You are on page 1of 4

GEOPHYSICAL RESEARCH LETTERS, VOL. 34, L15811, doi:10.

1029/2007GL029606, 2007
Click
Here
for
Full
Article

Significant differences in the cloud droplet effective radius between


nonprecipitating and precipitating clouds
Takahisa Kobayashi1
Received 7 February 2007; revised 12 April 2007; accepted 20 June 2007; published 15 August 2007.
[1] We examined the cloud-precipitation interaction by a the growth of cloud drops does not proceed uniformly but
combined use of radar and a solar/infrared radiometer advances in a stepwise manner at a specified size, i.e., is the
onboard the Tropical Rainfall Measuring Mission (TRMM). threshold size [Houze, 1993]. Cloud drops at the threshold
We found sudden changes in the cloud droplet effective size grow to raindrops rapidly. In other words, precipitation
radius in marine water clouds, and significant differences in is suppressed when cloud drops are reduced to below this
the effective radius between nonprecipitating clouds and size by increased concentration of aerosols. Models of this
precipitating clouds in which clouds and precipitation phenomenon have been developed in the last half a
coexist. The largest radius was limited to between 15 and century, although their results are inconsistent [Pruppacher
20 mm for nonprecipitating clouds, whereas it increased and Klett, 1997; Khain et al., 2000]. Measurements to
discontinuously to 30 mm for precipitating clouds. Detailed examine the threshold size in real rain are critical for
analysis suggests that the differences are likely due to the improving our understanding of the second indirect effect
rapid growth of cloud drops into raindrops. The radius in and developing an accurate parameterization of cloud-
nonprecipitating clouds indicates that threshold sizes exist precipitation processes for predicting global warming
below which precipitation hardly forms. This study revealed [Randall et al., 2003]. In particular, understanding of the
that the threshold sizes are clearly observed and are physical processes in low marine stratocumulus clouds is
consistent on a global scale, which improves our important because it has strong negative feedback and
understanding of the second indirect effects of aerosols. small changes in the amount could offset the warming
Citation: Kobayashi, T. (2007), Significant differences in the effects resulting from CO2 doubling [Albrecht, 1989; Randall
cloud droplet effective radius between nonprecipitating and et al., 1984]. However, few measurements of the cloud-
precipitating clouds, Geophys. Res. Lett., 34, L15811, precipitation interaction have been made because of limi-
doi:10.1029/2007GL029606. tations in remote sensing techniques.

1. Introduction 2. Method
[2] The cloud feedback problem is a critical issue to be [4] The Tropical Rainfall Measuring Mission (TRMM)
solved in climate studies. However, the progress has been satellite primarily intends to measure temporal and spatial
slow [Stephens, 2005] because clouds are a ‘many-body’ variations of precipitation and clouds [Kummerow et al.,
process and have mutual relations to aerosols and precipi- 2000; Masunaga et al., 2002]. The TRMM has been
tation. Increases in aerosol concentrations result in a de- continuously monitoring precipitation since the launch on
creased drop size and modify the cloud radiative forcing, November, 1997 and has five sensors, including precipitat-
which is known as the ‘first indirect effect’ [Breon et al., ing Radar (PR). The PR is a unique spaceborne sensor used
2002]. Decreased drop size may also suppress precipitation, to measure the three-dimensional rain structure with a
which is known as the ‘second indirect effect’ [Albrecht, horizontal resolution of 5 km at nadir and a swath of
1989; Rosenfeld, 2000] Precipitation is generated from 245 km. The vertical resolution is 125 m. The TRMM is
cloud drops by condensation and coalescence, which in also equipped with a visible and infrared scanner (VIRS) for
turn generates cloud drops by breakup [Kobayashi and measuring clouds, as well as the sea surface temperature and
Adachi, 2001], removes cloud drops and controls cloud aerosols. The VIRS has five channels and measures visible
amount. The interactions between aerosols and clouds have and infrared radiance with a horizontal resolution of 2.4 km
been studied extensively using numerical models and at nadir and a swath of 830 km. Both sensors sample almost
observations with airborne sensors [Asano et al., 1995] simultaneously and with similar spatial resolutions. Conse-
and spaceborne solar and infrared radiometers [Coakley quently, combined use of the PR and VIRS is well suited for
and Bernstein, 1987]. Despite extensive studies on cloud- studies of cloud-precipitation interactions. The orbit has an
precipitation interactions, our understanding is very limited inclination of 35° and the observed area is from 35° to
because of their complex nature. +35° in latitude and 0° to 360° in longitude. We selected the
[3] One of the key factors for better understanding the VIRS data for which the center position of the PR is within
second indirect effect is the threshold size of cloud drops the footprint of the VIRS and made match-up datasets from
below which precipitation hardly forms. Rain formation data for 11 days in December, 2002 and 10 days in June
models have suggested that, during the coalescence stage, 2003. The swath of the VIRS is wider than that of the PR.
The VIRS data extending beyond the swath of the PR were
1
Meteorological Research Institute, Tsukuba, Japan. excluded. Ultimately, we generated four global-segmented
datasets, each with data for 5 or 6 days. One segment box
Copyright 2007 by the American Geophysical Union. has a 0.5°  0.5° latitude-longitude spatial resolution.
0094-8276/07/2007GL029606$05.00

