Feb 27, 2000 - much solar energy is absorbed in the Earth's atmosphere? So far, a good deal of ..... and Alberta [1996] to compute the TOA albedo given the.
JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 105, NO. D4, PAGES 4743-4758, FEBRUARY
27, 2000
Absorption of solar radiation by the atmosphere as determined using satellite, aircraft, and surface data during the Atmospheric Radiation Measurement Enhanced Shortwave Experiment (ARESE) FranciscoP. J. Valero,• Patrick Minnis,2 ShellyK. Pope,• Anthony Bucholtz,• Brett C. Bush,1 David R. Doelling,3 William L. Smith Jr.,3 and Xiquan Dong3 Abstract. Data setsacquiredduring the AtmosphericRadiation MeasurementEnhanced ShortwaveExperiment(ARESE) usingsimultaneous measurements from five independent platforms(GOES 8 geostationary satellite,ER-2, Egrett and Twin Otter aircraft, and surface)are analyzedand compared.A consistentdata set can be built for selecteddays duringARESE on the basisof the observationsfrom theseplatforms.The GOES 8 albedosagreewith the ER 2, Egrett, and Twin Otter measuredinstantaneousalbedos within 0.013 _+0.016, 0.018 _+0.032, and 0.006 +_0.011, respectively.It is found that for heavyovercastconditionsthe aircraft measurementsyield an absorptanceof 0.32 _+0.03 for the layer betweenthe aircraft (0.5-13 km), while the GOES 8 albedoversussurface transmittanceanalysisgivesan absorptanceof 0.33 +_0.04 for the total atmosphere (surfaceto top). The absorptance of solarradiationestimatedby model calculationsfor overcastconditionsvariesbetween0.16 and 0.24, dependingon the model usedand on cloud and aerosolimplementation.These resultsare in general agreementwith recent findingsfor cloudyskies,but here a data set that bringstogetherindependent simultaneous observations (satellite,surface,and aircraft) is used.PreviousARESE results are reexaminedin light of the new findings,and it is concludedthat the overcast absorptancein the 0.224-0.68/xm spectralregion rangesbetween0.04 _+0.06 and 0.08 +_ 0.06, dependingon the particularcaseanalyzed.No evidenceof excessclear-sky absorptionbeyondmodel and experimentalerrors is found. 1.
Introduction
Disagreementsbetween radiative transfer models and observationshave persistedever sinceFritz [1951] suggested that the cloudyatmosphereabsorbsmore solar radiation than is predictedby theory [e.g., Stephensand Tsay 1990]. Recently, and nearlysimultaneously, Cesset al. [1995],Ramanathanet al. [1995], and Pilewskieand Valero [1995] reported a level of shortwavecloudabsorptionbeyondthe ability of any model to predict with conventionalmicrophysicalparameters. These three paperswere basedon a varietyof observationalsources, aircraft, surface, and satellite, and thus could not be refuted
simplyby questioninga singledata source,as had been done for past reported discrepancies. This stimulateda dedicated aircraftfield campaigncalledthe AtmosphericRadiation Measurement(ARM) EnhancedShortwave Experiment(ARESE), which seemedto confirm the finding of excess(relative to model predictions)absorptionof solar radiation by clouds [Valeroet al., 1997a, b; Zender et al., 1997; Cesset al., 1999;
1Atmospheric Research Laboratory, Centerfor Atmospheric Sciences,ScrippsInstitution of Oceanography,University of California, San Diego, La Jolla.
2Atmospheric Sciences Division,NASA LangleyResearch Center, Hampton, Virginia.
3AS&M,Inc.,Hampton,Virginia. Copyright2000 by the American GeophysicalUnion. Paper number 1999JD901063. 0148-0227/00/1999JD901063509.00
Arking, 1999]. Also, Wild et al. [1995] showedthat the best currentdataset(GlobalEnergyBalanceArchive(GEBA)) for global annual mean observedsolar radiation at the surface disagreessubstantiallywith radiative transfer models used in climatemodels.Nine climatemodelsproducea range of 164-
185Wm-2 for thisquantity, whilethe GEBA datasetyields 142 Wm -2.
