Instantaneous cloud overlap statistics in the tropical area revealed by ICESat/GLAS data

Similar documents
On the Satellite Determination of Multilayered Multiphase Cloud Properties. Science Systems and Applications, Inc., Hampton, Virginia 2

Characteristics of cirrus clouds from ICESat/GLAS observations

Validation of ECMWF global forecast model parameters using GLAS atmospheric channel measurements

Lecture 4b: Meteorological Satellites and Instruments. Acknowledgement: Dr. S. Kidder at Colorado State Univ.

Impact of the 2002 stratospheric warming in the southern hemisphere on the tropical cirrus clouds and convective activity

Cluster analysis of tropical clouds using CloudSat data

Estimation of ocean contribution at the MODIS near-infrared wavelengths along the east coast of the U.S.: Two case studies

Diurnal cycles of precipitation, clouds, and lightning in the tropics from 9 years of TRMM observations

Global characterization of cirrus clouds using CALIPSO data

6.6 VALIDATION OF ECMWF GLOBAL FORECAST MODEL PARAMETERS USING THE GEOSCIENCE LASER ALTIMETER SYSTEM (GLAS) ATMOSPHERIC CHANNEL MEASUREMENTS

Clouds and water vapor in the Northern Hemisphere summertime stratosphere

GLAS Team Member Quarterly Report

THIN CLOUD LENGTH SCALES USING CALIPSO AND CLOUDSAT DATA. A Thesis JEREMY EDWARD SOLBRIG

Changes in seasonal cloud cover over the Arctic seas from satellite and surface observations

Interannual Variations of Arctic Cloud Types:

REVISION OF THE STATEMENT OF GUIDANCE FOR GLOBAL NUMERICAL WEATHER PREDICTION. (Submitted by Dr. J. Eyre)

J12.4 SIGNIFICANT IMPACT OF AEROSOLS ON MULTI-YEAR RAIN FREQUENCY AND CLOUD THICKNESS

Global Cloud Climatologies from satellite-based InfraRed Sounders (TOVS, AIRS, IASI) +

Remote sensing of ice clouds

Global Daytime Distribution of Overlapping Cirrus Cloud from NOAA's Advanced Very High Resolution Radiometer

Cloud and radiation budget changes associated with tropical intraseasonal oscillations

A 2-d modeling approach for studying the formation, maintenance, and decay of Tropical Tropopause Layer (TTL) cirrus associated with Deep Convection

Study of the Influence of Thin Cirrus Clouds on Satellite Radiances Using Raman Lidar and GOES Data

Interpretation of Polar-orbiting Satellite Observations. Atmospheric Instrumentation

Cloud features detected by MODIS but not by CloudSat and CALIOP

Convective scheme and resolution impacts on seasonal precipitation forecasts

Improving the CALIPSO VFM product with Aqua MODIS measurements

Remote Sensing of Precipitation

Ice clouds observed by passive remote sensing :

Lecture 3. Background materials. Planetary radiative equilibrium TOA outgoing radiation = TOA incoming radiation Figure 3.1

Tropical stratospheric zonal winds in ECMWF ERA-40 reanalysis, rocketsonde data, and rawinsonde data

Comparison of the CALIPSO satellite and ground based observations of cirrus clouds at the ARM TWP sites

Probability of Cloud-Free-Line-of-Sight (PCFLOS) Derived From CloudSat and CALIPSO Cloud Observations

Changes in Cloud Cover and Cloud Types Over the Ocean from Surface Observations, Ryan Eastman Stephen G. Warren Carole J.

Cloud Microphysical and Radiative Properties Derived from MODIS, VIRS, AVHRR, and GMS Data Over the Tropical Western Pacific

REMOTE SENSING KEY!!