L15811 1 of 4
L15811 KOBAYASHI: SIGNIFICANT DIFFERENCES IN DROP SIZE L15811

Nakajima, 1995]. The retrieval is based on the difference of


water absorbing characteristics between solar reflectance at
Ch1 and Ch3. Ch3 includes contributions of thermal emis-
sions from the atmosphere and ground surface. The unde-
sired thermal emissions from the cloud tops were removed
from the measured radiance at Ch3. The cloud top temper-
ature was determined from the Ch4 radiance using LOW-
TRAN-7 [Kneizys et al., 1988] assuming the tropical model
for the atmosphere above the cloud top. This assumption
leads to some error in the derived Re. However, a difference
of 4 K between the model and the actual temperature profile
causes an error of less than 0.7mm in the retrieved Re [Han
et al., 1994; Kawamoto et al., 2001]. Use of the midlatitude
summer model for the atmosphere above the cloud top
instead of the tropical model results in a difference of
0.8 mm in the mean Re. Since we applied the method to
limited regions of latitude, this error is not serious. We
ignored the thermal emission from the ground surface
because we examined optically thick clouds whose optical
thicknesses are mostly (7080%) larger than 10 [King,
1987].

3. Results
[6] Figure 1 shows sequential plots of Re for nonpreci-
pitating clouds (NPC: black dots) and precipitating clouds
(PRC: red: stratiform, blue: convective) derived from the
datasets for June (Figure 1a) and December (Figure 1b). The
near-surface rain rate is also plotted (green line) and was
estimated from the reflectivity factor at altitudes of around
1 to 2 km. A very interesting feature is the sudden change in
the Re of PRC: a discrete stepwise increase in the largest Re
for PRC occurs compared to that for NPC. For NPC, most
drops are less than 20 mm in Re, whereas the largest drop
size increases to 30 mm for PRC. Significant increases are
observed even for clouds with weak rain. This Re jump
Figure 1. Sequential plots of the effective radius of cloud
suggests that drops larger than a threshold size of 15 to 20 mm
drops derived from two datasets (a) for June and
are unstable and have a short life in NPC. These drops grow
(b) December. For NPC (black dots), Re is arranged in
into raindrops or dissipate very quickly. The effective radius
order of small optical thickness. The optical thickness is
of 20 mm corresponds to the mode radius of 14 mm for the
about 15 at center position of NPC (data no. = 6000) and is
log-normal drop size distribution with a variance of 1.42.
larger than 30 at the boundary of NPC and PRC. For PRC
Similar results have been obtained from measurements of
(red: stratiform, blue: convective), Re is plotted in order of
urban plumes [Rosenfeld, 2000]. Raindrop formation mod-
small near surface rain rate (green line). The near surface
els suggest that raindrops are formed very quickly from
rain rate is estimated from the PR measurements.
initial cloud drops with a mean size of 13 mm while they are
formed very slowly from drops of 9 mm [Frossmann et al.,
1985]. In addition, the gravity-induced collision rate of
Data exist for at least 100 pixels in each grid box. The pixel cloud drops increases drastically for a drop radius larger
for precipitating at the highest rain rate was selected from than 20 mm [Jonas, 1996]. These model results suggest that
100 data. When no precipitating pixels occur in a box, the drops larger than the threshold size are absent for NPC
pixel for the highest reflection at Ch1 (0.63 mm) was because these large drops grow into raindrops very quickly
selected. In the analysis, pixels for which the brightness and the most intense turbulence is observed near cloud tops.
temperature at Ch4 (10.8 mm) of VIRS ranged from 273 to Turbulent flow has a significant effect on drop growth due
290 K and the brightness temperature difference between to collisions between drops [Vaillancourt and Yau, 2000].
Ch4 and Ch5 (12 mm) was less than 1 K were further- Cloud drops at the threshold size are in the transition stage
more selected to remove erroneous data of optically thin, to drizzle and raindrops. Since Figure 1 contains various
cold clouds and subpixel clouds [Rosenfeld and Gutman, stages of precipitation, the Re jump also suggests that the
1994]. drops larger than the threshold disappear quickly in the rain
[5] The reflected radiances at Ch1 (0.63mm) and Ch3 dissipating process as well as in the growth stage. Unlike
(3.75mm) measured with the VIRS were used to derive the large drops, sudden decreases in the number of small drops
optical thicknesses and cloud droplet effective radius (Re) appear for PRC. Precipitation may remove cloud drops
of clouds near the cloud tops [Han et al., 1994; Nakajima and smaller than 10 mm.

