2.12 NEW CERES DATA EXAMINED FOR EVIDENCE OF TROPICAL IRIS FEEDBACK Lin H. Chambers*, Bing Lin and David EYoung NASA Langley Research Center, Hampton, VA 1. INTRODUCTION measures thermal radiation between 8.10 and 11.79 #m, and the total (TOT) channel measures New data products from the Clouds and the the total energy leaving the Earth at all wave- Earth's Radiant Energy System (CERES; Wielicki lengths (0.3 - >100 #m, Priestley et al., 2000). The et al., 1996) instrument were recently released. Iongwave (LW) radiation is obtained by subtrac- The Single Scanner Footprint (SSF) data product tion: LW=TOT-SW. combines radiation budget data from CERES with CERES is primarily a climate instrument, so cloud property retrievals from an imager on the great attention has been paid to its calibration. As same platform to provide a vastly improved charac- a result, the radiative fluxes measured by CERES terization of the state of the atmosphere. In addi- have been both stable and accurate over the life- tion, the SSF incorporates new CERES angular time of the instrument. Uncertainties in measured distribution models (ADM; Loeb et al., 2002) based radiances are generally below the 0.5% level on improved scene identification to obtain more (Priestley et al., 2000). The errors in the instanta- accurate top-of-atmosphere fluxes from the satel- neous estimated fluxes of SW (13 W/m 2) and LW lite-measured radiances. Together these (4.3 W/m 2)radiation are mainly due to errors in the advances allow the study of radiative fluxes for application of the angular distribution models (Loeb specific cloud types with unprecedented accu- et al., 2002) which result from errors in the scene racy. SSF data products are now available using identification. the Visible and InfraRed Sensor (VIRS) for the Tropical Rainfall Measuring Mission (TRMM). 3. DATA ANALYSIS These data are used here to explore the Iris hypothesis recently proposed by Lindzen et al., A CERES SSF contains about 130 parameters 2001 (hereafter LCH). describing each CERES footprint. These include time, position, and viewing angles, surface infor- 2. THE CERES INSTRUMENT mation, filtered and unfiltered radiances, radiative fluxes at the surface and top of atmosphere, and a The CERES instrument is a bi-axial scanning variety of parameters describing the clear and radiometer (Wielicki et al., 1996). On the TRMM cloudy portions of the footprint. The latter are satellite, CERES operated in a quasi-periodic cycle obtained from imager (VIRS) information and from of 2 days in Fixed Azimuth Plane Scan (FAPS) ancillary inputs such as numerical weather predic- mode and 1 day in Rotating Azimuth Plane Scan tion models (http://asd-www.larc.nasa.gov/ATBD/ (RAPS) mode, with an occasional day of along- ATBD.html). The nominal 2-km VIRS-pixel derived track scanning. FAPS mode is basically a cross- properties are then convolved using the CERES track scanning mode and obtains maximum geo- scanner point spread function to obtain cloud prop- graphic coverage over the orbital swath. RAPS erties for up to two cloud layers within the -10 km mode scans in azimuth in order to obtain informa- CERES footprint. Currently these layers are dis- tion on the anisotropy of radiation in the full hemi- tinct, with no overlap properties identified. For this sphere. TRMM is a precessing spacecraft with an analysis, since radiative fluxes are only available at orbital inclination of about 35°. It samples all local the CERES footprint scale, the properties of the times over a 46 day period. two layers are area-weighted to obtain footprint- CERES measures radiative energy in 3 broad mean cloud properties. channels: the shortwave (SW) channel measures A set of screening criteria are applied to the reflected solar radiation, the window (WN) channel data to ensure that footprints with problems are not used. This leaves more than half a million region- * Corresponding author address: Lin H. Chambers, ally distributed footprints per