A new ArcGIS toolset for automated mapping of land surface temperature with the use of LANDSAT satellite data Jakub P. Walawenaer 1,2, Monika J. Hafto 3, Piotr Iwaniuk 4 1 Satellite Remote Sensing Centre, Institute oI Meteorology and Water ManagementNational Research Institute (IMGW-PIB), Krakow, Poland 2 Department oI Climatology, Institute oI Geography and Spatial Management, Jagiellonian University, Krakow, Poland 3 Department oI Air Pollution Monitoring and Modelling, Institute oI Meteorology and Water ManagementNational Research Institute (IMGW-PIB), Krakow, Poland 4 Technical Support Division, ESRI Polska, Warsaw, Poland ABSTRACT ThepaperpresentsanadditionalArcGISDesktop10 toolboxIorautomaticretrievaloIbrightnesstemperature (Ts),landsurIaceemissivity(LSE)andlandsurIace temperature(LST)IromLANDSATdata.TheLANDSAT TRS(7hermal5emote6ensing)Toolsemploy landsurIace temperatureestimationproceduredevelopedbyJ.C. Jimnez-MuozandJ.A.Sobrino|1,2,3and4|.Detailed description oI this algorithm is included. Special attention is paidtopracticaluseoIthelandsurIacetemperaturemaps derivedIromLANDSATdataintermsoIenvironmental research.NumerousexamplesarereIerredtoanddetailed descriptionoILANDSATlandsurIacetemperaturemaps applicationIorthermalconditionresearchinurbanareasis provided.AsampleLSTmapoIagglomerationoIKrakow in Southern Poland is included. ,QGH[7HUPVArcGISDesktop,LANDSAT,lanasurface temperature, TRS toolbox, automatea processing. 1. INTRODUCTION LandsurIacetemperature(LST)isanimportantparameter IormanyresearchapplicationsandArcGISDesktopisone oI the commonly used soItware among GIS users. ThereIore anewArcGISplug-inwascreatedbytheauthorstomake the complex LST retrieval procedure easier, Iaster and wider applicable. 2. TOOLS AND METHODS LANDSATTRSToolboxcanbeusedinordertoprocess bothLANDSATTMandETMdata.ItconsistsoIthree separatetoolscreatedwithPythonscriptinglanguage(Iig. 1).The LANDSAT TRS Toolbox consists oI the Iollowing tools: 1) Thermal Band DN to Brightness Temperature Tool The tool convertspixel values (Digital numbers DNs) oI LANDSATthermalband(band6)toat-sensorspectral radiance(Ls)inWm-2sr-1m-1usingappropriate conversion coeIIicients (gain and bias) according to equation 1 and then transIorms Ls to at-sensor brightness temperature Fig.1. LANDSAT TRS Toolbox 4371 978-1-4673-1159-5/12/$31.00 2012 IEEE IGARSS 2012(Ts) applying inverted Planck`s Law and speciIic calibration constants(K1andK2)asintheequation2.Thenecessary coeIIicients and constants are given by |5|. bias DN gain L s+=(1) ||.|

\|+=1 ln12ssLKKT(2) ThetoolinterIaceisshowninIig.2.Itworksonlywith LPGSdata.InputLANDSATthermalbandrasterdataset must be stored in one oI the ArcGIS supported Iormats (e.g. GeoTiIIorESRIGrid).Asubsetarea(polygonIeature layer)hastobedeIined.Bycheckingthecheckbox,output rasterdatasetsareautomaticallytransIormedtothe1992` coordinatesystemoIPoland(Projection:Gauss-Krger, Ellipsoid:GRS80).Thesensortype(Landsat-4/TM, Landsat-5/TM,Landsat-7/ETMinhighgain`modeand Landsat-7/ETMinlowgain`mode)canbeselectedIrom the dropdown list. Fig. 2. Thermal Bana DN to Brightness Temperature Tool Interface 2) NDVI Thresholds Tool ThetoolestimateslandsurIaceemissivityemploying NormalizedDiIIerenceVegetationIndexThresholds Method(NDVITHM)proposedby|6and7|todistinguish betweensoilpixels(NDVINDVIS),pixelsoIIull vegetation(NDVI~NDVIV)andmixedpixels(NDVIS_ NDVI _ NDVIV). The threshold values oI NDVIS 0.2 and NDVIV0.5makethemethodapplicableIorglobal conditions.TheNDVI is calculated Irom LANDSAT bands 3 and 4 (eq. 3) and then the NDVI value are converted to the LSE using modiIied NDVITHM. The emissivity () oI soil and Iull vegetation pixels can be deIined by the userthe deIault emissivityvaluesare:0.96Iorsoilpixels(S),0.99Ior Iullvegetationpixels(JC0.9850.005).The emissivity oI mixed pixels is computed by equation 4 taking intoaccountproportionoIvegetationineachpixel(PV) calculatedusingequation5|8|,andcavityeIIectwhichis due to surIace roughness (C) calculated by equation 6 using geometrical Iactor F` with the mean value 0.55. 3 43 4Bana BanaBana BanaNDJI+=(3) ( ) c c c C P PJ S J J+ += 1(4) 2||.|