2 of 4
L15811 KOBAYASHI: SIGNIFICANT DIFFERENCES IN DROP SIZE L15811

radius is crucial for improving our understanding of the


second indirect effect [Rotstayn, 1999]. However, variation
exists in the threshold effective radius as shown in Figure 1.
Theoretical studies suggest that factors other than effective
radius (e.g., droplet concentration, liquid water content, and
spectral shape of the cloud droplet size distribution) also
affect the transition from cloud droplets to raindrops [Liu
and Daum, 2004; Liu et al., 2005, 2006]. The TRMM data
here include various growing/dissipating stages of cloud,
which results in variation in these parameters and conse-
quently in the threshold size. The variation also results in an
overlap between the Re frequency distributions for NPC and
PRC as shown in Figures 1 and 2. Liu and Daum [2004]
derived the threshold radius as a function of LWC and N
(total number of cloud drops). The relation suggests the
threshold Re ranges from 15 to 20 mm depending on N,
LWC and the drop collection efficiency. Aircraft observa-
Figure 2. Sequential plots of the brightness temperature of tions also suggest that drizzle LWC increases rapidly for
clouds from two datasets for June. The near surface rain rate Re > 16mm [Gerber, 1996]. These results are similar to our
is plotted by solid line (right axis). results.

4. Conclusions
[7] We may have observed a very thin, high-altitude [11] We derived cloud drop effective size by a combined
cirrus clouds overlying lower water clouds. In retrieving use of the PR and VIRS onboard the TRMM satellite.
Re from space, the effects of cirrus contamination cannot be Significant differences in cloud drop effective sizes were
neglected. To avoid cirrus contamination, we selected found between precipitating and nonprecipitating clouds.
clouds with a brightness temperature Bt > 273K. This The maximum drop size was more than 30 mm for precip-
criterion may not enough to avoid cirrus contamination. itating clouds but 20 mm for nonprecipitating clouds. This
Therefore, we derived Re using Bt > 280 K and compared it Re jump indicates that an Re of 15 to 20 mm is the threshold
to the Re for Bt > 273 K [Han et al., 1994]. Essentially below which precipitation little forms. Precipitating clouds
similar results were obtained which suggests that cirrus with an Re larger than the threshold value are widely
contamination is likely removed using the first criterion. distributed throughout the observation area of the TRMM
The largest size for PRC may be too small even for the and are probably consistent with water clouds on a global
warm water clouds that we analyzed. This may have been scale. This fact is useful for examining the extent to which
due to the observational limit. the aerosol second indirect effect occurs.
[8] Figure 2 shows sequential plots of the brightness
temperature of clouds. The near surface rain rate is also
plotted (solid line: right axis). The brightness temperatures
for PRC are slightly lower than those for nonprecipitating
clouds. However, no significant differences are observed in
contrast to Figure 1. A larger Re is often observed for higher
clouds. Figure 2 suggests that the Re jump is not due to the
effects of cloud height.
[9] Figure 3 shows the frequency distributions of Re for
NPC (hatched) and PRC (black) estimated from the two
datasets for June. The distributions are normalized so that
the total amount of Re data is equal for both NPC and PRC.
As expected from Figure 1, significant differences occur
between NPC and PRC. For NPC, the distribution is
relatively narrow with a mode radius of 13 mm. Significant
decreases in the frequency appear at Re > 15 mm. Analysis
of the International Satellite Cloud Climatology Project
[Han et al., 1994] has also shown that cases with Re >
10– 20 mm decrease rapidly. For PRC, a wider distribution
is observed with the mode radius of 17 mm. Almost no cases
are found at Re < 10 mm. Similar results have been obtained
from model calculations [Lohmann et al., 2000].
[10] Significant differences in the Re between NPC and Figure 3. Normalized frequency distributions of Re for
PRC in Figures 1 and 3 suggest that the cloud drop radius NPC (hatched) and PRC (solid). Pixels of PRC, which is
mostly determines the transition from cloud droplets to about one third of those of NPC in number, are plotted by
raindrops. Therefore, an accurate estimation of the threshold multiplied a factor three so that number of pixels are equal
for NPC and PRC.

