scope and status of cloud property products claudia stubenrauch*, s. kinne, w. b. rossow + cloud...

Scope and status of cloud property Scope and status of cloud property

productsproducts Claudia Stubenrauch*,

S. Kinne, W. B. Rossow

+ Cloud Assessment Team

*Laboratoire de Météorologie Dynamique, IPSL/CNRS Ecole Polytechnique, France

Max Planck

Cloud AssessmentCloud Assessment

http://climserv.ipsl.polytechnique.fr/gewexca

Cloud Assessment

initiated in 2005 by GEWEX Radiation panel (GRP)

2005-2010: 4 workshops (Madison, New York, Berlin)

2009-2011: Preparation of common data base (monthly statistics in netCDF format)

2011: WCRP report, BAMS article & opening of data base to public

http://climserv.ipsl.polytechnique.fr/gewexca

co-chairs: C. Stubenrauch, S. Kinne

Assessments essential for climate studies & model evaluation:

Homogenized documentation on•sensor, calibration •retrieval method, ancillary data•sampling•evaluation

state strength & limitations & suitable applications for each data setmake clear statements for each of the cloud properties(global averages & distributions, regional variability, seasonal cycles, diurnal cycle, interannual variations, longterm anomalies)

April 2011 WOAP assessment workshop, Frascati

3

properties: (GCOS ECV’s)

• cloud amount CA + rel. cloud type amount

• pressure/ height CP/CZ

• temperature CT

• IR emissivity CEM

• eff cloud amount CAE (= cloud amount weighted by emissivity)

• VIS optical depth COD

• Water path CLWP/CIWP

• eff part. radius CRE

1° x 1° monthly statistics per obs time: ● averages, ●monthly variability, ● histogramshistograms(tot, High, Mid Low Water, Ice) CP< 440 hPa, CP>680hPa CT>260K, CT<260 / 230K

Cloud Assessment common data base

to facilitate assessments, climate studies & model evaluation


4

ISCCPISCCP GEWEX cloud dataset 1984-2007 (Rossow et al. 1983, 1999)

TOVS Path-BTOVS Path-B 1987-1994 (Stubenrauch et al. 1999, 2006) AIRS-LMDAIRS-LMD 2003-2009 (Stubenrauch et al. 2008, 2010)

MODIS-SMODIS-ScienceTTeameam 2001/3-2009 (Ackerman et al.; Platnick et al.) MODIS-CERESMODIS-CERES 2001/3-2006 (Minnis et al.)

MISRMISR 2001-2007 (DiGirolamo et al.)

relatively new retrieval versions:

PATMOS-xPATMOS-x (AVHRR) 1982-2009 (histos 96-09) (Heidinger et al.) ATSR-GRAPEATSR-GRAPE ERS, 1999-2002 / ENV, 2003-2009 (Poulsen et al.)

only CA or CAE & CT:

HIRS-NOAAHIRS-NOAA 1982-2008 (Wylie et al. 2005) CALIPSO-SCALIPSO-ScienceTTeameam 2007-2008 (Winker et al. 2007, 2009)

CALIPSO-GOCCP CALIPSO-GOCCP 2007-2008 (Chepfer et al. 2009)

POLDERPOLDER (O2 & Rayleigh) 2006-2008 (Riedi et al.)

Participating datasets:

complementary cloud information:


5

Cloud properties from space:1) multi-spectral cloud detection 2) cloud property retrieval

(based on radiative transfer)

Passive remote sensing (>1980) Active (A-Train, >2006)info on uppermost cloud layeruppermost cloud layer info on all cloud all cloud layerslayersgood spatial coverage sparse sampling (track/1000km)

•p/z, T (radiative height)(radiative height) VIS / IR

•horizontal extension•bulk microphysical properties

•z (top)(top) VIS

•vertical extension •microphys. prop. profiles

IR-NIR-VIS radiometers, IR Sounders, multi-angle VIS radiometers perceive clouds differently

cloud property accuracy scene dependent ! most difficult scenes: low contrast with surface (thin Ci, low cld, polar regions ), multi-layer Ci

Global : generalized vs optimized retrievals

lidar, radar perceive clouds differently

lidar : sensitive to thin (subvis) ci, apparent cloud base (COD<5)-> lidar – radar synergy


6

IR-NIR-VIS radiometers good spat res (1-5km), 1 to 5 radiometric channels: day-night1) COD,CT (assumption on microphys) 2) spectral diff (VIS-NIR) -> CRE, CWP IR Sounders spat res 15km, good spectral res in CO2 abs band (5-8 channels): sensitive to thin cirrus1) CP,CEM (nono assumption on microphys) 2) spectral diff (8-12m) -> CRE, CWP (only ci)

