遥感物理-红外遥感2009-3

1、第六章第六章热红外遥感热红外遥感u第一节第一节热红外遥感概论热红外遥感概论u第二节第二节海面温度遥感海面温度遥感u第三节第三节陆面温度遥感陆面温度遥感u第四节第四节热红外多角度遥感信息模型热红外多角度遥感信息模型第三节第三节陆面温度遥感陆面温度遥感6.3.1 陆面温度遥感反演概述陆面温度遥感反演概述6.3.2 陆面地表温度的分裂窗反演算法陆面地表温度的分裂窗反演算法6.3.3 地表比辐射率的遥感反演方法地表比辐射率的遥感反演方法1. 陆面温度遥感反演面临的主要问题陆面温度遥感反演面临的主要问题非同温混合象元占绝大多数，对这样的象元而言，定义象元的非同温混合象元占绝大多数，对这样的象元而言，定义

2、象元的有效平均温度也比较困难，关于这类非同温混合象元的陆面温有效平均温度也比较困难，关于这类非同温混合象元的陆面温度遥感问题需要专门讨论。度遥感问题需要专门讨论。对纯象元陆面温度的遥感问题，由于陆面目标的比辐射率明显对纯象元陆面温度的遥感问题，由于陆面目标的比辐射率明显小于小于“1”1”，所以需要考虑大气下行辐射的贡献与干扰。，所以需要考虑大气下行辐射的贡献与干扰。(1)(1)陆面目标的比辐射率往往受物理状况（如土壤比辐射率随土壤陆面目标的比辐射率往往受物理状况（如土壤比辐射率随土壤含水量而变），表面粗糙度、地表起伏造成传感器对地表实际含水量而变），表面粗糙度、地表起伏造成传感器对地表实际视角

3、的起伏等因子控制，所以一般只能作为未知量，不能事先视角的起伏等因子控制，所以一般只能作为未知量，不能事先设定。设定。6.3.1 陆面温度遥感反演概述陆面温度遥感反演概述大气效应纠正大气效应纠正 地表真实辐射亮度值地表真实辐射亮度值地表真实比辐射率值地表真实比辐射率值(4) (4) 陆面目标的比辐射随波段变化显著，这样导致方程组陆面目标的比辐射随波段变化显著，这样导致方程组的完备，因为第一个波段包含一个未知的比辐射率，的完备，因为第一个波段包含一个未知的比辐射率，N N个波个波段包含段包含N N个未知比辐射率，外加一个未知温度，所以未知数个未知比辐射率，外加一个未知温度，所以未知数总比独立方程数

4、多一个。总比独立方程数多一个。m2. 2. 陆面温度遥感反演回顾陆面温度遥感反演回顾陆面温度的遥感反演问题最早可追溯到陆面温度的遥感反演问题最早可追溯到TIROS2TIROS2上搭载的热红外上搭载的热红外辐射计，其波段为辐射计，其波段为8-14 8-14 。大家发现传感器得到的陆面温度和地面。大家发现传感器得到的陆面温度和地面实测的沙漠表面温度差异很大。实测的沙漠表面温度差异很大。Buettner and Kern(1965)Buettner and Kern(1965)通过测量沙通过测量沙子子( (石英含量高石英含量高) )的比辐射率，发现沙子的比辐射率明显小于，解释的比辐射率，发现沙子的比

5、辐射率明显小于，解释了这个矛盾。了这个矛盾。Nimbus 4Nimbus 4上的上的IRISIRIS测量结果也证实了沙地在测量结果也证实了沙地在9 9 附近辐附近辐射率明显小于射率明显小于(Prabhara and dalu 1976)(Prabhara and dalu 1976)。 Marlatt(1967)Marlatt(1967)可能是可能是第一次系统地野外测量了地表比辐射率对热辐射的影响。第一次系统地野外测量了地表比辐射率对热辐射的影响。 m陆面温度遥感反演的研究到陆面温度遥感反演的研究到NOAA/AVHRRNOAA/AVHRR第四、五通道在海温遥第四、五通道在海温遥感反演取得成功之