Franciset al. [1997], Hayasakaet al. [1995a,b], Imre et al. [1996], Li and Moreau [1996], Liet al. [1997], and Stephens [1996] disagreewith the findingsof excessabsorption.Both Franciset al. [1997] and Hayasakaet al. [1995a,b] presented analysesof field observations showingperhapssomebut much less excessabsorptionthan is found in the ARESE studies. Imre et al. [1996] and Stephens[1996] usedmethodsdifferent from thoseusedby Cesset al. [1995] and contendthat there is no excessabsorption.Li and Moreau [1996] andLi et al. [1997] usedsatellitedata and a clear-skyradiativetransfer algorithm to questionthe excessabsorption.They ascribeany observed, unexplained absorption to aerosols deriving from biomass burningand other sources. There is alsodebateoverwhetherthere is excessabsorption in clearskies.Arking [1996]foundclear-skyabsorptionexceeding that predictedby models,while others reached a similar conclusionon the basisof analysesof a largedatabasefrom the OklahomaARM site [Kato et al., 1997,Halthoreet al., 1998]. On the other hand, Zenderet al. [1997] and Valeroand Bush [1999]detectedno excessatmosphericabsorptionin clear skies beyond the bounds of the model and observationalerrors. Charlocket al. [1998] derivedmajor excessabsorptionfor both
4743
4744
VALERO
ET AL.:
ABSORPTION
OF SOLAR
RADIATION
BY THE
ATMOSPHERE
clear and cloudy skiesfrom analysesof Oklahoma ARM surface and satellite data. The various results showinganomalouslylarge clear-skyabsorptionare mainly based.on surface insolationmeasurements. Bushet al. [1999a]and R. D. Cesset al. (Consistencies and inconsistencies in measuredtotal, direct,
GOES-8
-- 35,800
ER-2
~ 20 km
km
and diffuse shortwave radiation at the surface, submitted to
Journalof Geophysical Research,1999,hereinafter referred to as Cesset al., submittedmanuscript,1999) find that it is likely that some surface insolationmeasurements,in particular the diffuseradiationdata, maybe affectedby significantsystematic errorsinherentto someof the instruments.Crisp[1997] maintains that the traditional
radiative
transfer models are incorrect
and that if correctlydone, they would showgreater absorption more in agreementwith observations. If the excessabsorptionby cloudsreally exists,it shouldbe manifested
as anomalous
near-infrared
to total
albedo
ratios
when comparedto traditional modelsof reflected irradiance unlessthe excessabsorptionin the near infrared and in the visibleis suchthat the ratio is unchanged.Collins[1998] found Twin Otter ~ 500 rn that the albedoratiosderivedfrom Nimbus7 (Earth Radiation Budget (ERB) observations)and the National Center for Atmospheric Research (NCAR) Community Climate Model, Ground Station Surface Version3 (CCM3) divergemonotonicallyasbroadbandalbedo and cloud cover increase.The discrepancyoccursat all lati- Figure 1. Observationalplatformsused in AtmosphericRatudes with ice-free oceansduring all seasonsand is highly diation Measurement (ARM) Enhanced ShortwaveExperistatistically significant for each year in the observational ment (ARESE). record.
These
results are consistent
with enhanced
short-wave
absorptionin cloudy,but not cloud-free,atmospheres.Absorption discrepancies are now seenby many,but major uncertainties remainregardingboth the magnitudeof the excessabsorption and the situations(clear, cloudy, or both) in which it occurs.
The traditionalpoint of view,basedon currenttheory,is that radiation models correctly predict that the atmosphere absorbs,on average,---20% [Kiehlet al., 1995]of the solarenergy arriving at the Earth and that the magnitudeof the (broadband, globallyannuallyaveraged)absorptionis minimally affected by clouds. Some of the recent studies,noted earlier, estimate that the averageatmosphericabsorptionis close to 28% and is greatly affected by clouds. The 8% difference between the two alternative energy budgetsis currently the largestuncertaintyin the entire climaticenergybudget. If the larger figure is true, it has major consequences for our understandingof rainfall, planetarycirculation,and indeed the entire Earth system. From the above discussionone may concludethat there is currently a major questionin climate studies.That is, How much solar energyis absorbedin the Earth's atmosphere? So far, a good deal of the scientific effort has involved studyingdiscrepanciesbetween observationsand trying to assignmeasurementerrorsto explainthe excessabsorption.The approachusedhere, in contrast,is to concentrateon the analysis of ARESE data for those casesthat show consistency between totally independentmeasurementsmade from the surface, a satellite, and three aircraft.
mental values, including retrievals of cloud properties using GOES 8 and surfacedata, are comparedwith calculationswith the purposeof checkingmodel consistency and model ability to predict the absorptionof solar radiation by the atmosphere. The results of previous studies are examined in light of the presentfindings.