CloudSat adding new insight into tropical penetrating convection

PARAMETERIZATION OF CLOUD FROM NWP TO CLIMATE MODEL RESOLUTION. Richard M. Forbes, 1. European Centre for Medium Range Weather Forecasts, Reading, UK

CORRELATION BETWEEN ATMOSPHERIC COMPOSITION AND VERTICAL STRUCTURE AS MEASURED BY THREE GENERATIONS OF HYPERSPECTRAL SOUNDERS IN SPACE

Relationships among cloud occurrence frequency, overlap, and effective thickness derived from CALIPSO and CloudSat merged cloud vertical profiles

ESA Cloud-CCI Phase 1 Results Climate Research Perspective

A statistical approach for rainfall confidence estimation using MSG-SEVIRI observations

A "New" Mechanism for the Diurnal Variation of Convection over the Tropical Western Pacific Ocean

A Suite of Retrieval Algorithms for Cirrus Cloud Microphysical Properties Applied To Lidar, Radar, and Radiometer Data Prepared for the A-Train

Cloud type climatology over the Tibetan Plateau: A comparison of ISCCP and MODIS/TERRA measurements with surface observations

Threshold radar reflectivity for drizzling clouds

On the global character of overlap between low and high clouds

Monitoring Climate Change from Space

Use of A Train data to estimate convective buoyancy and entrainment rate

Analysis of Cloud-Radiation Interactions Using ARM Observations and a Single-Column Model

The Climatology of Clouds using surface observations. S.G. Warren and C.J. Hahn Encyclopedia of Atmospheric Sciences.

Correcting Microwave Precipitation Retrievals for near- Surface Evaporation

Tropical cirrus and water vapor: an effective Earth infrared iris feedback?

Evaluating Cloud Frequency of Occurrence and Cloud-Top Height Using Spaceborne Lidar Observations

Reconciling Ground-Based and Space-Based Estimates of the Frequency of Occurrence and Radiative Effect of Clouds around Darwin, Australia

Use of A-Train Data to Estimate Convective Buoyancy and Entrainment. Rate

Final Report: NASA Award Number, NNX07AR95G, entitled, COMPARISON OF A - TRAIN CLOUD RETRIEVALS AND MULTI-INSTRUMENT ALGORITHM STUDIES.

Effect of clouds on the calculated vertical distribution of shortwave absorption in the tropics

ABSTRACT 2 DATA 1 INTRODUCTION

P2.12 Sampling Errors of Climate Monitoring Constellations

Retrieving cloud top structure from infrared satellite data

Journal of the Meteorological Society of Japan, Vol. 75, No. 1, pp , Day-to-Night Cloudiness Change of Cloud Types Inferred from

Implications of the differences between daytime and nighttime CloudSat. Chuntao Liu, Edward J. Zipser, Gerald G. Mace, and Sally Benson

A Preliminary Assessment of the Simulation of Cloudiness at SHEBA by the ECMWF Model. Tony Beesley and Chris Bretherton. Univ.

APPENDIX 2 OVERVIEW OF THE GLOBAL PRECIPITATION MEASUREMENT (GPM) AND THE TROPICAL RAINFALL MEASURING MISSION (TRMM) 2-1

2. Data and Methods LXXXXX

Radiative Climatology of the North Slope of Alaska and the Adjacent Arctic Ocean

Satellite-based estimate of global aerosol-cloud radiative forcing by marine warm clouds

Chapter Seven. Heating and Cooling Rates from CloudSat.

An Annual Cycle of Arctic Cloud Microphysics

Remote sensing of sea ice

Do aerosols affect lightning?: A global study of a relation between aerosol optical depth and cloud to ground lightning

Influence of Clouds and Aerosols on the Earth s Radiation Budget Using Clouds and the Earth s Radiant Energy System (CERES) Measurements

A description of hydrometeor layer occurrence statistics derived from the first year of merged Cloudsat and CALIPSO data

8.2 Numerical Study of Relationships between Convective Vertical Velocity, Radar Reflectivity Profiles, and Passive Microwave Brightness Temperatures

Ground-based Validation of spaceborne lidar measurements

Global Moderate Resolution Imaging Spectroradiometer (MODIS) cloud detection and height evaluation using CALIOP

Comparisons of satellites liquid water estimates to ECMWF and GMAO analyses, 20th century IPCC AR4 climate simulations, and GCM simulations

