Aerosol Single Sca.ering Albedo and Layer

Height using VIIRS and OMPS-NM

San$agoGassó,GSFC/ESSICRobertC.Levy,GSFCOmarTorres,GSFC

YingxiShi,GSFC/USRAMODIS/VIIRSScienceTeamMee$ng,October15–19,2018,SilverSpring,MD

•  AerosolAbsorp$on(AA)isoneoftheoutstandingchallengesinaerosolresearchandclimateapplica$ons

Mo$va$on:LargeVariabilityinAnnualMeanver$caldistribu$onofAbsorbingAerosolsinModels

•  Forcingandhea$ngratecomputa$onsarecri$callylinkedtoaerosolsinglescaWeringalbedo(SSA)andlayerheight(Z)

Kiplingetal.,2016,AeroCom

DustandSmokeMeanVer$calDistribu$on

Thereisaneedtoimprove

globalobserva$onsof

AerosolAbsorp$onandHeight

AnotherMo$va$on:BeWerAAinforma$onwillbenefitCurrentAerosolAlgorithms

RecentStudiesHighlightbiasesinAODretrievalsbecauselackofaerosolabsorp$onandheightinforma$on

3.2. Seasonal Analysis of MODIS and MISRAOD Retrievals[24] Any consistent seasonal shift in SSA has the potential

to impact satellite AOD retrievals and thus result in a season-ally varying bias which could be perceived as random noisein a large verification study. Retrieval of AOD from thestandard MODIS algorithms, for example, is dependent onthe a priori assumption of a value of aerosol absorption[Remer et al., 2005; Levy et al., 2007a, 2007b]. Deviationfrom the assumed values has a compounding effect onretrieval bias as AODs increase into multiple scatteringenvironments. AERONET measured AOD is highly accurate(~0.01 in the visible and near infrared) and therefore compar-ison to satellite retrievals allows for analysis of the likelyreasons for potential biases, primarily either from errors incharacterizing surface reflectance or in allowing for a realistic

enough aerosol model. Given the high AODs found in thesouthern Africa smoke plume, the seasonal variation inAERONET inverted SSA provides a natural laboratory totest the impact of SSA on satellite AOD retrieval error.Indeed, as we found no systematic seasonal microphysicalchanges in the retrievals other than absorption, the Mongusite allows the examination of a rarely occurring “partial de-rivative” of satellite retrieval microphysical models.[25] To begin, comparisons of AOD retrievals from both

Terra and Aqua satellite MODIS sensors against the MonguAERONET site are presented in Figures 11a and 11b, respec-tively. Comparisons of AOD are shown for 550 nm with theAERONET data being interpolated to 550 nm. This matchupwas described in section 2. To observe the influence ofseasonal variation of SSA, data are divided into the threecore months of burning, August, September, and October,

Figure 11. Comparison of MODIS retrievals (dark target algorithm) from both (a) Terra 2001–2010 and (b)Aqua 2003–2010 of AOD at 550nm to AERONET measurements at Mongu, Zambia, with data separated forthe primary burning season months of August, September, and October. Figures 11c and 11d are the same asFigures 11a and 11b but for the NRL data assimilation grade MODIS product of Hyer et al. [2011].

ECK ET AL.: TREND IN PARTICLE SSA IN SOUTHERN AFRICA

6423

(Ecketal.,2013)

Devia$onsfoundinSSAassump$on Similarlybyacustomizedaerosollayer

(Wuetal.,2017)

AtmosphericRadianceisdis$nc$velydifferentintheUV

ExamplefromtheCloudAerosolImager

680nm

ITOA=IATM+ISURFSmoke/dust

ISURF

VisibletoNear-IR

•  Surface:difficultto

modelorassume

•  SignificantSurfacecontribu$on

•  MinorSensi$vitytoAerosolabsorp$on

AtmosphericRadianceisdis$nc$velydifferentintheUV

382nm

•  AtmosphericSignalisdominatedbyRayleighscaWering(easytomodel)

