Retrieving Land Surface Temperature from Hyperspectral

Preparing to load PDF file. please wait...

0 of 0
100%
Retrieving Land Surface Temperature from Hyperspectral

Transcript Of Retrieving Land Surface Temperature from Hyperspectral

sensors
Article
Retrieving Land Surface Temperature from Hyperspectral Thermal Infrared Data Using a Multi-Channel Method
Xinke Zhong 1, Xing Huo 2, Chao Ren 3, Jelila Labed 1 and Zhao-Liang Li 4,*
1 ICube, UdS, CNRS, 300 Bld Sebastien Brant, CS10413, Illkirch 67412, France; [email protected] (X.Z.); [email protected] (J.L.)
2 School of Computer and Information, Hefei University of Technology, Hefei 230009, China; [email protected]
3 College of Geometics and Geoinformation, Guilin University of Technology, Guilin 541004, China; [email protected]
4 Key Laboratory of Agri-Informatics, Ministry of Agriculture/Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China
* Correspondence: [email protected]; Tel.: +86-10-8210-5077
Academic Editor: Assefa M. Melesse Received: 19 February 2016; Accepted: 9 May 2016; Published: 13 May 2016
Abstract: Land Surface Temperature (LST) is a key parameter in climate systems. The methods for retrieving LST from hyperspectral thermal infrared data either require accurate atmospheric profile data or require thousands of continuous channels. We aim to retrieve LST for natural land surfaces from hyperspectral thermal infrared data using an adapted multi-channel method taking Land Surface Emissivity (LSE) properly into consideration. In the adapted method, LST can be retrieved by a linear function of 36 brightness temperatures at Top of Atmosphere (TOA) using channels where LSE has high values. We evaluated the adapted method using simulation data at nadir and satellite data near nadir. The Root Mean Square Error (RMSE) of the LST retrieved from the simulation data is 0.90 K. Compared with an LST product from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) on Meteosat, the error in the LST retrieved from the Infared Atmospheric Sounding Interferometer (IASI) is approximately 1.6 K. The adapted method can be used for the near-real-time production of an LST product and to provide the physical method to simultaneously retrieve atmospheric profiles, LST, and LSE with a first-guess LST value. The limitations of the adapted method are that it requires the minimum LSE in the spectral interval of 800–950 cm´1 larger than 0.95 and it has not been extended for off-nadir measurements.
Keywords: land surface temperature; hyperspectral thermal infrared; multi-channel method; land surface emissivity

1. Introduction
Land Surface Temperature (LST) is a key parameter in climate systems. LST is used for Earth surface energy budget studies [1], numerical weather/climate forecasting [2], the retrieval of climate variables [3], soil moisture/evapotranspiration estimations [4], and generation of time-consistent LST product [5,6]. For severe weather forecasting application, the near-real-time LST can provide important diagnostic information [7]. Thermal infrared remote sensing has become an effective method to measure LST on large spatial scales [8,9].
Various methods to retrieve LST from satellite-based multispectral thermal infrared data include the following: the single-channel method [10], the Split-Window (SW) method [11–14], the multi-channel method [1–17], the multi-angle method [18], the physical-based day/night operational

Sensors 2016, 16, 687; doi:10.3390/s16050687

www.mdpi.com/journal/sensors

Sensors 2016, 16, 687

2 of 14

method [19], the Temperature and Emissivity Separation (TES) method [20], the multi-temporal physical method [21], the Kalman filter physical method [22], and the Two-Step Retrieval Method (TSRM) [23,24]. The single channel method requires good knowledge of the Land Surface Emissivity (LSE) at the channel used and an accurate atmospheric profile. The SW method requires accurate atmospheric water vapor content and LSE for land applications [8]. The use of the multi-channel method is limited by the uncertainty in LSE for the two mid-infrared channels (3–6 µm) being larger than that of the channels centered between 10 µm and 12 µm [25]. The multi-angle method suffers from the phenomenon of LSE and LST angular dependence [26]. The physical-based day/night operational method suffers from problems of geometry mis-registration, variations in the viewing zenith angle, and inaccurate atmospheric corrections [27]. The TES method, the multi-temporal physical method, and the Kalman-filter physical method require good atmospheric corrections [21,22,28]. The requirement of adequate channels and the TRSM method’s complex nature make it difficult to apply. The expected accuracy of a LST product from thermal infrared sensors is less than 1 K [29], however, this has not yet been achieved.
The hyperspectral thermal infrared data from sensors, such as the Infrared Atmospheric Sounding Interferometer (IASI) [30] and the Cross-track Infrared Sounder (CrIS) [31], have thousands of channels and provide a wealth of information on the atmosphere and the land surface. This type of data provides a new opportunity for methodological development in retrieving LST from satellite data.
The methods for retrieving LST from space-borne hyperspectral thermal infrared data can be classified into two types: empirical methods [32–38] and physical based methods [39–44]. The latter either require atmospheric profiles or are difficult to apply due to their complex nature. The empirical methods, which include the Artificial Neural Network (ANN) method and the principal component regression method, are based on a linear/nonlinear empirical relation between principal component amplitudes of the brightness temperature spectrum at the Top of the Atmosphere (TOA) and LST [32–37]. The principal component regression method and the ANN method are fast enough for near real-time applications [39]. However, the empirical methods require thousands of channels, which are not available for measurements because they contain damaged data at certain channels. It is required to develop a flexible multi-channel method for retrieving LST from hyperspectral thermal infrared data with less channels. Previously, we developed a multi-channel method to retrieve surface temperature for high emissivity surfaces from IASI data using brightness temperatures at TOA at 10 channels [38]. However, the multi-channel method for high emissivity surfaces mentioned above requires the assumption of blackbody LSE. The objective of this paper is to adapt the multi-channel method for retrieving LST from hyperspectral thermal infrared data for natural land surfaces while properly taking LSE into consideration.
The adapted multi-channel method can be used for retrieving LST for natural land surface from hyperspectral thermal infrared data containing damaged data at certain channels. The adapted multi-channel method is also promising for near-real-time production of LST products from hyperspectral thermal infrared data.
This paper is organized as follows: Section 2 adapts the multi-channel method using simulation data with typical LSE data of natural land surfaces. The sensitivity of the adapted method to instrumental noise and LSE is shown in Section 3. The evaluation of the adapted method using simulation data and satellite data is shown in Section 4, and the last section lists the conclusions.

