Comparison of Cloud Properties between CloudSat Retrievals and Airplane Measurements in Mixed-Phase Cloud Layers of Weak Convective and Stratus Clouds

QIU YujunThomas CHOULARTONJonathan CROSIERand Zixia LIU

1Key Laboratory of Meteorological Disaster of Ministry of Education, Nanjing University of Information Science&Technology,Nanjing 210044

2Key Laboratory for Aerosol-Cloud-Precipitation of China Meteorological Administration, Nanjing University of Information Science&Technology,Nanjing 210044

3Centre for Atmospheric Science,SEAES,University of Manchester,Manchester M13 9PL,UK

Comparison of Cloud Properties between CloudSat Retrievals and Airplane Measurements in Mixed-Phase Cloud Layers of Weak Convective and Stratus Clouds

QIU Yujun?1,2,Thomas CHOULARTON3,Jonathan CROSIER3,and Zixia LIU3

1Key Laboratory of Meteorological Disaster of Ministry of Education, Nanjing University of Information Science&Technology,Nanjing 210044

2Key Laboratory for Aerosol-Cloud-Precipitation of China Meteorological Administration, Nanjing University of Information Science&Technology,Nanjing 210044

3Centre for Atmospheric Science,SEAES,University of Manchester,Manchester M13 9PL,UK

Cloud microphysical properties including liquid and ice particle number concentration(NC),liquid water content(LWC), ice watercontent(IWC)and effective radius(RE)were retrieved from CloudSatdata fora weakly convective and a widespread stratus cloud.Within the mixed-phase cloud layers,liquid-phase fractions needed to be assumed in the data retrieval process, and one existing linear(p1)and two exponential(p2andp3)functions,which estimate the liquid-phase fraction as a function of subfreezing temperature(from?20?C to 0?C),were tested.The retrieved NC,LWC,IWC and RE usingp1were on average larger than airplane measurements in the same cloud layer.Functionp2performed better thanp1orp3in retrieving the NCs of cloud droplets in the convective cloud,while functionp1performed better in the stratus cloud.Functionp3performed better in LWC estimation in both convective and stratus clouds.The REs of cloud droplets calculated using the retrieved cloud droplet NC and LWC were closer to the values ofin situobservations than those retrieved directly using thep1function.The retrieved NCs of ice particles in both convective and stratus clouds,on the assumption of liquid-phase fraction during the retrieval of liquid droplet NCs,were closer to those of airplane observations than on the assumption of functionp1.

mixed-phase cloud,liquid water content,effective radius,ice particle

1.Introduction

Cloud droplets in mixed-phase clouds experience a complicated three-phase transformation,adding diff iculties in quantifying cloud properties.Hallettand Mossop(1974)conducted experiments in a cloud chamber and demonstrated the production of secondary ice crystals at slightly supercooled temperatures of between?3?C and?9?C,which was later referred to as the Hallett–Mossop process.This process has also been observed in natural clouds by airplane observations (e.g.,Hogan et al.,2002;Crosier et al.,2011;Zhang et al., 2011).Due to the complex transformation of liquid and ice particles,the properties of mixed-phase clouds are poorly represented in climate models(Hogan et al.,2004).Further observational studies need to be carried out to advance our understanding of cloud properties in mixed-phase clouds, and the knowledge gained should then be used to constrain GCM cloud parameterizations and reduce uncertainties in cloud feedbacks(Tsushima et al.,2006).

The 94-GHz(W-band)nadir-looking Cloud Prof i ling Radar(CPR),aboard the CloudSat satellite launched on 28 April 2006,has provided vast amounts of unprecedented high-resolution data to study cloud microphysical properties all over the world(Stephens et al.,2002).CloudSat releases the Level 2B Radar-only Cloud Water Content(2B-CWCRO)and Level 2B Radar-Visible Optical Depth Cloud Water Content(2B-CWC-RVOD)products,which provide profiles of cloud microphysical retrievals(Austin,2007).These products have been used to study clouds properties(Barker et al., 2008;Hu et al.,2010;Zhang et al.,2010;Devasthale and Thomas,2012;Gao et al.,2014;Wang et al.,2014).However,large discrepancies in cloud properties have been found between CloudSat retrievals and airplane observations inmixed-phase cloud layers.For example,Barker et al.(2008) found that the crystal number concentrations estimated from CloudSat retrieval products were larger by a factor of 5.0 compared toin situairplane observations in a mixed-phase boundary layer cloud.Protat et al.(2010)pointed out that CloudSat retrievals produce ice water content and extinction amounts in a much narrower range than a ground-based method,overestimating the mean vertical profiles of microphysical parameters below the height of 10 km by a factor of greater than 2.Devasthale and Thomas(2012)conducted sensitivity analyses on subfreezing-temperature clouds using different liquid and ice fractions and found that the liquid water content(LWC)under temperatures down to?20?C,estimated using a quadratic or sigmoid-shaped function,differed by 20%–40%over the tropics(in terms of seasonal means), by 10%–30%over the midlatitudes,and by up to 50%over high latitudes,compared to using a linear function.

