Page 1
Level-2LandSurfaceTemperatureandEmissivity
ProductStatus
GlynnHulley,RobertFreepartner,NabinMalakarJetPropulsionLaboratory,CaliforniaIns=tuteofTechnology
(c)2017CaliforniaIns3tuteofTechnology.Governmentsponsorshipacknowledged.ECOSTRESSScienceTeamMee/ng,UCDavis,CA,15-17May2017
Na/onalAeronau/csandSpaceAdministra/on
PrincipalInves;gator:SimonHook,JPLCo-Inves;gators:RickAllen,Univ.ofIdaho;MarthaAnderson,USDAJoshuaFisher,JPL;AndrewFrench,USDAGlynnHulley,JPL;EricWood,PrincetonUniv.Collaborators:ChristopherHain,Univ.Maryland
Page 2
Outline
1. Level-2Products2. Level-2AlgorithmandStatus3. Band4shiWuncertaintyanalysis4. GEOS-5FPImplementa/on/Tes/ng/Valida/on5. Summary
Page 3
Level-2FlowSchema/c
L1AProducts
L1BProducts
L2Products
PGEmodules
Intermediatedata
Otheragencydata
• PathRadiance• TransmiZance• SkyIrradiance
AtmosphericCorrec/on
TemperatureEmissivitySepara/on
CalibratedSensorRadiances
CloudMasking
MERRA2/NCEP/GEOS5Atmosphere
SurfaceRadianceCalcula/on
CloudMask
Space/Timeinterp.atmosphericfields:• SurfacePressure• AirTemperatureProfile• Rela/veHumidityProfile• Geopoten/alHeight
L2Products
Geoloca/onParameters
Code:C++Timing:~5minutesforoneECOSTRESSgranule(~25millionpixels)Runconfig:Mul/plerun/meop/ons(cloudthresholds,atmosphericdata,WVSmodel)
• TotalWaterVapor
L3
L3
**Retrieveonallpixelsregardlessofcloud
L3
Page 4
SDS Long Name Data type Units Valid Range
Fill Value
Scale Factor
Offset
Group L2 LSTE Data LST Land Surface
Temperature uint16 K 7500-
65535 0 0.02 0.0
QC Quality control for LST and emissivity
uint16 n/a 0-65535 0 n/a n/a
Emis1 Band 1 emissivity uint8 n/a 1-255 0 0.002 0.49 Emis2 Band 2 emissivity uint8 n/a 1-255 0 0.002 0.49 Emis3 Band 3 emissivity uint8 n/a 1-255 0 0.002 0.49 Emis4 Band 4 emissivity uint8 n/a 1-255 0 0.002 0.49 Emis5 Band 5 emissivity uint8 n/a 1-255 0 0.002 0.49 LST_Err Land Surface
Temperature error uint8 K 1-255 0 0.04 0.0
Emis1_Err Band 1 emissivity error
uint16 n/a 0-65535 0 0.0001 0.0
Emis2_Err Band 2 emissivity error
uint16 n/a 0-65535 0 0.0001 0.0
Emis3_Err Band 3 emissivity error
uint16 n/a 0-65535 0 0.0001 0.0
Emis4_Err Band 4 emissivity error
uint16 n/a 0-65535 0 0.0001 0.0
Emis5_Err Band 5 emissivity error
uint16 n/a 0-65535 0 0.0001 0.0
EmisWB Wideband emissivity uint8 n/a 1-255 0 0.002 0.49
ECOSTRESSLevel-2Products
Page 5
Hulley et al. 2012 (Uncertainty Analysis Study)
LST Uncertainty (K) Surface types
Samples MODTRAN Simulations
3-band TES
5-band TES
Dense vegetation, Water, Ice, Snow
8
660,096
2.19 1.63
Rocks
48 3,960,576
1.44
1.45 Soils
45 3,713,040
0.89
0.91 Sands
10 825,120
1.12
0.99 Total
111 9,158,832
1.49 K
1.13 K
LSTUncertaintyAnalysis
TES5-bandapproachmeets~1KaccuracycapabilityforECOSTRESS(Requirementis2K)
Page 6
LEVEL-2TASK Comple;onSimulateL1/L2Products(VIIRS,ASTER) 01-31-2016
L2Documenta/on(ATBD,PSD) 02-23-2016
L2CodeconversiontoC++ 07-22-2016
Installa/onofnecessarylibraries,radia/vetransfermodels,Ancillarydata(ASTER)
07-25-2016
Metadata,uncertain/es,cloudmask,errorlogs 07-29-2016
L2codetes/ngonsimulateddata 08-15-2016
BaselineL2PGEwithProcessControlSystem(PCS) 09-30-2016
IncorporateNCEPatmosphericdata(backup)
12-15-2016
ImplementWaterVaporScaling(WVS)Model 1-25-2016
Documenta/on(CloudATBD,ASD,UserGuide) Ongoing
CloudMaskevalua/on/refinement Ongoing
TestandimplementGEOS-5FPatmosphericdata Ongoing
Updatealgorithmcoefficients/dataforband4spectralshiW Ongoing
L2AlgorithmStatus
Page 7
SpectralResponseFunc/ons(bestes/mateasof8.25.2016)
Band CentralWavelength(µm)
Bandwidth(µm)
1 8.29 0.355
2 8.80 0.309
3 9.20 0.395
4 11.41(10.6) 0.553
5 12.09 0.610
Atmospheric‘windows’
O3H2O
H2OCO2