3 of 4
L15811 KOBAYASHI: SIGNIFICANT DIFFERENCES IN DROP SIZE L15811

[12] Acknowledgments. This study is partly supported by Grant-in Liu, Y., and P. H. Daum (2004), Parameterization of autoconversion pro-
Aids for Scientific Research 18540437, Japan Society for the Promotion of cess. Part I: Analysis formulation of the Kessler-type parameterization,
Science. The TRMM data were provided by the Japan Aerospace Explo- J. Atmos. Sci., 61, 1539 – 1548.
ration Agency. Liu, Y., P. H. Daum, and R. L. McGraw (2005), Size truncation effect,
threshold behavior, and a new type of parameterization, Geophys. Res.
References Lett., 32, L11811, doi:10.1029/2005GL022636.
Liu, Y., P. H. Daum, R. L. McGraw, and M. Miller (2006), Generalized
Albrecht, B. A. (1989), Aerosol, cloud microphysics, and fractional cloudi- threshold function accounting for effect of relative dispersion on thresh-
ness, Science, 245, 1227 – 1230. old behavior of autoconversion process, Geophys. Res. Lett., 33, L11804,
Asano, S., M. Shiobara, and A. Uchiyama (1995), Estimation of cloud doi:10.1029/2005GL025500.
physical parameters from airborne spectral reflectance measurements Lohmann, U., G. Tselioudis, and C. Tyler (2000), Why is the cloud albedo –
for stratocumulus clouds, J. Atmos. Sci., 52, 3556 – 3576. particle size relationship different in optically thick and optically thin
Breon, F.-M., D. Tanre, and S. Generoso (2002), Aerosol effect on cloud clouds?, Geophys. Res. Lett., 15, 1099 – 1102.
droplet size monitored from satellite, Science, 295, 834 – 838. Masunaga, H., T. Y. Nakajima, T. Nakajima, M. Kachi, and K. Suzuki
Coakley, J. A., Jr., and R. L. Bernstein (1987), Effect of ship-track effluents (2002), Physical properties of maritime low clouds as retrieved by com-
on cloud reflectivity, Science, 237, 1020 – 1022. bined use of Tropical Rainfall Measuring Mission (TRMM) Microwave
Frossmann, A. I., W. D. Hall, and H. R. Pruppacher (1985), A theoretical Imager and Visible/Infrared Scanner: 2. Climatology of warm clouds and
study of the wet removal of atmospheric pollutants. Part I: The redistri- rain, J. Geophys. Res., 107(D19), 4367, doi:10.1029/2001JD001269.
bution of aerosol particles captured through nucleation and impaction Nakajima, T. Y., and T. Nakajima (1995), Wide-area determination of cloud
scavenging by growing cloud drops, J. Atmos. Sci., 42, 583 – 606. microphysics properties from NOAA AVHRR measurements for FIRE
Gerber, H. (1996), Microphysics of marine stratocumulus clouds with two and ASTEX regions, J. Atmos. Sci., 52, 4043 – 4059.
drizzle modes, J. Atmos. Sci., 53, 1649 – 1662. Pruppacher, H. R., and J. D. Klett (1997), Microphysics of Clouds and