Course outline, objectives, workload, projects, expectations

Modulation of the diurnal cycle of tropical deep convective clouds

Evaluation of cloud thermodynamic phase parametrizations in the LMDZ GCM by using POLDER satellite data

Advantageous GOES IR results for ash mapping at high latitudes: Cleveland eruptions 2001

A Climatology of Surface Cloud Radiative Effects at the ARM Tropical Western Pacific Sites

VERIFICATION OF MERIS LEVEL 2 PRODUCTS: CLOUD TOP PRESSURE AND CLOUD OPTICAL THICKNESS

The MODIS Cloud Data Record

Inferring Cloud Feedbacks from ARM Continuous Forcing, ISCCP, and ARSCL Data

Relation of atmospheric humidity and cloud properties to surface-near temperatures derived from GOME satellite observations

Cloud type comparisons of AIRS, CloudSat, and CALIPSO cloud height and amount

GPS RO Retrieval Improvements in Ice Clouds

HIRDLS and CALIPSO observations of tropical cirrus

Sensitivity Study of the MODIS Cloud Top Property

Prospects for radar and lidar cloud assimilation

Spaceborne Hyperspectral Infrared Observations of the Cloudy Boundary Layer

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, D00H32, doi: /2009jd012334, 2010

Clouds, Haze, and Climate Change

Consistent estimates from satellites and models for the first aerosol indirect forcing

Backscatter Color Ratios of Cirrus Clouds Measured by the Cloud Physics Lidar

How TRMM precipitation radar and microwave imager retrieved rain rates differ

Improved rainfall and cloud-radiation interaction with Betts-Miller-Janjic cumulus scheme in the tropics

Condensation: Dew, Fog, & Clouds. Chapter 5

Transcription:

GEOPHYSICAL RESEARCH LETTERS, VOL. 33, L15804, doi:10.1029/2005gl024350, 2006 Instantaneous cloud overlap statistics in the tropical area revealed by ICESat/GLAS data Likun Wang 1,2 and Andrew E. Dessler 1,3 Received 10 August 2005; revised 18 April 2006; accepted 27 June 2006; published 4 August 2006. [1] This study uses ICESat/GLAS instantaneous observations from 29 September to 18 November 2003 to investigate cloud overlap statistics between 10 S 20 N. The results show that 75.1% of profiles detect clouds: 46.5% are single layer and 28.6% multilayer clouds (cloud layers are separated by 0.5km). Using a definition of cloud type based on cloud heights and laser attenuation information, cloud overlap statistics are derived by analyzing 96.4% of the cloudy profiles. The most frequent overlap occurs between cirrus clouds and boundary layer clouds, which accounts for 31.88% of boundary clouds. 23.8% of deep convection has overlying cirrus clouds. We find that differences exist between the cloud overlap fraction from the GLAS observations and one calculated from the random overlap assumption commonly used by climate models. Citation: Wang, L., and A. E. Dessler (2006), Instantaneous cloud overlap statistics in the tropical area revealed by ICESat/GLAS data, Geophys. Res. Lett., 33, L15804, doi:10.1029/2005gl024350. ground-based cloud radars that provide vertical distribution of cloud fraction. These data, however, are limited to specific locations and time periods [Hogan and Illingworth, 2000; Mace and Benson-Troth, 2002]. In addition, due to the dependence of radar reflectivity on the sixth moment of the particle size distribution, microwave radar measurements sometimes are insensitive to thin cirrus, and are also obscured when heavy rain is present at lower levels [Sassen and Mace, 2002]. The last source is measurements by passive remote sensors on satellites. While in recent years great improvements have been made in detecting multiple layer clouds using these data, these algorithms only work well in specific situations (e.g., cirrus overlapping water clouds [Chang and Li, 2005a] and multilayer cirrus clouds [Ou et al., 1996]) and are still by and large experimental. [4] In this paper, we focus on cloud overlap statistics for the tropics derived from a new data source: the Geoscience Laser Altimeter System (GLAS), carried aboard the Ice, Cloud, and Land Elevation Satellite (ICESat). 1. Introduction [2] Cloud overlap variations greatly influence atmospheric heating/cooling rates, with important implications for the radiative balance of the surface-troposphere system [Stephens et al., 2004]. General circulation models (GCMs) incorporate empirical overlap assumptions, which are mostly without rigorous physical basis or conclusive validation against observations [Stephens et al., 2004], mainly due to the lack of accurate, unbiased cloud overlap observations. [3] Before now, the statistics of cloud overlap have been obtained primarily from four sources. The first source is observations by surface weather observers [Warren et al., 1985]. However, these results underestimate the frequency of upper clouds overlapped with lower clouds because the surface observer s view can be obscured by the lower thick cloud deck. The second source is the analysis of vertical profiles of relative humidity from radiosondes [Wang et al., 2000] clouds are assumed to occur when relative humidity reaches saturation. However, these balloon humidity measurements are not reliable at temperatures well below the freezing point, making the inference of clouds in the upper troposphere inaccurate [Guichard et al., 2000]. The third source relies on continuous vertically-pointing 1 Earth System Science Interdisciplinary Center, University of Maryland at College Park, College Park, Maryland, USA. 2 Now at QSS Group Inc, Lanham, Maryland, USA. 3 Now at Department of Atmospheric Sciences, Texas A&M University, College Station, Texas, USA. Copyright 2006 by the American Geophysical Union. 0094-8276/06/2005GL024350 2. Data and Methods [5] The GLAS is a diode-pumped Q-switched Nd:YAG laser operating with the near infrared (1064 nm) and visible (532 nm) channels. Measurements of attenuated backscatter from the instrument are processed by the science team to produce high-quality measurements of cloud properties over most of the globe [Dessler et al., 2006; Spinhirne et al., 2005]. For a thorough discussion of the GLAS data, see the papers in the special issue of Geophysical Research Letters devoted to ICESat/GLAS atmosphere data (volume 32, issues 21 and 22, 2005). During the period analyzed in this paper, 29 September to 18 November 2003 (the so-called Laser 2a period), the satellite obtained measurements in the tropics around 8 am and 8 pm local solar time with a filed of view of 70 m. [6] We only use the cloud tops and bottoms retrieved from the 532-nm channel in the released GLA09 data product. Their retrieval algorithm averages four seconds (160 laser shots) of attenuated backscatter data and then applies a single-pass threshold technique that defines cloud layer top (bottom) as the height where two consecutive GLAS data bins of 76.8 m exceed (are less than) the threshold for each profile (from the top to bottom) (S. P. Palm et al., Geosciences Laser Altimeter System (GLAS) atmospheric data products, algorithm theoretical basis document, 2002, available at: http://www.csr.utexas.edu/glas/ atbd.html). The inherent uncertainty of cloud top measurements is less than 200 m [Dessler et al., 2006]. In this study, cloud base heights are not used to classify cloud types once full attenuation of the laser beam happens. For the cloud layers that are separated by less than 0.5 km, we treat them L15804 1of5