•  MinorSurfacecontribu$on

•  SignificantSensi$vitytoAerosolAbsorp$on

UV

IATM~IRAY+IAEROSOL

ITOA=IATM+ISURF

RayleighScaWering

Smoke/dust

ExamplefromtheCloudAerosolImager

680nm

ITOA=IATM+ISURFSmoke/dust

ISURF

VisibletoNear-IR

•  Surface:difficultto

modelorassume

•  SignificantSurfacecontribu$on

•  MinorSensi$vitytoAerosolabsorp$on

80’sand90’s:AVHRR/GOESandTOMS

SimultaneousSpa$alResolu$onUV-NearIR

VisibleNear-IR

5008002100

Near-UV

330400

~50x50km

TOMS

~1km

AVHRR/GOES

It’sbeenalongwaytoachieveUVtonear-IRaerosolretrievals

Pastmissionsdidnothavetherightfeatures

It’sbeenalongwaytoachieveUVtonear-IRaerosolretrievals

Pastmissionsdidnothavetherightfeatures

80’sand90’s:AVHRR/GOESandTOMS

SimultaneousSpa$alResolu$onUV-NearIR

VisibleNear-IR

5008002100

Near-UV

330400

~50x50km

TOMS

~1km

AVHRR/GOES

SimultaneousSpa$alResolu$onü UV-NearIR

2000’s:AquaandAura

Near-UV

330400~20x13km

Aura-OMI

Visible Near-IR

4005008002100

0.5kmAqua-MODIS

~7–15min

Near-UVVisibleNear-IR

0.5km

~50x50km

VIIRS&OMPS-NM

ü SimultaneousSpa$alResolu$onü UV-NearIR

Near-UVVisNear-IR

3805002100

~1km

PACE

ü Simultaneousü Spa$alResolu$onü UV-NearIR

ThePresentisgoodandtheFuturebodesverywell

IntheUV,aerosol τ,ωoandZmodulatetheatmosphericpathradiance

ITOA=IATM(τ,ωo,z)

AerosolUVTop-of-atmosphereRadianceisanUnconstrainedRetrieval

AerosolUVretrievalsrelyonobserva$onsattwobandsandtheITOAismodulatedby

3parameters(τ,ωo,z)

IntheUV,aerosol τ,ωoandZmodulatetheatmosphericpathradiance

ITOA=IATM(τ,ωo,z)=IATM(τ ,ωo,z)

AerosolUVTop-of-atmosphereRadianceisanUnconstrainedRetrieval

AerosolUVretrievalsrelyonobserva$onsattwobandsandtheITOAismodulatedby

3parameters(τ,ωo,z)

IntheUV,aerosol τ,ωoandZmodulatetheatmosphericpathradiance

ITOA=IATM(τ,ωo,z)=IATM(τ ,ωo,z)=IATM(τ ,ωo,z)

AerosolUVTop-of-atmosphereRadianceisanUnconstrainedRetrieval

AerosolUVretrievalsrelyonobserva$onsattwobandsandtheITOAismodulatedby

3parameters(τ,ωo,z)

Allthesescenarioshavesameupwelling

radiance

IntheUV,aerosol τ,ωoandZmodulatetheatmosphericpathradiance

ITOA=IATM(τ,ωo,z)=IATM(τ ,ωo,z)=IATM(τ ,ωo,z)

AerosolUVTop-of-atmosphereRadianceisanUnconstrainedRetrieval

AerosolUVretrievalsrelyonobserva$onsattwobandsandtheITOAismodulatedby

3parameters(τ,ωo,z)

DeriveAerosolOp$calDepthwithaVISsensor(MODIS/VIIRS)anduseitto

constraintheUVretrieval(2obsand2unknowns,ωoandz)

Allthesescenarioshavesameupwelling

radiance

MODIS:DarkTarget Aerosolmodelsw/variablemodeconc.&MixedSSA

Outputs τVIS,τNIR

OMI: DeriveSSAandZ

Aerosolmodelsw/Mix

modeconc.&variableSSA Outputsz, τUV,ωUV

ExtrapolateMODISτtothenear-UV Input τNUV=f(τVIS,τNIR)

MODIS OMI-AerosolIndex

Exis$ngMethodologyusingMODISandOMIsynergy

MODIS:DarkTarget Aerosolmodelsw/variablemodeconc.&MixedSSA

Outputs τVIS,τNIR

OMI: DeriveSSAandZ

Aerosolmodelsw/Mix

modeconc.&variableSSA Outputsz, τUV,ωUV

ExtrapolateMODISτtothenear-UV Input τNUV=f(τVIS,τNIR)

MODIS OMI-AerosolIndex

OMI-MODIS

GassóandTorres,Theroleofcloudcontamina/on,aerosollayerheightand

aerosolmodelintheassessmentoftheOMInear-UVretrievalsovertheocean(2016)

Sa/sfactoryresultswhencomparedwithLidar

Satheeshetal.,Improvedassessmentofaerosolabsorp1onusingOMI-MODISjointretrieval(2009)