VIS, multi-angle radiometers only day, 1/20km res, only sensitive to clouds with COD>2: Ci over low cld -> low cldmulti-angle scattering -> cld top Polarization -> CT independent phase


7

IR,NIR,VIS radiometers:IR,NIR,VIS radiometers:

ISCCPISCCP geo+leo, 3h / 5km; VIS-IR -> CT, COD; AVHRR VIS-NIR -> CRE,CWP• only dataset that directly resolves diurnal cycle & covers only dataset that directly resolves diurnal cycle & covers whole globewhole globe• most thoroughly evaluated most thoroughly evaluated ((≥ 37 articles)≥ 37 articles)• has proven its extreme usefulness in many cloud studieshas proven its extreme usefulness in many cloud studiescld detection based on spatial and temporal variability; operational TOVS atm profiles

PATMOSxPATMOSx 7:30 /1:30; 5km; 11m, 12m -> CT,CEM; 0.6m,3.8m -> COD,CWP,CRE)7 scene types; OE cld retrieval; NCEP atm profiles

ATSR-GRAPEATSR-GRAPE 10:30; 1km; 0.7,0.9,1.6,11,12m ->COD,CRE,CP,Tsurf,phaseOE cld retrieval; ECMWF atm profiles

Particularities of datasets:


8

IR Sounders:IR Sounders: 5-8 channels around 14mCO2 abs band -> CP, CEM sensitive to thin cirrus (COD > 0.1), also in case of multi-layer clouds; day and night

HIRS-NOAAHIRS-NOAA (7:30/1:30; 17km, <32°)ISCCP like cloud setection, CO2 slicing for CP < 650 hPa, else opaque CT (CT-ST<-3/-5K o/l)NCEP atm profiles

TOVS Path-BTOVS Path-B (7:30/1:30; 17/100km); AIRS-LMDAIRS-LMD (1:30;13km)multi-spectral cld detection, weighted 2 method, providing unbiased CP for hgh & low cloudsretrieved atm profiles

active lidar:active lidar:CALIPSO CALIPSO (1:30, 70mx5km)

ST:ST: hor av: 5/20/80km for cld detection -> includes subvisible Ci, CT at top

GOCCP:GOCCP: vert av



9

IR,NIR,VIS + CO2 abs band:IR,NIR,VIS + CO2 abs band:

MODISMODIS (10:30/1:30;1km)

ST:ST: multispectral cld detection, CO2 slicing -> CP,CT,CEM; NCEP atm profilessubsample of well determined clds: COD,CWP,CRE (2.1m)

CECE: multispectral cld detection (0.7,2.1,3.8,11,12m), CT-COD, (VIS-3.8m)CWP,CRE; GMAO atm profiles

multi-anglemulti-angle VIS radiometers: VIS radiometers:

MISRMISR (10:30; 1/17km) COD>2, CZ at top (multi-angle reflectance)

POLDERPOLDER (1:30; 6/20km) O2 abs band -> CP (below midlevel of cld) ; Rayleigh scat CP (only 2/3 of clds (scatt angle 80-120°))polarization -> phase



10

Global Climate Observing System Target Requirements: Cloud Essential Climate Variables (2011 update)

Variable Horizontal Res

Vertical Res

Temporal Res

Accuracy Stability (/decade)

CA 50km N/A 3hr 0.01 – 0.05 0.003 – 0.03

CP 50km NA 3hr 15hPa – 50hPa 3hPa -15hPa

CT 50km NA 3hr 1K – 5K 0.2K – 1K

CWP 50km NA 3hr 25% 5%

CRE 50km NA 3hr 5-10% 1-2%

assuming cloud feedback similar to rad forcing of 0.3Wm-2 (~ 20% of current GHG forcing)

radiative forcing depends on CAE (and not CA)

=> target ranges (low for opt thick/ low clouds - high for opt thin cirrus (CEM=0.2))

based on NISTIR 7047 report (March 2004)


11

Quantity ± instantaneous error / mean error (Rossow)

• CA ± 12% 5% • CT ± 3-6 K 2 K• CZ ± 0.5-2 km 0.3 km • COD ± 25% 10%• CLWP ± 10% 10% • CIWP ± 30% 100% • CREL ± 1 m 1 m• CREI ± 3 m 10 m

• CT base ± 3-6K 2K (SOBS, Warren)

• CZ base ± 0.5-1 km 0.3 km

ISCCP cloud property uncertainty estimates


12

CA: 0.05 – 0.15largest over deserts & Antarctica

CP: ~40-50 hPa slightly smaller for AIRS

hgh / low clds:

