Estimating precipitation susceptibility in warm marine

Atmos Chem Phys 18 1763ndash1783 2018httpsdoiorg105194acp-18-1763-2018copy Author(s) 2018 This work is distributed underthe Creative Commons Attribution 40 License

Estimating precipitation susceptibility in warm marine clouds usingmulti-sensor aerosol and cloud products from A-Train satellitesHeming Bai12 Cheng Gong3 Minghuai Wang12 Zhibo Zhang4 and Tristan LrsquoEcuyer5

1Institute for Climate and Global Change Research and School of Atmospheric Sciences Nanjing University Nanjing China2Collaborative Innovation Center of Climate Change Jiangsu China3Institute of Atmospheric Physics Chinese Academy of Science Beijing China4Physics Department University of Maryland Baltimore County (UMBC) Baltimore MD USA5Department of Atmospheric and Oceanic Sciences University of Wisconsin Madison WI USA

Correspondence Minghuai Wang (minghuaiwangnjueducn)

Received 23 September 2017 ndash Discussion started 12 October 2017Revised 9 December 2017 ndash Accepted 3 January 2018 ndash Published 7 February 2018

Abstract Precipitation susceptibility to aerosol perturba-tion plays a key role in understanding aerosolndashcloud inter-actions and constraining aerosol indirect effects Howeverlarge discrepancies exist in the previous satellite estimatesof precipitation susceptibility In this paper multi-sensoraerosol and cloud products including those from the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation(CALIPSO) CloudSat Moderate Resolution Imaging Spec-troradiometer (MODIS) and Advanced Microwave Scan-ning Radiometer for the Earth Observing System (AMSR-E) from June 2006 to April 2011 are analyzed to estimateprecipitation frequency susceptibility SPOP precipitation in-tensity susceptibility SI and precipitation rate susceptibilitySR in warm marine clouds We find that SPOP strongly de-pends on atmospheric stability with larger values under morestable environments Our results show that precipitation sus-ceptibility for drizzle (with a minus15 dBZ rainfall threshold) issignificantly different than that for rain (with a 0 dBZ rain-fall threshold) Onset of drizzle is not as readily suppressedin warm clouds as rainfall while precipitation intensity sus-ceptibility is generally smaller for rain than for drizzle Wefind that SPOP derived with respect to aerosol index (AI) isabout one-third of SPOP derived with respect to cloud dropletnumber concentration (CDNC) Overall SPOP demonstratesrelatively robust features throughout independent liquid wa-ter path (LWP) products and diverse rain products In con-trast the behaviors of SI and SR are subject to LWP or rainproducts used to derive them Recommendations are further

made for how to better use these metrics to quantify aerosolndashcloudndashprecipitation interactions in observations and models

1 Introduction

Aerosolndashcloud interactions play an important role in the cli-mate system and affect the global energy budget and hydro-logical cycle The effective radiative forcing from aerosolndashcloud interactions (ERFaci) which includes the instanta-neous effect on cloud albedo from changes in cloud conden-sation nuclei (CCN) or ice nuclei and all subsequent changesto cloud lifetime and thermodynamics remains one of thelargest uncertainties in our estimates of anthropogenic ra-diative forcing (Boucher et al 2013) Over the past fewdecades numerous methodologies have been developed tounderstand and quantify the impacts of aerosolndashcloud inter-actions on the climate system A unique method is to use theso-called ldquosusceptibilityrdquo to explain and predict how cloudsand precipitation would respond if there were some aerosolperturbations Susceptibility is defined as the derivative ofcloud andor precipitation properties with respect to aerosol-related properties For example Platnick and Twomey (1994)proposed a cloud albedo susceptibility as Sλ = partApartCDNCwhere A is cloud albedo and CDNC is cloud droplet numberconcentration to quantify the cloud albedo effect of aerosol

Precipitation susceptibility has been proposed to evaluateaerosolndashcloudndashprecipitation interactions and to further con-strain the cloud water response to aerosol perturbations in cli-

Published by Copernicus Publications on behalf of the European Geosciences Union

1764 H Bai et al Estimating precipitation susceptibility in warm marine clouds

mate models (Feingold and Siebert 2009 Terai et al 2012Wang et al 2012) It was first proposed by Feingold andSiebert (2009) and was defined as

S0 =minusdlnR

dlnCDNC (1)

where R is precipitation intensity (precipitation rate for rainyclouds) and CDNC is cloud droplet number concentration(Feingold and Siebert 2009) Sorooshian et al (2009) furtherestimated S0 by replacing CDNC with aerosol index (AI)Wang et al (2012) proposed an alternative metric the pre-cipitation frequency susceptibility defined as

SPOP =minusdlnPOPdlnAI

where POP is the probability of precipitation SPOP has beenshown to strongly correlate with cloud water response toaerosol perturbations in global climate models (Wang et al2012 Ghan et al 2016) Terai et al (2012 2015) furtherextended the definition of precipitation susceptibility

SX =minusdlnX

dlnCDNC (3)

where X can represent precipitation intensity (I precipita-tion rate from rainy clouds only) precipitation fraction (POPor f ) or precipitation rate (R = POPtimes I mean precipitationrate from both rainy and non-rainy clouds) Depending onwhether I POP or R are used in Eq (3) precipitation in-tensity susceptibility (SI) precipitation frequency suscepti-bility (SPOP or Sf) or precipitation rate susceptibility (SR)are therefore defined accordingly Since R can be decom-posed into the product of POP and I SR asymp SPOP+ SI (Teraiet al 2012 2015) In addition some other studies substi-tute aerosol concentration (NA) or cloud condensation nu-clei (CCN) concentration (NCCN) for CDNC to calculate SX(Terai et al 2012 Mann et al 2014)

The behavior and magnitude of aforementioned precipita-tion susceptibility metrics varies a lot in different studies Forinstance SR and SPOP using NA as an aerosol proxy fromTerai et al (2012) both noticeably decrease with increasingLWP whereas SI is flat in the same study Additionally pre-vious satellite studies (Wang et al 2012 Terai et al 2015Michibata et al 2016) show SX calculated with respect toCDNC is higher than that with respect to AI The diverse def-initions of precipitation susceptibility make it challenging tounderstand susceptibility discrepancies in different studiesAn important objective of this study is to derive these suscep-tibilities using the same observations in the same context andto better understand their differences through comparisons

Another source of uncertainty in the estimation of pre-cipitation susceptibility is the uncertainty associated withthe observation Among many others the Advanced Mi-crowave Scanning Radiometer for the Earth Observing Sys-tem (AMSR-E) and Moderate Resolution Imaging Spectro-

radiometer (MODIS) are two widely used satellite cloud re-trieval products in aerosolndashcloud interaction studies For in-stance Sorooshian et al (2009) and Wang et al (2012) bothused the AMSR-E LWP product to estimate SI and SPOP withrespect to AI respectively Terai et al (2015) and Michibataet al (2016) used the MODIS LWP product to estimate SIwith respect to CDNC Both products have their advantagesand limitations and are both subject to various retrieval un-certainties AMSR-E has a coarser spatial resolution thanMODIS Its LWP retrievals are available for both daytimeand nighttime but suffer from instrument noise cloud detec-tion issues and beam filling effect (Greenwald et al 2007Horvaacuteth and Gentemann 2007 Seethala and Horvaacuteth 2010)MODIS LWP retrievals are available only during daytimeThe main uncertainty sources in MODIS LWP retrievals in-clude instrument noise subpixel cloud inhomogeneity three-dimensional radiative effects and uncertainties in ancillarydata (Cho et al 2015 Platnick et al 2017 Zhang and Plat-nick 2011) A recent study by Seethala and Horvaacuteth (2010)revealed several significant differences between AMSR-Eand MODIS LWP products which could contribute to theaforementioned discrepancy of precipitation susceptibility inthe literature

Additionally different definitions of rain events andor dif-ferent methods to derive rain rates could also lead to discrep-ancy in observation-based estimation of precipitation suscep-tibility For example the rain rate used in Terai et al (2015)and Michibata et al (2016) is simply estimated based on aZndashR relationship from CloudSat radar reflectivity profile mea-surements In contrast Sorooshian et al (2009) and Wanget al (2012) used the rain rate reported in the CloudSat opera-tional product which makes use not only of radar reflectivitybut also path-integrated attenuation in the retrieval process(Haynes et al 2009) The primary satellite data sets used inthe previous studies for estimating precipitation susceptibil-ity are listed in Table 1 To account for the discrepancy insusceptibility as shown in Table 1 it is important to examinehow different LWP and rain data sets affect the estimates ofprecipitation susceptibility

Here we estimate precipitation susceptibility using multi-sensor cloud and aerosol products from A-Train satellitesThe main objective of this study is to compare precipitationsusceptibility estimates based on different retrieval productsand to better understand discrepancies documented in previ-ous studies As previous studies have shown that aerosol indi-rect effect and its uncertainties vary in different cloud dynam-ical regimes (LrsquoEcuyer et al 2009 Wang et al 2012 Zhanget al 2016) we further examine how precipitation suscepti-bility might be different under different atmospheric stabil-ity conditions Section 2 introduces different satellite prod-ucts and methods used to calculate the susceptibility Sect 3compares precipitation susceptibility estimates from differ-ent satellite products and explores how atmospheric stabil-ity affects precipitation susceptibility finally the discussionsare made in Sect 4 followed by the summary in Sect 5

Atmos Chem Phys 18 1763ndash1783 2018 wwwatmos-chem-physnet1817632018

H Bai et al Estimating precipitation susceptibility in warm marine clouds 1765

Table 1 The summary of previous satellite studies for estimating precipitation susceptibility

Studies Rain Aerosol Thresholds Behavior Satellitevariables proxies data sets

Sorooshian I AI Surface 1 mmhminus1 SI 2C-PRECIP-COLUMNet al (2009) AMSR-E L2B-Ocean

MYD08-D3

Wang et al POP AI Rain certaina SPOP lt 02 2C-PRECIP-COLUMN(2012) AMSR-E L2B-Ocean

MYD08-D3

Terai et al RPOPI CDNC minus15 dBZ of Zmaxb SR 2B-GEOPROF

(2015) SPOP MYD06_L2SI

Michibata I CDNC minus15 dBZ of Zmaxb SI 2B-GEOPROF

et al (2016) MYD06_L2

a ldquoRain certainrdquo is a flag of the 2C-PRECIP-COLUMN product which is equivalent to greater than the attenuation-correctedreflectivity threshold of 0 dBZb Zmax the maximum column radar reflectivity from the 2B-GEOPROF product The symbols () represent the increasing(decreasing) trend of susceptibility with increasing LWP

Table 2 Satellite products employed to estimate aerosol and cloud properties in this study

Parameter Product Subset Horizontal Sensor Satelliteresolution

AI MYD08_D3 Aerosol_Optical_Depth_Land_Ocean_Mean 1 MODIS AquaAerosol_AE1_Ocean_JHisto_vs_Opt_Depth