6、后，变的越来越有感反演取得成功之后，变的越来越有“诱惑力诱惑力”。Price(1984)Price(1984)首先把首先把海温遥感的分裂窗口方法引用到农田地区的温度反演中来。他在仔细海温遥感的分裂窗口方法引用到农田地区的温度反演中来。他在仔细分析了各种误差来源之后，预计反演精度约分析了各种误差来源之后，预计反演精度约 3K3K。在。在 1 1时，他给出时，他给出的反演公式为：的反演公式为： PricePrice指出当温度为指出当温度为300K300K时，时， 误差误差0.010.01可引起可引起2K2K的温度误差。的温度误差。sTTTTT .().()445454533335450750ABe

7、cker(1987)Becker(1987)考虑考虑AVHRRAVHRR第四、五通道的地表反射率之差对温度第四、五通道的地表反射率之差对温度反演的影响，并提出了一个模型解释热红外测量温度和地表热力学温反演的影响，并提出了一个模型解释热红外测量温度和地表热力学温度的差别。度的差别。Becker(1990a)Becker(1990a)接着在辐射传输方程线性近似的基础上，进接着在辐射传输方程线性近似的基础上，进一步讨论了地表比辐射率对温度反演的影响，给出了一个一步讨论了地表比辐射率对温度反演的影响，给出了一个“局地分裂局地分裂窗口窗口”法的反演公式。法的反演公式。 =constant=constan

8、t 是一个与是一个与 无关的独立常数，无关的独立常数，P P和和M M与与 有关，可以通过大气有关，可以通过大气辐射传输程序辐射传输程序Lowtran 7Lowtran 7用最小二乘法回归确定系数用最小二乘法回归确定系数 。sTATTTTPM04545220AP 112 ()2)1 ( MA0, , , Becker(1990b)Becker(1990b)进一步把进一步把 NOAA/AVHRRNOAA/AVHRR第三通道的信息考虑进来，第三通道的信息考虑进来，提出一个与温度无关的独立因子。提出一个与温度无关的独立因子。Li(1993)Li(1993)在此基础上讨论了用这个在此基础上讨论了用这个

9、概念反演地表比辐射率的可行性。他也同时指出，这种方法要能实际概念反演地表比辐射率的可行性。他也同时指出，这种方法要能实际应用还有许多工作要做，其中两个最大的制约因素是：应用还有许多工作要做，其中两个最大的制约因素是：(1)(1)太阳的中红外辐射受大气衰减比较严重，如何找到一个合理太阳的中红外辐射受大气衰减比较严重，如何找到一个合理的方法或模型来估算太阳辐射对第三通道的中红外波段的贡献；的方法或模型来估算太阳辐射对第三通道的中红外波段的贡献；(2)(2)地表在第三通道的中红外波段的双向反射率特性比第四、五地表在第三通道的中红外波段的双向反射率特性比第四、五通道更强烈，有必要做更多的野外实测和理论

10、工作以建立地表在第三通道更强烈，有必要做更多的野外实测和理论工作以建立地表在第三通道的双向反射率模型。通道的双向反射率模型。 Gillespie et al.(1986,1987)Gillespie et al.(1986,1987)也讨论了把地表比辐射率和地表也讨论了把地表比辐射率和地表温度对辐射测量的影响分离开的问题。温度对辐射测量的影响分离开的问题。Wan Wan 和和 Dozier(1989)Dozier(1989)把遥测地表温度当作一个地球物理学的反把遥测地表温度当作一个地球物理学的反演问题演问题, ,通过通过LowtranLowtran程序进行数值模拟程序进行数值模拟, ,评价了温