day to be analyzed. Atmospheric Sciences, Mail Stop 420, NASA Langley CERES TRMM SSF data are available for the Research Center, Hampton, Virginia 23681-2199; e-mail period January 1 to August 31, 1998. This <l.h.chambers@larc. nasa.gov>. includes the peak and decay of the strong 1997-98 ElNifio.Shortperiodsofdataareavailableafter albedo for the cloudy moist portion of the model this,butarenotincludedinthisstudybecauseof are too low, while the corresponding properties of theprecessinogrbit.Along-tracdkaysarealsonot the dry portion are too high. Hartmann and Mich- includedd,uetopoorspatiaslamplingT.oprevent elsen (2002) have called into question the interpre- temporaslamplingbiases,the TRMMdataare tation of the observed cloud cover versus SST analyzedin46-dayintervalstogetradiativeprop- relation. ertiesoverthefullrangeoflocatlimes.Thisisnec- From the SSF, radiative properties are essaryto computea correctinsolation-weighted obtained for the dry region, which LCH assume albedo.Detailsoftheaveraginmgethodaregiven covers half the Tropics, from those footprints hav- inChamberestal.,2002a.Theradiativeproper- ing the highest 50% of outgoing LW flux. This cor- tiesofaparticulaMr-typefootprinatrefoundtobe responds to emission from the lowest levels of the verystablefromoneprecessiocnycletothenext, atmosphere, as occurs with very low upper tropo- suggestintghattheresultsareinfactrepresenta- sphere humidity. Properties for the moist region tive of that cloud/atmosphertyepe;and not are obtained from the remaining pixels. These are respondintgoaseasonaolrElNifiosignal. compared to the 3.5 box properties of LCH in ThreeTropicaOl ceanregionsbetween30° S- Table 1. The final row combines the LCH Cloudy 30° Nlatitudeareconsideredt:heTropicaWl est and Clear Moist values for comparison. Note the Pacific(TWP,130°to190°EastLongitudec,orre- near-constant net flux measured by CERES, as spondingtotheregionstudiedbyLCH);theEast opposed to the large range of net flux in the LCH Pacific(200° to280° EastLongitude)a;ndthe values. entireTropicaOl cean.Whencalculatintgheradia- tivepropertiees,achregionwasbrokenupinto10° Table 1: Radiative Properties for the Tropics latitudezones.Sincesimilarresultsareobtained forallthreeregionso,nlythosefortheentireTropi- Area SW OLR, Net Flux, calOceanaregiveninthispaper.Fortheland Region Coverage Albedo W/m2 W/m2 mask,SSFfootprintsarescreenedby surface scenetypeforcompletceoveragebywater. CERES Fluxes for Highest and Lowest 50% of LW 4. RADIATIVE PROPERTIES Dry 50% 0.14 292 51.4 Moist 50% 0.32 231 42.4 4.1 The original 3.5 box Iris model LCH Values The Iris hypothesis of LCH relied on examina- tion of the variation of cirrus cloud coverage with Dry 50% 0.21 303 12.8 cloud-weighted Sea Surface Temperature (SST) over a large area. This was done using the 11 and Cloudy 22% 0.35 138 123. 12 pm split window channels on Japan's Geosta- Moist tionary Meteorological Satellite, GMS-5. LCH used Clear 28% 0.21 263 52.8 these observations to develop a 3.5 box climate Moist model consisting of the extratropics and the dry and moist Tropics, the latter further divided into a Moist 50% 0.27 208 84.3 cloudy-moist region containing upper-level cirrus (Cldm+ cloud and a clear-moist region without this high cir- Clrm) rus. The GMS instrument provides only limited information on the cloud properties. Thus, LCH 4.2 Improved Cloudy Moist Definitions used the brightness temperature at 11 pm, Tll < 260 K, as an indicator of the cloudy moist region. The use of a brightness temperature threshold Further, LCH used a fairly subjective approach to to identify upper level cirrus cloud is problematic. select the radiative properties of each box in the When the brightness temperature threshold is set climate model, subject to the constraint of match- very low, to remove all water clouds, it will also ing Earth Radiation Budget Experiment (ERBE, miss thinner cirrus clouds. When the threshold is Barkstrom et al., 1989) global and Tropical mean too