Table 1. Definitions of Cloud Type Occurring in the Tropical Region a Cloud Base, km Cloud Top, km Lasers Pass Through? Cirrus > = 7.0 - Yes Boundary clouds - < = 4.0 Both Deep convection - >10.0 No Shallow convection - >4.0 and < = 10.0 No Mid-level clouds >4.0 < = 7.0 Yes a Those not falling into any of the categories are unclassified. as one layer because current GCMs with a relatively lower vertical resolution can not resolve them. This treatment decreases multilayer cloud amount. However, it does not significantly change the overlap statistics by doing sensitive test. [7] The geographic region of our focus in this paper is in the tropics. We choose this region because it is dominated by boundary layer clouds and cirrus with relatively small amounts of clouds in the middle atmosphere [Comstock and Jacob, 2004; Chang and Li, 2005b], making it easy to discriminate cloud type based on cloud-height values alone. Measurements of outgoing longwave radiation indicate that most convection occurred between 10 S and 20 N during the Laser-2a period [Dessler et al., 2006], so we concentrate on this latitude range in this paper. We divide GLAS observations of clouds into five types: cirrus, mid-level clouds, deep convective clouds, shallow convective clouds, and boundary clouds. The first determination that we make is whether the cloud fully attenuates the GLAS laser this generally happens for clouds with optical depths above about three to four [Hart et al., 2005]. For clouds that do not completely attenuate the laser, we define cirrus clouds to be those clouds with bases above 7.0 km [Comstock et al., 2002], and mid-level clouds to be those with tops between 4.0 and 7.0 km (comprised primarily of altocumulus or altostratus). [8] For the thick clouds that completely attenuate the laser beam, we classify them as deep convection if their cloud tops are above 10.0 km or shallow convection if their cloud tops are between 4.0 and 10.0 km. Clouds with tops below 4.0 km are defined as boundary clouds (e.g., cumulus congestus) regardless of whether they attenuate the laser. This partition of thick clouds follows the trimodal distribution of tropical convection identified by Johnson et al. [1999]. Clouds not falling into these five categories are unclassified. We should note that the deep convection clouds based on our definition include both convective cores and extended thick anvils. Shallow convective clouds, which extend up to 10 km, include clouds that some might consider moderate deep convection. The cloud types are summarized in Table 1. [9] The most common cloud overlap scheme in current GCMs is called random-maximum overlap. It assumes that cloud layers separated by any clear layers are randomly overlapped while vertically-continues cloud layers overlap maximally [Stephens et al., 2004]. Cloud layers identified in this study are separated by at least 0.5 km, and thus we focus on the overlap of separated layers. Given the fractional cloud coverage of two separated cloud layers, C 1 and C 2, their overlap fraction is arbitrarily assumed by the products of C 1 and C 2. The total cloud fraction C random under the random overlap is given by: C random ¼ C 1 þ C 2 C 1 C 2 : On the other hand, if we know their overlap fraction, C overlap, the observed cloud fraction C real is defined as: C real ¼ C 1 þ C 2 C overlap : By comparing C overlap with the products of C 1 and C 2,we can evaluate how well random overlap can characterize the overlap of two separated cloud layers. [10] Note that cloud fraction in GCMs is defined as the fraction of a simulated grid box that is covered by clouds. However, cloud fraction in this study represents the occurrence frequency of a particular cloud type in the dataset. While different, it is of interest to examine the statistical overlap properties of different cloud types at a large spatial and temporal scale, which do represent GCM grid boxes. 3. Results and Discussion [11] Figure 1 gives the cloud layer (separated by 0.5 km) occurrence frequency in the tropical area, with the statistics broken down by whether the surface is detected or not. 75.1% of the samples in the tropics detect one or more cloud layers. 46.5% detect only a single layer, 20.6% detect two layers, 6.3% detect three layers, and 1.6% detect four or more cloud layers. Note that our results on cloud layer frequency can be considered a lower limit because the GLAS cannot see below clouds with optical depths above 3 4 and therefore missed underlying clouds. In the rest of the paper, we limit our analysis to the observations of one layer, two layer, and most of three layer clouds that are mainly double layer cirrus overlapped with other clouds (82.9% of three layer clouds), which account for 96.4% of the cloudy observations. [12] Figure 2 shows the frequency of observations of two layer clouds as a function of the top layer s cloud base and the second layer s cloud top. The bin size is 0.25 km, and the plot separates observations that detect the ground and those that do not. Using the cloud type definitions in Table 1, Table 2 gives the frequency of each cloud type Figure 1. Cloud layer (separated by 0.5 km) occurrence frequency between 10 S and 20 N during GLAS Laser-2a period. ð1þ ð2þ 2of5