MYD04_L2 Optical_Depth_Land_And_Ocean 10 kmAngstrom_Exponent_1_Ocean

CAL_LID_L2_05kmALay Column_Optical_Depth_Aerosol_532 5 km CALIOP CALIPSOColumn_Optical_Depth_Aerosols_1064

CDNCLWP MYD06_L2a Cloud_Effective_Radius 5 km MODIS AquaCloud_Optical_Thickness

LWP AE_Ocean_L2B High_res_cloud 12 km AMSR-E Aqua

POPR 2B-GEOPROFa CPR_Cloud_mask 5 km CPR CloudSatRadar_Reflectivity

2C-PRECIP-COLUMN Precip_rate 14kmtimes 17kmPrecip_flag

2C-RAIN-PROFILE Rain_ratePrecip_flag

a The original horizontal resolution of MYD06_L2 and 2B-GEOPROF products is 1 km and 14 kmtimes 17 km respectively Since these products both are obtained from thecaltrack product collocated to the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) subtrack the resolution is resampled to 5 km Detailed information is provided athttpwwwicareuniv-lille1frprojectscalxtractproducts

2 Methods

21 Satellite data sets

This study mainly uses cloud and aerosol property retrievalproducts from MODIS on Aqua AMSR-E on Aqua theCloud Profiling Radar (CPR) on CloudSat and the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) onthe Cloud-Aerosol Lidar and Infrared Pathfinder SatelliteObservation (CALIPSO) All of these satellites operate inthe framework of the A-Train constellation (LrsquoEcuyer and

Jiang 2010 Stephens et al 2002) Considering most of thewarm rainfall occurs in the marine areas (Muumllmenstaumldt et al2015) and that satellite retrievals often suffer large uncertain-ties in the polar regions (Seethala and Horvaacuteth 2010) thestudy region is limited to 60 S to 60 N over global oceanscovering the period June 2006 to April 2011 Since MODIScloud LWP retrieval is only available for daytime we restrictour analysis to clouds observed in daytime (1330 LT) TheMODIS cloud product and CPR radar reflectivity observa-tions used in this study are both provided from the caltrack

wwwatmos-chem-physnet1817632018 Atmos Chem Phys 18 1763ndash1783 2018

data sets which resample observations from many sensorsunder CALIOP subtrack with the horizontal resolution of5 km (see httpwwwicareuniv-lille1frprojectscalxtractproducts for more information) For other aerosol and cloudproducts including MODISCALIOP aerosol products andAMSR-E cloud products they are further collocated into theCALIOP subtracks in the caltrack data set For each cloudpixel in the caltrack data set the closest aerosolcloud re-trieval sample within a 1 grid box (1times 1) centered at thiscaltrack cloud pixel is chosen To reduce the uncertainty incloud retrievals only samples where MODIS cloud fractionis equal to 100 are selected The main satellite data setsused in this study are briefly listed in Table 2

211 AI and CDNC

Three aerosol products are used in the study the MODISLevel 3 daily mean atmosphere product (MYD08_D3 Col-lection 6) MODIS Level 2 aerosol product (MYD04_L2Collection 6) and CALIOP Level 2 aerosol layer prod-uct (CAL_LID_L2_05kmALay Version 301) The 1

daily mean product of MYD08_D3 is aggregated fromMYD04_L2 with 10 km horizontal resolution (Hubankset al 2016) This MODIS Level 3 data set has been usedin previous studies to examine aerosolndashcloudndashprecipitationinteractions (eg LrsquoEcuyer et al 2009 Wang et al 2012)and is compared here with results from the MODIS Level2 aerosol product to examine how aerosol homogeneitymight affect precipitation susceptibility estimates The hor-izontal resolution of column aerosol optical depth from theCAL_LID_L2_05kmALay product is 5 km Aerosol prop-erty in this data set is obtained by averaging the 16 aerosolextinction profiles with 333 m of native resolution alongtrack (Young and Vaughan 2009)

Since AI is a better proxy for CCN concentrations ascompared to AOD (Nakajima et al 2001) AI is calcu-lated as one of the proxies for CCN based on the defini-tion of AI= AODtimesAE where AOD and AE are aerosoloptical depth and the Aringngstroumlm coefficient respectively ForMODIS AODs at 055 microm reported from MYD08_D3 andMYD04_L2 products are based on the Dark Target algo-rithm over the ocean (Kaufman et al 1997 Tanreacute et al1997 Levy et al 2013) For CALIOP AOD at wavelengthof 0532 microm is obtained from the CAL_LID_L2_05kmALayproduct (Vaughan et al 2004) Unlike MODIS AE whichis directly reported in aerosol products AE measurement forCALIOP is calculated based on AOD at 1064 and 0532 micromfrom the CAL_LID_L2_05kmALay product (Breacuteon et al2011) Our data screening for CAL_LID_L2_05kmALayfollows a previous study by Kim et al (2013)

Three aerosol products used in this study are listed in Ta-ble 2 It should be noted that all aerosol samples are undercloud-free conditions and are selected in close proximity tocloud pixels Retrievals of aerosol properties from passivesensors and lidar observation are both affected by clouds near

the aerosol and thereby result in overestimation for aerosolproperties (Chand et al 2012 Christensen et al 2017 Tack-ett and Di Girolamo 2009) The extent of this overestima-tion may be different among different sensors and dependson how far aerosol pixels are chosen from the correspond-ing cloud pixels (Christensen et al 2017) This effect how-ever would likely impact all metrics in a similar way andwe would not expect this effect to impact qualitative com-parisons between different metrics

CDNC is derived from the cloud optical thickness τ andcloud top effective radius reff both reported in the MODISLevel 2 cloud product (namely MYD06_L2) based on thefollowing formula (Bennartz 2007 Quaas et al 2006)

CDNC= ατ 05rminus25eff (4)

where the coefficient α = 137times 10minus5 mminus05 is estimatedbased on the assumption that cloud vertical structure followsthe classic adiabatic growth model (Quaas et al 2006) Toreduce the uncertainty when deriving CDNC cloud pixels(identified by the caltrack-MODIS cloud product with thehorizontal resolution of 5 km) where cloud optical depth isless than 3 and cloud fraction is less than 100 are excluded(Cho et al 2015 Zhang and Platnick 2011) Additionallywe limit our analysis to warm clouds by screening cloud pix-els with cloud top temperature warmer than 273 K Underthese screening criteria our results show that 94 of warmclouds are single layered (93 in Kubar et al 2009) There-fore our analysis mainly focuses on single-layer clouds

212 LWP

Cloud LWP for MODIS is diagnosed from solar reflectanceobservations of reff and τ as (Platnick et al 2003)

LWP= aρwτreff (5)

where ρw denotes the liquid water density and a is a con-stant determined by the assumed vertical variation in clouddroplet size (Greenwald 2009) For a vertically homoge-neous cloud a = 23 (Bennartz 2007) and a = 59 whenthe adiabatic assumption is applied (Szczodrak et al 2001)A recent study by Miller et al (2016) provides a system-atic investigation of the impacts of cloud vertical structure onMODIS LWP retrievals To be consistent with the adiabaticassumption used in Eq (4) for estimating CDNC a = 59 isapplied here

The other LWP retrieval comes from the AMSR-E Level2B Global Swath Ocean Product (Wentz and Meissner2004) Unlike retrieving from solar reflectance of visi-ble near-infrared (VNIR) for MODIS LWP for AMSR-Eis directly derived from brightness temperatures based onliquid-sensitive 37 GHz channel measurements (Seethala andHorvaacuteth 2010) More information on the retrieval techniqueof AMSR-E LWP is documented in Wentz and Meissner(2000) Horizontal resolution of the AMSR-E LWP prod-

uct (12 km) is also different from the MODIS LWP product(5 km)

213 Precipitation

Precipitation data sets used in this study are derivedfrom three different products from the CloudSat CPRnamely 2B-GEOPROF 2C-PRECIP-COLUMN and 2C-RAIN-PROFILE All the estimates are limited to cloudy pro-files by using 2B-GEOPROF cloud mask which is set togreater than 20 (King et al 2015) For the 2B-GEOPROFproduct (Marchand et al 2008) the maximum radar reflec-tivity for each cloudy profile is used to define rain events andto estimate rain rate More specifically rain rate is obtainedby employing the reflectivityndashrainfall (ZndashR) relationship atcloud base (Z = 25R13 from Comstock et al 2004) anda radar reflectivity threshold is used to distinguish betweendrizzling and non-drizzling clouds (Terai et al 2012 2015)

The empirical ZndashR relationship however does not ac-count for multiple scattering by raindrops and attenuationdue to both gases and hydrometeors which poses major chal-lenges for calculation of rain rate especially surface rain rate(Lebsock and LrsquoEcuyer 2011) To address those challengesHaynes et al (2009) introduced a full rainfall retrieval algo-rithm which is the basis of the 2C-PRECIP-COLUMN prod-uct The algorithm first makes use of path-integrated attenu-ation (PIA) derived from measurements of radar backscat-ter over the ocean surface in conjunction with surface windspeed and sea surface temperature Surface rain rate is thenestimated based on a simple algorithm using the PIA For the2C-PRECIP-COLUMN product a rain event is identified byusing rain likelihood mask Here we use the flag of ldquoraincertainrdquo to define a rain event which means attenuation-corrected reflectivity near the surface is above 0 dBZ (Hayneset al 2009)

2C-PRECIP-COLUMN assumes a constant vertical rainprofile in the precipitating column (Haynes et al 2009)which may not be suitable for warm rain where vertical vari-ation of the rain profile is significant (Lebsock and LrsquoEcuyer2011) To address this issue CloudSat developed a thirdrain product 2C-RAIN-PROFILE that utilizes the completevertically resolved reflectivity profile observed by the CPRand incorporates a subcloud evaporation model 2C-RAIN-PROFILE also uses MODIS cloud visible properties to con-strain cloud water in its retrieval algorithm (Lebsock andLrsquoEcuyer 2011) Note that the 2C-RAIN-PROFILE algo-rithm directly uses the precipitation occurrence flag from 2C-PRECIP-COLUMN to define rain events Thus the prob-ability of precipitation (POP) is the same for both rainproducts Note that surface rain rates are only retrieved forthose pixels that identified as rain certain in the 2C-RAIN-PROFILE product (Lebsock and LrsquoEcuyer 2011) Overallthree rain rate data sets in this study are significantly dif-ferent rain rate directly estimated from 2B-GEOPROF rep-resents the maximum rainfall rate precipitation from 2C-

PRECIP-COLUMN is the column-mean rainfall rate andrain rate from 2C-RAIN-PROFILE stands for surface rain-fall rate