The purpose of the present study is to evaluate,by comparing with airplane observations,the cloud properties retrieved from CloudSat data for mixed-phase convective and stratus clouds,and to improve the retrieval algorithms in mixed-phase clouds.The knowledge gained from the study is expected to be useful in improving mixed-phase cloud microphysics parameterizations in meteorological models.

2.Data description

2.1.CloudSat data and retrieval algorithm

CloudSat carries a 94 GHz CPR that provides vertically resolved information on clouds at a resolution of 240 m and with a footprint of 1.4 km cross-track by 2.5 km along-track (Mace et al.,2007).Usinga priorivalues(for the liquid and ice particle size distribution parameters in each cloudy bin),CloudSat RO(using radar only)retrievals mainly provide cloud microphysical properties,including cloud LWC and ice water content(IWC),liquid and ice droplet number concentration(NC),and droplet effective radius(RE),based on the Forward Model algorithm and assuming a log-normal size distribution.Simultaneously,retrievals from CloudSat RVOD(Radar-Visible Optical Depth),which combines the radar ref l ectivity factor and visible optical depth,provide the same cloud microphysics parameters usinga priorivalues. The NC,LWC and RE of liquid cloud droplets,and the NC, IWC and RE of crystal particles,are def i ned below,in Eqs. (1)–(3)and(4)–(6),respectively.A detailed description of the Forward Model algorithm can be found in Wood(2008) and Austin et al.(2009).

In the above equations,NTis the droplet number density,ris the droplet radius,rgis the geometric mean radius,Dis the crystal particle radius,Dgis the crystal particle geometric mean radius,σgis the geometric standard deviation,andρwandρiare the densities of water and ice crystals,respectively.

2B-CWC-RO and 2B-CWC-RVOD retrievals are performed separately for liquid and ice phases,assuming in each case that the radar prof i le is due to a single phase of water. The resulting separate liquid and ice profiles are then combined using a scheme based on temperature.In this scheme, the portion of the prof i le colder than?20?C is deemed pure ice,warmer than 0?C pure liquid,and that between is partitioned linearly into ice and liquid phases according to

wherep1is the liquid-phase fraction,Tis the temperature of cloud layers,Tminis?20?C,andTmaxis 0?C.

The distributions of LWC and IWC versus temperature in mixed-phase clouds are complex,as demonstrated byin situmeasurements(e.g.,You and Liu,1995;Fleishauer et al., 2002;Korolev et al.,2003),mainly because the riming of ice particles has a nonlinear relationship with cloud temperature, as in the Hallett–Mossop process.The production of secondary ice particles appears to have a peak value at aroundT=?4?C toT=?8?C,as measured in a cloud chamber (Hallett and Mossop,1974).Therefore,besides the above linear partition function[p1in Eq.(7)],two exponential functions[p2in Eq.(8)andp3in Eq.(9)]were also used,as sensitivity tests,for the estimation of the liquid-phase fraction:

The liquid-phase fractions estimated from the three partitioning functions as a function of temperature are shown in Fig. 1.Fractions calculated using the exponential functions(p2andp3)were lower than those from using the linear function (p1)at the same sub-zero temperatures.The largest difference was on the order of a factor of 2.0,at aroundT=?8?C.

2.2.In situ observation

On 18 February 2009,the UK BAe146 Facility for Airborne Atmospheric Measurement(FAAM)airplane observation was conducted as part of the APPRAISE-Clouds project(Crosier et al.,2011).The airplane f l ew in mixedphase clouds in the vicinity of the Chilbolton Facility for Atmospheric and Radio Research(CFARR)ground site (51.1145?N,1.4370?W),which is located in southern England.The f l ight route is illustrated in Fig.2.A 3 GHz Doppler-Polarisation Radar(Chilbolton Advanced Meteorological Radar-CAMRa),CFARR,performed range height indicator(RHI)scans along the 253?radial.The BAe146 airplane f l ew several runs at different altitudes in mixed-phase cloud layers.A summary of the airplane maneuvers during the f l ight is provided in Table 1.