Han, Q., W. B. Rossow, and A. A. Lasis (1994), Near-global survey of Precipitation, 2nd ed., 771 pp., Oxford Univ. Press., New York.
effective droplet radii in liquid water clouds using ISCCP data, J. Clim., Randall, D. A., J. A. Coakley, C. W. Fairall, R. A. Kropfli, and D. H.
7, 465 – 497. Lenschow (1984), Outlook for research on subtropical marine stratiform
Houze, R. A., Jr. (1993), Cloud Dynamics, 573 pp., Academic, San Diego, clouds, Bull. Am. Meteorol. Soc., 65, 1290 – 1301.
Calif. Randall, D. A., M. Khairoutdinov, A. Arakawa, and W. Grabowski (2003),
Jonas, P. R. (1996), Turbulence and cloud microphysics, Atmos. Res., 40, Breaking the cloud parameterization deadlock, Bull. Am. Meteorol. Soc.,
283 – 306. 6, 1547 – 1564.
Kawamoto, K., T. Nakajima, and T. Y. Nakajima (2001), Wide-area deter- Rosenfeld, D. (2000), Suppression of rain and snow by urban and industrial
mination of cloud microphysics properties from NOAA AVHRR mea- air pollution, Science, 287, 1793 – 1796.
surements for FIRE and ASTEX regions, J. Clim., 14, 2054 – 2068. Rosenfeld, D., and G. Gutman (1994), Retrieving microphysical properties
King, M. D. (1987), Determination of the scaled optical thickness of clouds near tops of potential rain clouds by multispectral analysis of AVHRR
from reflected solar radiation measurements, J. Atmos. Sci., 44, 1734 – data, Atmos. Res., 34, 259 – 283.
1751. Rotstayn, L. D. (1999), Indirect forcing by anthropogenic aerosols: A glo-
Khain, A., M. Ovtchinnikov, M. Pinsky, A. Pokrovsky, and H. Krugliak bal climate model calculation of the effective-radius, J. Geophys. Res.,
(2000), Notes on the state-of-the-art numerical modeling of cloud micro- 104, 9369 – 9380.
physics, Atmos. Res., 55, 159 – 224. Stephens, G. L. (2005), Cloud feedbacks in the climate system: A critical
Kneizys, F. X., E. P. Shettle, L. W. Arbeu, J. H. Chetwynd, G. P. Andersen, review, J. Clim., 18, 237 – 273.
W. O. Gallery, J. E. Selby, and S. A. Clough (1988), Users guide to Vaillancourt, P. A., and M. K. Yau (2000), Review of particle-turbulence
LOWTRAN-7, Tech. Rep. AFGL-TR-88 – 0177, Air Force Geophys. interactions and consequences for cloud physics, Bull. Am. Meteorol.
Lab., Bedford, Mass. Soc., 81, 285 – 298.
Kobayashi, T., and A. Adachi (2001), Measurements of raindrop breakup
by using UHF wind profilers, Geophys. Res. Lett., 28, 4071 – 4074.
Kummerow, C., et al. (2000), The status of the Tropical Rainfall Measuring T. Kobayashi, Meteorological Research Institute, Tsukuba, 305-0052,
Mission (TRMM) after two years in orbit, J. Appl. Meteorol., 39, 1965 – Japan. (kobay@mri-jma.go.jp)
1982.

4 of 4

You might also like