22 Meteorological data sets

Aerosolndashcloudndashprecipitation interactions and precipitationsusceptibility have been shown to depend on cloud regimes(LrsquoEcuyer et al 2009) Following Klein and Hartmann(1993) we use the lower-tropospheric static stability (LTSS)which is defined as the difference in potential temperaturebetween 700 hPa and the surface to separate different atmo-sphere thermodynamic regimes In this study unstable andstable environments are defined as LTSS less than 135 K andLTSS larger than 18 K respectively Pixels where LTSS isbetween 135 and 18 K are defined as the midstable environ-ment (Wang et al 2012) The European Centre for Medium-Range Weather Forecasts auxiliary (ECMWF-AUX) prod-uct as an ancillary CloudSat product that contains temper-ature and pressure within each CPR bin is used to calculateLTSS

23 Precipitation susceptibility calculation

Following previous studies (Feingold and Siebert 2009Sorooshian et al 2009 Wang et al 2012 Terai et al 20122015) precipitation susceptibility is generally defined as

SX_Y =minusdlnXdlnY

where X can be substituted by POP (precipitation fre-quency) I (precipitation intensity) or R (R = POPtimes I precipitation rate) and Y indicates AI or CDNC Conse-quently six different precipitation susceptibilities can be ob-tained from the observations described above To constrainthe cloud macrophysical environment all samples are sortedaccording to their LWP values first and then divided into10 LWP bins The ratio of the number of pixels in each binto the total pixels ranges from 5 to 14 For each LWP binsamples are sorted by AI or CDNC and 10 AICDNC binsare equally divided to calculate POP mean I R AI andCDNC within each AICDNC bin Finally the values of SX_Yare derived by linear regression in logndashlog space

3 Results

31 SX_AI vs SX_CDNC

SX_AI and SX_CDNC as a function of LWP are shown in Fig 1Here LWP from MODIS and rain data from 2B-GEOPROFwith a rain threshold of minus15 dBZ are used to better com-pare with other satellite studies (Terai et al 2015 Michibataet al 2016) Here AI is estimated by using the MYD04 dataset and detailed comparison among different aerosol prod-ucts will be made in Sect 32

MODIS LWP [g m-2]50 150 250 350 450 550

Figure 1 SPOP SI and SR as a function of MODIS LWP with (a) AI and (b) CDNC Red squares green upward triangles and blue downwardtriangles stand for SR SPOP and SI respectively Error bars are based on 95 confidence intervals for the susceptibility estimates AI isderived from MYD04MODIS and CDNC is estimated from MYD06MODIS Intensity and probability of precipitation are based on the2B-GEOPROF product with a minus15 dBZ threshold The total amounts of data samples for the left panel and right panel are about 21 and 31million

Consistent with previous studies SX_AI are generallymuch smaller than SX_CDNC as shown in Fig 1 SPOP_AIfrom Wang et al (2012) is less than 02 over all LWP binswhile Terai et al (2015) showed that SPOP_CDNC decreaseswith increasing LWP ranging from 1 to 0 and SR_CDNC ismaintained at around 05 Figure 1b further shows SI_CDNCmonotonically increases with LWP followed by a slight de-crease Although the SI_CDNC peak (around 06 with LWP350 gmminus2) is not significant in Fig 1b SI_CDNC would de-crease distinctly after the peak if the upper bound of LWPand the number of LWP bins both increase (not shown) Thisturning point may correspond to the conversion process shift-ing from the autoconversion to accretion regime (Michibataet al 2016)

To account for discrepancy between SX_AI and SX_CDNCwe use the condition probability method (Gryspeerdt et al2016) to explore relationships between AI and CDNC AsFig 2a shows the majority of CDNC values concentrate onthe intervals between 20 to 100 cmminus3 representing an up-ward tendency with increasing AI over global oceans Thesimilar feature of CDNC with AI is also shown in differ-ent LTSS conditions (Fig 2bndashd) Note that fluctuation of thecurve at high AI results from the small number of effectivepixels especially in the unstable condition

To formally account for the relationship between CDNCand AI SX_AI can be decomposed into two parts

SX_AI =minusdlnXdlnAI

=minusdlnX

dlnCDNCdlnCDNC

= SX_CDNCdlnCDNC

dlnAI (7)

where dlnCDNCdlnAI is the link between SX_AI andSX_CDNC SX_AI is expected to be smaller than SX_CDNCif d lnCDNCdlnAI is smaller than 1 Figure 3 shows

dlnCDNCdlnAI over global oceans which is calculatedby logndashlog linear regressions in each MODIS LWP bind lnCDNCdlnAI is smaller than 04 which explains whySX_AI is generally smaller than SX_CDNC Table 3 furthershows the LWP-weighted mean of d lnCDNCdlnAI SX_AIand SX_CDNC over global oceans Our results are consis-tent with the previous satellite observations For instanceSPOP_AI is equal to 011 in our results obtained from AMSR-E LWP close to the value of 012 in Wang et al (2012)and our SR_CDNC derived from MODIS LWP is 074 simi-lar to that (06) in Terai et al (2015) Since the global meandlnCDNCdlnAI is about 03 we would expect SX_AI tobe about one-third of SX_CDNC according to Eq (7) Table 3shows that this relationship is generally true for SPOP but lessso for SI especially for SI calculated based on MODIS LWP

Table 3 further demonstrates that SR asymp SI+ SPOP is gen-erally true for different LWP products and over different sta-bility regimes consistent with Terai et al (2015)

32 SX_AI from different aerosol products

Now we explore how precipitation susceptibility esti-mates might be different from different aerosol products(ie MYD04 MYD08 and CAL_LID_L2_05kmALay) Asshown in Fig 4 despite differences in their horizontal res-olutions (10 km vs 1) SX_AI calculated from MYD04 andMYD08 agrees well (Fig 4a and b) which may result fromthe fact that aerosol layers are likely homogeneous overrelatively large spatial scales less than 200 km (Andersonet al 2003) especially over global oceans In additionMcComiskey and Feingold (2012) found that the statistics(ie min max and variance) of AOD are constant betweenMYD04 and MYD08 products over the northeast PacificOcean for a given day Although not shown here the proba-

Figure 2 The probability of CDNC under given AI over the (a) global ocean (b) unstable (c) stable and (d) midstable conditions In eachfigure the red line represents change in average CDNC with AI and the lower and upper lines stand for mean CDNC for the 25th percentileand 75th percentile of samples AI and CDNC are estimated from MYD04 and MYD06 respectively

Table 3 The LWP weighted-mean values of precipitation susceptibility SX_Y and dlnCDNCdlnAI over global oceans under differentstability regimes The statistics are based on the 2B-GEOPROFCPR product using the cloud baseZndashR relationship and aminus15 dBZ threshold

SR_AI SI_AI SPOP_AI SR_CDNC SI_CDNC SPOP_CDNC dlnCDNCdlnAI

MODIS LWP Global 005 minus002 008 074 047 044 028Unstable minus004 minus009 004 052 030 026 022Stable 022 013 012 084 048 060 030Midstable 001 minus005 007 066 039 035 029

AMSR-E LWP Global 017 007 011 047 016 037 032Unstable 014 006 008 021 004 018 025Stable 029 017 015 067 023 055 033Midstable 013 004 010 040 013 029 034

bility distributions of AI derived from MYD04 and MYD08products are qualitatively similar over global oceans In com-parison with the results based on MODIS retrievals SX_AI

obtained from CALIOP (Fig 4c) is smaller and relatively flatacross all LWP bins Further testing shows that SX_AI usingCALIOP AOD but MYD04 AE agrees better with that based

LWP [g m-2]50 150 250 350 450 550

Global oceanUnstable conditionStable conditionMidstable condition

Figure 3 dlnCDNCdlnAI obtained by linear regression oflnCDNC and lnAI under MODIS LWP bins Red lines denotethe global ocean Green blue and cyan stand for unstable stableand midstable conditions respectively AI and CDNC are estimatedfrom MYD04 and MYD06

on MODIS aerosol products (Fig 4d) This suggests that dif-ferences in AE estimates from MODIS and CALIOP largelyexplain the discrepancy between two aerosol products Pre-vious studies indicate that MODIS and CALIOP AOD arepoorly correlated (eg Costantino and Breacuteon 2010 Kimet al 2013 Kittaka et al 2011 Ma et al 2013) Our re-sults suggest that differences in AOD retrievals can lead todifferences in AE estimates and further affect AI and precip-itation susceptibly estimates Given that AI from MODIS hasbeen widely used in previous studies for examining aerosolndashcloudndashprecipitation interactions for the rest of the paper AIfrom MYD04 is used unless otherwise stated

33 SX_Y from different LWP data sets

Figure 5 shows the behavior of SPOP and SI based on differentLWP data sets (ie AMSR-E and MODIS LWP) Estimatesof rain rate and rain events are based on 2B-GEOPROF witha minus15dBZ threshold as mentioned in Sect 213 Here wefocus on characteristics of SPOP and SI since SR asymp SI+SPOPas mentioned in Sect 31 As shown in Fig 5a SPOP_CDNCbased on MODIS LWP is similar to that calculated based onAMSR-E LWP This consistency is also found for SPOP_AIIn contrast SI_CDNC and SI_AI calculated based on twoLWP products are quite different (Fig 5b) SI_CDNC basedon MODIS LWP is significantly larger than that based onAMSR-E LWP over all LWP bins (see upward triangles inFig 5b) while SI_AI from two LWP products shows an op-posite pattern SI_AI based on MODIS LWP is lower than thatbased on AMSR-E LWP (see downward triangles in Fig 5b)These features of discrepancies in SI between MODIS andAMSR-E LWP are still applicable to SPOP though the mag-

nitude is much smaller and is not statistically significant(Fig 5a)

Figure 5b shows that LWP value where SI_CDNC peaksbased on MODIS LWP is larger than that based on AMSR-ELWP Large eddy simulation analysis by Duong et al (2011)showed a similar shift in LWP with changing spatial res-olutions which is attributed to reduction in mean LWP atcoarser resolutions However Fig 6 shows that there is nosystematic shift in the frequency distribution of LWP be-tween two LWP products regardless of precipitation or non-precipitation samples

To better understand the discrepancy in precipitation sus-ceptibility estimates from two LWP products in Fig 5 weplot POP and intensity as a function of CDNCAI in logspace for each LWP bin obtained from MODIS and AMSR-E Figure 7andashd show that the relationships between POP andCDNC (AI) from MODIS LWP are similar to those fromAMSR-E LWP In contrast intensity vs CDNC (AI) be-tween two LWP products shows large differences (Fig 7endashh) Figure 7f shows that intensity is positively correlatedwith CDNC at low CDNC for high AMSR-E LWP binswhich helps to explain why SI_CDNC from AMSR-E LWP issmaller than that from MODIS especially at high LWP bins(Fig 5b)

Combining Eqs (4) and (5) CDNC from MODIS can bereformulated as a function of LWP and reff