\|=S JSJNDJI NDJINDJI NDJIP(5) ( ) ( )J J SP F C = 1 1' c c(6) Additionally,thetoolletstheuserdeIineemissivityIor surIace waters (0.995 by deIault). Water bodies mask as well as a subset area (polygon Ieature layer) must be deIined. The tool interIace is shown in Iig. 3. Fig. 3. NDJI Thresholas Tool Interface 3) Retrieve LST (Single-Channel Algorithm) Tool ThetoolestimateslandsurIacetemperaturevalueswith Single-Channel (S-C) Algorithm |1, 2, 3 and 4| according to equations:7a,7band7c.Thenecessaryinputdataare:the calculatedbrightnesstemperature(Ts)andthelandsurIace emissivity()datasetsaswellassomespeciIic atmospheric Iunctions(AF).TheAF`s(1,2and3)areusedIor 4372correctionoItheatmosphereinIluencewhichisvery important partoI theLST retrieval algorithm. The AF`s are computedIromatmosphericparameters(atmospheric transmissivity - , up-welling atmospheric radiance - L and down-welling atmospheric radiance - L) using equations 8a, 8band8c.ThesitespeciIicatmosphericparameterscanbe calculatedbymeansoIanatmosphericradiativetransIer model(e.g.ACPC-$tmospheric&orrection3arameter &alculatorIreelyaccessibleweb-basedMODTRAN interIace |9 and 10|). The S-C Algorithm uses alsoPlanck`s radiation constants (c1 1.19104 108 W m4 m-2 sr-1; c2 1.43877 104 m K) and the eIIective wavelength () oI LANDSATTM/ETMband6.ThetoolinterIaceisshown in Iig. 4. o c + + + = | ) (1|3 2 1 sL LST(7a) 114221)`((

+ =cLTL csss (7b) s sT L + = o(7c) t11=(8a) t|! =LL2(8b) != L3(8c) Fig. 4. NDJI Thresholas Tool Interface The three presented tools correspond to the main three parts oI the whole procedure oI land surIace temperature retrieval IromLANDSATdata.Bydividing the procedure intothree parts it is possible to apply only a part oI it, Ior example: to calculateonlythebrightnesstemperatureortoemployan emissivitydatasetcomputedwithanothermethod/toolin order to calculate the LST. This makes the LANDSAT TRS ToolsmoreIlexibleandadoptabletotheuserneeds.The LST retrieval procedure implemented in the TRS Tools is in constantdevelopment.Numerousapplicationsworldwide conIirmed that it gives very good results. The TRS Toolsis Ireely available Ior scientiIic use. In case oIaninterest,newusersarekindlyrequestedtocontact authors directly. 3. APPLICATIONS Remotely-sensed land surIace temperatureis widelyused in meteorologicalandclimatologicalresearch,especiallyIor exploringurbanheatisland(UHI)eIIect|11|.TheLSTis alsoakeyparameterIornumerousenvironmentalresearch suchas:monitoringevapotranspiration|12|,drought assessment|13|,landscapestudies|14|,healthrisk determination|15|,airqualityinvestigations|16|and geothermal heat Ilux estimation |17|.AdvantagesoItheLANDSATdataandtheSingle-Channel Algorithm in estimation oI land surIace temperature patterns in urban areas were also examined by the authors themselves |18|.ExampleoILANDSATthermalbandimageand recalculatedLSTmapshowingverysigniIicantspatial variability oI land surIace temperature in the city oI Krakow (Poland) are presented in Iigure 5.ItcanbeseeninthemapthatspatialresolutionoI LANDSATthermalimages(TM:120mx120m,ETM: 60m x 60m) is suIIicient enough to:a)recognize single emitters oI artiIicial heat. b)examinecomplexstructureoIthermalmosaicoIurban surIaces in relation to spatial variability oI diIIerent land use / land cover types 4. REFERENCES |1| J. C. Jimnez-Muoz and J. A. Sobrino, 'A generalized single-channel method Ior retrieving land surIace temperature Irom remote sensing data, J.Geophys.Res.,vol.108,no.D22,p.4688,2003.DOI: 10.1029/2003JD003480. |2|J.A.Sobrino,J.C.Jimnez-MuozandL.Paolini,'LandsurIace temperature retrieval Irom Landsat TM 5, Remote Sens. Environ., vol. 90, no. 4, pp. 434-440, 2004. |3| J. Cristobal, J. C. Jimnez-Muoz, J. A. Sobrino, M. Ninyerola and X. Pons,'ImprovementsinlandsurIacetemperatureretrievalIromthe Landsatseriesthermalbandusingwatervapourairtemperature,J. Geophys. Res., vol. 114, no. D08103, 2009. DOI: 10.1029/2008JD010616. 