Cloud droplets(2μm

Table 1.Summary of the constant altitude runs conducted by the FAAM BAe146 airplane along a CFARR radial on 18 February 2009.

2.3.Convective and stratus cloud regions

On 18 February 2009 the UK experienced high pressure conditions that resulted in large-scale descent of relatively warm and dry air and a large-scale supercooled cloud with a cloud top temperature of approximately?12?C(Crosier et al.,2011).The cold frontal system initially moved slowly to the west,and then remained stationary during the entire course of the fl ights.The stationary front marked a boundary line separating warm air to its west and cold air to its east, with the boundary roughly aligned in the north–south direction.As a result,an extensive layer of supercooled mid-level stratus cloud formed.Ice particles falling from the supercooled layer were observed to evaporate at a height of 2.5 km.

The airplane fl ew fi ve horizontal legs in mixed-phase cloud layers during the period 1157 to 1319 UTC 18 February 2009(Table 1).Due to the restrictions of airplane operations by Air Traf fi c Control,the majority of the airplane’s fl ight time was spent along a radial of 253?from CFARR at distances ranging from 0 km(overpass)to 100 km.A local weak convective cloud was observed at 15–25 km west of the CFARR Radar site station during the CFARR Radar RHI scan.The weak convective cloud is in agreement with the CloudSat Radar identi fi ed(see Fig.3).Figure 3 presents the CloudSat radar re flectivity,CAMRa RHI scan,and average radar re flectivity pro files.The CAMRa RHI scan was performed half way through the airplane constant altitude run.

Figures 3b and c show the CAMRa scan images at the beginning and end of the airplane observational period in the mixed-phase cloud layers.The three images are broadly consistent with the convective and stratiform regions.The mean pro files of both cloud regions shown in Figs.3d and e also indicate the good level of agreement between CloudSat and ground-based measurements,albeit some discrepancies are apparent in the re flectivity values of the two average profiles.The discrepancy might have been caused by the different scanning modes and wavelengths of the two radars. The 94 GHz CloudSat radar scans from the top of the atmosphere to the ground,while the 3 GHz site radar scans from the ground to the atmosphere.The differences in radar sensitivity and spatiotemporal dislocation likely contributed to differences in radar re flectivities.The weakly convective and stratiform regions showed robust and homogeneous features (Crosier et al.,2011).

3.Results and discussion

3.1.Comparison of liquid droplet properties between CloudSat retrievals and airplane observations

3.1.1.Airplane run 2(R2)

Considering thatthe complete data from f l ightleg R2contain information on both the convective and stratus regions, measurements from R2were used to evaluate the CloudSat retrievals of NC,LWC and RE.It should be noted that the airplane measurements were conducted about half an hour earlier than the CloudSat overpass,but the cloud regimes maintained a steady state during this time period,as shown by the CFAAR radar(Fig.3).Figure 4 shows the comparisons as a function ofdistance from CFARR.The retrievals and airplane observations in the convective region were observed to f l uctuate more sharply than those in the stratus cloud region.The trends in the time series of cloud property parameters were similar in the two datasets,although discrepancies of more than 20%existed in the actual values of the parameters,suggesting that the CloudSat retrievals are still effective data to study cloud microphysical characteristics.

At the distance of 18–20 km,located in the convective region,there were no effective 2B-CWC-RO RE,NC,LWC retrievals,while 2B-CWC-RVOD RE,NC,LWC retrievals showed peaks consistent with the airplane observations.This implies that,in convective cloud regions of this nature,retrievals using both the radar ref l ectivity factor and visible optical depth are better than using the radar ref l ectivity factor only.In addition to the default value,2B-CWC-RO retrievals showed retrievals that were nearly identical to the 2B-CWCRVOD retrievals at the same cloud layer height.

In the convective regime,99%of data samples showed larger average droplet NCs from retrievals than from airplane observations,and the overall average differed by a factor of 2.9.Similarly,the average LWC from the CloudSat retrievals was about 2.6 times larger than that from the airplane observations.The difference in RE between retrieved and airplane observations was much smaller,e.g.,only 20%larger from the retrieval.A similar phenomenon was also found for the stratus region,with the retrieved average droplet NC,LWC and RE being 1.5,5.6 and 2.2 times larger,respectively,than those from the airplane measurements.