2. Methodology

2.1. Multi-Channel Method
According to the multi-channel method for retrieving LST from hyperspectral thermal infrared data for high emissivity surfaces [38], LST can be retrieved from:

ÿ

Ts “ w0 `

wi Tbi

(1)

i“1:p

Sensors 2016, 16, 687

3 of 14

where Ts is the LST, wi (i = [1, p]) are the regression coefficients, and Tbi is the brightness temperature at TOA at channel i. The number of channels is p, and the center wavenumbers at channel
i (i = [1, p]) and the wi (i = [0, p]) coefficients can be determined using simulation data with LSE of unity. The determined central wavenumbers of the channels are in the spectral interval of 800–1200 cm´1.
The specifics for determining the wi coefficients and the central wavenumbers of the channels are shown in [38]. This multi-channel method relies on the assumption that LSE is equal to one and can
not be directly used for applications on natural land surfaces.

2.2. Adaptation for Natural Land Surfaces

2.2.1. Adapted Multi-Channel Method

The previous multi-channel method is limited to applications on high emissivity surfaces. To extend the previous multi-channel method, we developed the adapted multi-channel method for retrieving LST from hyperspectral thermal infrared data for natural land surfaces. In the adapted multi-channel method, with the assumption that the LSE has high values in the spectral interval of [νa, νb] cm´1, LST can be retrieved from:

ÿ

Ts “ α0 `

pαs,i Tbs,i ` αw,i Tbw,i q

(2)

i“1:n

where α0, αs,i (i = [1:n]) and αw,i (i = [1:n]) are the regression coefficients (also called αi in this paper), and Tbs,i (i = [1:n]) and Tbw,i (i = [1:n]) are the brightness temperatures at TOA at a channel-pair centered in the spectral interval of [νa, νb]. A channel-pair includes two nearby channels where the absorption of water vapor is strong in one channel and is weak in another channel. The α0, αs,i„ and αw,i (i = [1:n]) coefficients and the central wavenumbers of the channels are determined using simulation data as described below. As a trade-off between the loss of information and the number of required channels, we only used the brightness temperatures at the channel-pairs, which represent the main feature of the brightness temperature spectrum at TOA in the spectral interval of [νa, νb].

2.2.2. Analysis of the Variation in LSE
To determine the spectral interval where LSE has high values, we used typical LSE data from the Advanced Spaceborne Thermal Emission Radiometer (ASTER) emissivity library to study the variation in LSE. The wavelength of the LSE data in the ASTER library is from 714 cm´1 to 25,000 cm´1. The materials in the LSE data in the ASTER library include rocks, minerals, soils, vegetation, water bodies, meteorites, and manmade materials. Because pure pixels of rocks, minerals, meteorites, and manmade materials are rare in hyperspectral thermal infrared data with a spatial resolution of 12 km, we did not use the LSE data for these four types of materials for this analysis. Specifically, the materials of the LSE data used include water, snow, ice, three vegetation types, and 41 soils. The absorption of atmospheric ozone is strong in the spectral interval of 985–1071 cm´1. To eliminate the effect of atmospheric ozone, the spectral intervals used for this analysis were 800–985 cm´1 and 1071–1200 cm´1. The criteria for determining the spectral interval of [νa, νb] in Equation (2) is that the mean values of the channel LSEs are larger than 0.95 and the standard deviations of the channel LSEs are not larger than 0.01.
The mean values of the channel LSEs and the standard deviations of the channel LSEs as a function of wavenumber are shown in Figure 1. The mean channel LSEs in the spectral interval of 800–950 cm´1 are larger than 0.95, and the corresponding standard deviations of the channel LSEs are approximately 0.01. The mean channel LSE decreases to about 0.943 in the spectral interval of 1071–1200 cm´1 and the standard deviation of the channel LSE increases to high values in the spectral interval of 1071–1200 cm´1, ranging between 0.03 and 0.045. We only considered the channels in the spectral interval of 800–950 cm´1 in the determination of the central wavenumbers of the channels.