2.
Experimental Description
The experimentalemphasisof ARESE involvedthe acquisition of radiometricdata by multiple coordinatedaircraft and from satellitesand surfacesites.Three aircraft measuredupwelling and downwelling solar radiative fluxes at altitudes rangingfrom the lower stratosphereto the low troposphere. Flux measurements
were also made from the ARM
cloud and
radiation testbed (CART) central facility located at 97.48ø longitudeand 36.59ø latitude and from secondarysurfacestations (extendedfacilities) that are part of the ARM Southern Great Plains (SGP) site. Broadbandshortwavetop of the atmosphere(TOA) albedoswere derived from visible channel radiances
from
GOES
8.
The ARESE observationalsystemis illustrated schematically in Figure 1. The aircraft, stackedat different altitudes, flew tracks over the surface stations, while the GOES 8 data
were taken at a 1 km resolutionalongthe flight tracksevery 15 rain In this manner it was possibleto obtain coevalmeasurements of radiative fluxes from which the absorptionof radiation by the atmosphericlayer betweentwo altitudes(the flux divergence)was determinedas the differencebetweenthe net fluxesat eachlevel in the atmosphere[Valeroet al., 1997a,b].
In summary,this paper presentsthe resultsof an analysisof data acquiredsimultaneouslyby 15 different instrumentsfrom five independentplatforms.The following sectionspresent a descriptionof the observationalsystemand of the data from The net radiative flux is the difference between the downthe different platformsaswell as a discussionand comparison welling and upwellingfluxesat each altitude. Surfaceobservaof the data setswith the purposeof investigatingconsistency tionsprovidedthe radiativeflux transmittedthroughthe total and establishingerror estimates.Absorptanceis computedfor column, and the GOES 8 data determined the albedo of the cloudy and clear skies using different methods. The experi- entire column.
VALERO
ET AL.:
ABSORPTION
OF SOLAR
The basicinstrumentsrequired to meet the ARESE objectives, in addition to the CART site facilities and GOES 8, are
includedin the RadiationMeasurementSystem(RAMS). The RAMS is an array of radiometerscoveringthe spectrumfrom the near ultravioletto the far (thermal)infrared.Components of the RAMS vary dependingon the purpose and platform being used. The experimentalapparatusand methods to acquire airborneand surfacedata from the RAMS were detailed by Valeroet al. [1997a,b]. In this study,pairs of the RAMS total solarbroadbandradiometer(TSBR) and fractionalsolar broadbandradiometer(FSBR) were usedfor simultaneously viewingin the zenith and nadir directionsfrom eachplatform. The TSBR and FSBR cover the spectralrangesfrom 0.224 to 3.91 /•m and from 0.68 to 3.3 /•m, respectively[Valeroet al., 1997b]. Radiative flux measurementsin the stratosphere,the upper troposphere,and the lower tropospherewere made usingthe RAMS on the NASA ER 2, a Grob Egrett, and a Twin Otter aircraft, respectively.The Egrett also carried a nadir-viewing
scanningspectralpolarimeter(SSP) [Parrainet al., 1998;Stephens et al., 1999]. Additionally, the RAMS systemwas installedat three ARM CART surfacesites[Bushet al., 1999a]. The RAMS broadband radiometers were calibrated before,
during, and after the experiment. The calibration included power calibration,angular responsecalibration,and spectral responsecalibration [Valeroet al., 1997b]. The calibration accuracy of the broadband radiometers is -1%. The in-flight accuracyis somewhatlower, -1.5%, becauseof the uncertainties introducedby the pitch and roll movementsof the aircraft that affectmostlythe direct downwardcomponentof the solar flux. A correction is applied to minimize such uncertainties [Hammeret al., 1991; Valeroet al., 1997b]. The precisionof the airborne measurementwas tested in flight during ARESE [Valero et al., 1997b]. The Egrett and
RADIATION
BY THE