CT: 4K/ 2-4K

CEM: 0.05 / 0.10-0.25

CEMH

CEML

CTH

CTL

CA CP

AIRS

TOVSB

TOVS PathB/AIRS-LMD cloud property uncertainty estimates


13

ISCCP cloud property bias estimates

missing thin Ci, under/over-estimation polar summer / winter

25-30% of midlevel clouds = thin Ci (night or over low cld)

CT of very thin Ci too low in tropics


1414

Interprétation des propriétés nuageuses

global CA 60-70% (+ 5% subvisible Ci) : 40% high , 40% single layer low

CAHR (hgh clds out of all clds) depends on sensitivity to thin Ci (misidentified as midlevel clouds by ISCCP, ATSR, POLDER)

CAE (effective CA=CA weighted by cld emissivity) agrees better

global monthly variability of CA: 20%-30% of CAE: 0.25-0.30

Interpretation of cloud properties from satellite observations

CALIPSO only considers uppermost layers to better compare with the other data sets


1515

Interprétation des propriétés nuageuses

15% more clouds over ocean than over land (low clouds),

whereas over land there are more high and midlevel clouds

The latter are optically thinner over land, so that effective cloud amount of those is similar

Differences ocean - land


16

Relative High cloud amount (CAHR)

CAHR depends on sensitivity to thin Ci:CALIPSO > TOVS/AIRS MODIS/PATMOS > ISCCP > MISRgeographical distributions similar(except HIRS, MISR )


17

Relative Low cloud amount

geographical distributions similar(except desert )


18

latitudinal & seasonal

variations similar (except HIRS CALR in SHtrp)

HC

A/C

ALC

A/C

A

Latitudinal & seasonal variation of uppermost cloud layers


19

CALIPSO: including subvis Ci, T(cld top) passive remote sensing: T(rad. cld height)

=> CTH(CALIPSO) should be lowest & nearest to tropopause,largest latitudinal variability(PATMOSX should not be like CALIPSO for high clouds)

CT distributions reflect decrease of vertical extent of troposphere from tropics to poles

SHtrp

SHmid

SHpol

K

Cloud temperature:Cloud temperature:latitudinal variationlatitudinal variation & & distributionsdistributions


20

Tropical high clouds: Tcld distributions

TOVS-BTOVS-BISCCPISCCPPATMOSXPATMOSXMODIS-CEMODIS-CEAIRSAIRSCALIPSO:CALIPSO:all all >0.1 >0.1 >0.2 O>0.2 O

0°-30°

SH

Tcld h

igh

(K

)

ocean ISCCPISCCP CALIPSO CALIPSO TOVSTOVS AIRSAIRS

Clouds in tropics have diffuse cloud tops:

CALIPSO max backscatter ≥ 1km below cloud top

AIRS-LMD:___ cld > 0.95

--- 0.95 > cld > 0.50…. 0.50 > cld > 0.05

midlatitudes


21

10° x 10° regions of typical climate regimes with increasing small scale variations:(1 – <COD(rad)>/<COD(lin)>)

Specific regions, Specific regions, compared to globecompared to globeRossow et al. J. Clim. 2002

1: SH Str Africa 2: SH Str America3: SH midlat 4: NH EPacific 5: NAtlantic storms6: SH Ci off America 7: SH Ci Amazon8: SH Cb Africa 9: NH Cb Indonesia10: ARM Southern Great Plain

Strcum regions (1,2): average CAHR, but optically thin

Storm regions (3,4,5): largest CANAtlantic (5): smaller CAHR & monthly CT variability

ITCZ (8,9): largest CAHR (small CEMH, linked to Ci) & largest monthly CT variability


22

Whereas CA, CEM,CT, CP of the data base are well understood,differences in CRE and CWP have still to be further explored

Global averages of CREW / CREI(H) agree quite well with 15m / 25m

IR sounders determine CREIH, CIWPH only for a subsample: semi-transparent ice cloudsCIWPH is much smaller (25 gm-2) than averaged over all ice clouds (~100gm-2)VIS-IR methods: MODIS-ST / ATSR-GRAPE much larger values than ISCCP / MODIS-CE / PATMOSX

distributions are not Gaussian ….