SSeennssoorrss22001166,, 1166,, 668877

44ooff1144

Figure 1. The mean values of the channel LSEs and the standard deviations of the channel LSEs as a Ffuignucrteio1n. oTfhwe amveeannumvableure.s(sotfdth=estcahnadnanredl dLeSvEisatainodn othfechstaannndealrLdSdEe).viations of the channel LSEs as a function of wavenumber. (std = standard deviation of channel LSE).
22..22..33.. DDeetteerrmmiinnaattiioonnooffCCeennttrraallWWaavveennuummbbeerrss ooff tthhee CChhaannnneellss TToo ddeetteerrmmiinnee tthhee cceennttrraall wwaavveennuummbbeerr ooff ssttrroonngg--aabbssoorrppttiioonn cchhaannnneellss aanndd wweeaakk--aabbssoorrppttiioonn
cchhaannnneellss iinn EEqquuaattiioonn ((22)),, wwee ssiimmuullaatteedd ddaattaa uussiinngg tthhee OOppeerraattiioonnaall rreelleeaassee ffoorr AAuuttoommaattiizzeedd AAttmmoosspphheerriicc AAbbssoorrppttiioonn AAttllaass ((44AA//OOPP))[4[455,4,466]]wwitihthththeeMMididddlele-l-alatittiutuddee--ssuummmmeerraattmmoosspphheerriiccpprrooffiillee eexxttrraacctteedd frforommthethMe OMDOTRDATNRAsoNftwsaorfetw. Tahree.atTmhoespahtemriocspprhoefirliec copnrsoifsitlse ofcoatnmsiosstsphoefricattemmopseprhaeturirce taenmdpmeroaitsutruereanpdrofimleosistthuartehparvoefi3le5slatyhearts hfraovme 31501l3ayhePras tforo0m.011h0P1a3. hTPhae ttoota0l.0p1rehcPipai.taTbhlee wtoattaelr vapor and the bottom temperature of the atmospheric profile are 2.91 g/cm2 and 294.2 K, respectively. precipitable water vapor and the bottom temperature of the atmospheric profile are 2.91 g/cm2 and 2T9h4e.2LKS,TreasnpdectLivSeElyf.oTr hteheLSsTimanudlaLtiSoEn fworetrhee2s9im9.u7 laKtioanndwuerneit2y9,9r.7esKpeacntdivuenlyi.ty,4rAes/pOePctiwvealsy.u4sAed/OtPo wsiamsuulsaetde tthoesiamtmuolastpehthereicatrmadoisaptihveeritcerrmadsiaftoirvemteearsmusrefomremntesaasut rneamdeirnwtsiatht nthaediMr wOiDtherthateeMreOsoDluertaioten raetsmoloustpiohneraictmToRsAphNesrmicisTsRioAnNasnmdisrsaiodniananced (rMadOiaDnTceRA(MNO) DsoTfRtwAaNre) sboufitlwd-airne pbruoifildle-isn. pTrhoefilsepse. cTthrael interval and spectral sampling frequency for the simulation were 800–950 cm´1 and 0.25 cm´1, spectral interval and spectral sampling frequency for the simulation were 800–950 cm−1 and 0.25 cm−1, rreessppeeccttiivveellyy.. TThheeraraddiaiatitviveetrtraannssfeferreeqquuaatitoionnwwaass uusseedd ttoo ssiimmuullaattee tthhee bbrriigghhttnneessss tteemmppeerraattuurree ssppeeccttrruummaattTTOOAAfofronrandairdoirbsoebrsveartvioantisownsithwtihthe athtme oastpmhoesrpichreardiciartaivdeiatetirvmestferrommstfhreo4mA/tOheP4mAo/dOelP. Wmiotdhelt.hWe ibthrigthhetnbersigshttenmespsetreamtupreeradtautrae dsiamtausliamteudlautesdinugsitnygpitcyaplicaatlmaotmspohseprhicerpicrpofriolefiledadtaat,a,wweeffiirrsstt sseelleecctteeddtthheewweeaakk--aabbssoorrppttiioonncchhaannnneellssffrroommeeaacchhmmiiccrroo aattmmoosspphheerriicc wwiinnddooww iinn tthheessppeeccttrraalliinntteerrvvaall ooff 880000––995500 ccmm−´1.1.ThTehne,n,wwe esesleelcetcetdednneeaarbrbyysstrtoronngg-a-abbsosorprptitoionncchhaannnneelsl,s,wwhhicichhhhaadd aa bbrriigghhttnneessss tteemmppeerraattuurree ddififfeferreenncceelalragregretrhatnhacnKccoKmpcaormedpwarietdh twheimthsetlhveems asseltvheesdaetsertmheineddetsetrrmoning-eadbssotrrpotniognacbhsaonrnpetliso.nTchheavnanleules.oTf hcefovratlhuies osifmcufloartetdhibsrsigimhtunleastesdtebmripgehrtantuesrse tdeamtapiesraantuerme dpairtiacaisl vaanlueemopfi2ri.5ca. l valueTohfe2d.5e.termined central wavenumbers of the channels and the simulated brightness temperature specTtrhuemdeatterTmOinAedarceensthroalwwnavinenFuimgubreers2o. f tThheech3a6ncnhealsnannedlsthceenstimeruedlatiendtbhreigshptnecetsrsatleminpteerrvaatul roef s8p0e0c–t9r5u0mcmat´T1OaAreaurseesdhionwtnhiisnsFtuigduy.re 2. The 36 channels centered in the spectral interval of 800–950 cm−1 aTroeaunsaeldyzine tthheisrsetpurdesye.ntativeness of the selected channels, we simulated brightness temperature dataTaos aanbaolvyezeutshinegretphreestreonptaictiavleannedsstohfethsuebs-epleocltaerd-wchinatnenrealtsm, woespshimeruiclapterdofiblreigdhattnaesesxtteramcpteedraftruorme dMaOtaDaTsRaAboNv.eThuesitnogtatlhperetcrioppiitcaablleawndattehrevasupobr-ps ofolarrt-hweitnrotepricaatlmporospfihleearincdptrhoefisluebd-paotalaerx-wtrianctteerdpfrroofimle MarOe D4.0T8RgA/Ncm. T2haendto0ta.4l2pgr/eccimp2it,arbelsepewcatitveerlyv.aTphoersbofottrotmhetetmroppeircaatluprersoffioler tahnedtrtohpeicsaulbp-rpoofillaer-awnidntthere psurobf-ipleolaarre-w4.i0n8tegr/pcmro2fialendar0e.42299g.7/cKm2a,nrdes2p5e7c.t2ivKe,lyre.sTpheectbivoetltyo.mThteismspimerualtautreeds bforirgthhtenetsrsopteicmapl eprraotfuilree adnadtatihseshsuowb-npoinlaFr-iwguinrete2r. pFrroofmileFiagruer2e929,.7weKcaanndse2e5t7h.2atKth, erebsrpigehcttinveeslsy.teTmhpisersaimtuurelasteadt thberigsehltencetesds tcehmapnnerealstuarreedreaptareissesnhtoawtivneinofFtihgeureen2ti.rFerbormighFitgnuesrse t2e,mwpeecraantusreeestpheactttrhuembrinig8h0t0n–e9ss85tecmmp´e1rafoturrtehseaset tthyepiscealleacttemdocsphahnernieclscoanrdeitrieopnrse.sentative of the entire brightness temperature spectrum in 800–985 cm−1 for these typical atmospheric conditions.

Sensors 2016, 16, 687 SSeennssoorrss 22001166,, 1166,, 668877

5 of 14 5 of 14

Figure 2. The central wavenumbers of the channels illustrated with the simulated brightness

temFipgeurraetu2r.e Figure 2.

TsTphheeectccreeunnmttrraaallt

wTwOaavAveen(nTuuommpbbeoerfrssAotomff otthhsepe hccehhraaenn).nnTeellhsse

iibllllouutsstttorrmaatteetddemwwpiietthhrattthuheeressiiismm2uu9ll4aa.tt2eeddK,bbarrniiggdhhttthnneeessLssST

(LatteenmmdppSeeurrraafttauucrreee TssppeemecctptrreuurmmatuaatrteTT)OOisAA2(9T(9To.op7pKoof.fAAtmtmosopshpehreer)e. )T. hTehbeotbtootmtotmemtepmerpaeturaretuirse2i9s4.229K4.,2aKnd, athnedLthSTe

L(LSaTn(dLaSnudrfaScuerfTaecme Tpeemraptuerraet)uisre2)9i9s.729K9..7 K.

2.2.4. Determination of the αi Coefficients

22..22T..44o.. DDdeeettteeerrrmmmiiinnnaaettiitoohnne ooαffi ttchhoeeeαfαfiiicCCiooeneefftfifsi,cciiweennettsssimulated a large amount of data using 4A/OP with typical