ATMOSPHERE
4745
cover. During September 25, October 13, and November 1, there were periods of clear sky, broken clouds,and overcast skies.The GOES 8 imagein Figure 7 showsthe cloudcoveron November 1, for example. The combinationsof cloud conditions observedduring the same flight are very useful for comparing the relative effectsof varyingcloud cover on the radiationfieldwhile minimizingthe potentialinstrumentalchanges betweenobservations.On suchflightsthe differencesbetween cloudyand clear conditions,which constitutethe fundamental point of this research,becomerelative measurementsdependent mostlyon precision,rather than absolutemeasurements, demandingprecisionplus accuracy. The data gapsin Figures3-6 correspondto timeswhen the Twin Otter was refueling or when the Egrett was turning to maintain coordinationwith the Twin Otter. At the Egrett's altitude of 13 km its ground speedis greater than that of the Twin Otter, so the Egrett made periodic 360ø turns in order to maintain a horizontal separationbetweenthe two aircraft of 1 km or less.These maneuverswere describedby Valero et al. [1997a,b]. These 360ø turns can be seenas loopsin Figure 2. All data points outsidethe 1 km range were eliminatedfrom the analysisto assurecollocation.The cumulativeaveragingof the data (section6.2) takes care of small collocationoffsets within the i km requirement[Marshaket al., 1997]. The fluxes measured
from aircraft
are used to estimate
at-
mosphericabsorptancefor comparisonto model calculations. S. K. Popeand F. P. J. Valero (Observationsand modelsof flux p....... , column transmittance,and column reflectanceduring ARESE, submittedto Journalof GeophysicalResearch,1999, hereinafter referred to as Pope and Valero, submittedmanuscript,1999) comparedcalculatedand measureddownwelling fluxesat 13 km in order to examinethe observedday-to-day and shorter timescalevariability in the downwellingfluxes measured from the Egrett. For both day-to-dayand instantaTwin Otter were flown at the same altitude and as close to each neousvariabilitythey found agreementbetweenmodeled and other as possiblefor a side by side comparisonof the zenith measureddownwellingfluxes.They confirmedthat suchvariand nadir radiometers. The observed differences between corations are largely a consequenceof multiple scattering,which respondingradiometersin each aircraftwere not larger than 5 Wm-2. The averagedifferencefor all testedradiometers was in turn dependson the optical depth and optical depth vari-2-3 Wm-2. For operational reasons it wasimpossible to fly abilityaboveand belowthe aircraft flight level.The creationof the ER 2 side by side with the other aircraft. However, the substantialdiffuseupwellingradiation by a high-albedosurface causesadditional downwellingradiation by the backscattering radiometers were rotated between aircraft and the surface duringthe experiment,providingat leasta partial checkon the action of the gasesand any aerosolsin the atmosphereabove the aircraft. Pope and Valero (submittedmanuscript,1999) ER 2 radiometers. showed that even at 13 km, the multiple-scatteringeffects of An additionalcheckof instrumentaccuracywas done during ARESE by periodiccomparisons with an absolutecavityradi- the atmosphereabove the aircraft on the downwellingfluxes andcanreachvaluesof 10-20 Wm-2 andposometer, the same one used for the power calibrations and are significant siblyhigher. The magnitudeof the enhancementand variabiltraceable to the World Radiation Reference Standard. Further RAMS broadbandaccuracycheckswere made duringthe Sub- ity in the downwellingfluxesdependson the atmosphericconsonicAircraft Contrail and CloudEffectsSpecialStudy(SUC- ditions not only above but also below the aircraft (clouds, CESS) [Valeroand Bush,1999]by Wiscombe et al. [1998] and aerosolloading,surfaceand cloudalbedo,etc.). The Pope and Valero (submittedmanuscript,1999) analysis,together