SRFalgorithmicdependencies:• Emissivitycalibra/oncurve• BrightnesstemperatureLUT’s• RadianceconversionLUT’s• RTmodelcoefficientfiles• Uncertaintyes/mates
TransmiZance
Page 8
SpectralResponseFunc/ons(ExpectedshiW05.2017)
Atmospheric‘windows’
O3H2O
H2O
Band CentralWavelength(µm)
Bandwidth(µm)
1 8.29 0.355
2 8.80 0.309
3 9.20 0.395
4 11.41(10.6) 0.553
5 12.09 0.610
Page 9
SpectralResponseFunc/ons
Page 10
SpectralResponseFunc/ons
Page 11
LST RMSE Surface types
V1*
V2* (band 4 shift)
Graybody (Veg, Water, Snow)
0.555K 0.542K
Barren (sands, soils, rocks)
0.974K 0.958K
Total 0.79K 0.78K
TESAlgorithmperformancewithBand4shiW
• MODTRANRadianceSimula/ons• ECOSTRESSTESalgorithm• GlobalRadiosondeprofiles(SeeBor)• ASTERspectrallibraryforemissivity
*V1=ECOSTRESSspectralresponsefrombestes/mate08/25/16V2=ECOSTRESSV1withsimulatedband4shiWfrom11.2to10.6micron
Page 12
NCEPGDAS MERRA-2 GEOS-5FP
Descrip/on Reanalysisfrom1979-present
Reanalysisfrom1979-present
Analysesandforecastsinnearreal-/meusingrecentGEOS-5model
Spa/alResolu/on
1.0degree(~100km)
0.5degree(~50km)
0.25degree(~25km)
TemporalResolu/on
6hourly 6hourly 3hourly
Ver/calResolu/on
26pressurelevels 42pressurelevels 42pressurelevels
Latency Daily 1month Daily
NumericalWeatherPredic/on(NWP)ModelsusedforECOSTRESSAtmosphericcorrec/on
Page 13
MERRA-2TotalWaterVapor(mm)0.5x0.625deg(~50km)
Page 14
GEOS5-FPTotalWaterVapor(mm)0.25deg(~25km)
Page 15
Redwood
TexasGrassland
LST&EValida/onSites(Stage1,‘Tes/ng’)
Page 16
9%reduc/oninerror
Insitu
HurrianceOdile
Page 17
30%reduc/oninerror
Insitu
Page 18
Summary• ECOSTRESSL2Products:– LandSurfaceTemperature(LST)– SpectralEmissivity(5bands)– BroadbandEmissivity– CloudMask
• Welldefinedandstrongalgorithmheritage(ASTER/MODIS/VIIRS/MASTER)
• L2CcodetestedandbaselinedinSDS• Band4shiWhasnoeffectonL2accuracy• Usinghigherresolu/onGEOS5-FPresultsin10-30%higheraccuracyofretrievedLST
Page 19
CloudMaskOutput
• 8-bitproduct
Table3.8bitLevel2CloudMaskProduct.
X Uncalibratedband–dynamicthresholdrequiredpersceneMayoveres/mateovermostlandsurfaces
Carefulconsidera/onoflandsurfaceemissivityrequired
Page 20
GoogleEarth:08/28/14MASTERLST:08/26/2014
319.2K323.8K
1. Signatures of vegetation stress are manifested in the LST signal before any visible deterioration of vegetation cover occurs.
2. The surface moisture state can be deduced directly from the remotely sensed LST.
400m
Page 21
Theore/calBasis:SurfaceTemperature
⎥⎥⎦
⎤
⎢⎢⎣
⎡−⎟⎟⎠
⎞⎜⎜⎝
⎛=
1exp 25
1
sTccBλλ
λ
EmissivityIrradianceSky gDownwellin
Reflection Surface eTemperatur Brightness Surface)
eTemperatur Surface wavelength=
radiance spectralblackbody :where
=
=
=
=
=
=
↓
λ
λ
λ
ε
ρ
λ
L
(θTT
λ
s
B
AsthetemperatureincreasesthepeakinthePlanckfunc/onshiWstoshorterandshorterwavelengths
⎟⎟⎠
⎞⎜⎜⎝
⎛ −=
↓−
λ
λλλ ε
ρθ LLBTs)(1
0
20
40
60
80
100
Ra
dia
nc
e (
W/m
*m*m
)/1
.0e
6
4 6 8 10 12 14 16 18 20 Wavelength (micrometers)
450K
350K
273.15K
ThermalInfrared(TIR)
LandSurface(‘Skin’)Temperature(LST)
( ))()( 1 θθ λλλ LBT −=
Radiometric(‘Brightness’)Temperature PlanckFunc/on
Page 22
22
TemperatureEmissivitySepara;on(TES)Algorithm‘ASTERapproach’
EmissivityCalibra/onCurve
T-E separation is under-determined: If have N measurements, always have N+1 unknowns: Radiance Band 1 = T + emissivity1 Radiance Band 2 = T + emissivity2 Radiance Band 3 = T + emissivity3 Radiance Band 4 = T + emissivity4 Radiance Band 5 = T + emissivity5 …………………………. ………………………
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0
Emissivity(0
-1)
Wavelength(micrometers)
Emisssivitiesforvarietyofmaterials
GraniteBasaltGrassWaterSandstoneInseptisol
Page 23
↑↓+⎥⎦⎤
⎢⎣⎡ −+= λλλλλλ εθτθ LLLL surf )1()()( )(TsBLsurf λλ ε=
Itera;velysolveforsurfaceradiance+Temperature/Emissivity