CDNC= α(aρw)minus05LWP05rminus3

eff (8)

where α a and ρw are all constant Accordingly reff de-creases with increasing CDNC for any given MODIS LWPbin and larger CDNC leads to smaller reff which furtherresults in reduction in precipitation efficiency as shown inFig 7e The CDNCndashreff relationship still holds when data arebinned by AMSR-E LWP and reff decreases with increasingCDNC even at larger LWP AMSR-E LWP bins (Fig 8a)We would then expect that rain intensity still decreases withincreasing CDNC for the AMSR-E LWP at low CDNC Sothen what might lead to increases in precipitation intensitywith increasing CDNC at low CDNC when data are binnedaccording to constant AMSR-E LWP (Fig 7f) Our analysissuggests that this might come from the discrepancies in twoLWP products under low CDNC Figure 8b shows that underconstant AMSR-E LWP MODIS LWP significantly varieswith CDNC (Fig 8b) In particular MODIS LWP rapidly in-creases with CDNC at low CDNC which might explain whyrain intensity increases with increasing CDNC at low CDNCunder constant AMSR-E LWP which further leads to muchsmaller SI_CDNC from AMSR-E LWP Our results further in-dicate that rain intensity retrieval from CloudSat might bemore consistent with LWP retrieval from MODIS than thatfrom AMSR-E as under constant AMSR-E LWP rain inten-sity increases with increasing MODIS LWP at low CDNC(Figs 7f and 8b)

It is interesting to note that for rainy pixels the differ-ence in LWP between MODIS and AMSR-E varies with

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP AOD but MYD04 AE

Figure 4 Susceptibilities (SX_AI) as a function of MODIS LWP The rain product used is the same as that in Fig 1 AI is derived from (a)MYD04MODIS (b) MYD08MODIS and (c) CAL_LID_L2_05kmALayCALIOP products Panel (d) is the same as (c) but using MYD04AE

LWP [g m-2]50 150 250 350 450 550

LWP-gtMODISLWP-gtAMSR-E

Figure 5 (a) SPOP_Y and (b) SI_Y as a function of LWP The subscript Y denotes different aerosol proxies corresponding to AI (downwardtriangles) and CDNC (upward triangles) Blue (red) represents LWP derived from MODIS (AMSR-E) The rain product used is the same asthat in Fig 1 AI and CDNC are estimated from MYD04 and MYD06 respectively

MODIS CDNC Under constant AMSR-E LWP (larger than200 gmminus2) MODIS LWP dramatically increases with in-creasing CDNC at lower CDNC (ltsim 25 cmminus3) These fea-tures are also applicable to non-rainy samples (not shown)Further studies are needed to understand the aforementioneddiscrepancy

34 SX_Y from different rainfall definitions

Given that rainy samples may be dominated by differentprecipitation process (eg autoconversion vs accretion pro-cesses) with increasing threshold for defining a rainfallevent (Jung et al 2016) precipitation susceptibility may

LWP [g m-2]0 100 200 300 400 500 600 700

MODISMODIS precipMODIS nonprecipAMSR-EAMSR-E precipAMSR-E nonprecip

Figure 6 Distribution of frequency of LWP derived from MODISand AMSR-E under different scenarios namely all samples non-precipitation and only-precipitation samples

be changed when we apply different rainfall thresholds Toexamine this we plot SPOP and SI under different thresh-olds (ie minus15 and 0 dBZ of maximum radar reflectivity)used to define a rain event based on 2B-GEOPROF productsThese thresholds of minus15 and 0 dBZ correspond to approxi-mate precipitation rates of 014 and 2 mmdminus1 respectively(Comstock et al 2004) Hence precipitation susceptibili-ties under these two thresholds can be referred to as drizzle(gtminus15 dBZ) and rain (gt 0 dBZ) susceptibilities As Fig 9indicates the difference in SX_AI between drizzle and rain isat first glance less evident compared to SX_CDNC This canbe partly attributed to the low values of SX_AI themselvesRelative differences in SX_AI are even larger than those ofSX_CDNC at low AMSR-E LWP (not shown) Figure 9a and cshow that rain SPOP_AI is higher than that of drizzle overmost LWP bins which is consistent with results from Wanget al (2012)

The rainfall definition significantly impacts SPOP_CDNCand SI_CDNC increasing the threshold results in reduc-tion of SI_CDNC over all LWP and by contrast leads toa distinct increase in SPOP_CDNC especially at moderateLWP (see Fig 9) These overall changes in SI_CDNC andSPOP_CDNC after increasing the threshold are consistent withTerai et al (2015) The observational results from Mannet al (2014) have also shown an evident increase in SPOPwith respect to NCCN at high LWP with increasing thresh-olds The systematic increase in SPOP_CDNC may result froma larger proportion of non-drizzling samples with an increas-ing threshold The reduction of SI_CDNC is in agreement withprevious studies (Duong et al 2011 Jung et al 2016) Al-though not shown here for a fixed threshold there is nosignificant discrepancy between the results of SI_CDNC andSR_CDNC based on different ZndashR relationships (Z = 25R13

and Z = 302R09 are used both from Comstock et al 2004

which aim at cloud base and surface rain rate respectively)which is consistent with the result from Terai et al (2012)

Overall our results show that SPOP and SI are both sen-sitive to the rainfall definition and that SPOP is greater forrain while SI is greater for drizzle Our results further im-ply that the onset of drizzle is not as readily suppressed inwarm clouds as rainfall (ie SPOP is greater for rain than fordrizzle) By contrast SR is not affected by the rainfall def-inition since the mean rain rate R for a given LWPCDNCor LWPAI bin is calculated based on both rainy and non-rainy clouds and does not depend on rainfall thresholds (notshown)

While the response of precipitation susceptibility tochange in threshold shows the same pattern between MODISand AMSR-E LWP (Fig 9) the extent to which susceptibil-ity changes with an increasing threshold is quite different be-tween these LWP products Overall sensitivity of SX_CDNCto different thresholds using MODIS LWP is larger than thatbased on AMSR-E LWP this pattern is opposite for SX_AIIt is interesting to note that while the difference in SPOP be-tween MODIS and AMSR-E LWP is small with theminus15 dBZthreshold (Figs 5 9a and c) the difference is relatively largerfor the 0 dBZ threshold (Fig 9a and c) especially at largerLWP bins

35 SX_Y from different precipitation data sets

The diverse rain data sets allow us to explore differencesin precipitation susceptibility estimates from different rainproducts In Fig 10 we illustrate SPOP and SI for dif-ferent rain data sets namely 2B-GEOPROF 2C-PRECIP-COLUMN and 2C-RAIN-PROFILE (Marchand et al 2008Haynes et al 2009 Lebsock and LrsquoEcuyer 2011) productsHere we use LWP derived from MODIS and use the raincertain flag for rain definition reported in the latter two rainproducts Since the precipitation flags used in these two rainproducts are identical (Lebsock and LrsquoEcuyer 2011) onlySPOP based on 2C-PRECIP-COLUMN is plotted in Fig 10For 2B-GEOPROF the threshold of 0 dBZ radar reflectiv-ity is used to define a rain event and rain rate is estimatedby using Z = 25R13 suggested by Comstock et al (2004)Note that using the rain certain flag or a threshold of 0 dBZto identify rain events for those rain products would resultin a reduction of rain events across all LWP bins especiallyat low LWP bins therefore we expand bounds of low LWPbins to include enough rain samples at low LWP binsSPOP exhibits a similar dependence on LWP among these

three rain products but SPOP based on 2B-GEOPROF is sys-tematically larger than that based on 2C-PRECIP-COLUMN(this is also true for SI) It is unclear what might lead to higherSPOP and SI from 2B-GEOPROF The vertical structure ofclouds may play a role here as the maximum radar reflec-tivity is used from 2B-GEOPROF and surface rain rates areused from the other products

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

Figure 7 POP and I as a function of CDNC (AI) for each LWP bin obtained from (a c e g) MODIS and (b d f h) AMSR-E The dataused here are the same as those in Fig 5

The most significant discrepancy occurs in SI_CDNC andSI_AI (see Fig 10b) Figure 10b shows that SI_CDNC and SI_AIare both near zero for LWPlt 400 gmminus2 which may be at-tributed to high thresholds used among the three rain prod-

ucts This indicates that precipitation intensity with a highthreshold is insensitive to CDNC and AI at moderate LWPThis result is consistent with Terai et al (2015) who sug-gested that heavy drizzle intensity is insensitive to CDNC As

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

Figure 8 (a) reff and (b) MODIS LWP as a function of CDNC for each AMSR-E LWP bin Only rainy samples defined by a minus15 dBZthreshold are used in here Different color lines represent different AMSR-E bins corresponding to Fig 7f

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

Y-gtAI threshold of -15 dBZY-gtAI threshold of 0 dBZ

AMSR-E LWP [g m-2]50 150 250 350 450 550

Y-gtCDNC threshold of -15 dBZY-gtCDNC threshold of 0 dBZ

Figure 9 (a c) SPOP_Y and (b d) SI_Y as a function of LWP The subscript Y denotes different aerosol proxies corresponding to AI(downward triangles) and CDNC (upward triangles) The 2B-GEOPROF product is used here Blue and red symbols represent minus15 dBZ and0 dBZ thresholds respectively The top and bottom panels stand for MODIS and AMSR-E LWP respectively

Fig 10b shows SI_CDNC based on the 2C-RAIN-PROFILEproduct (red squares in Fig 10b) with subcloud evapora-tion model incorporated is higher than that based on the2C-PRECIP-COLUMN product (blue squares in Fig 10b)at high LWP (above 300 gmminus2) Hill et al (2015) showedthat when considering rain evaporation SI_CDNC based onsurface rain rate is larger than that based on cloud baseand column maximum rain rate at LWPgt 400 gmminus2 How-ever their difference is more obvious than our results which

may result from the threshold used (001 mm dayminus1 in Hillet al (2015) vs surface the 0 dBZ in 2C-RAIN-PROFILEand 2C-PRECIP-COLUMN products) It is interesting tonote that the sign of SI_CDNC at large LWP is different fromthat of SI_AI (Fig 10b) which is not true for AMSR-E LWP(not shown) This warrants further investigation in the future

MODIS LWP [g m-2]50 150 250 350 450 550

Rain data-gt2B-GEOPRain data-gt2C-COLURain data-gt2C-PROF

MODIS LWP [g m-2]50 150 250 350 450 550

Figure 10 (a) SPOP_Y and (b) SI_Y as a function of MODIS LWP The subscript Y denotes different aerosol proxies corresponding toAI (point) and CDNC (square) Different color symbols stand for different rain products 2B-GEOPROF (2B-GEOP green) 2C-PRECIP-COLUMN (2C-COLU blue) and 2C-RAIN-PROFILE (2C-PROF red) See text for further details