with Bushand Valero [1999]. the "wing to wing" comparisonsdescribedby Valero et al. [1997b]and the followingGOES 8 comparisons,constitutesa checkfor the sensitivity,dynamicrange, precision,and accu3. RAMS Data racyof the RAMS aircraft instruments,whichprovidemuch of The aircraft flight tracks over the CART site during the 4 the data used in the following analyses. For surfaceradiative fluxeswe use the RAMS TSBR, spedaysselectedfor this studyare shownin Figure 2. Figures3-6 are time seriesof 1 s mean net fluxesfrom the Egrett and Twin cially installed at the CART central facility during ARESE, Otter that were used to determine the absorptancefor clear- rather than the CART siteradiometersthat were alsooperated skyand cloudyconditions.Figure3 (October 11) depictsclear- during ARESE. This decision was based on studies by skyconditions,while Figures4-6 (September25, October 13, Morikofer [1939], Bener [1950], Drummond and Roche andNovember1, respectively)displayvaryingdegreesof cloud [1965], Robinson [1966], Rodskjer [1971], Gulbrandsen
4746
VALERO
ET AL.' ABSORPTION
OF SOLAR
RADIATION
BY THE
ATMOSPHERE
'
37.4
37.4
37.2
37.2
37.0
37.0
36.8
36.8
36.6
I
(c)-
36.6
36.4
-99.5
-99.0
-98.5
-98.0
I
36.4
-97.5
,
,
,
-99.5
,
I
,
,
,
,
-99.0
longitude
36.6
37.2
36.4
37.0
36.2
36.8
36.0
36.6
I
•
-99.5
[
[
,
I
,
,
,
-99.0
,
I
,
I
,
[
[
[
,
,
,
-98.0
I
-97.5
,
-98.5
,
,
I
]
,
-98.0
,
[
I
'
'
'
'
I
'
'
'
'
I
....
I
....
I
ß
37.4
Oct. 11, 1995
35.8
,
longitude
I
36.8
I
-98.5
(d)_' Nov. 1, 1995
36.4
-97.5
-99.5
-99.0
-9-8.5
longitude
-98.0
-97.5
longitude
Figure 2. Aircraftlatitudeis plottedversuslongitudeto illustratethe flightpathsfor the 4 daysanalyzedin thisstudy:(a) September 25,1995,(b) October11,1995,(c) October13,1995,and(d) November1, 1995.The SouthernGreatPlainscloudandradiationtestbed(CART) siteis indicatedby a solidsquare. [1978],and more recently,Bushet al. [1999b]and Cesset al. (submitted manuscript,1999). These studiesindicate the potential for significantthermally generated errors that affect measurementsacquiredwith radiometersof designsimilar to those used at the CART site. For such reasons, data
4.
GOES
8 Data
To match the aircraft observations, GOES 8 visible radi-
ancesfrom 1 km pixelswereaveragedalonga 10 min legof the aircraftflight path. The flight legwascenteredon the GOES 8
from these instruments were used only for comparison observingtime. The mean GOES 8 visibleradianceswere then convertedto broadbandalbedosfollowingthe procedures outpurposes.
9OO
9OO Oct. 11, 1995
Sept. 25, 1995
700
8OO
V..
5oo
.
/.
700
300
Otter
600 ß
•
•
E
œ.
,;• ß
:
600
.
.i
5OO
400
o
-.-..'
400
•
•
200 o
55000
60000
65000
70000
GMT seconds
55000
.,•'•
ß
,,. ,•j.• 60000
65000
GMT seconds
Figure 3. A time seriesof net fluxesfrom the Egrett and Twin Otter that were usedto determinethe absorptance for
Figure 4. A time seriesof net fluxesfrom the Egrett and Twin Otter that were usedto determinethe absorptance for
October 11, 1995. The data shown are 1 s means.
September25, 1995. The data shownare 1 s means.
VALERO
ET AL.: ABSORPTION
OF SOLAR
RADIATION
0ct. 15, 1995
•oo 500
300
between
i
/""'"i. r • ..'.:
ATMOSPHERE
4747
day. The instantaneousGOES 8 albedoswere 0.022 _+0.053 greater than those derived from ERBS WFOV data during ARESE [Doellinget al., 1998]. The mean difference is not statisticallydifferent from zero. A more reliable comparison usesall of the ERBE data through 1998. The mean difference
900
400
BY THE
.
:
the ERBE
WFOV
and GOES
8 between
1994 and
1998 is -0.012 _+ 0.033 [Doellinget al., 1999]. Much of the uncertaintyin the GOES WFOV differencesresultsfrom errors in matchingthe data setsand the small samplenumbers.
;I ß
The mean difference
between
GOES
8 and the CERES
on the
Tropical Rainfall Measuring Mission satellite (TRMM) is -0.001 ___ 0.030 for the period between Januaryand August 60000
70000
65000
1998.