11、度反演的可行性并提评价了温度反演的可行性并提出了合理的波谱段范围出了合理的波谱段范围, ,认为通过多波谱同时反演地表温度和地表比辐认为通过多波谱同时反演地表温度和地表比辐射率是可行的。射率是可行的。Wan Wan 和和 Dozier(1996)Dozier(1996)通过大气传输模型进一步模拟计算指出：通过大气传输模型进一步模拟计算指出：1)1)统计回归的系数与传感器的视角有关；统计回归的系数与传感器的视角有关；2)2)为了提高反演精度，模拟为了提高反演精度，模拟计算回归系数时有必要把大气含水量、大气低层温度和地表温度考虑计算回归系数时有必要把大气含水量、大气低层温度和地表温度考虑进去进去,

12、,而不能在所有的情况下都用相同的系数来反演地表温度。而不能在所有的情况下都用相同的系数来反演地表温度。 Prata(1991) Prata(1991)在线性假设的基础上，用在线性假设的基础上，用NOAA/TOVSNOAA/TOVS反演大气水汽反演大气水汽和温度廓线作大气修正，建立了一个地表温度反演的理论公式。和温度廓线作大气修正，建立了一个地表温度反演的理论公式。Prata(1993)Prata(1993)进一步在理论推导的基础上给出可地表温度的双通进一步在理论推导的基础上给出可地表温度的双通道反演公式：道反演公式： 分别为第四、五通道的透过率，分别为第四、五通道的透过率， 为大气平均温度，可

13、以近为大气平均温度，可以近似用似用 代替，代替， 为大气下行辐射的改变值，为大气下行辐射的改变值，PrataPrata给出了不同时给出了不同时间和地区的间和地区的 值。值。TaTbTds45at111454 bt 545111()dcIBTabTBTBTTT414411 ()( )1145tt45tt45,TTT45或IIVidal(1994)Vidal(1994)把第四、五通道的温度先转化为把第四、五通道的温度先转化为8-14 8-14 宽通道的宽通道的黑体亮度温度，再根据地表比辐射率改正得到地表温度，系数通过回黑体亮度温度，再根据地表比辐射率改正得到地表温度，系数通过回归得到，相关系数归得

14、到，相关系数 =0.89 =0.89 。mR230015020.2958. 345. 554BBsBBTTTTTSobrino(1994)Sobrino(1994)考虑大气透过率的非线性特点，利用考虑大气透过率的非线性特点，利用Klesspies Klesspies 和和McMillin(1990)McMillin(1990)提出的估算大气第四、五通道的透过率之比的方法。提出的估算大气第四、五通道的透过率之比的方法。把大气两各通道透过率之比和地表比辐射率的影响同时考虑进来，改把大气两各通道透过率之比和地表比辐射率的影响同时考虑进来，改进了地表温度反演的分裂窗口方法。进了地表温度反演的分裂窗口方

15、法。其中：其中： R R为第四、五通道大气透过率之比。通过模拟计算，为第四、五通道大气透过率之比。通过模拟计算，SobrinoSobrino指指出，在出，在 已知的条件下，这种方法的精度可以达到已知的条件下，这种方法的精度可以达到0.4K0.4K以内。以内。BTRBTRBTRs41244344556()()()()()()iiii0142451 ()()RTTTTTTkkkNkkN()()()440550144021为了提高分裂窗口法反演地表温度的精度，近年来的工作越来越为了提高分裂窗口法反演地表温度的精度，近年来的工作越来越把大气状态作为温度反演中的信息。而放弃那种建立对任何大气模式都把大气

16、状态作为温度反演中的信息。而放弃那种建立对任何大气模式都适用的适用的“全能全能”模型。模型。FrancoisFrancois和和Ottle(1996)Ottle(1996)指出在指出在 已知的条件下，考虑水汽修正已知的条件下，考虑水汽修正可使地表温度反演的精度达到可使地表温度反演的精度达到0.20.1K0.20.1K。这以。这以Francois(1996)Francois(1996)，Wan Wan (1996)(1996)，Sobrino(1994)Sobrino(1994)所作的工作为代表。所作的工作为代表。总之，陆面温度反演的研究多以分裂窗口方法为基础，为了提高总之，陆面温度反演的研究多