The simple assumption of the linear function of ice/liquid phase partition under sub-zero temperatures in the RO and RVOD retrieval algorithms might have caused the large discrepancies between the retrievals andin situobservations in the mixed-phase cloud layers of the convective and stratus regions,because such a relationship may not accurately capture the mixed-phase cloud structure(e.g.,Mazin,1995;Nasiri and Kahn,2008;Yin et al.,2011).The distributions of LWC and IWC versus temperature in mixed-phase clouds are complex,mainly because the riming of ice particles has a nonlinear relationship with the cloud temperature,as in the Hallett–Mossop process.This assumption is validated in the next section.

3.1.2.Cloud droplet property pro files in the convective regime

The cloud property retrievals from CloudSat were acquired by averaging the values in the convective regime.Note thatthismay have caused the cloud base and top heights identified from CPR Cloud mask to possess some discrepancies with the airplane observations.This study focuses mainly on comparing the values of cloud properties in similar cloud layers between the airplane observations and CloudSat retrievals.Ice particles formed within the supercooled layer started evaporating as they fell below 2.5 km,as determined by the airplane observations.Mixed-phase cloud layersabove 2.5 km were considered during airplane passes R2to R5,as shown in Figs.5,6,8 and 9.

Since the profiles of RO retrievals only differed slightly from those of RVOD retrievals,based on thep1assumption, we only compared the RVOD retrievals with the airplane observations,but using all of the assumptions(p1,p2andp3). In addition,NCvaluesranged by more than one orderofmagnitude,and thus the averaged values from airplane observations for the same convective cloud layers were used when comparing with retrievals from CloudSat

Figure 5a shows that the assumption of exponential functions(p2andp3)was better than that of the linear function (p1)in retrieving the NCs of cloud droplets.The NCs of cloud droplets retrieved based on thep1assumption were close to the maximum values of measurements at the heights of the R2,R4and R5f l ight legs,whilep2andp3produced much closer retrievals to the airplane observations in general. In addition,the NCs retrieved based on thep2assumption were also better than those based on thep3assumption.

The retrieved LWC increased with decreasing altitude, opposite to what was measured by the airplane.The highest proportion of liquid droplet water was typically found in the top levels within the cloud,based on airplane observations,similar to what has been found in midlatitude mixedphase clouds and Arctic clouds(Hobbs et al.,2001;McFarquhar et al.,2007;Carey et al.,2008).However,the three retrieved profiles of LWC show that the largest values occurred at the height of the melting layer.The ice crystals within the melting layer were potential factors impacting the Cloud-Sat radar ref l ectivity because melting ice crystals would have caused excessively large radar ref l ectivity,resulting in larger retrievals than thein situobservations.Bouniol et al.(2008) suggested a multiple scattering enhancement of at least 2.5 dB in the melting layer of convective systems.In general, the retrieved LWCs based onp3were the closest to the airplane observations,as compared to those retrieved based onp1orp2.The retrieved LWCs were closer to thein situobservations in the top layers than in layers near the melting layer.The retrieved LWC based onp3at the base of the melting layer achieved the maximal value of the airplane observations,which provides a direction to revise the signal of cloud radar in the melting layer of convective clouds.

The liquid droplet RE was calculated using Eqs.(1)–(3) with NCs and LWCs retrieved based on thep2orp3function (Fig.5c).The results show that the REs calculated based on NC and LWC retrievals were closer to the average values of the airplane observations,with less variance than the Cloud-Sat RO and RVOD RE retrieved based onp1.

3.1.3.Cloud droplets property pro files in the stratus region

Comparisons of NC,LWC and RE between thein situairplane observations and CloudSat retrieval in the widespread mixed stratus cloud are presented in Fig.6.The RVOD retrieved–NC using thep1function was closer to the average value of the airplane observations than that usingp2orp3.The values calculated usingp2were smaller than half of the average values from the airplane observations.This implies that the NC of RVOD retrievals can represent the overall level of NC in stratus.

It is clear that the LWCs retrieved from RVOD usingp1were too large—larger,even,than the maximal value of the fl ight campaign in stratus cloud.The LWCs observed from the airplane showed larger values at the top layer and smaller values at the base layer.The retrieved value usingp3was the closest to the average value fromin situmeasurements.The value calculated usingp2was close to the maximum value in stratus layers observed by the airplane.

Based on the retrieved NC usingp1and the LWC usingp3,the RE was calculated using Eqs.(1)–(3).Similar to the convective region,the calculated RE was closer to the airplane measurements than the CloudSat-retrieved RE usingp1.

Note that the retrieved NC and LWC profiles were smoother in the stratus region than in the weak convective region.This implies that the CloudSat radar data can be used for retrieving cloud properties in wide ranges of stratus clouds and presenting the overall characteristics of cloud properties.