Figure 2. Frequency distribution of two layer clouds as function of top layer s cloud base and second layer s cloud top (the bin size is 0.25 km). and their overlap occurrence. The statistics over ocean are given in parentheses. [13] We note first that cloud occurrence in the tropical region is dominated by cirrus clouds and boundary clouds (observed in 47.04% and 20.83% of the total GLAS observations). A similar statistical result was also found by Chang and Li [2005b] and Comstock and Jacob [2004]. Among two layered clouds (20.6% of the GLAS observations and separated by 0.5 km), 31.8% are cirrus overlying other cirrus, 1.0% are boundary clouds overlying other boundary clouds, and 0.3% are mid-level clouds overlying other mid-level clouds. The overlap from the same cloud type is not important for passive remote sensing or numerical climate simulation because they have similar cloud properties and generally similar temperatures when they are not separated too far. Treating them as a single layer cloud may not introduce severe errors in the calculation of their cloud properties. [14] We are most interested in the overlap between different cloud types. The most common overlap is between cirrus and boundary clouds, which account for 6.64% of the GLAS observations and 31.88% of boundary clouds. By applying a new method to the Moderate Resolution Imaging Spectroradiometer (MODIS) data, Chang and Li [2005b] also concluded that high thin cirrus clouds often overlie boundary-layer clouds in the tropical region. [15] Deep convection accounts for 8.78% of the total observations, which is consistent with the results from other satellite observations [Alcala and Dessler, 2002; Gettelman et al., 2002]. Of the total observations, 2.09% have overlying cirrus clouds. Other analyses have found that tropical cirrus clouds can be stacked above deep convection vertically [Garrett et al., 2004]. The potential importance of this kind of overlap has been discussed by Hartmann et al. [2001]. If sufficient cirrus clouds at the tropopause frequently lie above convective anvil clouds, they can lead to net radiative cooling accompanied by subsidence and contribute to a mechanism for drying the air entering the stratosphere. However, our results indicate that only 23.8% of deep convection lies below cirrus clouds. Such small overlap fraction probably is not responsible for the mechanism of stratospheric drain suggested by Hartmann et al. [2001]. [16] Finally, we use the GLAS observations to evaluate cloud overlap schemes used by GCMs. Given the occurrence frequency of any pair of cloud types, their overlap fraction can be inferred by their products under random overlap, which is compared with that from the GLAS observations (Table 3). The results show that differences exist, though random cloud overlap is extensively thought to better characterize cloud overlap behavior than minimum overlap and maximum overlap. For example, for cirrus with Table 2. Cloud Type Frequencies Between 10 S and 20 N During GLAS Laser-2a Period (96.4% of the Cloudy Observations) a Given the Cloud Type Single Layer (Total 46.5 (36.5)) Cirrus With Lower Single Cloud Layer (Total 20.6 (15.1)) Boundary Clouds Deep Convection Shallow Convection Midlevel Clouds Unclassified With Upper Double Layer Cirrus 5.22(3.64) Total Cloud Cover Cirrus 22.24 (16.57) 6.55 (4.60) 5.31 (4.41) 1.89 (1.32) 2.79 (1.98) 1.86 (1.20) 1.18 (1.13) 1.07 (0.72) 47.04 (34.50) Boundary 13.42 (12.26) - 0.20 (0.17) - - - - 1.33 (1.00) 20.83 (18.31) Deep convection 6.69 (4.86) - - - - - - 0.20 (0.15) 8.78 (6.34) Shallow convection 1.78 (1.12) - - - - - - 1.27 (0.92) 5.97 (4.10) Mid-level clouds 1.44 (0.65) - 0.43 (0.36) - 0.08 (0.04) 0.06 (0.03) 0.03 (0.03) 0.88(0.56) 4.83 (3.10) Unclassified 0.96 (0.87) - 0.14 (0.11) - 0.05 (0.03) 0.05 (0.03) 0.01 (0.01) 0.47 (0.46) 2.89 (2.67) a The numbers are a percentage of the total observations. Boldfaced values indicate the overlap from the same cloud type, and those in parentheses occur over ocean. 3of5