17、以分裂窗口方法为基础，为了提高温度的反演精度，不同的作者主要从多个方面进行了研究。但由于陆面温度的反演精度，不同的作者主要从多个方面进行了研究。但由于陆面温度反演问题的复杂性，迄今为止，陆面温度反演的研究主要仍以可行温度反演问题的复杂性，迄今为止，陆面温度反演的研究主要仍以可行性研究为主，大气效应的纠正、地表比辐射率性研究为主，大气效应的纠正、地表比辐射率 未知和地表温度的皮肤未知和地表温度的皮肤效应问题仍制约着陆面温度遥感反演方法在实际中的应用。效应问题仍制约着陆面温度遥感反演方法在实际中的应用。 单通道法单通道法Single infrared channel methodSingle in

18、frared channel method 多通道法多通道法Multi-channel (Split-Window) methodMulti-channel (Split-Window) method 多角度法多角度法Multi-angle methodMulti-angle method 多通道与多角度结合法多通道与多角度结合法Multi-channel and multi-angle methodMulti-channel and multi-angle method6.3.1 陆面温度遥感反演概述陆面温度遥感反演概述6.3.2 陆面地表温度的分裂窗反演算法陆面地表温度的分裂窗反演算法6.

19、3.3 地表比辐射率的遥感反演方法地表比辐射率的遥感反演方法1. Quantity Measured by Radiometry from Space SatelliteAtmosphereOuterofatmosphereSunSurface),(),(),(),(),(iisaiiiRRRI),()cos(),()(),(1 ()(),(),(ssisissbslatisiiiERRTBRiii4765123Spectral Radiance at Satellite Level at viewing angle ),(),(),(),(),(iisaiiiRRRI),()cos(),()

20、(),(1 ()(),(),(ssisissbslatisiiiERRTBRiiiSpectral Radiance at Ground LeveliI,v:thespectralradianceobservedatgroundlevelv:thethermalpathatmosphericradiance,v:thepathradianceresultingfromscatteringofsolarradiation:theradiationemitteddirectlybysurface.:thedownwardatmosphericthermalradiationandsolardiff

21、usionradiationreflectedbysurface.:thedirectsolarradiationreflectedbysurface.v ,Ts:surfacespectralemissivityandsurfacetemperature,v:thedownwardatmosphericthermalradiance,v:thedownwardsolardiffusionradiation,v:thebi-directionalsurfacespectralreflectivity,v:thesolarirradianceatthetopoftheatmosphere.iRa

22、iRsiRiR123atiRsliRbiiE123During the nighttime or for channels in the 8-14 mThe contribution of solar radiation to the spectral radiation at the TOA is negligible, radiative transfer equation can be simplified as ),(),(),(),(iatiiiRRIiatisiiiRTBR),(1 ()(),(),(At Satellite LevelAt Ground LevelDefiniti

23、on of brightness temperature Ti),(),(iiiITBDefinition of brightness temperature Tgi),(),(igiiRTBuImpact and influence of atmosphere on surface radiationuAbsorption and Re-emissionIn TIR, the main absorption gases in the atm. are H2O, CO2 and O3In 10-12m, the main absorption gas in the atm. is H2OuIm

24、pact and influence of atmosphere on surface radiationu is not only a function of W, but also depends on the distribution of H2O in the atmosphere and the profiles of Ta and PressuresuImpact and influence of atmosphere on surface radiationiataiiRTB1)(aT)()()(iiiiiTTTBTBTBTBRRTTTTiatiiiaiiiiSi)(1)(1 A