4373Fig.5.LANDSAT/ETM8-bit,greyscalethermal[bana610,4-12,5m{imageofSouthernPolanaregistereaonJuly26,2000(left)ana the corresponaing lana surface temperature map of Krakow ana the nearest surrounaings of the city (right) |4|J.C.Jimnez-Muoz,J.Cristobal,J.A.Sobrino,G.Soria,M. NinyerolaandX.Pons,'RevisionoItheSingle-ChannelAlgorithmIor Land SurIace Temperature Retrieval From LandsatThermal-InIrared Data, IEEE Trans. Geosci. Remote Sens., vol. 47, no. 1, pp. 339-349, 2009. DOI: 10.1109/TGRS.2008.2007125. |5|G.Chander,B.L.MarkhamandD.L.Helder,'SummaryoIcurrent radiometric calibration coeIIicients Ior Landsat MSS, TM, ETM and EO-1 ALI sensors, Remote Sens. Environ., vol. 113, pp. 893-903, 2009. |6|J.A.Sobrino,Z-L.LiandM.P.Stoll,'SigniIicanceoItheremotely sensedthermalinIraredmeasurementsobtainedoveracitrusorchard, ISPRS J. Photogramm. Remote Sens., vol. 44, pp. 345-354, 1990. |7|J.A.Sobrino,J.C.Jimnez-Muoz,G.Soria,M.Romaguera,L. Guanter,J.Moreno,A.PlazaandP.Martinez,'LandsurIaceemissivity retrievalIromdiIIerentVNIRandTIRsensors,IEEETrans.Geosci. RemoteSens.,vol.46,no.2,pp.316-327,2008. DOI:10.1109/TGRS.2007.904834. |8|T.N.CarlsonandD.A.Ripley,'OntherelationbetweenNDVI, IractionalvegetationcoverandleaIareaindex,RemoteSens.Environ., vol. 62, no. 3, pp. 241-252, 1997. |9|J.A.Barsi,J.L. BarkerandSchottJ.R., 'AnAtmosphericCorrection ParameterCalculatorIoraSingleThermalBandEarth-Sensing Instrument,inProc.IEEEIGARSS,Toulouse,France,2003,pp.3014-3016. |10|J.A.Barsi,J. R.Schott, F. D.Palluconi and S. J. Hook,'Validation oIaWeb-BasedAtmosphericCorrectionToolIorSingleThermalBand Instruments, in Proc. SPIE, Bellingham, WA. 2005, vol. 5882. |11|C.J.Tomlinson,L.Chapman,J.E.ThornesandC.Baker,'Remote sensinglandsurIacetemperatureIormeteorologyandclimatology:a review, Meteorol. Appl., vol. 18, pp. 296-306, 2011. |12|M.C.Anderson,R.G.Allen,A.MorseandW.P.Kustas,'UseoI Landsatthermalimageryinmonitoringevapotranspirationandmanaging waterresources,RemoteSens.Environ.,(inpress),2012.DOI: 10.1016/J.RSE.2011.08.025. |13|Z.Gao,W.GaoandN.Chang,'Integratingtemperaturevegetation drynessindex(TVDI)andregional waterstress index(RWSI) Iordrought assessmentwiththeaidoILANDSATTM/ETMimages,Int.J.Appl. EarthObs.GeoinI.,vol.13,pp.495-503,2011.DOI: 10.1016/J.JAG.2010.10.005. |14|M.HaisandT.Kucera,'TheinIluenceoItopographyontheIorest surIacetemperatureretrievedIromLandsatTM,ETMandASTER thermal channels, ISPRS J. Photogramm. Remote Sens., vol. 64, pp. 585-591, 2009. DOI: 10.1016/J.ISPRSJPRS.2009.04.003. |15|Y.Kestens,A.Brand,M.Fournier,S.Goudreau,T.Kosatsky,M. MaloleyandA.Smargiassi,'ModellingthevariationoIlandsurIace temperatureasdeterminantoIriskoIheat-relatedhealthevents,Int.J. Health Geogr., vol. 10:7, 2011. DOI: 10.1186/1476-072X-10-7. |16| L. Wald and J.-M. Baleynaud, 'Observing air quality over the city oI NantesbymeansoILandsatthermalinIrareddata,Int.J.RemoteSens., vol. 20, no. 5, pp. 947-959, 1999. |17|M.B.Mia,C.J.BromleyandY.Fujimitsu,'MonitoringheatIlux usingLandsatTM/ETM thermal inIrared dataA case study at Karapiti (CratersoItheMoon`)thermalarea,NewZealand,J.Jolc.Geotherm. Res.,vol.235-236,pp.1-10,2012.DOI: 10.1016/J.JVOLGEORES.2012.05.005. |18|J. Walawender, M. Hajto, 2009,Assessment oI thermal conditions in urbanareaswithuseoIdiIIerentsatellitedataandGIS,EUMETSAT Meteorological Satellite ConIerence Proceedings, P.55 4374