high, water clouds will be selected along with values. Previous studies (Fu et al., 2002 and Lin et the cirrus clouds. Chou et al. (2002) indicated that al., 2002) suggest that the LCH LW flux and SW T11 < 260K was merely an index for the variation of thecloudymoistregiona,ndLCHactuallyusedan of ways of defining the cloudy moist region. These areafractionabouttwicethatoftheT11<260K include cloud temperature thresholds, Tc < -15 C regionforthecloudymoistbox.TheCERESSSF and Tc < -30 C and phase thresholds of Phase > containsvastlyimprovedinformationon cloud 1.5 (more than half of cloud in footprint is ice) and propertiesin,cludingcloudtemperaturaendcloud Phase > 1.9 (cloud in footprint is predominantly particlecondensatiopnhase(i.e.,wateror ice), ice). Since cirrus over cumulus is expected to usingassociateVdIRSimagerdata.Asa result, occur often in the Tropics, a second set of tests is retrievedcloudpropertiesthemselvescan be also performed using only the properties of the examinetdoobtaina betteridentificatioonfcirrus upper cloud layer, if that layer covers more than cloud.Twobasicapproacheasreconsiderehdere: half the CERES footprint. This captures any 1) TOtestsfind CERESfootprintswherethe occurrence of extensive and identifiable high cirrus retrievedcloudtemperatur(eaccountinfgorthe cloud in the same footprint with identifiable low emissivityof thinclouds)is suggestiveof ice cloud. The net flux is quite insensitive to the defini- clouds2;)cloudwater/icephasetestsfindCERES tion used, including restriction of the results by footprintisnwhichthecloudismostlyorpredomi- footprint cloud fraction. About 85% of the cloudy nantlyice. moist pixels are 99% or more cloud covered, con- InitialvalidationstudiesoftheCERESSSF firming that cirrus cloud tends to be spatially exten- cloudpropertiessuggestthatthe TOretrieval sive. placessinglelayercirruscloudswithin1kmoftheir actuahl eightw; hilethephaseretrievaisl excellent 5. CLOUD FEEDBACK CALCULATIONS forsinglelayerclouds(Minnis,2002,personal communication).Comparisonto groundtruth The radiative properties and area fractions resultsshowsnoevidencethattheCERESalgo- from the CERES SSF analysis can be inserted in rithmismissinganythingmorethanisolatedthin the simple climate model used by LCH. As in their cirrus(0-5%cloudfraction;Chamberset al., study, the tropical cloudy moist area is varied +/- 2002b). Forcirrusoverlowwatercloud,the 30%, as Acldm=Acldm0(l+B), where p=-0.3 to 0.3. retrievailsnotsowellunderstoodS.omehighcir- The clear moist area, Acm=Acm0(l+T# ), may vary rusoverlowcloudmaybemissedormis-classifiedthrough a range of dynamics, from following the inthisanalysis.Howevegr,iventhelackofsensi- change in the cloudy moist area (7=1) to remaining tivityintheresultsbelow,itdoesnotappearthat constant (7=0). thiscouldbeamajoirssue. 29O 140 -- 7=1 .... 7=0.5 ....... 7=0 Tc<-15 C 120 _ _ Tc<-30C 289 " _ Ph> 1.5 _ Ph> 1.9 t S: _100 _ LCH QE.. t S" --¢ -- -D-- - Tc<-15C 1-288 iT 80 _ -- -A-- - T¢<-30C -- -D- - Ph>l.5 o :_ [ -- _ - Ph>l.9 ,s 4 287 LCH Vahles{ _ 40 20 286 i i i i i i i i i i i 0.1 0.2 0.3 Tropical Cloudy MoistFraction Threshold CloudFraction Fig. 2 Surface temperature response in 3.5 box model using various means of defining radiative properties. Fig. 1 Netflux computed foravariety ofdefinitions of the cloudy moist region. Solid symbols for results from foot- The global surface temperature response from print-mean cloud properties. Open symbols for results a change inthe cloudy moist area fraction is shown from upper layer properties. in Fig. 2. The strong negative feedback found by LCH is reduced to a much weaker feedback. This Figure 1shows net radiative fluxes for avariety feedbackmaybeeitherslightlypositiveorslightly negativedependinognwhichdefinitioinsusedfor 7. REFERENCES thecloudymoistregion,butinallcasesismuch smalletrhanthatfoundbyLCH. Barkstrom, B., E. Harrison, G. Smith, R. Green, J. Incontrasttotheresultsof LCH,thisstudy Kibler, R. Cess, and the ERBE Science