atmospTThooedrdieectteeprrmrmoifinnileeestthhfeeroααmii ccooteehffefificcTiieehnnettrssm,,wwoedesysinimmaumullaiactteedIdnaiatlilaaarlrggGeeauamemsosouuRnntettoorfifeddvaaattalau(uTssiIniGngRg4)4AAd//aOOtaPPbwwasiietthh[t4tyy7pp,4iicc8aa]ll as meaanttmmtiooonssppedhheearrbiiccovpperroo. fifTillheeess affrrtomommosttphhheeeTTrihhceeprrmmrooofiddleyynndaaammtaiiccanIInndiitttiihaaell GGLSuuTeessdssaRRtaeettfrroiieervvtaahlle((sTTiIImGGuRRl))atddioaattnaabbtoaassdeee[[t44e77rm,,4488in]] eaassthe αi mcmoeeennffttiiocoinneeenddtsaabwbooevvreee.. tTThhheeesaaattmmeoossappshhteehrroiiccspeprrdooefifisllceerddibaaettaadaainndd[3tthh8ee].LLFSSoTTr ddeaatctaahffosoirrmtthhueelasstiiimmonuullcaaottiinoodnnittooioddnee,ttteehrrmemLiinnSeeEtthhdeeata forααtiihcceooeesffififmicciuieelnnatttssiowwneewrreeattshhteehsseaamdmaeetaaassretthhfeoorssreeedddeetssoccrriinibbeSeddeciitnnio[[3n3882]]..2FF.1oo.rrAeeaarccahhnssdiimmomuullaanttoiiooisnneccwoonnitddhiittaiiooNnn,,ottihhsee LLESSqEEuiddvaaattlaaent TeffmoorrpttehhreeatssuiimmreuuDllaaittfiifooennrewwnaacses tt(hhNeeEdd∆aaTttaa) roreefffe0err.1rreeKdd twtooaiinsnaSSdeecdctteiioodnnt2o2..22th..11e.. AsAimrraaunnlddaootemmd nnbooriisgseehwtwniietthshsaateNNmooipisseeerEEaqtquuuriievvaadlleaentnatt at TOTTAeemm. UppeserirnaatgtuutrrheeeDDliaiffrffegerereensnicmceeu((NNlaEEti∆∆oTTn))dooafft00a..,11wKKewwdaeastseaarddmddieenddedttootththheeeαssiiimcmouuellfaaftitceeiddenbbtrrsiiggihnhttEnneqessussattteeimomnpp(ee2rr)aatwtuuirrteehddtaahtteaalaaettast squTTOaOrAAe..mUUesstiihnngogdtthh. ee llaarrggee ssiimmuullaattiioonn ddaattaa,, wwee ddeetteerrmmiinneedd tthhee ααii ccooeefffificciieennttss iinn EEqquuaattiioonn ((22)) wwiitthh tthhee lleeaasstt
ssqquuTaahrreee mdmeeetttehhroomddi..ned αi coefficients are shown in Figure 3. The αi (i = [1, 36]) coefficients vary over a small raTTnhhgeeeddfeerttoeemrrmm−iinn1e.e5ddtααoii2cc.oo0ee.fffificciieennttss aarree sshhoowwnn iinn FFiigguurree 33.. TThhee ααii ((ii == [[11,, 3366]])) ccooeefffificciieennttss vvaarryy oovveerr aa
ssmmaallll rraannggee ffrroomm ´−11..55ttoo22..00..

FigFFuiiggrueurre3e.33T.. hTTehheαe iααcii occeoofeeffifcffiiicceiineentnsttssaaannndddthtthheee ccceeennntttrrraaalll wwwaavveennuummmbbbeeerrrsss oooffftththheeecchchhaanannnnneellesslsddedetteeetrremmrmiinnieenddeduussuiinnsgginttghheethe simulation data.
simsiumlautliaotniodnadtaat.a.

Sensors 2016, 16, 668877

66 ooff 1144

3. Sensitivity Analysis 3. Sensitivity Analysis
3.1. Sensitivity to Land Surface Emissivity 3.1. Sensitivity to Land Surface Emissivity
To analyze the sensitivity of the adapted method to LSE, we retrieved LSTs from independent simulTaotiaonnadlyaztae tuhseinsgenthsietiavditaypotefdthme eatdhaopdteadndmaenthaolydzteodLtShEe,vwareiarteitornievoef dthLeSeTrsrofrroomf tihnederpeterniedveendt LsiSmTuslwatiitohntdhaetamuinsiinmgutmheLaSdEapvtaedlume ienthtohde asnpdecatrnaallyinzteedrvthael ovfar8i0a0ti–o9n50ofctmhe−1.erTrhoer oaftmthoesrpehtreireivcepdrLofSiTles dwaittah athnedmthineimLSuTmdLatSaEfvoarlutheeinintdheepsepnedcetrnatl siinmteurlvaatlioonf 8c0a0n–b95e0focmun´d1.inTh[3e8a]t.mTohsepthoetrailcpprreocfiipleitdabaltea wanadtetrhveaLpSoTr odfatthaefosreltehceteidndaetmpeonsdpehnetriscimpruolfaitlieosnracnangebdeffrooumnd0 ign/c[m382]t.oT5heg/tcomta2l. pTrheeciLpSitEabdlaetawaantedr tvhaepionrstorfutmheensteallecntoeidseaftomrotshpehseimricuplartoiofinleasrreatnhgoesde mfroemnti0onge/dcmin2Steoct5iogn/c2m. T2o. Tcohme pLuStEe dthaetastaantdisttihcse oinfstthruemLSenTtaelrrnooriss,ewfoerctlhaesssiifmieudlatthieonsiamreutlhatoisoenmcaesnetsioinnetdo ifnouSrecdtaiotanb2a.sTeos caocmcoprduitnegthteostthaetismtiicnsiomfuthme LLSSET verarloures,fowrewchlaiscshifithede mthiensiimmuumlatLioSnEcvaasleuseisnatorefosuhrowdantaibnaTseasblaecc1o. rTdhinegertrootrhseomf tihneimreutmrieLvSeEd vLaSlTuse ffoorr ewahchicshimthuelamtiionnimduatmabLaSseEavsaalufeusnactrieonshoofwthneimn Tinaibmleu1m. TLhSEe evrarlouresionftthheesrpeetrcitervaleidntLeSrTvsalfoorf 8e0a0ch– 9s5im0 ucmlat−i1oanredsahtaobwanseinasFaigfuurnect4i.on of the minimum LSE value in the spectral interval of 800–950 cm´1 are shown in Figure 4.

Table 1. Minimum channel LSE values for the four databases for analysis of the sensitivity of the aTdaabplete1d. mMeitnhiomdutmo LcShEa.nnel LSE values for the four databases for analysis of the sensitivity of the adapted method to LSE.

No. of Database

1

2

3

4

MiNniom. oufmDacthaabnanseel LSE values 1 [0.95, 0.96] 2[0.96, 0.97] [0.397, 0.98] [0.98, 40.99]

Minimum channel LSE values

[0.95, 0.96]

[0.96, 0.97]

[0.97, 0.98]

[0.98, 0.99]