Bulk microphysical properties:Bulk microphysical properties: eff. liquid / ice particle radiuseff. liquid / ice particle radius & & liquid / ice water pathliquid / ice water path

------ monthly mean variability


23

CREW distributions agree quite well, with a large peak around 11m, small peak at 42 m from ISCCP (perhaps water cloud misidentification) CREIH: IR sounders, ISCCP: large peak at 32 m,second peak of ISCCP at 18 m (at top for opt thick clouds)peaks of MODIS-ST and ATSR-GRAPE at 27 m

eff. particle radiuseff. particle radius & & water pathwater pathdistributionsdistributions

CLWP: large peak at 80 gm-2

CIWPH: AIRS, TOVS compact distribution between 5 & 100 gm-

2; ISCCP, PATMOSX large peak at 4 gm-2 (regions with low clouds clouds?)further investigations

necessary!

m

liquid

ice

gm-2

liquid

ice


24

climate change studiesclimate change studies: be aware of temporal changes in coverage!MODIS at high latitudes more than 1 orbit passages, all others have kept only 1 passage ISCCP nearly 100% coverage – MISR / ATSR 20% - CALIPSO 5%Interannual variability increases with decreasing Interannual variability increases with decreasing Earth coverage!Earth coverage!

Monitoring of Earth coverageMonitoring of Earth coverage

0900AM0130AM

1030AM

0730AM

1030AM

0130AM


25

Global CA anomalies

global CA within global CA within ±±2.5% 2.5% ((~ ~ interannual & monthly mean interannual & monthly mean variability)variability)possible origins of variability:possible origins of variability:• changing average view angle changing average view angle (decreasing with nb of covering geo (decreasing with nb of covering geo sat for ISCCP) sat for ISCCP) • satellite drift satellite drift (for NOAA polar afternoon satellites)(for NOAA polar afternoon satellites)• change in Earth coverage, …..


26

trend analysis -> synergy of different variables

Europe

5

7,5

10

12,5

15

17,5

20

22,5

25

87,5 88 88,5 89 89,5 90 90,5 91 91,5 92 92,5 93 93,5 94 94,5 95

time (years)

eff.

clou

d am

ount

pot. Contrailsclr+thin Ci

Stubenrauch and Schumann GRL 2005

use TOVS upper tropospheric relative humidity to extract meteorological situations favorable for contrail formation

example: persistent contrails

-> detection of increase of thin cirrus over Europe (& North Atlantic Flight corridor)2.8-3.5% (±1.5%) per decade ~0.19% - 0.25% per decade (all situations)


27

example: tropical convection penetrating into the lower stratosphere

cluster analysis of ISCCP DX data (Rossow & Pearl GRL 2007)

mostly larger, organized, convective systems penetrate stratosphere

Longterm datasets -> explore rare events

pene

trat

ion

freq

uenc

y

cluster size

pene

trat

ion

freq

uenc

y

cluster size


28

diurnal cycle of clouds Cairns, Atm. Res. 1995

ISCCP C2, Complex Empirical Orthogonal Functions, project. on distorted diurnal harmonics

Low clouds over land: significant diurnal cycle, max early afternoon

Low clouds over ocean: max in early morning

High clouds: max in evening

Mid clouds: max in early morning or late at night (-> cirrus : TOVS analysis)

Stubenrauch et al. 2006

near noon

early evening


29

ConclusionsConclusionsSatellite instruments:

unique possibility to study cloud properties over long period

•only dataset that directly resolves diurnal cycle & covers only dataset that directly resolves diurnal cycle & covers whole globewhole globe• most thoroughly evaluated most thoroughly evaluated ((≥ 37 articles)≥ 37 articles)• has proven its extreme usefulness in many cloud studieshas proven its extreme usefulness in many cloud studies

To produce a common data base is challenging (GEWEX Cloud Assessment activity not funded)

However, once the data base is reliable, it provides a wealth of information for climate studies & model evaluation

So far statistical analyses:geographical distributions, latitudinal & seasonal variations agree quite welldifferences can be mostly understood by different sensitivities to cirrus,

(problems in retrieval methods, misidentification water-ice clouds)monitoring of cloud properties very difficult (need synergy of different variables)

ESA Cloud_CCI project (Climate Change Initiative) includes assessment activities& another cloud assessment workshop is foreseen at the end of the project (2013)


30

Longterm datasets:good statistics on seasonal cycles; explore rare events; trend analysis difficult (synergy of variables)

Synergy of different data sets & variables very important:

Evaluation of climate models by comparing correlations!

70% (+10%) clouds: ~ 40% high clouds & ~40% single-layer low clouds

geographical cloud structures and seasonal cycles agree quite wellabsolute values depend on instrument sensitivity (& retrieval method)IR sounders passive instruments most sensitive to cirrus

A-Train: unique possibility to evaluate IR sounder retrieval & to give insight into vertical structure of different cloud types

scope and status of cloud property products claudia stubenrauch*, s. kinne, w. b. rossow + cloud...

Documents

rossow cloud assessment

gcos ecvs cloud

frascati cloud properties

uppermost cloud layerinfo

multispectral cloud

ct isccp gewex cloud

cae ct

vis radiometers good