Team, findsthelargesetffecwt hen_=0andthesmallest 1989: Earth Radiation Budget Experiment effecwt hen_=1.Thisisaresulot ftheverydiffer- (ERBE) Archival and April 1985 Results, Buff. ent propertiesof the dry regionin the two Amer. Meteor. Soc., 70, 1254-1262. approachesL.CHassumedaverysmallnetflux Chambers, L. H., B. Lin, and D. F. Young, New inputinthedryregions,odecreasinegithercloudy CERES data examined for evidence of Tropical moistorclearmoistareacoverageresultedina Iris feedback, submitted to J. Clim., 2002a. verystrongnegativefeedback.Incontrastt,he Chambers, L. H., D. E Young, P. K. Costulis, P.T. propertiesof thedry regionas obtainedfrom Detweiler, D. B. Stoddard, R. Sepulveda, Joyce CERESmeasuremen(tTsable1)resultina net D. Watkins, and A. Falcone, 2002b: The fluxwhichisnearlythesameasthatinthecloudy CERES S'COOL Project, submitted to Buff. andclearmoisrtegionsA. saresulti,ncreasintghe Amer. Meteor. Soc., February. amounotfdryareahaslittleeffect. Chou, M. D., R. S. Lindzen, and A. Y. Hou, 2002: Response to "The Iris hypothesis: A negative 6.CONCLUSIONS or positive cloud feedback?", J. Clim., in press. Fu, Q., M. Baker, and D. L. Hartmann, 2002: Trop- Newdataproductsareavailablefromthe ical cirrus and water vapor: An effective earth CERESinstrumenta, partof the NASAEarth infrared iris? Atmos. Chem. Phys., 2, 31-37. ObservingSystem.TheSSFproductcombines Hartmann, D. L., and M. L. Michelsen, 2002: No radiativefluxeswithextensiveinformatioonnthe evidence for Iris. Bull. Amer. Meteor. Soc., 83, cloud conditionsin the footprint,which are 249-254. retrievedusingtheco-orbitinigmagerinstrument. Lin, B., B. A. Wielicki, L. H. Chambers, Y. Hu, and Thesedatahavebeenanalyzedtomoreaccu- K.-M. Xu, 2002: The Iris hypothesis: A negative ratelydefinetheradiativepropertiefsorthevarious or positive cloud feedback?, J. Clim., 15, 3-7. regionsoftherecently-proposaeddaptiveinfrared Lindzen, R. S., M.-D. Chou, and A. Y. Hou, 2001: Iris(Lindzenetal.,2001).Avarietyofwaysof Does the Earth have an adaptive infrared Iris?, definingthecloudymoistregionwereexamined. Bull. Amer. Meteor. Soc., 82,417-432. AccordintgoCERESt,henetradiativefluxforthe Loeb, N. G., N. Manalo-Smith, S. Kato, W. E Miller, cloudymoistregionrangesbetween28and54W/ S. Gupta, R Minnis, and B. A. Wielicki, 2002: m2dependinognthespecificdefinitionused.This Angular distribution models for top-of-atmo- isincontrasttothevalueof123W/m2whichwas sphere radiative flux estimation from the somewhastubjectivealyssignedbyLCH. Clouds and the Earth's Radiant Energy Sys- Thissimplemodeml aymissmanyfeedbacks tem instrument on the Tropical Rainfall Mea- intheclimatesystemb,utitshouldprovidearough suring Mission Satellite. Part I: Methodology, rangeoftheclimatevariationisfthephysicosfthe Submitted to J.Appl. Meteor. Irisiscorrect.Thereissomequestionwhethetrhe Priestley, K. J., B. R. Barkstrom, R. B. Lee III, R. N. changeincloudymoistareawithcloud-weighted Green, S. Thomas, R. S. Wilson, R L. Spence, SSTactuallyrepresentas usefulquantity(Hart- J. Paden, D. K. Pandey, and A. AI-Hajjah, mannandMichelsen2,002)a,ndwhetheerxtrapo- 2000: Postlaunch radiometric validation of the latingit froma regionavl ariationto a global Clouds and the Earth's Radiant Energy Sys- responseto warmerclimateis appropriate. tem (CERES) Proto-Flight Model on the Tropi- Regardlessth,ecurrenrtesultsshowthatthepro- cal Rainfall Measuring Mission (TRMM) posedIrisfeedbackis verymuchweakerwhen spacecraft through 1999. J. Appl. Meteor., 39, objectively-determinreaddiativepropertiesare 2249-2258. usedinthemodel. Wielicki, B. A., B. R. Barkstrom, E. E Harrison, R. B. Lee III, G. L. Smith, and J. E. Cooper, 1996: Acknowledgments. The CERES data were obtained Clouds and the Earth's Radiant Energy Sys- from the Atmospheric Sciences Data Center at the tem (CERES): An Earth Observing System NASA Langley Research Center. Discussions with Experiment, Buff. Amer. Meteor. Soc., 77, 853- Bruce A.Wielicki on CERES data were of great benefit. 868.