Table 3. Overlap Cloud Fraction of Any Pair of Cloud Type From the GLAS Observations (C overlap ) and the Random Overlap Assumption (C 1 C 2 ), and Their Relative Difference a boundary clouds and cirrus with deep convective clouds, their overlap fractions are overestimated by 47.57% and 97.38%, respectively, by applying random overlap. The overestimation also happens for mid-level clouds with boundary clouds and mid-level clouds with shallow convection. The differences decrease for the overlap occurring over oceans. On the other hand, the overlap of cirrus with mid-level clouds and cirrus with shallow convection are underestimated by 17.08% and 30.83%, with increasing the underestimation over oceans. This indicates that land surface effects may intensify the tendency of the overlap toward minimum overlap. [17] This analysis of the GLAS data set suffers from two uncertainties. First, the GLAS provides no information about clouds sitting beneath thick, laser-attenuating clouds (optical depths > 3 4). Convective clouds generally extend vertically from the boundary to the middle or upper troposphere, so the chance that other clouds stand below can exist but is very small. For boundary clouds that block the laser beam, the possible lower layer is still boundary cloud. It does not bring uncertainty to the overlap statistics of different cloud types. Secondly, the GLAS laser views the nadir only, so on any orbit it only observes a small fraction of the planet. This sparse sampling leads to the possibility of sampling biases. We calculated the standard deviation of daily zonal cloud fraction over the sampling time period that is 2.9%. Since we focus on the statistics over a large spatial (zonal) and temporal (eight weeks) scale with a large number of samples, we believe that the sampling biases are very small. For both of these fundamental limitations, we expect data from the Cloud-Aerosol LIDAR and Infrared Pathfinder Satellite (CALIPSO) and CloudSat missions to provide a more thorough data set for analysis of these issues [Stephens et al., 2002]. 4. Conclusion C 1 C 2 C overlap Difference b Cirrus + Boundary 9.80 (6.32) 6.64 (5.13) 47.57 (23.14) Cirrus + Deep 4.13 (2.19) 2.09 (1.48) 97.38 (48.09) Cirrus+ Mid-level 2.27 (1.07) 2.74 (1.77) 17.08 ( 39.43) Cirrus + Shallow 2.81 (1.41) 4.06 (2.90) 30.83 ( 51.29) Mid-level + Boundary 1.01 (0.57) 0.43 (0.36) 135.76 (57.16) Mid-level + Shallow 0.29 (0.13) 0.08 (0.04) 281.86 (197.44) a The numbers are a percentage of the total observations, and those in brackets occur on ocean surfaces. b Calculated from (C 1 C 2 C overlap )/C overlap 100%. [18] We have used the ICESat/GLAS GLA09 cloudheight product to investigate the cloud overlap statistics in the tropical areas. Between 29 September and 18 November 2003 and between 10 S and 20 N, the results show that 75.1% of the GLAS observations detect clouds - 46.5% of the observations detect single-layer clouds and 28.6% detect more than one cloud layer (separation 0.5 km). This indicates that clouds often overlap with each other and passive remote sensing algorithms must consider cloud overlap situations. [19] For the overlap of different cloud types, the most frequent overlap occurs between cirrus and boundary clouds. We also find that only 23.8% of deep convection has overlying cirrus clouds. Applying our analysis results to evaluate random cloud overlap assumption used by GCMs, the differences exist between the overlap fraction from the GLAS observations and one calculated using random overlap assumption. [20] We should point out that the actual overlap of cloud layers is a strong function of the cloud spatial scale, cloud type, and synoptic situation, which should be characterized by their locations and time periods (e.g., season) [Warren et al., 1985; Mace and Benson-Troth, 2002]. Deriving cloud overlap statistics also depends on assumptions that one made (e.g., cloud fraction, cloud layer separation). We are expecting future missions to provide new measurements to address this issue again. [21] Acknowledgments. We appreciate the comments from anonymous reviewers. We also thank Steven Sherwood, Sun Wong, and Zhien Wang for their helpful comments. This work was supported by NASA EOS/ IDS grant to the University of Maryland. References Alcala, C. M., and A. E. Dessler (2002), Observations of deep convection in the tropics using the TRMM precipitation radar, J. Geophys. Res., 107(D24), 4792, doi:10.1029/2002jd002457. Chang, F.