Exis$ngMethodologyusingMODISandOMIsynergy

However,OMIandMODISoverlapsarenotquiteadequateforasystema$cretrievalapproach

Both,$medifference(cloud

forma$on)andviewing

geometryareanimpediment

forasystema$cUVtoNear-IR

retrieval

Yellownumbersingrid:AerosolIndex(len)

MODISimagewithOMIgridoverlap

OMIAODretrieved(right)

Near-UVVisibleNear-IR

0.5km

~50x50km

VIIRS&OMPS-NM

ü SimultaneousSpa$alResolu$onü UV-NearIR

Near-UVVisNear-IR

3805002100

~1km

PACE

ü Simultaneousü Spa$alResolu$onü UV-NearIR

OMPS-NMHighResolu$onmodeandVIIRSareanimprovement

Near-UVVisibleNear-IR

0.5km

~50x50km

VIIRS&OMPS-NM

ü SimultaneousSpa$alResolu$onü UV-NearIR

Near-UVVisNear-IR

3805002100

~1km

PACE

ü Simultaneousü Spa$alResolu$onü UV-NearIR

Near-UVVisibleNear-IR

0.5km~12x5km~12x12km

VIIRS&OMPS-NM-HighResMode

ü Simultaneousü Spa$alResolu$onü UV-NearIR

OMPS-NMHighResolu$onModeandVIIRSareanimprovement

OMPS-NMhastwoHighResolu$onModes

~12x12km~12x5km

•  ~4000orbitsspanning4years(1day/week)

•  “High”res~12x12km,similartoOMI

•  “Superhigh”res~12x5km,BeWerthanOMI!!

ASynerge$cUVtoNear-IRAlgorithm:FusionDarkTarget–OMAERUVapproach

AconsistentmethodologyforretrievalsofAOD,SSAandZusing

Near-UVtonear-IRradiances

ExtrapolateOMIωototheVIS

Input:Z&ωVIS=f(ωUV,AE)

VIIRS:DarkTarget Aerosolmodelsw/variablemodeconc.&MixedSSA

Outputs:τVIS,τNIR

OMPS: DeriveSSAandZ

Aerosolmodelsw/Mixmodeconc.&variableSSA

Outputs:z, ωUV

ExtrapolateMODISτtothenear-UV

Input:τNUV=f(τVIS,τNIR)

ASynerge$cUVtoNear-IRAlgorithm:FusionDarkTarget–OMAERUVapproach

Methodologyalreadydeveloped(GassóandTorres,2016)

ToBedevelopedinthisproject

AconsistentmethodologyforretrievalsofAOD,SSAandZusing

Near-UVtonear-IRradiances

ExtrapolateOMIωototheVIS

Input:Z&ωVIS=f(ωUV,AE)

VIIRS:DarkTarget Aerosolmodelsw/variablemodeconc.&MixedSSA

Outputs:τVIS,τNIR

OMPS: DeriveSSAandZ

Aerosolmodelsw/Mixmodeconc.&variableSSA

Outputs:z, ωUV

ExtrapolateMODISτtothenear-UV

Input:τNUV=f(τVIS,τNIR)

ExpectedOutcomeandSummary

•  GlobalMapsofSSAandZfromcombinedOMPS-NMandVIIRS

•  UnifiedMethodologyforAerosolretrievalsofAOD,SSAandZ

fromtheUVtoNearIR

Misc

Figure 3:Illustration of the standard (or operational, Torres et al., 2013) and hybrid (Satheesh et al., 2009) retrieval schemes. The standard algorithm computes the pair of AOD and SSA at five assumed aerosol heights for the pixel’s viewing geometry (blue solid lines and circles). In a prior step, it selected an aerosol model and surface albedo used in the computation. Each triplet (Z, SSA and AOD) has a corresponding upwelling radiance matching the observed radiance by OMI. To select the final (or retrieved) AOD and SSA for the pixel,the operational algorithm uses a climatological height (Zc-clm, red arrow and red dashed lines). The hybrid method uses a VIS AOD (extrapolated to 388nm) as entry point (black arrow and black dashed lines) to determine the Z and SSA using the triplets from the lookup table.

ZC-CLM

ZHYB τMOD

τOMI

ωHYB ωOMI

ExampleofUVRadiancesaggregatedatdifferentpixelsizeovertheNevadaDesert(fromCAIBand1)

0.5-1kmx0.5-1km20kmx13km40-60kmx40-60km

TOMS,OMPS OMI CAI-2,PACE,ACE?

TEMPO,OMPS(JPS1)TropOMI

3-10kmx3-10km

1970s-1990s 2000s 2017> 2020s>

Spa$alresolu$oninnearUVsensorshasbeenimprovingsteadily