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of -15 dBZ

Y-gtAI unstableY-gtAI midsatableY-gtAI stable

MODIS LWP [g m-2]50 150 250 350 450 550

Y-gtCDNC unstableY-gtCDNC midsatableY-gtCDNC stable

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

Figure 11 (a c) SPOP_Y and (b d) SI_Y as a function of MODIS LWP The subscript Y denotes different aerosol proxies correspondingto AI (point) and CDNC (square) Blue red and green symbols stand for stable unstable and midstable regimes respectively Rain datacome from 2B-GEOPROF The top panels are for results based on the rain threshold of minus15 dBZ and the bottom panels are based on the rainthreshold of 0 dBZ

36 SX_Y under different stability regimes

Here we examine precipitation susceptibility under dif-ferent atmospheric stability regimes as aerosolndashcloudndashprecipitation interactions have been shown to differ under

different stability regimes (eg LrsquoEcuyer et al 2009 Zhanget al 2016 Michibata et al 2016) Based on the MODISLWP and 2B-GEOPROF product with a minus15 dBZ thresholdFig 11a and b suggest that both SPOP and SI increase witha more stable environment This pattern for SPOP is consis-

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

Figure 12 Distribution of (andashb) SPOP_Y and (cndashd) SI_Y as a function of MODIS LWP and LTSS Rain data are from 2B-GEOPROF with athreshold of minus15 dBZ Each LTSS bin contains on average the same amount of pixels

tent with the findings of LrsquoEcuyer et al (2009) who showedthat suppression of precipitation was largest at lower LWP instable environments Terai et al (2015) also found maximumSPOP_CDNC occurred in regions where a stable regime waspredominant The distribution of the precipitation suscepti-bility with respect to LTSS and LWP shown in Fig 12 usingthe 2B-GEOPROF product with the minus15 dBZ rain thresholdis consistent with Fig 11a and b SPOP increases with in-creasing LTSS with the exception of high LWP Although notshown here SPOP_AI based on the 2C-PRECIP-COLUMNand AMSR-E LWP product produces a similar pattern withthe result of (LrsquoEcuyer et al 2009) who showed the slopebetween POP and AI is small both at low and high LWP butthis magnitude tends to increase at intermediate LWP andhigh LTSS

Rain definition significantly affects spread of SPOP and SIunder different stability regimes As rain threshold increasesthe discrepancy in SPOP among different LTSS conditions ismore significant (Fig 11c vs a) while discrepancy in SI be-comes smaller LTSS dependence of SI is even reversed atlow LWP with the 0 dBZ threshold compared to that usingthe minus15 dBZ threshold (Fig 11d vs b)

The abovementioned features of LTSS dependency arealso true in terms of LWP-weighed mean value as shownin Fig 13 For all those cases based on different rain prod-ucts and LWP products the LWP-weighed mean of SPOP isgenerally larger under stable conditions compared with un-stable conditions Yet this feature does not hold true for SIexcept the case based on the 2B-GEOPROF data set with theminus15 dBZ threshold Our results also suggest that it is impor-tant to account for the influence of atmospheric stability dueto the clear dependence of SPOP on metrics like LTSS thoughit is acknowledged that LTSS alone is an imperfect metric forisolating cloud regimes (eg Nam and Quaas 2013) Dif-ferent metrics associated with cloud regimes should be ex-amined in the future to better understand the effect of cloudregimes on precipitation susceptibility For instance LTSScan be combined with vertical pressure velocity to distin-guish between different cloud types (Zhang et al 2016)

4 Discussion

Figure 14 shows the range of precipitation susceptibility es-timated from different LWP and rain products Here the

Case1 Case2 Case3 Case4

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

05 Case1-gt2B-GEOP with -15 dBZCase2-gt2B-GEOP with 0 dBZ

Case3-gt2C-COLUCase4-gt2C-PROF

AMSR-E LWP

Global meanstableMidstableUnstable

Figure 13 The LWP-weighed mean values of (a b e f) SPOP and (c d g h) SI under different stability regimes for four cases Case1 andcase2 are both based on the 2B-GEOPROF product but use the threshold of minus15 and 0 dBZ respectively Case3 and case4 use 2C-PRECIP-COLUMN and 2C-RAIN-PROFILE products respectively The top two panels use MODIS LWP and the bottom two panels use AMSR-ELWP Error bars are based on the LWP-weighed mean values of 95 confidence intervals for the susceptibility estimates

threshold of 0 dBZ of maximum radar reflectivity is used forthe 2B-GEOPROF product and the rain certain flag is usedfor 2C-PRECIP-COLUMN and 2C-RAIN-PROFILE prod-ucts It shows that uncertainties in SPOP (Fig 14a) as a resultof using different LWP andor rain products are smaller thanthe uncertainties associated with SI and SR (Fig 14b and c)The uncertainties in SPOP are mainly attributed to differentLWP products as described in Sect 34 (see red symbols inFig 9a and c)

Our results may help to reconcile some of the differ-ences in previous estimates of precipitation susceptibilityFor example our results show that SX_AI asymp 03SX_CDNC(Table 3 and Fig 1) which explains why SPOP_CDNC inTerai et al (2015) is much larger than SPOP_AI in Wanget al (2012) Previous studies are also different in how pre-cipitation susceptibility varies with increasing LWP Our re-sults show that SI generally increases with LWP at low andmoderate LWP and then decreases with increasing LWPat moderate and high LWP consistent with results from

LWP [g m-2]100 200 300 400 500

Proxy for CDNCProxy for AI

LWP [g m-2]100 200 300 400 500

Figure 14 (a) SPOP_Y (b) SI_Y and (c) SR_Y as a function of LWP The subscript Y denotes different aerosol proxies corresponding to AI(light green) and CDNC (light blue) Shade areas show the range of precipitation susceptibility from different rain products (same as Fig 10)and different LWP products (MODIS and AMSR-E)

Feingold et al (2013) Michibata et al (2016) and Junget al (2016) The monotonic increase of SI_CDNC with in-creasing LWP in Terai et al (2015) is mainly because theLWP range in their study is relatively narrow (from 0 tosim 400 gmminus2) and our results suggest that when the upperbound of LWP is extended to sim 800 gmminus2 the ldquodescendingbranchrdquo (S decreases with increasing LWP) noted in Fein-gold et al (2013) appears though the exact LWP value whereSI_CDNC peaks depends on LWP and rain products used aswell as the rainfall threshold choices

Interestingly SI tends to be negative at low LWP bothfor AMSR-E and MODIS LWP (Fig 5b) This is closelyassociated with positive correlation between conditional-mean rainfall intensity and CDNC (AI) at low LWP binswhere CDNC (AI) is high (Fig 7endashh) More negative val-ues are captured when SI is estimated using 2C-PRECIP-COLUMN and 2C-RAIN-PROFILE products and using highrainfall thresholds (Figs 9b d 10b and 13) Furthermorebased on these rain products SI_CDNC is all negative at lowand intermediate LWP regardless of the LWP data set used(Fig 10b) and almost all of mean SI_CDNC is significantlynegative regardless of stability regimes (Fig 13) Dependingon the LWP products adopted using AI instead of CDNCin estimating SI can make it less negative (for AMSR-ELWP) or more negative (for MODIS LWP) (Fig 13) Terai

et al (2015) also found negative values of SI_CDNC at lowLWP and high CDNC In their study the sign andor mag-nitude of SI_CDNC at low LWP are distinct across differentregions In addition Koren et al (2014) found a positiverelationship between AOD and rain rate over pristine areaswith warm and aerosol-limited clouds which was attributedto aerosol invigoration effect As SI shows large differencesunder different stability regimes (Fig 13) it would be highlyinteresting to analyze regional variation in SI to further un-derstand negative SI in the future especially under unstableregimes

Furthermore our results show that drizzle intensity ismore susceptible to aerosol perturbations than rain intensity(see Fig 9b and d) which might help to explain why nega-tive values of SI_CDNC occur more frequently with increas-ing rainfall thresholds Jung et al (2016) found more nega-tive values of SI_CDNC with increasing threshold (see Fig B2in Jung et al 2016) In addition rain products used in ourstudy are all derived from CPR aboard CloudSat With in-creasing thresholds rainfall becomes heavy and uncertaintyin rain rate retrieval can grow as CPR is insensitive to heavyprecipitation (Haynes et al 2009) Thus a combination ofdifferent rain satellite products (eg CloudSat and the Trop-ical Rainfall Measuring Mission ndash TRMM) would be helpfulfor better understanding negative SI

It should be noted that precipitation susceptibility in ourstudy is based on Eq (6) and is derived by linear regressionbetween precipitation fields and CDNCAI in logndashlog spaceThe negativepositive correlation between precipitation fre-quencyintensity and aerosols may not be readily explainedlike aerosol effects on precipitation For example a nega-tive correlation between precipitation frequency and aerosolsmay come from the wet scavenging effects of aerosols (moreprecipitation leads to less aerosols) but not aerosol suppres-sion of precipitation However in our study we not only cal-culate precipitation susceptibility with respect to AI (SX_AI)but also with respect to CDNC (SX_CDNC) and the latter isexpected to be less affected by the wet scavenging effectsThe broad consistency between these two estimates shownin our results (Fig 13) especially for the estimate of SPOPlends support to the limited influence of wet scavenging inour estimate Further support for this comes from the factthat precipitation susceptibility estimates based on the 1 L3MODIS aerosol products are similar to those based on the10 km L2 MODIS aerosol products (Fig 4) as we wouldexpect the wet scavenging effects are more important atsmaller scales if the wet scavenging effects are a dominatingfactor Nevertheless the effects of wet scavenging can still beimportant in satellite studies of aerosolndashcloudndashprecipitationinteractions and should be better quantified in future perhapsin combination with model simulations

5 Summary

In this paper we estimate precipitation susceptibility onwarm clouds over global oceans based on multi-sensoraerosol and cloud products from the A-Train satellites in-cluding MODIS AMSR-E CALIOP and CPR observationscovering the period June 2006 to April 2011 In additionto different aerosol cloud and rain products we also ana-lyze other factors that have potential influence on suscepti-bility such as different definitions of precipitation suscepti-bility (six different susceptibilities defined by Eq 6) stabil-ity regimes and different thresholds for defining a rain event(ie minus15 and 0 dBZ of maximum radar reflectivity for 2B-GEOPROF) The primary goal of the study is to quantifyuncertainties in precipitation susceptibility estimates fromsatellite observations

In general SPOP is a relatively robust metric throughoutdifferent LWP and rain products and its estimate is less sen-sitive to different data sets used (Figs 13 and 14) SPOP_CDNCshows overall a monotonic decreasing trend with respect toLWP SPOP_AI increases to a maximum at low LWP and thendecreases with higher LWP In contrast SI differs consider-ably among different LWP and rain products (Figs 13 and14) Interestingly SI_CDNC and SI_AI differ between thoseLWP products with the opposite pattern SI_CDNC based onMODIS LWP is higher than that using AMSR-E LWP andthe reverse is true for SI_AI (Fig 13) Negative SI is found

in our study especially at low LWP However the extentof negative SI depends on LWP and rain products rainfallthresholds and whether SI is calculated with respect to AI orCDNC (Fig 13) More negative values are found when SI iscalculated based on 2C-PRECIP-COLUMN and 2C-RAIN-PROFILE products and SI based on rain samples (with a0 dBZ threshold) tends to be more negative Further studies(regional variation in SI combination of different rain satel-lite products etc) are needed to understand this issue