Figure 4. The errors of the retrieved LSTs for each simulation database as a function of the minimum FLiSgEurveal4u.eTihnetehrerosrpseoctfrtahleinrteetrrvieavleodf 8L0S0T–s9f5o0rcemac´h1.si(mRMulSaEtio=nRdoaotat bMaeseanasSqaufuarnectEiorrnoro)f. the minimum LSE value in the spectral interval of 800–950 cm−1. (RMSE = Root Mean Square Error).
AAss tthhee mmiinniimmuumm LLSSEE vvaalluuee iinn tthhee ssppeeccttrraall iinntteerrvvaall ooff880000––995500ccmm´−11ggrroowwssffrroomm aapppprrooxxiimmaatteellyy 00..9955 ttoo aapppprrooxxiimmaatteellyy 00..9988,, tthhee bbiiaass ooff tthhee rreettrriieevveedd LLSSTTss ffoorr tthhee ssiimmuullaattiioonn ddaattaabbaassee wwiitthh tthhee ccoorrrreessppoonnddiinngg LLSSEE ccoonnddiittiioonn ggrroowwss ffrroomm −´00.2.2KKttoo00..33KK aanndd tthhee ccoorrrreessppoonnddiinngg RRoooott MMeeaann SSqquuaarree EErrrroorr ((RRMMSSEE)) ooff tthhee rreettrriieevveedd LLSSTTss ddeeccrreeaasseess ffrroomm 11..2255 KK ttoo 00..8855 KK.. TThhee LLSSTT rreettrriieevveedd bbyy tthhee aaddaapptteedd mmuullttii--cchhaannnneell mmeetthhoodd hhaass llaarrggeerr eerrrroorrss wwhheenn tthhee mmiinniimmuumm LLSSEE vvaalluuee iiss llooww.. NNoottee tthhaatt tthhee RRMMSSEE ooff tthhee rreettrriieevveedd LLSSTTss ffoorr tthhee ssiimmuullaattiioonn ddaattaa wwiitthh aa mmiinniimmuumm LLSSEE ooff 00..9955 iiss lleesssstthhaann11..2255KK..
TThhee LLSSTT rreettrriieevveedd bbyy tthhee aaddaapptteedd mmuullttii--cchhaannnneell mmeetthhoodd iiss aann iinnccrreeaassiinngg ffuunnccttiioonn ooff bbrriigghhttnneessss tteemmppeerraattuurreess aatt TTOOAAaattththeesesleelcetcetdedchcahnannenleslasnadndbribgrhigtnhetnssestesmtepmerpaetruarteuaret TaOt ATOaAt aacthaancnhealninieslailsios aalnsoinacnreianscirnegasfiunngctfiuonnctoifonLSoEf aLtStEhiast cthhaisncnheal;nanseal; raessualrte, sthuelt,btihase obfiatsheofrethtreierveetrdieLvSeTdiLnScrTeainscersewasieths wthiethgrthowe ginrogwofintgheomf tihneimmuinmimchuamnncehlaLnSnEelinLStEheintrtehnedt.rend.

Sensors 2016, 16, 687 Sensors 2016, 16, 687

7 of 14 7 of 14

3.2. Sensitivity to Instrumental Noise

3.2. STenosictoivnidtyutcot Itnhsitsrusemnesnitiavl iNtyoiasne alysis, we created three simulation databases by adding noise to noiseTloescsoInAdSuIcdtatthaiswsietnhsNitiEv∆itTy =an0a.1lyKsi,s0,.w2 Ke ,craenadte0d.3thKr.eTehseimnuoilsaetiloesnsdIAatSaIbdasaetas bwyeraeddcrienagtendoiusseintog nthoeisienldesespIeAnSdIendtataatmwoitshphNeEri∆cTpr=o0fi.l1eKd,a0ta.2, tKh,eaLnSdT0d.3aKta., Tahnde nthoeisLelSeEssdIaAtaSImdeantatiowneerde icnreSaetcetdiounsi3n.1g. tThheeinmdeetpheonddfeonrt aadtmdionsgphneoriisceptroofithlee dnaotias,etlhesesLsSiTmdualattai,oannddatthaecLaSnEbdeaftoaumnednitnion[3e8d].inThSeecatdioanp3te.1d. Tmhueltmi-cehthaondneflomr eatdhdoidngwansouisseedtotothreetnrioeivseelLeSssTsfirmomultahtieotnhrdeaetsaimcaunlabteiofnoduantdabianse[3s8. F].igTuhree 5addaepptiectds mthueletir-rcohrasnonfetlhme LetShTosdrewtraisevuesdedfrtoomreetraicehvesiLmSuTlafrtioomn dthaetatbharesee asismaufulanticotinondaotfatbhaeseins.stFriugmureen5tadl enpoiicstes. the errors of the LSTs retrieved from each simulation database as a function of the instrumental noise.

Figure 5. The errors of the LSTs retrieved from each noise-added simulation database as a function of tFhieguinrsetr5u. mTheenetarlronrosisoef.the LSTs retrieved from each noise-added simulation database as a function of the instrumental noise.
WWhheenn tthheeNNEE∆∆TTfofor rthtehseimsiumlautliaotniodnadtaabtaasbeawseaws eaqsueaql utoalthtaot tuhsaetdutoseddevtoeldopevtehleoapdtahpeteaddmaputletdimchualntin-cehl amnentehlomde(0th.1oKd )(,0t.h1eKR),MthSeERoMf tShEe roeftrthieevreedtrLieSvTesdfoLrStThsefsoirmthuelastiimonudlaattiaobnadseatwabaass0e.8w5aKs.0W.8h5eKn. WthheeNnEt∆hTe NfoEr ∆thTe fsoimr tuhleatsioimn udlaattaiobnasdeawtaabsaisnecrweaasseidncbryea0s.1edKbaynd0.10.K2 Ka,ntdhe0.R2MKS, EthseofRtMheSErestroifevthede rLeStrTiesvfeodr LtShTes cfoorrrtehsepcoonrdreinspgonsidminuglastiimonuldataitoanbadsaetaibnacsreeainscerdeabsyed0b.3y50.K35 aKndand0.505.55KK, ,rreessppeeccttiivveellyy.. TThheerreeffoorree,, tthhee aaccccuurraaccyy ooff tthhee LLSSTT rreettrriieevveedd uussiinngg EEqquuaattiioonn ((22)) iiss nnoott ssiiggnniifificcaannttllyy aaffffeecctteedd bbyy tthhee iinnssttrruummeennttaall nnooiissee..
4. Evaluation 4. Evaluation
4.1. With Independent Simulation Data 4.1. With Independent Simulation Data
We evaluated the accuracy of the adapted multi-channel method with the independent simulation data Wmeenetivoanleudateind Stehcetioancc3u.1ra. cyThoef cethnetraaldwapatveednummublteir-cshoafntnheel cmhaenthnoedls wanitdh ththeeαiincdoeepffiecnidenentst csaimlcuullaattieodn idnaStaecmtioennti2onweedreinusSeedctitoonr3e.t1r.ieTvheeLcSeTntfrraolmwtahveeniundmebpeernsdoefntthseimchualantnioenls daantda.thTehαei actomefofiscpiehnetrsiccaplrcoufilaletesdfoinr tSheectiinodne2pwenedreenutsseidmtuolraetitorinevweeLrSeTdfirffoemrenthtefrionmdeptheendaetmntossipmhuelraictiopnrodfialteas. mTheentaitomnoedspihneSreicctpioronfi2l.es for the independent simulation were different from the atmospheric profiles mentTiohneeedrrionrsSoecfttihoenL2S. Ts retrieved using Equation (2) for the independent simulation data are shown in FigTuhree 6e.rrTohres RoMf tShEeoLfSthTes rreettrriieevveeddLuSsTisnfgorEtqhueaitniodnep(e2n)dfeonrttshiemiunldateiponenddaetnatissi0m.9u0lKat.iTonhedeartraoraoref tshheorwentriienveFdigLuSreT 6is. cTohnesiRsMtenStEwoifththtehereLtSriTeveerdrorLsSmTsenfotirotnheediinndoetpheenrdreencet nstimstuuldaiteiosn[2d1,a3t5a].is 0.90 K. The error of the retrieved LST is consistent with the LST errors mentioned in other recent studies
[21,35].