3.2.Comparison of ice particle properties between retrievals and airplane observations

3.2.1.Ice crystal NC distribution

The average RE of ice particles from the airplane observations was about 63μm(in the size range of 2DS,from 55μm to 65μm);55μm is taken as the smallest size of ice particles here.The ice particle NC in the size range of less than 165μm differed significantly between the convective and stratus clouds.This size range is referred to as the small size range,while the size range larger than 165μm is referred to as the large size range.The NC in the small size range was at least one order of magnitude higher than that in the large size range.For example,during the airplane observations of R2in the convective cloud,the average NC in the small size range was about 2.6 L?1μm?1,but was only 0.1 L?1μm?1in the large size range.Similarly,the average NC was3.3 and 0.08 L?1μm?1in the smalland large size ranges, respectively,in the stratus cloud.The difference in the NC distribution between the convective and stratus cloud implies that the mixing effect from turbulence increases the NC of large sizes in convective clouds,which potentially has an impact on retrievals because the uncertainty is significantly amplif i ed due to the fact that radar ref l ectivity is the sixth power of the droplet diameter,although this is not necessarily true (sixth power)for large crystals in the W-Band due to non-Rayleigh scattering.

Figure 7 shows signifi cant differences in the number distributions of ice particles between the convective and stratus clouds.In the convective cloud,the NC peaks appeared in the size range from 165 to 1305μm,except in the top cloud layer.The NC varied more widely in the convective cloud than in the stratus cloud.A peak region of NC in the convective cloud was apparent in the radius range of 165–300 μm.Zhang et al.(2011)observed a similar phenomenon in Shandong Province,China,through more than 10 airplane deployments during 2006–08.The only exception was in the top layer of the convective cloud,where NCs were similar in the two types of clouds.This implies a weaker effect of turbulence mixing in the top layer than in the lower layers of convective cloud.The log-normal distribution,which the CloudSat RO and RVOD retrievals assumed,for NCs in the large size range,fits better in the convective layers than in the stratus layers.

3.2.2.Ice crystal profiles in the convection region

Many studies have focused on combining lidar and radar data to retrieve ice cloud properties(Delano¨e et al.,2013; Deng et al.,2013;Heymsf i eld et al.,2014),but discrepancies still exist among different products of retrievals.Radar ref l ectivity is sensitive to ice particle shape,size and distribution(Molthan and Petersen,2011),which affects IWC retrieval(McFarquhar and Heymsf i eld,1998).The estimation of IWC from particle size data requires an assumption of particle mass-dimension,which can potentially cause error on the order of tens of percent(Carey et al.,2008;Protat et al., 2009).If ice particles are modeled as oblate spheroids rather than spheres for radar scattering data,the retrieved IWC is reduced by 50%on average,with a re flectivity factor larger than 0 dBZ(Stein et al.,2011)in clouds.The Cloudsat profiles in convective cloud need to be corrected for attenuation by supercooled liquid water and ice aggregates/graupel particles and multiple scattering prior to their quantitative use (Protat et al.,2009).In the present study,comparison between airplane 2DS data and retrievals was explored by only considering the NCs of ice crystals.The retrieved IWC profiles are shown in Fig.8.

As reported in section 3.1.2,a better retrieval effect of liquid droplet NCs was obtained in the convective region when usingp2compared top1orp3;for LWC,overall,p3performed the best.So,the NC of ice particles and IWC were retrieved based onp2andp3,respectively(see Fig.8).From thein situobservations,the peak NC of large ice particles appeared in the upper layers,which was opposite to the case of the small size range.Note that the NC pro files of the RO retrieval showed a similar tendency to that of small size particles from thein situobservations,but the average value of the former was about 1.9 times that of the latter.The maximum values of the RO NC retrieval were typically observed to be located in the lower half of the mixed-phase layer,which is similar to thein situobservations reported by Carey et al. (2008).

The NC pro files of the RVOD retrieval maintained a similar tendency as thein situobservations of large crystals, which dominated the radar signal and,as a result,affected the corresponding retrievals significantly.The average retrieved NC values of RVOD usingp1andp2were about 0.7 and 1.2 times,respectively,of the values fromin situobservations. This result shows that the NCs of RVOD retrievals usingp2are the closest to observations.