75000
GMT seconds
Figure 5. A time seriesof net fluxes from the Egrett and Twin Otter that were used to determine the absorptancefor
While few of the satellitecomparisonswere performed for the ARESE period, they representthe typicalperformancefor the GOES 8 broadband albedos. Furthermore, the uncertain-
ties are generallyvery closeto thoseexpectedfrom the process of convertingradianceto flux with anisotropiccorrectionfactors. Thus no systematicerrors are expectedto occur in the
October 13, 1995. The data shown are 1 s means.
linedbyMinniset al. [1995a]andupdatedbyMinnisand Smith GOES 8 albedos unless the visible channel calibration varies in [1998].This processusesa regressionfunctionbetweenOcto- an unpredictablefashion.Ayerset al. [1998] and Nguyenet al. ber 1986 GOES 6 visible albedos and the Earth Radiation [1999] used the NOAA 14 advancedvery high resolutionraBudgetExperiment(ERBE) Earth RadiationBudgetSatellite diometer(AVHRR) asa referencefor calibratingthe GOES 8 (ERBS) broadbandshortwave(0.2-5.0 /xm) scanner(35 km visible channel. This calibration source has been tracked conresolution)albedos.Thus the GOES-derivedalbedosshould tinuouslysincethe launchof the NOAA 14, and its calibration be representative of the valuesexpectedfrom ERBE. Minniset coefficientshave been updated to accountfor degradationof al. [1995a]and Minnis and Smith [1998] discussin detail the the sensor.The accuracyof this calibrationwas confirmed use of a singleregressionfunctionfor both clear and cloudy independentlyseveraltimesusinga varietyof sourcesincludskies.The proceduresfollowed include the use of separate ing the Antarctic snow surface [Loeb, 1997] as well as the bidirectionalcorrectionfactorsfor clear and cloudypixelsand AdvancedTroposphericScanningRadiometer (N. Rao, peralso include the determination of cloud fraction from the satsonalcommunication,1998) and the Visible Infrared Scanner ellite data. on TRMM [Nguyenet al., 1999].These latter instrumentsuse Preliminaryestimatesof the uncertaintiesin the broadband onboard calibration systemsthat permit views of the Sun shortwave albedos derived with GOES visible channel data throughwell-characterizeddiffuserplates.Thus the degradataken over the ARM SGP domain havebeen determinedusing tion of the NOAA 14 visible channelsensitivityis well estabseveralapproaches [Doellinget al., 1998,1999]includinginitial lished.By performingcalibrationsagainstNOAA 14 everyfew comparisons with Egrett, Twin Otter, and ER 2 RAMS data months,Ayerset al. [1998] showedthat the GOES 8 visible taken duringARESE. Three independentsatellitedata sets, channelhas degradedin a predictablelinear fashionsinceit includingthe 1000 km resolutionERBS wide field of view beganoperationsin 1995.The GOES 8 gain (0.656) usedhere (WFOV) and the Cloudsand Earth Radiant Energy System is based on an update of that calibration degradationcurve (CERES) albedos[Wielickiet al., 1998],were alsomatchedto usingthe latestNOAA 14 AVHRR calibration.The AVHRR coincident GOES 7 and GOES 8 broadband albedos at various GOES 8 calibrationclosestto the ARESE period that contribtimesbetween1994and 1998.During the ARESE period,only uted to the degradationcurve was performed during August 5 ERBE datapointscouldbe matchedwith GOES 8 duringthe 1995.The slopeof the degradationcurveis preciseto within +5%
between
1995 and 1998.
To establish that there were no short-term
• •)
v
900
.........
I
.........
I
Nov. 1,1995
'
'
700
300
• 'E
-
•
=•
ß-•
400
200 , •0000
;•F":"••. ,• : ''
•'
'1
3
• '
v.
:
,
...... :.
•:' •: . .... • ', •
" • :' '
•5000
' •
ß 70000
GMT seconds
Figure 6. A time series of net fluxes from the Egrett and Twin Otter that were used to determine the absorptancefor November 1, 1995. The data shown are 1 s means.
variations
in the
GOES-8 gain,the clear-skyvisiblechannelalbedowasderived over the CART site everyday duringARESE when possible. The resultsplotted in Figure 8 showthat exceptfor October3 the albedo varied by