Sensors 2016, 16, 687
Sensors 2016, 16, 687 Sensors 2016, 16, 687

8 of 14
8 of 14 8 of 14

FiFgiugurere6.6.TThheeeerrororrooffththeeLLSSTTreretrtrieievveedduussiinnggEEqquuaattiioonn ((22)) ffrroomm tthhe iinndependent simmuullaattiioonn ddaattaa (L(SLTSF_Tirg_eurterrteireiv6e.evdTedh=e=tehtrhereorreretortfireitevhveeeddLLSLSTSTTraeatnrndidevLLeSSdTT_u_tstrriunuege==Eqtthhueeatttiroune (L2S) Tfr)o. m the independent simulation data
(LST_retrieved = the retrieved LST and LST_true = the true LST).
4.42..2W. WitihthSSataetlelliltieteDDaatata 4.2. With Satellite Data TThheTeshismeimsuiumlalautitloiaontniomnmomoddeoeldlietilstseitelsflefhlhfaahssausununcnceecrrettaratiiannitntyyt;y; t;thhtheerererefefooforeree, ,wweeeevvaalluuaatteedd iittss aaccccuurrraaacccyyybbbyyycccooommmpppaaarririninnggg
thteheLthSLeTSLTrSerTtertrireeivterveiedevdbedbyybthytheteahdeadaapadptaetpdetdemdmumultulit-lict-ich-hcahannannnenelelmlmmeetethhtohododdffrfroromommMMMeeettotooppp---AAA///IIIAAASSSIII wwwiiittthhh ttthhheeeLLLSSSTTTppprrroooddduuucctctt frformofmrothmtehetShpSepinSinpniinninngginEEgnnhEhananhncaecnedcdeVdVisViisbiisblielbealeannadndIdnInIfnrfrafarrareerdeddImImImaagagegererr((S(SESEEVVVIIRIRRII)I))ooonnnMMMeeettteeeooosssaaattt...
TThheTethateragrtgeatregtaearteraaearseswawsewerereetrhtehetehSSeaahShaaahrraaarDaDeDesseeesrrett,r,ttt,hhteheeIIbbIebereriiraainannPPPeennininssuulala,,aannddaaffoorreesstt aaarrreeeaaaiiinnnttthhheeesssooouuutththhwwweeseststt ofofForFafrnaFncreacneacnaendadnthtdhetenhneoorntrohthrothoffSoSfppSaapiniani(nF(Fi(giFguiugrureer7e7))7..)T.ThTheheeSSaSahahhaaararaDDeesseerrtthhaass llaattiittuudddeeesss rrraaannngggiiinnngggfffrrrooommm777..00.0WWWttotoo 292.92.22E9E.a2naEndadlnoldonnglogintiutguditdeusedsrearsanrngaginnigngignfrgforfomrmom11771.2.72.N2NNttooto33333.3.55.5NNN.. T.TThhheeeddeesseerrttssuurrffaacceeiiss aa hhhooommmooogggeeennneeeooouuussslllaaannndddsssuuurrfrfafaaccecee... ThTehTeShaSehahaSraaahraaDrDaeseDesreetsrtewrwtaawsssaseselseleeclcteetcedtdebdbeebccaeacuuassueesLeLSLSTTSTrreertterrtiireeivevveededdffrrforoommmhhhyyypppeeerrrssspppeeeccctttrrraaall tthheerrmmaaallliiinnnfffrrraaarrreeeddddddaaattataaooovvveererr dedseesdreetrssteusrrtufsraufcarecfseascheahsvahevaelvaleragrleagreugnuecnuecnretcaretirantitaniietnistei[se2s[82[]82. ]8T.]h.TeThheIbeIebIrbeiearrinaianPnePnPeiennnisinunssluaullhaaahhsaalssolnloognnigtguiittduueddseerssarrnaagnnigngiignngfgroffrrmoomm9.969..6W6 toW0W.t6oEt0o.a60nE.6daElnaadtnitlduatdliatetuistdureadsnersgairnnagngignfirgnogfmrform3o6m.3163.N61.1NtoNt4oto44.4044.0N.0N.NT. .hTTehhelealnlaandnddsussurufrraffacaceceesssiininnttthhhiiisss aaarreeaa aarrreee mmmaaaiiinnnllylyyssosooiilill susrufrsafucaercfseasacenasndadsnpsdpasarpsreasleryslyevlvyeegvgeeetgateatettaedtdeldalanlnaddnsdsuusrurffaracfcaeecsse..sTT. hThihissisaarareeraeaawwwaaasssssseeellleeecccttteeedddbbbeeeccaauussee ((11)) ttthhheeessskkkyyyiisissffrfrereeqqquuueenennttltlyyly clcelaecralreoavorveorevrthetrhisitshairaseraear;e;aaan;ndadn(d2(2)()v2v)aarviraoirouiuossulsalanlnaddndssuusrurffarafcaceecettyytpyppeesessccacaannnbbbeeeuuussseeedddfffooorrrttthhheee eeevvvaalluuaattiiooonnn... TTThhheeetttaaarrrgggeeetteteeddd fofroerfseotsraterasetraeaarhehaasahslalstaittliauttuditdeusedsrearsanrngaginnigngignfgrforfomrmom44224.7.27.N7NNtotot4o4545.54.4.4NNNaanandnddlloolonnngggiititutuudddeeesssrrraaannnggiinngg ffrroomm 33..44 WWW tttooo000...333EEE...
(a(a) ) Figure 7. Cont.

Sensors 2016, 16, 687 Sensors 2016, 16, 687

9 of 14 9 of 14

(b)
FFiigguurree 77.. TThhee ttaarrggeett aarreeaass aanndd ttaarrggeett ppooiinnttss iinn ((aa)) EEuurrooppee aanndd iinn ((bb)) NNoorrtthh AAffrriiccaa uusseedd ttoo eevvaalluuaattee tthhee aaddaapptteedd mmuullttii--cchhaannnneell mmeetthhoodd wwiitthh ssaatteelllliittee ddaattaa ppllootttteedd oonn aa llaanndd ccoovveerr ttyyppee mmaapp..