3.2.3.Ice crystal profiles in the stratus region

Section 3.1.3 demonstrated that the best retrievals of liquid droplet NCs and LWCs were achieved usingp1andp3, respectively,in the stratus cloud.The crystal NCs were retrieved based on thep1function,which CloudSat RO and RVOD assumed.In this section,the NCs of crystals and IWC are compared using various functions,as shown in Fig.9. The average NCs of the RVOD and RO retrievals were 1.9 and 4.9 times,respectively,of the average observations.The average NC of the RO retrieval was close to the maximum prof i le of thein situobservations.The comparison here implies that the values of the RVOD retrieval usingp1are more appropriate than those of the RO retrieval.

However,the NC profiles of the RVOD retrieval showed a different tendency to those obtained through thein situobservations,with the minimum of the former appearing in the melting layer where the maximum values were observed by the airplane.A small difference was obtained between the stratus and convective clouds in terms of their NC values of the RVOD and RO retrievals;however,this was not the case in the airplane observations.The contrasting results between the retrievals and airplane observations suggest that the retrieval method for the NCs of ice particles still needs furtherimprovement.

4.Summary

The CloudSat retrievals of cloud droplet properties(NC, LWC and RE),based on the assumption of the liquid and ice phase fraction partitioned as a linear function of cloud temperature,were on average larger than airplane observations in both the convective and stratus clouds.The magnitude of the differences between the satellite retrievals and airplane observations depended heavily on the cloud type.The average NC,LWC and RE from the retrievals were about 2.9,2.6 and 1.2 times of those from the airplane observations in the convective cloud,and were 1.5,5.6 and 2.2 times of in the stratus cloud.The large discrepancies between the CloudSat retrievals and the airplane observations suggest that the existing linear function used for ice/liquid phase partitioning in retrieving cloud microphysical properties needs further improvement.

In mixed-phase cloud layers,the relationship between the liquid-phase fraction and temperature is complex due to ice–liquid transformation processes.The exponential function performed better than the linear function when used for retrieving the NC and LWC in the convective cloud,and the LWC in stratus cloud.On the other hand,the linear function was found to be appropriate for retrieving the NC in stratus cloud.The REs calculated based on the LWC and NC retrievals were closer to the airplane observations than those from CloudSat RO and RVOD retrievals using the linear function.Overall,the exponential functionp3is recommended for use in mixed-phase cloud layers when retrieving the NC,LWC and RE,if the cloud type is not clearly identifi ed.

Large differences in ice particle NCs appeared between the RO and RVOD retrievals when using the linear function, especially in the melting layer and in the upper layers of both convective cloud and stratus cloud.The NCs of the RVOD retrievals were closer to the airplane observations than those of the RO retrievals.Furthermore,the NCs of ice particles retrieved based on the exponential function were closer to thein situobservations than those based on the linear function, in convective cloud layers.However,crystal NC retrievals in the convective cloud deviated signi ficantly from the airplane observations in the melting layer.The attenuation in the melting layer of the CloudSat measurements needs to be corrected.Different weighting coef fi cients according to the scale ofcloud particles and cloud temperatures mightbe helpful to eliminate the in fl uence of large particles on the process of retrieving cloud microphysical parameters.

Acknowledgements.We would like to acknowledge the support of CloudSat DPC for providing the data.We would also like to thank Facility for Ground-based Atmospheric Measurement (FGAM),Facility for Airborne Atmospheric Measurement(FAAM) and Direct Flight in obtaining the airplane dataset,and the support of the CFARR measurements.This study was funded by the National Natural Science Foundation of China(Grant No.41475035), the Natural Science Foundation of Jiangsu Province(Grant No. BK20131433),the Foundations from KLME of NUIST(Grant No.KLME1206),and the Key Laboratory for Aerosol–Cloud–Precipitation of China Meteorological Administration of NUIST (Grant No.KDW1203).

REFERENCES

Austin,R.,2007:Level 2B radar-only cloud water content (2B-CWC-RO)process description document.Version:5.1, CloudSat Project Report,A NASA Earth System Science Path fi nder Mission,1–24.[Available online at http://www. cloudsat.cira.colostate.edu/sites/default/ files/products/ files/2BCWC-RO PDICD.P R04.20071021.pdf.]

Austin,R.T.,A.J.Heyms fi eld,and G.L.Stephens,2009:Retrievalofice cloud microphysicalparameters using the Cloud-Sat millimeter-wave radar and temperature.J.Geophys.Res., 114(D8),D00A23,doi:10.1029/2008JD010049.

Barker,H.W.,A.V.Korolev,D.R.Hudak,J.W.Strapp,K.B. Strawbridge,and M.Wolde,2008:A comparison between CloudSat and aircraft data for a multilayer,mixed phase cloud system during the Canadian CloudSat-CALIPSO Validation Project.J.Geophys.Res.,113(D8),D00A16,doi:10.1029/ 2008JD009971.