25、tmospheric equivalent temperature First order of approximation of Planck function From radiative transfer equationAtmospheric effectsEmissivity effectTgi250255260265270275280285290295300305310315320Tgi-Ti-2-10123456789101112TropicalMidlatitude sum m erMidlatitude winterChannel 4uTemperature Correcti

26、on (Deficit) due to Absorption and Emission of AtmosphereatmT)(1aiiiigatmTTTTTiFor dry atmosphere, or for inversion of Ta profile, , is small, otherwise for wet atmosphere, is large. 1iaTTatmTatmTFor a given atmosphere, temperature correction is linearly correlated to Ti or ground brightness tempera

27、ture Tgi atmTTropical atm: W=4.11g/cm2, Ta(0)=299.7KMidlatitude Summer: W=2.99g/cm2, Ta(0)=294.2KMidlatitude Winter: W=0.85g/cm2, Ta(0)=272.2Kand are constant for a reference temperature uImpact of surface emissivityuVariation of surface emissivityuIn 8-13 m， may vary from 0.90 to 0.99uIn 3-5 m，the

28、range of variation may be very large Power expression of the Plancks functioniniiTTB)(iin5 . 411mn137 . 3mn Temperature correction caused by surface emissivityTiigiinTTi11)(iigiatiTBRFor a given temperature, when W increases, increases decreases If no atmosphere, , reaches its maximum, thus, the dri

29、er the atmosphere is, the bigger the impact of surface emissivity on Ts retrieval is iT0iTFor a given atmosphere, the higher Ts is, the smaller is and the bigger is, otherwise the smaller is.iTTBecause is inversely proportional to , thus for a given , the longer the wavelength of channel is, the big

30、ger is.TinTTBRRTTTTiatiiiaiiiiSi)(1)(1i0,890,900,910,920,930,940,950,960,970,980,991,00Ts-Tgi0123456TropicalMidlatitude summerMidlatitude winterChannel 4different slopes correspond to different Ts (20 C variation) for same atmosphereError on Ts retrieval resulted from error on emissivity6 . 02 . 0至T

31、Data pre-processingMeasurement of Land Surface Temperature From SpaceSatellite dataEmissivity RetrievalTemperature RetrievalRadiometric CalibrationImage NavigationGeometric CorrectionsCloud ScreeningAtm. Correction Vis/NIRAtm. Correction TIRNDVI or Classifi.methodsTISI Method (day/night)Split Window

32、Radiative transfer2. Multi-channel (Split-Window) methodThismethodisbasedonthedifferentialabsorptioninthetwoadjacentchannels,centeredat11and12m,tocorrectforatmosphericeffects.cbTaTTjisCoefficientspre-determinedtominimizetheerrorinTsdeterminationBrightnesstemperaturemeasuredintwoadjacentchannelsi and

33、jSurfacetemperatureTheaccuracyofTsretrievalisdependentonthecorrectchoiceofthecoefficientsa, bandcThistypeofmethodhasbeeninitiallydevelopedwithsuccesstodetermineseasurfacetemperature,ithasbeenextendedinthelaterof1980sandinthebeginningof1990storetrievesurfacetemperatureoverlandLand Surface Temperature

34、 Retrieval Local Split Window),(),(),(121211111211WcTWbTWaTsSplit-WindowmethodisextremelysimpleandhasproventobeveryefficientandaccurateforretrievingSSTSplit-WindowcannotbeusedforretrievingLSTwithoutmodificationsLSTcanbeexpressedasalinearcombinationofT11andT12The coefficients a, b and c are generally

35、 function of W in the atmosphere, view angle as well as surface emissivities in these two adjacent channelsThe values of a, b and c are sensor dependent. They depend also on the radiative transfer code and the variation of surface and atmospheric parameters used to derive them For relative drier atm

36、osphere (w2.0g/cm2), a, b and c are local coefficients which depend only on surface emissivities Land Surface Temperature Retrieval Local Split WindowAn algorithm for AVHRR/NOAA14),(),(),(121211111211WcTWbTWaLST2),(2),(),(1211121112111211TTWQTTWPWcLSTor baPbaQModtran4.0;Viewzenithangle0-67;Ts:250-33