Precipitation susceptibility for drizzle (with a minus15 dBZrainfall threshold) is significantly different from that for rain(with a 0 dBZ rainfall threshold) (Figs 9 and 13) Our re-sults suggest that onset of drizzle is not as readily suppressedby increases in AI or CDNC in warm clouds as rainfall (ieSPOP is smaller for drizzle than for rain especially at mod-erate LWP Fig 9) This may partly come from the fact thatPOP of drizzle is close to 100 at moderate and high LWPregardless of CDNC or AI values (Fig 7andashd) which makesit insensitive to perturbations in CDNC or AI and results insmaller SPOP at these LWP bins compared with SPOP for rain(Fig 9) On the other hand precipitation intensity suscep-tibility is generally smaller for rain than for drizzle This isconsistent with our expectation that when precipitation inten-sity increases accretion contributes more to the productionof precipitation which makes precipitation intensity less sen-sitive to perturbation in CDNC or AI as accretion is less de-pendent on CDNC compared with autoconversion (Feingoldet al 2013 Wood 2005) In addition the extent of thesedifferences between drizzle and rain depends on the LWPproducts usedSX_AI based on aerosol products at different spatial reso-

lutions (ie 10 km vs 1) is consistent with the others Chenet al (2014) also found that aerosol indirect forcing derivedfrom satellite observations was similar from AI observationsat different resolutions (ie 20 km vs 1) This suggests thataerosol layers over oceans are relatively homogeneous im-plying that aerosol properties at coarse resolution may besuitable for studying aerosolndashcloud interactions over oceansSPOP strongly depends on LTSS with larger values under

a more stable environment This dependence is evident overall LWP bins especially at low and moderate LWP bins andis more significant for rain than for drizzle (Figs 11 and 13)These features however are less robust for SI throughout dif-ferent LWP and rain products as SI estimates show large un-certainties from different data sets (Fig 13) Only in the caseof SI estimated from the 2B-GEOPROF product for drizzle(with a minus15 dBZ threshold) does the LTSS dependence ofSI hold for both MODIS and AMSR-E LWP The pattern ofSPOP_AI under different stability conditions from our paper(Fig 13b and f) is consistent with the findings of LrsquoEcuyeret al (2009) In addition Terai et al (2015) found that maxi-mum SPOP_CDNC occurred in regions where a stable regime ispredominant Lebo and Feingold (2014) calculated precipita-tion susceptibility for stratocumulus and trade wind cumulusclouds using large eddy simulations (LESs) and included an

overview of precipitation susceptibility estimates based onLESs in the literature However their results focus on the re-lationship between precipitation susceptibility and cloud wa-ter response to aerosol perturbations and did not examinehow precipitation susceptibility might be different for cloudsunder different cloud regimes The physical mechanisms be-hind the strong dependence of SPOP on stability are still un-clear and warrant further investigation in the future

The results presented here show that the discrepancy inmagnitude between SX_AI and SX_CDNC can be mainly at-tributed to the dependency of CDNC on AI On the globalscale our results show that SX_AI is about one-third ofSX_CDNC This relationship is more applicable to SPOP andis less applicable to SI In addition SR asymp SI+ SPOP is gener-ally true for different LWP products and over different LTSSconditions

As SPOP demonstrates relatively robust features across dif-ferent LWP and rain products this makes it a valuable met-ric for quantifying aerosolndashcloudndashprecipitation interactionsin observations and models For instance it would be highlyinteresting to examine why SPOP strongly depends on at-mospheric stability and how well this dependence is repre-sented in a hierarchy of models (eg large eddy simulationscloud resolving models regional climate models and globalclimate models) We also note that SPOP_CDNC is generallyless uncertain compared to SPOP_AI and that a relativelyrobust relationship between SPOP_CDNC and SPOP_AI exists(ie SX_AI asymp 03SX_CDNC) (Fig 13 and Table 3) Giventhat aerosol retrievals near clouds are still challenging andaerosolndashcloud relationships in satellite observations can beaffected by aerosol retrieval contaminations from clouds werecommend to first thoroughly quantify SPOP_CDNC in ob-servations and models As SPOP_CDNC is derived based onCDNC instead of AI SPOP_CDNC is also not influenced bywet scavenging Only after SPOP_CDNC is thoroughly quanti-fied can we then combine it with CDNC dependence on AIto better quantify SPOP_AI

On the other hand SI estimates strongly depend on satel-lite retrieval products Uncertainties in SI estimate are par-ticularly large when SI is estimated based on rain samples(gt 0 dBZ) rather than drizzle samples (gtminus15 dBZ) It wouldthen be desirable to use drizzle samples to estimate SI How-ever satellite retrieval of the precipitation rate for drizzle canbe highly uncertain It is therefore recommended to furtherimprove the retrieval accuracy of the precipitation rate fordrizzle in satellite observations in order to better use satelliteestimates of SI to quantify aerosolndashcloud precipitation inter-actions Alternatively long-term ground and in situ obser-vations with high-accuracy precipitation rate retrievals canbe used to provide better estimate SI and to further quantifyaerosolndashcloudndashprecipitation interactions

Data availability MYD08_D3 and MYD04_L2 products areavailable through LAADS the Level 1 and Atmosphere Archive

and Distribution System (httpladswebnascomnasagov)MYD06_L2 and 2B-GEOPROF data both collocated to CALIOPsubtrack are obtained from the ICARE Data and Services Cen-ter (httpwwwicareuniv-lille1frprojectscalxtractproducts)2C-PRECIP-COLUMN and 2C-RAIN-PROFILE data sets areavailable from the CloudSat Data Processing Center (httpcloudsatatmoscolostateedudata) CAL_LID_L2_05kmALaydata are obtained from the Atmospheric Science Data Cen-ter (ASDC httpseosweblarcnasagov) The AMSR-EAquaL2B Global Swath Ocean product can be obtained from theNASA Distributed Active Archive Center (DAAC) at NSIDC(httpnsidcorgdaac)

Competing interests The authors declare that they have no conflictof interest

Acknowledgements Minghuai Wang was supported by theNational Natural Science Foundation of China (41575073 and41621005) and by the Jiangsu Province specially appointed profes-sorship grant the One Thousand Young Talents Program We thankJohannes Quaas and an anonymous reviewer for their construc-tive comments which allowed us to further improve the manuscript

Edited by Michael SchulzReviewed by Johannes Quaas and one anonymous referee

References

Anderson T L Charlson R J Winker D M Ogren J A andHolmeacuten K Mesoscale variations of tropospheric aerosols JAtmos Sci 60 119ndash136 httpsdoiorg1011751520-0469(2003)060lt0119MVOTAgt20CO2 2003

Bennartz R Global assessment of marine boundary layer clouddroplet number concentration from satellite J Geophys Res-Atmos 112 D02201 httpsdoiorg1010292006JD0075472007

Boucher O Randall D Artaxo P Bretherton C Feingold GForster P Kerminen V-M Kondo Y Liao H Lohmann URasch P Satheesh S K Sherwood S Stevens B andZhang X Y Clouds and aerosols in Climate Change 2013The Physical Science Basis Contribution of Working Group I tothe Fifth Assessment Report of the Intergovernmental Panel onClimate Change edited by Stocker T F Qin D Plattner G-K Tignor M Allen S K Boschung J Nauels A Xia YBex V and Midgley P M Cambridge University Press Cam-bridge UK and New York NY USA 2013

Breacuteon F M Vermeulen A and Descloitres J An eval-uation of satellite aerosol products against sunphotome-ter measurements Remote Sens Environ 115 3102ndash3111httpsdoiorg101016jrse201106017 2011

Chand D Wood R Ghan S J Wang M Ovchinnikov MRasch P J Miller S Schichtel B and Moore T Aerosol op-tical depth increase in partly cloudy conditions J Geophys Res-Atmos 117 d17207 httpsdoiorg1010292012JD0178942012

Chen Y-C Christensen M W Stephens G L and Sein-feld J H Satellite-based estimate of global aerosol-cloud ra-diative forcing by marine warm clouds Nat Geosci 7 643ndash646httpsdoiorg101038ngeo2214 2014

Cho H-M Zhang Z Meyer K Lebsock M Platnick SAckerman A S Di Girolamo L C-Labonnote L Cor-net C Riedi J and Holz R E Frequency and causes offailed MODIS cloud property retrievals for liquid phase cloudsover global oceans J Geophys Res-Atmos 120 4132ndash4154httpsdoiorg1010022015JD023161 2015

Christensen M W Neubauer D Poulsen C A Thomas G EMcGarragh G R Povey A C Proud S R and GraingerR G Unveiling aerosolndashcloud interactions ndash Part 1 Cloudcontamination in satellite products enhances the aerosol indi-rect forcing estimate Atmos Chem Phys 17 13151ndash13164httpsdoiorg105194acp-17-13151-2017 2017

Comstock K K Wood R Yuter S E and Brether-ton C S Reflectivity and rain rate in and below driz-zling stratocumulus Q J Roy Meteor Soc 130 2891ndash2918httpsdoiorg101256qj03187 2004

Costantino L and Breacuteon F M Analysis of aerosol-cloud inter-action from multi-sensor satellite observations Geophys ResLett 37 1ndash5 httpsdoiorg1010292009GL041828 2010

Duong H T Sorooshian A and Feingold G Investigating po-tential biases in observed and modeled metrics of aerosol-cloud-precipitation interactions Atmos Chem Phys 11 4027ndash4037httpsdoiorg105194acp-11-4027-2011 2011

Feingold G and Siebert H Cloud-Aerosol Interactions from theMicro to the Cloud Scale MIT Press Cambridge MA 319ndash3382009

Feingold G McComiskey A Rosenfeld D and Sorooshian AOn the relationship between cloud contact time and precipitationsusceptibility to aerosol J Geophys Res-Atmos 118 10544ndash10554 httpsdoiorg101002jgrd50819 2013

Ghan S Wang M Zhang S Ferrachat S Gettelman A Gries-feller J Kipling Z Lohmann U Morrison H Neubauer DPartridge D G Stier P Takemura T Wang H andZhang K Challenges in constraining anthropogenic aerosol ef-fects on cloud radiative forcing using present-day spatiotem-poral variability P Natl Acad Sci USA 113 5804ndash5811httpsdoiorg101073pnas1514036113 2016

Greenwald T J A 2 year comparison of AMSR-E and MODIScloud liquid water path observations Geophys Res Lett 362ndash7 httpsdoiorg1010292009GL040394 2009

Greenwald T J LrsquoEcuyer T S and Christopher S AEvaluating specific error characteristics of microwave-derivedcloud liquid water products Geophys Res Lett 34 L22807httpsdoiorg1010292007GL031180 2007