Bouniol,D.,A.Protat,A.Plana-Fattori,M.Giraud,J.-P.Vinson, and N.Grand,2008:Comparison of airborne and spaceborne 95-GHz radar re flectivities and evaluation of multiple scattering effects in spaceborne measurements.J.Atmos.Oceanic Technol.,25(11),1983–1995.

Carey,L.D.,J.G.Niu,P.Yang,J.A.Kankiewicz,V.E.Larson, and T.H.V.Haar,2008:The vertical pro fi le of liquid and ice watercontentin midlatitude mixed-phase altocumulusclouds.J.Appl.Meteor.Climatol.,47(9),2487–2495.

Crosier,J.,and Coauthors,2011:Observations of ice multiplication in a weakly convective cell embedded in supercooled mid-level stratus.Atmos.Chem.Phys.,11,257–273.

Delanoe¨,J.,A.Protat,O.Jourdan,J.Pelon,M.Papazzoni,R. Dupuy,J.-F.Gayet,and C.Jouan,2013:Comparison of airborne in situ,airborne radar-lidar,and spaceborne radar-lidar retrievals of polar ice cloud properties sampled during the polarcat campaign.J.Atmos.Oceanic Technol.,30(1),57–73.

Deng,M.,G.G.Mace,Z.E.Wang,and R.P.Lawson,2013: Evaluation of several a-train ice cloud retrieval products with in situ measurements collected during the SPARTICUS campaign.J.Appl.Meteor.Climatol.,52(4),1014–1030.

Devasthale,A.,and M.A.Thomas,2012:Sensitivity of cloud liquid water content estimates to the temperature-dependent thermodynamic phase:A global study usingCloudSatdata.J.Climate,25(20),7297–7307.

Fleishauer,R.P.,V.E.Larson,and T.H.V.Haar,2002:Observed microphysical structure of midlevel,mixed-phase clouds.J. Atmos.Sci.,59(11),1779–1804.

Gao,W.H.,C.-H.Sui,and Z.J.Hu,2014:A study of macrophysical and microphysical properties of warm clouds over the Northern Hemisphere using CloudSat/CALIPSO data.J. Geophys.Res.,119(6),3268–3280.

Hallett,J.,and S.C.Mossop,1974:Production of secondary ice particles during the riming process.Nature,249,26–28.

Heyms fi eld,A.,D.Winker,M.Avery,M.Vaughan,G.Diskin,M. Deng,V.Mitev,and R.Matthey,2014:Relationships betweenice water content and volume extinction coeff i cient from in situ observations for temperatures from 0?to-86?:Implications for spaceborne lidar retrievals.J.Appl.Meteor.Climatol.,53(2),479–505.

Hobbs,P.V.,A.L.Rangno,M.Shupe,and T.Uttal,2001:Airborne studies of cloud structures over the Arctic Ocean and comparisons with retrievals from ship-based remote sensing measurements.J.Geophys.Res.,106(D14),15 029–15 044.

Hogan,R.J.,P.R.Field,A.J.Illingworth,R.J.Cotton,and T. W.Choularton,2002:Properties of embedded convection in warm-frontal mixed-phase cloud from aircraft and polarimetric radar.Quart.J.Roy.Meteor.Soc.,128(580),451–476.

Hogan,R.J.,M.D.Behera,E.J.O’Connor,and A.J.Illingworth, 2004:Estimate of the global distribution of stratiform supercooled liquid water clouds using the LITE lidar.Geophys. Res.Lett.,31(5),L05106,doi:10.1029/2003GL018977.

Hu,Y.X.,S.Rodier,K.M.Xu,W.B.Sun,J.P.Huang,B.Lin,P. W.Zhai,and D.Josset,2010:Occurrence,liquid water content,and fraction of supercooled water clouds from combined CALIOP/IIR/MODIS measurements.J.Geophys.Res.,115 (D4),D00H34,doi:10.1029/2009JD012384.

Korolev,A.V.,G.A.Isaac,S.G.Cober,J.W.Strapp,and J. Hallett,2003:Microphysical characterization of mixed-phase clouds.Quart.J.Royal Meteor.Soc.,129(587),39–65.

Mace,G.G.,R.Marchand,Q.Q.Zhang,and G.Stephens,2007: Global hydrometeor occurrence as observed by CloudSat: Initial observations from summer 2006.Geophys.Res.Lett., 34(9),L09808,doi:10.1029/2006GL029017.