The L1c data from Metop-A IASI has 8461 thermal infrared channels in the spectral interval of 645–2T7h6e0Lcm1c−1dwatiathfraosmpeMcteratol psa-AmpIAlinSgI hfraesq8u4e6n1cythoefr0m.2a5l cinmfr−1a.rTehdecshpaantniaellrseisnoltuhteiosnpeocftaranlIiAnSteIrivmaal goef 6a4t 5th–2e7n6a0dcimr p´o1inwtitish 1a2skpmec.trTahlesasmcapnliannggflereaqtutehnecyenodf 0o.f2e5accmh ´sc1a. nThlienespisat4ia8l.9r8e°s.oIlAuStiIononofMaentoIpA-SAI ismcaangsetahtethMeendaidteirrrpaonienatnisa1re2akimn.mTihde-mscoarnnainnggloerabtitths eeveenrdyodfaeya.cOh nsclyanIAliSnIedisat4a8.w98it˝h. IaAvSiIewoninMgezteonpi-tAh sacnagnlse tlhesesMtheadnit1e5rr°awneaasnuasreedaiinntmhiisds-tmuodryn.ing orbits every day. Only IASI data with a viewing zenith angleSlEeVssIRthIaonn15th˝ ewMaseutesoedsaitnhthasis esitguhdty.infrared channels in the spectral interval of 769 cm−1 to 2564 ScEmV−1I.RSIEoVnIRthI escManestetohseathehmasisepihgehrticinEfarartrhedsucrhfaanceneelvseirny t1h5emspinecitnragleionstteartvioanl aorfy7o6r9bictmw´it1htoa 2sp56a4tiaclmre´s1o.luStEioVnIRoIf s3caknms. tThheehSeEmVisIRpIh/eMriecteEoasratthLsSuTrfpacroedeuvcetryis 1u5semdinasina greefoesrteantcioentaoreyvoarlubiattewtihthe aacscpuartaicayl roesfotlhuetioLnSTof r3etkrmiev. eTdhebySEtVhIeRIa/dMapetteedosmatuLltSiT-chparnodneulctmisetuhsoedd afrsoamretfheereInAceSItodeavtaa.luTahtee tShEeVaIRccIu/MraectyeoosfatthLeSTLSpTrordeutrciteviserdetbryievtheed audsianpgtetdhemguenltei-rcahliaznedneslpmlite-twhionddofrwommeththeoIdA[S4I9]dwatiath. LTShEe SaEs VinIpRuI/t Mdaettae.oTshateLreStTripevroadl uocftLiSsEreitsribeavseedduosnintghethVeeggeentaetriaolnizCedovsperliMt-wetihnoddow[50m].ethod [49] with LSE as inpTuhtedfaivtae.-mThineurteetrMieveatol po-fALSIEASisI bLa1scedimoangtehseoVnegthetraeteiocnleCaor vdearyMs efothroedac[5h0t]a. rget area and their matcThhede SfiEvVe-ImRIi/nMueteteMoseattoipm-aAgeIAs wSIerLe1ucsiemdafgoersthoins etvharleueactiloeanr. TdhaeyssefnosrinegacthimtearogfetthearseealeacnteddtIhAeSirI mimaatcgheesdfoSrEtVhIeRSI/ahMaertaeoDseasteirmt,afgoerstwheerfeoruessetdarfoearst,haisnedvfaolur athtieonIb. eTrhiaensePnesniningstuimlaeisosfhthoewsnelienctTeadbIlAe S2I. iTmhaegdeisfffeorrenthcee bSeathwaeraenDtehseesrte,nfsoirngthteimfoereosftaanreIAasS,Iainmdagfoeratnhde tIhbaetrioafnthPeenminatscuhleadisSsEhVoIwRIniminaTgaebwlea2s. Tlehses tdhiaffnerfeinvecembientuwteese.nIAthSeI speinxseilnsgwtiitmhemoofraenthIAanSI95im%acgleeaarn-sdkythSaEt VofIRthI epimxealtschweedreSEuVseIdRIfoimr tahgies wevaasluleastsiotnh.aTnhfievcermiteirniuatefosr. IsApaStIiaplilxyelms awtcithhinmgoSrEeVthIRanI p95ix%elscleaanrd-skIAySSIEpVixIRelIspiisxethlsawt tehree udsisetdanfocer tbheitsweeveanlutahteiocne.nTtehreocfriItAerSiaI fpoirxeslpsaatinadllythmeactecnhtienrgoSfESVEIVRIIRpIixpeilxsealsndisIlAesSsI pthixaenls6iskmth.atTthhee mdiasttcahnecde bpeotiwnteseanntdhethceenlatenrdocfoIAveSrI tpyipxeelms aanpdatrheescheonwtenr oinf SFEigVuIrReI7p.iTxehles liasnledsscothvaenr t6ykpmes. oTfhtehme matachtcehdedpoairnetas ainndthtehIeblearniadncPoevneirntsyuplae wmearpe aprreimshaorwilynsionilFsiugrufraece7s. Tanhde lsapnadrsceolyvevretgyeptaetseodfstuhrefamceast.chTehde alarnead icnovtheer Itbyepreiaonf Ptheenitnasrugleatewdefroerpesritmaraerailywsaosilmsuarinfalyceesvaenrdgrsepeanrsneelyedvleegleeatfatfeodressut.rfIancetos.taTl,h1e5l4a2ndmcaotvcheredtycpaeseosf twheertearugseetdedfofrortehset eavreaaluwataisonmwaiintlhytehveesrgatreelelnitenedeadtale.leaf forest. In total, 1542 matched cases were used for the evaluation with the satellite data.

Table 2. ThTeasbelnes2in. gThtiemseenosfitnhgetsimeleecotefdthIeAsSeIleimctaegdeIsAfSoIritmheagthersefeortatrhgeettharreeeasta. rget areas.