Gryspeerdt E Quaas J and Bellouin N Constraining the aerosolinfluence on cloud fraction J Geophys Res-Atmos 121 3566ndash3583 httpsdoiorg1010022015JD023744 2016

Haynes J M LrsquoEcuyer T S Stephens G L Miller S DMitrescu C Wood N B and Tanelli S Rainfall retrievalover the ocean with spaceborne W-band radar J Geophys Res-Atmos 114 D00A22 httpsdoiorg1010292008JD0099732009

Hill A A Shipway B J and Boutle I A How sensi-tive are aerosol-precipitation interactions to the warm rain

representation J Adv Model Earth Sy 7 987ndash1004httpsdoiorg1010022014MS000422 2015

Horvaacuteth Aacute and Gentemann C Cloud-fraction-dependent biasin satellite liquid water path retrievals of shallow non-precipitating marine clouds Geophys Res Lett 34 L22806httpsdoiorg1010292007GL030625 2007

Hubanks P Platnick S King M and Ridgway B MODISAtmosphere L3 Gridded Product Algorithm Theoretical Ba-sis Document (ATBD) amp Users Guide Collection 006 Ver-sion 42 NASA-GSFC Greenbelt Md available at httpsmodis-imagesgsfcnasagov_docsL3_ATBD_C6pdf (last ac-cess 3 February 2018) 1 2016

Jung E Albrecht B A Sorooshian A Zuidema P and Jon-sson H H Precipitation susceptibility in marine stratocumu-lus and shallow cumulus from airborne measurements AtmosChem Phys 16 11395ndash11413 httpsdoiorg105194acp-16-11395-2016 2016

Kaufman Y J Tanreacute D Remer L A Vermote E F Chu Aand Holben B N Operational remote sensing of tropo-spheric aerosol over land from EOS moderate resolution imag-ing spectroradiometer J Geophys Res 102 17051ndash17067httpsdoiorg10102996JD03988 1997

Kim M H Kim S W Yoon S C and Omar A HComparison of aerosol optical depth between CALIOPand MODIS-Aqua for CALIOP aerosol subtypes overthe ocean J Geophys Res-Atmos 118 13241ndash13252httpsdoiorg1010022013JD019527 2013

King J M Kummerow C D van den Heever S C andIgel M R Observed and modeled warm rainfall occurrence andits relationships with cloud macrophysical properties J AtmosSci 72 4075ndash4090 httpsdoiorg101175JAS-D-14-036812015

Kittaka C Winker D M Vaughan M A Omar A and RemerL A Intercomparison of column aerosol optical depths fromCALIPSO and MODIS-Aqua Atmos Meas Tech 4 131ndash141httpsdoiorg105194amt-4-131-2011 2011

Klein S A and Hartmann D L The sea-sonal cycle of low stratiform clouds J Cli-mate 6 1587ndash1606 httpsdoiorg1011751520-0442(1993)006lt1587TSCOLSgt20CO2 1993

Koren I Dagan G and Altaratz O From aerosol-limited to in-vigoration of warm convective clouds Science 344 1143ndash1146httpsdoiorg101126science1252595 2014

Kubar T L Hartmann D L and Wood R Understanding theimportance of microphysics and macrophysics for warm rain inmarine low clouds Part I Satellite observations J Atmos Sci66 2953ndash2972 httpsdoiorg1011752009JAS30711 2009

Lebo Z J and Feingold G On the relationship between responsesin cloud water and precipitation to changes in aerosol AtmosChem Phys 14 11817ndash11831 httpsdoiorg105194acp-14-11817-2014 2014

Lebsock M D and LrsquoEcuyer T S The retrieval of warmrain from CloudSat J Geophys Res-Atmos 116 1ndash14httpsdoiorg1010292011JD016076 2011

LrsquoEcuyer T S and Jiang J H Touring the atmo-sphere aboard the A-Train Phys Today 63 36ndash41httpsdoiorg10106313653856 2010

LrsquoEcuyer T S Berg W Haynes J Lebsock M and Take-mura T Global observations of aerosol impacts on precipitation

occurrence in warm maritime clouds J Geophys Res-Atmos114 1ndash15 httpsdoiorg1010292008JD011273 2009

Levy R C Mattoo S Munchak L A Remer L A Sayer AM Patadia F and Hsu N C The Collection 6 MODIS aerosolproducts over land and ocean Atmos Meas Tech 6 2989ndash3034 httpsdoiorg105194amt-6-2989-2013 2013

Ma X Bartlett K Harmon K and Yu F Comparison of AODbetween CALIPSO and MODIS significant differences over ma-jor dust and biomass burning regions Atmos Meas Tech 62391ndash2401 httpsdoiorg105194amt-6-2391-2013 2013

Mann J A L Chiu J C Hogan R J Oconnor E JLecuyer T S Stein T H M and Jefferson A Aerosol im-pacts on drizzle properties in warm clouds from ARM MobileFacility maritime and continental deployments J Geophys Res119 4136ndash4148 httpsdoiorg1010022013JD021339 2014

Marchand R Mace G G Ackerman T and Stephens GHydrometeor detection using Cloudsat ndash an earth-orbiting94-GHz cloud radar J Atmos Ocean Tech 25 519ndash533httpsdoiorg1011752007JTECHA10061 2008

McComiskey A and Feingold G The scale problem in quantify-ing aerosol indirect effects Atmos Chem Phys 12 1031ndash1049httpsdoiorg105194acp-12-1031-2012 2012

Michibata T Suzuki K Sato Y and Takemura T Thesource of discrepancies in aerosolndashcloudndashprecipitation inter-actions between GCM and A-Train retrievals Atmos ChemPhys 16 15413ndash15424 httpsdoiorg105194acp-16-15413-2016 2016

Miller D J Zhang Z Ackerman A S Platnick S andBaum B A The impact of cloud vertical profile on liquid wa-ter path retrieval based on the bispectral method A theoreticalstudy based on large-eddy simulations of shallow marine bound-ary layer clouds J Geophys Res-Atmos 121 4122ndash4141httpsdoiorg1010022015JD024322 2016

Muumllmenstaumldt J Sourdeval O Delanoeuml J and Quaas J Fre-quency of occurrence of rain from liquid- mixed- and ice-phaseclouds derived from A-Train satellite retrievals Geophys ResLett 42 6502ndash6509 httpsdoiorg1010022015GL0646042015

Nakajima T Higurashi A Kawamoto K and Penner J EA possible correlation between satellite-derived cloud andaerosol microphysical parameters Geophys Res Lett 281171ndash1174 httpsdoiorg1010292000GL012186 2001

Nam C C W and Quaas J Geographically versus dynamicallydefined boundary layer cloud regimes and their use to evalu-ate general circulation model cloud parameterizations GeophysRes Lett 40 4951ndash4956 httpsdoiorg101002grl509452013

Platnick S and Twomey S Determining the susceptibil-ity of cloud albedo to changes in droplet concentrationwith the Advanced Very High Resolution Spectrometer JAppl Meteorol 33 334ndash347 httpsdoiorg1011751520-0450(1994)033lt0334DTSOCAgt20CO2 1994

Platnick S King M D Ackerman S A Men-zel W P Baum B A Rieacutedi J C and Frey R AThe MODIS cloud products Algorithms and exam-ples from terra IEEE T Geosci Remote 41 459ndash472httpsdoiorg101109TGRS2002808301 2003

Platnick S Meyer K G King M D Wind G Amarasinghe NMarchant B Arnold G T Zhang Z Hubanks P A

Holz R E Yang P Ridgway W L and Riedi J The MODIScloud optical and microphysical products collection 6 updatesand examples from Terra and Aqua IEEE T Geosci Remote 55502ndash525 httpsdoiorg101109TGRS20162610522 2017

Quaas J Boucher O and Lohmann U Constraining the to-tal aerosol indirect effect in the LMDZ and ECHAM4 GCMsusing MODIS satellite data Atmos Chem Phys 6 947ndash955httpsdoiorg105194acp-6-947-2006 2006

Seethala C and Horvaacuteth Aacute Global assessment of AMSR-E and MODIS cloud liquid water path retrievals in warmoceanic clouds J Geophys Res-Atmos 115 1ndash19httpsdoiorg1010292009JD012662 2010

Sorooshian A Feingold G Lebsock M D Jiang H andStephens G L On the precipitation susceptibility of cloudsto aerosol perturbations Geophys Res Lett 36 L13803httpsdoiorg1010292009GL038993 2009

Stephens G L Vane D G Boain R J Mace G GSassen K Wang Z Illingworth A J OrsquoConnor E JRossow W B Durden S L Miller S D Austin R TBenedetti A and Mitrescu C The cloudsat mission andthe A-Train A new dimension of space-based observations ofclouds and precipitation B Am Meteorol Soc 83 1771ndash1790httpsdoiorg101175BAMS-83-12-1771 2002

Szczodrak M Austin P H and Krummel P B Variability of op-tical depth and effective radius in marine stratocumulus clouds JAtmos Sci 58 2912ndash2926 httpsdoiorg1011751520-0469(2001)058lt2912VOODAEgt20CO2 2001

Tackett J L and Di Girolamo L Enhanced aerosolbackscatter adjacent to tropical trade wind clouds revealedby satellite-based lidar Geophys Res Lett 36 1ndash5httpsdoiorg1010292009GL039264 2009

Tanreacute D Kaufman Y J Herman M and Mattoo S Remotesensing of aerosol properties over oceans using the MODISEOSspectral radiances J Geophys Res-Atmos 102 16971ndash16988httpsdoiorg10102996JD03437 1997

Terai C R Wood R Leon D C and Zuidema P Does pre-cipitation susceptibility vary with increasing cloud thickness inmarine stratocumulus Atmos Chem Phys 12 4567ndash4583httpsdoiorg105194acp-12-4567-2012 2012

Terai C R Wood R and Kubar T L Satellite esti-mates of precipitation susceptibility in low-level marine strat-iform clouds J Geophys Res-Atmos 120 8878ndash8889httpsdoiorg1010022015JD023319 2015

Vaughan M a Young S Winker D M Powell K Omar aLiu Z Hu Y and Hostetler C Fully automated analysisof space-based lidar data an overview of the CALIPSO re-trieval algorithms and data products Proc SPIE 5575 16ndash30httpsdoiorg10111712572024 2004

Wang M Ghan S Liu X LrsquoEcuyer T S Zhang K Morri-son H Ovchinnikov M Easter R Marchand R Chand DQian Y and Penner J E Constraining cloud lifetime effectsof aerosols using A-Train satellite observations Geophys ResLett 39 L15709 httpsdoiorg1010292012GL052204 2012

Wentz F and Meissner T AMSR Ocean Algorithm Theoreti-cal Basis Document (ATBD) RSS Tech Doc Santa Rosa CA2000