-L., and Z. Li (2005a), A new method for detecting overlapped high-and-low clouds and determining their optical properties using the MODIS data, J. Atmos. Sci., 18, 4752 4771. Chang, F.-L., and Z. Li (2005b), A near-global climatology of single-layer and overlapped clouds and their optical properties retrieved from Terra/ MODIS data using a new algorithms, J. Clim., 18, 4752 4771. Comstock, J. M., and C. Jacob (2004), Evaluation of tropical cirrus cloud properties derived from ECMWF model output and ground based measurements over Nauru Island, Geophys. Res. Lett., 31, L10106, doi:10.1029/2004gl019539. Comstock, J. M., T. P. Ackerman, and G. G. Mace (2002), Ground-based lidar and radar remote sensing of tropical cirrus clouds at Nauru Island: Cloud statistics and radiative impacts, J. Geophys. Res., 107(D23), 4714, doi:10.1029/2002jd002203. Dessler, A. E., S. P. Palm, W. D. Hart, and J. D. Spinhirne (2006), Tropopause-level thin cirrus coverage revealed by ICESat/GLAS, J. Geophys. Res., 111, D08203, doi:10.1029/2005jd006586. Garrett, T. J., et al. (2004), Convective generation of cirrus near the tropopause, J. Geophys. Res., 109, D21203, doi:10.1029/2004jd004952. Gettelman, A., M. L. Salby, and F. Sassi (2002), The distribution and influence of convection in the tropical tropopause region, J. Geophys. Res., 107(D10), 4080, doi:10.1029/2001jd001048. Guichard, F., D. Parsons, and E. Miller (2000), Thermodynamic and radiative impact of the correction of sounding humidity bias in the tropics, J. Clim., 13, 3611 3624. Hart, W. D., et al. (2005), Height distribution between cloud and aerosol layers from the GLAS spaceborne lidar in the Indian Ocean region, Geophys. Res. Lett., 32, L22S06, doi:10.1029/2005gl023671. Hartmann, D. L., J. R. Holton, and Q. Fu (2001), The heat balance of the tropical tropopause, cirrus and stratospheric dehydration, Geophys. Res. Lett., 28, 1969 1972. Hogan, R. J., and A. J. Illingworth (2000), Deriving cloud overlap statistics from radar Observations, Q. J. R. Meteorol. Soc., 126, 1 7. Johnson, R. H., et al. (1999), Trimodal characteristics of tropical convection, J. Clim., 12, 2397 2418. Mace, G. G., and S. Benson-Troth (2002), Cloud-layer overlap characteristics derived from long-term cloud radar data, J. Clim., 15, 2505 2515. Ou, S. C., K. N. Liou, and B. A. Baum (1996), Detection of multilayer cirrus cloud systems using AVHRR data: Verification based on FIRE-II IFO composite measurements, J. Appl. Meteorol., 35, 178 191. Sassen, K., and G. G. Mace (2002), Ground based remote sensing of cirrus clouds, in Cirrus, edited by D. Lynch et al., pp. 168 195, Oxford Univ. Press, New York. Spinhirne, J. D., et al. (2005), Cloud and aerosol measurements from GLAS: Overview and initial results, Geophys. Res. Lett., 32, L22S03, doi:10.1029/2005gl023507. 4of5

Stephens, G. L., et al. (2002), The CloudSat mission and the A-Train: A new dimension of space-based observations of clouds and precipitation, Bull. Am. Meteorol. Soc., 83, 1771 1790. Stephens, G. L., N. B. Wood, and P. M. Gabriel (2004), An assessment of the parameterization of subgrid-scale cloud effects on radiative transfer: Part I. Vertical overlap, J. Atmos. Sci., 61, 715 732. Wang, J., et al. (2000), Cloud vertical structure and its variations from a 20- year global rawinsonde dataset, J. Clim., 13, 3041 3056. Warren, S. G., et al. (1985), Simultaneous occurrence of different cloud types, J. Clim. Appl. Meteorol., 24, 658 667. A. E. Dessler, Department of Atmospheric Sciences, Texas A&M University, College Station, TX 77843, USA. L. Wang, QSS Group, Inc., NOAA, 5200 Auth Road, Room 810, Camp Springs, MD 20746, USA. (likun.wang@noaa.gov) 5of5