Target Area

Sensing Time 1

TtharegeStaAhareraa desert Sens2i0n1g4T-0im8-e011 08:15

the Stahheafroardeessteartreas 20142-0081-50-10048-:0155 21:00

tthhee fIobreersitaanrePaesninsula 20152-0041-40-50251-:0050 10:10

the Iberian Peninsula

2014-05-05 10:10

Sensing Time 2 Se2n0s1i4n-g05T-i0m3e029:15
2021041-055--0083-0099:1250:55 2021051-048--0089-2009:5150:30
2014-08-09 10:30

Sensing Time 3 20S1e4n-s1i1n-g0T2i0m9e:330
20201154--1101--2052 2091::3002 20201145--1110--1205 1210::0020
2014-11-10 10:00

The comparison of the LST retrieved by the adapted multi-channel method from the IASI data withTthhee LcoSmT pparroidsounctoffrothme LSESVTIrReItriisevshedowbyn tihneFaigduaprete8d. Tmhueltrio-cohtamnneaenl msqeuthaoreddfrioffmerethneceIAbeStIwdeaetna wthiethLtShTe LreStTripevroeddufcrtofmromtheSEIVAISRIIdisatsahoawndn itnheFiLgSuTrep8r.oTdhuectroforot mmeSaEnVsIqRuIarise d1.i6ff6erKen, caenbdettwheeemnetahne
difference between the two LST datasets is 0.1 K. On the whole, there is no large difference between

Sensors 2016, 16, 687 Sensors 2016, 16, 687

10 of 14 10 of 14

Sensors 2016, 16, 687

10 of 14

LthSeTtwreotriLeSvTeddafrtoamsettsh.eOIuArSfIinddaitnagainsdcothnesiLstSeTntpwroidthucatrfercoemntSlyErVeIpRoIritsed1.6fi6nKdi,nagnodnththeims LeSaTn ddiiffffeerreennccee tb[h4ee4tw]t.weoenLtShTedtwatoasLeStsT. Odautrasfientsdiins g0.i1s Kco. nOsnisttehnetwwhiothlea, trheecerentilsynroeplaorrgteeddiffifnedreinngceobnetthwiseeLnSTthdeiftfweroeLnSceT [d4a4t]a. sets. Our finding is consistent with a recently reported finding on this LST difference [44].

FFiigguurree 88.. CCoommppaarriissoonn ooff tthhee LLSSTT rreettrriieevveedd bbyy tthhee aaddaapptteedd mmuullttii--cchhaannnneellmmeetthhooddffrroommIIAASSII//MMeettoopp--AA Fddiaagttuaarwewi8tih.thCthotehmeSpEaSVrEiIVsRoIInR/MIo/fMettheeteoesLoaSstaTLtSrLeTtSrpTireovpderudocdbtuyocntthtehonraedetahcprleteeaedr cdmlaeuyaslrtifd-ocarhyeasancnfhoetrlamregaeectthaortdeaarfgr(oSeEmt VaIIrAReaSI I=/(MSSEpeVitnoInRpi-InAg= dESapntihananwnicnietghdEVtnhihesaibnSlEceeVadInRVdIi/sIMnibfelretaeraoensdadtIImnLfaSrgTaerrep;drLoISdmTuacIgtAeoSr;nIL=tShtThreIeAerSectIlre=iaetrvheeddaryeLtsSriTfeo;varendedaLcLShSTT;taaSrnEgdeVtLIRaSrITe=aSEt(hVSeEIRSVEIIVR= IItRh=eI SLSpESiTVnnIpRirnIogLdSuETcnth)p.arnocdeudctV).isible and Infrared Imager; LST IASI = the retrieved LST; and LST SEVIRI = the SEVIRI LST product).
The spatial pattern of the difference between the LST retrieved from the IASI data by the adapted multTTi-hhceheasspnpnaatetiilaamll ppeaatthtttoeerdrnnaoonffdtthhtheeeddLiiffSffeTerrepennrcoceedbubeecttwwfreeoeemnn ttShhEeeVLLISRSTTI orreenttrr2iieeNvvoeedvdeffmrroobmmertthh2e0e1IIA4AoSSvII dedraatthaaebbySyatthhaeeraaaddDaapepstteedrdt mmisuusllhttiio--ccwhhnaanninnneellFmmigeeuttrhheoodd9.aaTnnhddetthhlaeerLLgSSeTrT pLprSrooTdduduccifttffefrrrooemmncSeSEsEVVaIrIRRe IInooennar22 NtNhooevvceelmmoubbdeer-rc22o00n11t44amoovvienerratthehdeeSSaaarhehaar,raaaDnDdeessteehrrtet iimss assxhhioomwwunnmiinnLSFFTiiggduuirfrfeeer99e..nTcTehheiesl1laa1rrggKee.rrALLsSSeTTxpddeiifcfftfeeerdree,nntchceeessaaadrraeepntneeedaarrmttuhhleeti-cccllhoouaunddn--cceoolnnmttaaemmthiionndaattceeaddnnaaorreteaab,,eaaannpddplttihheede mmtoaacxxlioimmuudum-mcoLLnSStTaTmddiinffffaeetrreeednnccheeyipisse11r1s1pKKe..cAtArsasleetxxhppeeercmctteeaddl ,i,ntthhfreeaaraedddaappdttaeetdda.mmTuhulelttili-a-ccrhhgaaennennrerelol mrmoeeftththhooeddLccSaaTnnnnreootttrbibeeevaeapdpppflrliioeemdd ttIooAccSllIooufuoddr--ccoloonnuttdaamymiainntmaatteoeddsphhhyyeppreiecrrsscppoeenccdttrirtaaillottnhhseeirrsmmaaalslloiinnrfferrpaaroreerddteddaaittnaa..aTTrhheeceellnaartrggseetueedrrryroo[rr3oo0ff].tthhee LLSSTT rreettrriieevveedd ffrroomm IIAASSII ffoorr cclloouuddyy aattmmoosspphheerriicc ccoonnddiittiioonnss iiss aallssoo rreeppoorrtteedd iinn aa rreecceenntt ssttuuddyy [[3300]]..
FFiigguurree 99.. DDiiffffeerreennccee bbeettwweeeenn tthhee LLSSTT rreettrriieevveedd bbyy tthhee aaddaapptteedd mmuullttii--cchhaannnneell mmeetthhoodd ffrroomm IIAASSII aanndd FtthhigeeuLLrSeSTT9.ppDrroiofddfeuurccettnffcrreoombmetSSwEEeVVeInIRRtIIh((eIIAALSSITI--SSrEEeVtVrIiIReRvII)e)dppllbooyttttetehddeooanndaaapqqtueudaallmiittyyulmmtia-acpphaoonffnthtehelemSSEeEVthVIoRIRdI/If/rMMomeetteeIooAssSaaItt aLLnSSdTT tpphrreoodLduSucTcttopovrvoeedrruaacttyypfpriocicamal lpSpaEarVtrtoIRof Iftht(ehIAeSSaSIha-aShrEaaVraDIReDsIe)esrpetlrootntote2ndN2ooNnvoeavmeqbmuearble2itr0y124m01(aL4pS(TLoSfIATthSIeIA=SSEIthV=eItRrheIet/rMrieeetvtreeiedovsLeaSdt TLLS(SKTT) pa(Krnod)daLunScdtTLoSvSETeVrSIaREtVIy=IpRtiIcha=eltSphEaeVrStIERoVIf LItRhSeITLSpSarThoadpruraocDtd(ueKsce)t)r(.tKo)n). 2 November 2014 (LST IASI = the retrieved LST (K) and LST SEVIRI = the SEVIRI LST product (K)).
ChannelsLseIntervalLstData