Wentz F and Meissner T AMSR-EAqua L2B GlobalSwath Ocean Products derived from Wentz Algorithm Ver-sion 2 NASA National Snow and Ice Data Center Dis-

tributed Active Archive Center Boulder Colorado USAhttpsdoiorg105067AMSR-EAE_OCEAN002 2004

Wood R Drizzle in stratiform boundary layer clouds PartII Microphysical aspects J Atmos Sci 62 3034ndash3050httpsdoiorg101175JAS35301 2005

Young S A and Vaughan M A The retrieval of pro-files of particulate extinction from Cloud-Aerosol Lidar In-frared Pathfinder Satellite Observations (CALIPSO) data al-gorithm description J Atmos Ocean Tech 26 1105ndash1119httpsdoiorg1011752008JTECHA12211 2009

Zhang S Wang M Ghan S J Ding A Wang H Zhang KNeubauer D Lohmann U Ferrachat S Takeamura T Get-telman A Morrison H Lee Y Shindell D T PartridgeD G Stier P Kipling Z and Fu C On the character-istics of aerosol indirect effect based on dynamic regimes inglobal climate models Atmos Chem Phys 16 2765ndash2783httpsdoiorg105194acp-16-2765-2016 2016

Zhang Z and Platnick S An assessment of differences betweencloud effective particle radius retrievals for marine water cloudsfrom three MODIS spectral bands J Geophys Res-Atmos116 d20215 httpsdoiorg1010292011JD016216 2011

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation

Meteorological data sets

Precipitation susceptibility calculation

Results

SX_ AI vs SX_ CDNC

SX_ AI from different aerosol products

SX_ Y from different LWP data sets

SX_ Y from different rainfall definitions

SX_ Y from different precipitation data sets

SX_ Y under different stability regimes

Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

mate models (Feingold and Siebert 2009 Terai et al 2012Wang et al 2012) It was first proposed by Feingold andSiebert (2009) and was defined as

S0 =minusdlnR

dlnCDNC (1)

where R is precipitation intensity (precipitation rate for rainyclouds) and CDNC is cloud droplet number concentration(Feingold and Siebert 2009) Sorooshian et al (2009) furtherestimated S0 by replacing CDNC with aerosol index (AI)Wang et al (2012) proposed an alternative metric the pre-cipitation frequency susceptibility defined as

SPOP =minusdlnPOPdlnAI

where POP is the probability of precipitation SPOP has beenshown to strongly correlate with cloud water response toaerosol perturbations in global climate models (Wang et al2012 Ghan et al 2016) Terai et al (2012 2015) furtherextended the definition of precipitation susceptibility

SX =minusdlnX

dlnCDNC (3)

where X can represent precipitation intensity (I precipita-tion rate from rainy clouds only) precipitation fraction (POPor f ) or precipitation rate (R = POPtimes I mean precipitationrate from both rainy and non-rainy clouds) Depending onwhether I POP or R are used in Eq (3) precipitation in-tensity susceptibility (SI) precipitation frequency suscepti-bility (SPOP or Sf) or precipitation rate susceptibility (SR)are therefore defined accordingly Since R can be decom-posed into the product of POP and I SR asymp SPOP+ SI (Teraiet al 2012 2015) In addition some other studies substi-tute aerosol concentration (NA) or cloud condensation nu-clei (CCN) concentration (NCCN) for CDNC to calculate SX(Terai et al 2012 Mann et al 2014)

The behavior and magnitude of aforementioned precipita-tion susceptibility metrics varies a lot in different studies Forinstance SR and SPOP using NA as an aerosol proxy fromTerai et al (2012) both noticeably decrease with increasingLWP whereas SI is flat in the same study Additionally pre-vious satellite studies (Wang et al 2012 Terai et al 2015Michibata et al 2016) show SX calculated with respect toCDNC is higher than that with respect to AI The diverse def-initions of precipitation susceptibility make it challenging tounderstand susceptibility discrepancies in different studiesAn important objective of this study is to derive these suscep-tibilities using the same observations in the same context andto better understand their differences through comparisons

Another source of uncertainty in the estimation of pre-cipitation susceptibility is the uncertainty associated withthe observation Among many others the Advanced Mi-crowave Scanning Radiometer for the Earth Observing Sys-tem (AMSR-E) and Moderate Resolution Imaging Spectro-

radiometer (MODIS) are two widely used satellite cloud re-trieval products in aerosolndashcloud interaction studies For in-stance Sorooshian et al (2009) and Wang et al (2012) bothused the AMSR-E LWP product to estimate SI and SPOP withrespect to AI respectively Terai et al (2015) and Michibataet al (2016) used the MODIS LWP product to estimate SIwith respect to CDNC Both products have their advantagesand limitations and are both subject to various retrieval un-certainties AMSR-E has a coarser spatial resolution thanMODIS Its LWP retrievals are available for both daytimeand nighttime but suffer from instrument noise cloud detec-tion issues and beam filling effect (Greenwald et al 2007Horvaacuteth and Gentemann 2007 Seethala and Horvaacuteth 2010)MODIS LWP retrievals are available only during daytimeThe main uncertainty sources in MODIS LWP retrievals in-clude instrument noise subpixel cloud inhomogeneity three-dimensional radiative effects and uncertainties in ancillarydata (Cho et al 2015 Platnick et al 2017 Zhang and Plat-nick 2011) A recent study by Seethala and Horvaacuteth (2010)revealed several significant differences between AMSR-Eand MODIS LWP products which could contribute to theaforementioned discrepancy of precipitation susceptibility inthe literature

Additionally different definitions of rain events andor dif-ferent methods to derive rain rates could also lead to discrep-ancy in observation-based estimation of precipitation suscep-tibility For example the rain rate used in Terai et al (2015)and Michibata et al (2016) is simply estimated based on aZndashR relationship from CloudSat radar reflectivity profile mea-surements In contrast Sorooshian et al (2009) and Wanget al (2012) used the rain rate reported in the CloudSat opera-tional product which makes use not only of radar reflectivitybut also path-integrated attenuation in the retrieval process(Haynes et al 2009) The primary satellite data sets used inthe previous studies for estimating precipitation susceptibil-ity are listed in Table 1 To account for the discrepancy insusceptibility as shown in Table 1 it is important to examinehow different LWP and rain data sets affect the estimates ofprecipitation susceptibility

Here we estimate precipitation susceptibility using multi-sensor cloud and aerosol products from A-Train satellitesThe main objective of this study is to compare precipitationsusceptibility estimates based on different retrieval productsand to better understand discrepancies documented in previ-ous studies As previous studies have shown that aerosol indi-rect effect and its uncertainties vary in different cloud dynam-ical regimes (LrsquoEcuyer et al 2009 Wang et al 2012 Zhanget al 2016) we further examine how precipitation suscepti-bility might be different under different atmospheric stabil-ity conditions Section 2 introduces different satellite prod-ucts and methods used to calculate the susceptibility Sect 3compares precipitation susceptibility estimates from differ-ent satellite products and explores how atmospheric stabil-ity affects precipitation susceptibility finally the discussionsare made in Sect 4 followed by the summary in Sect 5

MYD08-D3

et al (2016) MYD06_L2

2 Methods

211 AI and CDNC

212 LWP

LWP= aρwτreff (5)

213 Precipitation

SX_Y =minusdlnXdlnY

3 Results

31 SX_AI vs SX_CDNC

MODIS LWP [g m-2]50 150 250 350 450 550

=minusdlnX

dlnCDNCdlnCDNC

= SX_CDNCdlnCDNC

dlnAI (7)

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

MYD08-D3

et al (2016) MYD06_L2

2 Methods

211 AI and CDNC

212 LWP

LWP= aρwτreff (5)

213 Precipitation

SX_Y =minusdlnXdlnY

3 Results

31 SX_AI vs SX_CDNC

MODIS LWP [g m-2]50 150 250 350 450 550

=minusdlnX

dlnCDNCdlnCDNC

= SX_CDNCdlnCDNC

dlnAI (7)

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

211 AI and CDNC

212 LWP

LWP= aρwτreff (5)

213 Precipitation

SX_Y =minusdlnXdlnY

3 Results

31 SX_AI vs SX_CDNC

MODIS LWP [g m-2]50 150 250 350 450 550

=minusdlnX

dlnCDNCdlnCDNC

= SX_CDNCdlnCDNC

dlnAI (7)

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

213 Precipitation

SX_Y =minusdlnXdlnY

3 Results

31 SX_AI vs SX_CDNC

MODIS LWP [g m-2]50 150 250 350 450 550

=minusdlnX

dlnCDNCdlnCDNC

= SX_CDNCdlnCDNC

dlnAI (7)

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

MODIS LWP [g m-2]50 150 250 350 450 550

=minusdlnX

dlnCDNCdlnCDNC

= SX_CDNCdlnCDNC

dlnAI (7)

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

LWP [g m-2]50 150 250 350 450 550

eff (8)

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

MODIS LWP [g m-2]50 150 250 350 450 550

CALIOP

MODIS LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]50 150 250 350 450 550

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

LWP [g m-2]0 100 200 300 400 500 600 700

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

CDNC [cm-3]

101 102

MODIS LWP

CDNC [cm-3]

101 102

AMSR-E LWP

AI10-1 100

MODIS LWP

AI10-1 100

AMSR-E LWP

CDNC [cm-3]

101 102

MODIS LWP

MODIS LWP (e)

CDNC [cm-3]

101 102

AMSR LWP

AMSR-E LWP (f)

AI10-1 100

101MODIS LWP (g)

AI10-1 100

101AMSR-E LWP (h)

164 g m-2343 g m-2

55 g m-2 70 g m-290 g m-2

110 g m-2134 g m-2

215 g m-2 272 g m-2500 g m-2

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

CDNC [cm-3]0 100 200 300

r eff [

microm

30(a) 55 g m-2

70 g m-2

90 g m-2

110 g m-2

134 g m-2

164 g m-2

215 g m-2

272 g m-2

343 g m-2

500 g m-2

CDNC [cm-3]0 100 200 300

400(b)

MODIS LWP [g m-2]50 150 250 350 450 550

AMSR-E LWP [g m-2]50 150 250 350 450 550

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g m-2]50 150 250 350 450 550

Threshold of 0 dBZ

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

SPOP_AI

MODIS LWP [g cm-3]50 100 200 300 500

SI_CDNC

MODIS LWP [g cm-3]50 100 200 300 500

4 Discussion

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

15(a) MODIS LWP

04(b) MODIS LWP

05(c) MODIS LWP

02(d) MODIS LWP

15(e) AMSR-E LWP

06(f) AMSR-E LWP

AMSR-E LWP

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

LWP [g m-2]100 200 300 400 500

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

5 Summary

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

Abstract

Introduction

Methods

Satellite data sets

AI and CDNC

LWP

Precipitation



Results

SX_ AI vs SX_ CDNC






Discussion

Summary

Data availability

Competing interests

Acknowledgements

References

Documents

Estimating precipitation susceptibility in warm marine