3
22.356 3
1.412 3

In this study, lidar ratios and their standard deviations

0.
0.
0.
0.

for the continental, maritime, and desert aerosol with

their dependence on relative humidity are presented.
The variability of a distinct aerosol type is accounted
for by variations of the mixing ratios of the correspond-
10212
10211
10214
10216
1021
1023
1025
1026
1027
101

ing aerosol components. All computations are per-

355 nm

formed for three Nd:YAG wavelengths: 355, 532, and

3
3
3
3
3
3
3
3
3
3

1064 nm. The model input is summarized in Tables 1

4.252
4.400
27.877
1.395
7.881
21.472
8.581
1.350
22.899
23.411

and 2, and the main results are plotted in Figs. 3–5.

Although a wide range of variability of tropospheric
aerosols was covered in this study, there are remaining
a10
a1
a2
a3
a4
a5
a6
a7
a8
a9

difficulties that can make the lidar ratio determination

1050 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 15

doubtful even if the relative humidity along the mea- as well as for their fruitful comments. The comments
surement path is known. of the two anonymous reviewers significantly improved
First, as is clear from Fig. 2, there exist atmospheric an earlier version of the paper.
conditions and hence mixing ratio combinations whose
resulting lidar ratios differ considerably from an as- REFERENCES
sumed mean value. This is especially true for a high
soot amount and low relative humidities, as in the case Bissonnette, L. R., 1986: Sensitivity analysis of lidar inversion al-
gorithms. Appl. Opt., 25, 2122–2125.
of continental aerosol. This problem can only be elim- Bohren, C. F., and D. R. Huffman, 1983: Absorption and Scattering
inated by using larger wavelengths, where the influence of Light by Small Particles. Wiley and Sons, 530 pp.
of soot is slightly diminished. d’Almeida, G., P. Koepke, and E. P. Shettle, 1991: Atmospheric Aer-
Second, the simultaneous existence of different aero- osols: Global Climatology and Radiative Characteristics. A.
Deepak, 561 pp.
sol types in different heights above the measurement Fitzgerald, J. W., 1984: Effect of relative humidity on the aerosol
site prevents the evaluation of the lidar signal over the backscattering coefficient at 0.694- and 10.6-mm wavelengths.
whole range domain. For example, over the oceans, the Appl. Opt., 23, 411–418.
occurrence of the sea salt components is mainly limited Hänel, G., 1976: The properties of atmospheric aerosol particles as
to the PBL. Above the PBL, the aerosol composition is functions of the relative humidity at thermodynamic equilibrium
with the surrounding moist air. Advances in Geophysics, Vol.
expected to be similar to that of continental aerosol but 19, Academic Press, 73–188.
with a small amount of insoluble particles. In this case, Hess, M., P. Koepke, and I. Schult, 1998: Optical Properties of Aer-
the height of the PBL must be estimated from the lidar osols and Clouds: The software package OPAC. Bull. Amer.
signal or the relative humidity profile. Accordingly, both Meteor. Soc., 79, 831–844.
Hughes, H. G., J. A. Ferguson, and D. H. Stephens, 1985: Sensitivity
parts of the measurement path, that is, the PBL and the of a lidar inversion algorithm to parameters relating atmospheric
free troposphere, are evaluated separately. Using the backscatter and extinction. Appl. Opt.,
data of Hess et al. (1998) for a free troposphere with Klett, J. D., 1981: Stable analytical inversion solution for processing
50% relative humidity, the computed lidar ratios are lidar returns. Appl. Opt., 20, 211–220.
56.9, 58.4, and 51.3 sr for the wavelengths of 355, 532, , 1985: Lidar inversion with variable backscatter/extinction ra-
tios. Appl. Opt., 24, 1638–1643.
and 1064 nm, respectively. Kovalev, V. A., 1993: Lidar measurement of the vertical aerosol
A similar problem can occur when mineral aerosol is extinction profiles with range-dependent backscatter-to-extinc-
transported from its source in the deserts to regions tion ratios. Appl. Opt., 32, 6053–6065.
where other aerosol types are present. In that case the McCormick, M. P., J. D. Lawrence, and F. R. Crownfield, 1968: Mie
total and differential backscattering cross sections at laser wave-
mineral particles form an aerosol layer above the aerosol lengths for Junge aerosol models. Appl. Opt., 7, 2424–2425.
layer of the PBL (Hess et al. 1998). By sedimentation, Pratt, R. W., 1985: Review of radiosonde humidity and temperature
the aerosol components of both layers are externally errors. J. Atmos. Oceanic Technol., 2, 404–407.
mixed and thus the microphysical properties of the orig- Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery,
inal aerosol layer are considerably changed. Under these 1992: Numerical Recipes in FORTRAN. Cambridge University
Press, 963 pp.
conditions, a lidar ratio determination is ambiguous and Rosen, J. M., R. G. Pinnick, and D. M. Garvey, 1997: Measurement
useful results cannot be expected. For the mineral-trans- of the extinction-to-backscatter for near-surface aerosols. J. Geo-
ported component alone, from the data of Hess et al. phys. Res., 102, 6017–6024.
(1998), the calculated lidar ratios are 22.0 (355 nm), Salemink, H., P. Schotanus, and J. B. Bergwerff, 1984: Quantitative
lidar at 532 nm for vertical extinction profiles in the lidar so-
12.5 (532 nm), and 26.6 sr (1064 nm). lution. Appl. Phys., 34B, 187–189.
Another source of error is the inaccuracy of the rel- Sasano, Y., and E. V. Browell, 1989: Light scattering characteristics
ative humidity determined with the hygrometer of the of various aerosol types derived from multiple wavelength lidar
radiosonde. These inaccuracies are mainly caused by observations. Appl. Opt., 28, 1670–1679.
, , and S. Ismail, 1985: Error caused by using a constant
the thermal lag and the solar heating of the instrument extinction/backscattering ratio in the lidar solution. Appl. Opt.,
during the ascent. Both effects provoke an overesti- 24, 3929–3932.
mation of f, and their strength depends on the actual Shettle, E. P., and R. W. Fenn, 1979: Models for the aerosols of the
state of the atmosphere. The mean relative humidity lower atmosphere and the effects of humidity variations on their
errors are 6% between 1013 and 700 hPa and 10% be- optical properties. Air Force Geophysics Laboratory Environ-
mental Research Papers 676, AFGL-TR79-TR79-0214, 94 pp.
tween 700 and 500 hPa (Pratt 1985). Thus, especially [Available from Air Force Geophysics Laboratory, Hanscom
for large relative humidities, the accuracy of the lidar AFB, Massachusetts 01731.]
ratio determination is not only limited by insufficient a Spinhirne, J. D., J. A. Reagan, and B. M. Herman, 1980: Vertical
priori knowledge about the composition of the aerosols distribution of aerosol extinction cross section and inference of
aerosol imaginary index in the troposphere by lidar technique.
but also by the uncertainty of the relative humidity de- J. Appl. Meteor., 19, 426–438.
termination. Takamura, T., Y. Sasano, and T. Hayasaka, 1994: Tropospheric aerosol
optical properties derived from lidar, sun photometer, and optical
particle counter measurements. Appl. Opt., 33, 7132–7140.
Acknowledgments. Peter Koepke and Peter Völger Waggoner, A. P., N. C. Ahlquist, and R. J. Charlson, 1972: Mea-
from the University of Munich are gratefully acknowl- surement of the aerosol total scatter-backscatter ratio. Appl. Opt.,
edged for the delivery of aerosol data and the Mie code, 11, 2886–2889.