Mazin,I.P.,1995:Cloud water content in continental clouds of middle latitudes.Atmospheric Research,35(2-4),283–297.

McFarquhar,G.M.,and A.J.Heymsf i eld,1998:The def i nition and significance of an effective radius for ice clouds.J.Atmos.Sci.,55(11),2039–2052.

McFarquhar,G.M.,G.Zhang,M.R.Poellot,G.L.Kok,R.Mc-Coy,T.Tooman,A.Fridlind,and A.J.Heymsf i eld,2007:Ice properties of single-layer stratocumulus during the Mixed-Phase Arctic Cloud Experiment:1.Observations.J.Geophys. Res.,112,D24201,doi:10.1029/2007JD008633.

Molthan,A.L.,and W.A.Petersen,2011:Incorporating ice crystal scattering databases in the simulation of millimeterwavelength radar ref l ectivity.J.Atmos.Oceanic Technol., 28(3),337–351.

Nasiri,S.L.,and B.H.Kahn,2008:Limitations of bispectral infrared cloud phase determination and potential for improvement.J.Appl.Meteor.Climatol.,47(11),2895–2910.

Protat,A.,and Coauthors,2009:Assessment of CloudSat ref l ectivity measurements and ice cloud properties using groundbased and airborne cloud radar observations.J.Atmos. Oceanic Technol.,26(9),1717–1741.

Protat,A.,J.Delanoe¨,E.J.O’Connor,and T.S.L’Ecuyer,2010: The evaluation of CloudSat and CALIPSO ice microphysical products using ground-based cloud radar and lidar observations.J.Atmos.Oceanic Technol.,27(5),793–810.

Stein,T.H.M.,J.Delanoe¨,and R.J.Hogan,2011:A comparison among fourdifferentretrievalmethods forice-cloud properties using data fromCloudSat,CALIPSO,and MODIS.J. Appl.Meteor.Climatol.,50(9),1952–1969.

Stephens,G.L.,and Coauthors,2002:The CloudSat mission and the a-train:A new dimension of space-based observations of clouds and precipitation.Bull.Amer.Meteor.Soc.,83(12), 1771–1790.

Tsushima,Y.,and Coauthors,2006:Importance of the mixedphase cloud distribution in the control climate for assessing the response of clouds to carbon dioxide increase:A multimodel study.Climate Dyn.,27(2–3),113–126.

Wang,L.,C.C.Li,Z.G.Yao,Z.L.Zhao,Z.G.Han,and Q.Wei, 2014:Application of aircraft observations over Beijing in cloud microphysical property retrievals from CloudSat.Adv. Atmos.Sci.,31(4),926–937,doi:10.1007/s00376-013-3156-2.

Wood,N.,2008:Level 2B radar-visible optical depth cloud water content(2B-CWC-RVOD)process description document. Version 5.1,CloudSat Project Report,A NASA Earth System Science Path fi nder Mission,1–26.[Available online at http:// www.cloudsat.cira.colostate.edu/sites/default/ files/products/ files/2B-CWC-RVOD PDICD.P R04.20081023.pdf.]

Yin,J.F.,D.H.Wang,and G.Q.Zhai,2011:Long-termin situmeasurements of the cloud-precipitation microphysical properties over East Asia.Atmospheric Research,102(1–2),206–217.

You,L.G.,and Y.G.Liu,1995:Some microphysical characteristics of cloud and precipitation over China.Atmospheric Research,35(2–4),271–281.

Zhang,D.G.,X.L.Guo,D.L.Gong,and Z.Y.Yao,2011:The observationalresults of the clouds microphysicalstructure based on the data obtained by 23 sorties between 1989 and 2008 in Shandong Province.Acta Meteorologica Sinica,69,195–207. (in Chinese)

Zhang,D.M.,Z.E.Wang,and D.Liu,2010:A global view of midlevel liquid-layer topped stratiform cloud distribution and phase partition from CALIPSO and CloudSat measurements.J.Geophys.Res.,115(D4),D00H13,doi:10.1029/2009JD 012143.

:Qiu,Y.J.,T.Choularton,J.Crosier,and Z.Liu,2015:Comparison of cloud properties between CloudSat retrievals and airplane measurements in mixed-phase cloud layers of weak convective and stratus clouds.Adv.Atmos.Sci.,32(12), 1628–1638,

10.1007/s00376-015-4287-4.

23 January 2015;revised 2 May 2015;accepted 2 June 2015)?

QIU Yujun Email:qyj@nuist.edu.cn