37、0K;:0.90-1.00;:-0.020.02Wc)sec41. 004. 0(sec42. 445. 2 121111sec0161. 00479. 0sec0108. 03012. 01sec32 . 90101. 0sec31 . 61918. 0)sec01974. 09907. 0(WWeeP121111sec49. 110. 81sec72. 175. 4)sec48. 011. 0()sec09. 061. 3(WQ834Atmospheres,W5.0g/cm2with3. Determination of Atmospheric Water Vapor Satellite

38、Sounding (TOVS, MODIS, AIRS, IASI) Meteorological Station (Radiosounding) Satellite Measurements Near-infrared techniquesRequirements: at least one channel in the water absorption band (0.94 m), and one nearby channel in the atmospheric windows (0.86 m, 1.05 m and 1.24 m). Passive microwave techniqu

39、es Thermal infrared techniquesRequirements: Two nearby thermal infrared channels one channel near to the water vapor absorption band (12-13 m), and one nearby channel in the atmospheric windows (10-11 m) . .Determination of Atmospheric Water Vapor from Split WindowChannel Measurements Principle of t

40、he MethodiatiiatiisiiiiiRRTBTBI1Ii : radiance measured from space in channel i : brightness temperature in channel i at satellite level, : surface emissivity in channel i, : surface temperature : total atmospheric transmittance in channel i, : atmospheric upwelling and downwelling radiances in chann

41、el i, iTisTatiRatiRiAssumptionsAtmosphere and are unchanged over the N neighboring pixels Or effects of their spatial variations are not larger than the effect of instrument noise over the N neighboring pixels Only surface temperature changes over the N neighboring pixels iRelationship between and P

42、rinciple of MethodsksiiikiTTTT,iTsTk:pixelkoftheNneighboringpixels :mean(orthemedian)TioftheNneighboringpixels :mean(orthemedian)TsoftheNneighboringpixels iTsTFor measurements in channel jsksjjjkjTTTT,0,jkjijijikiTTTTRelationship between and iTjTPrinciple of MethodjijiijRNkikijkjNkikijiTTTTTTR12,1,)

43、()( By least-squares analysis, one gets withijijjiRNkjkjjkjNkikiijTTTTTTR12,1,)()(withvRemarks:Thenumeratoranddenominatoronther.h.s.represent,respectively,thecovarianceandthevarianceofTdirectlymeasuredbythesatellitethetransmittanceratioscanbederiveddirectlyfromsatellitedataprovidedthattheemissivityr

44、atiooftwochannelsareknown Principle of MethodSquare of the linear correlation coefficient of two measurements ijjiNkjkjNkikijkjNkikiRRTTTTTTTTr12,12,2,1,2)()()(12jjiiiijjijjiRRr ThisindicatesthatthetransmittanceratioderivedfromaboveequationsisfeasibleonlyifthebrightnesstemperaturesTi andTjmadeinthet

45、wosplit-windowchannelsiandjoverNpixelsareperfectlycorrelated(r=1)oralmostperfectlycorrelated.ThisconstraintcanbeusedtocheckwhethertheassumptionsmadeinthederivationoftransmittanceratioarefulfilledSample No.02040608010012014016018020022024011/12.92.93.94.95.96.97.98.991.001.011.02Igneous RocksMetamorp

46、hicMeteoriteMineralsSedimentarySoilsVegetation+ SnowWaterRocksSoils+Veg.+WaterEmissivityratiosofchannels11mand12mofATSR2calculatedfromthehemisphericreflectancesmeasuredinlaboratorybySalisburyandDAria1992fordifferenttypesofnaturalsurfaces.11211jiijR formostsurfacesatscaleof1kmx1kmSimulation with Modt

47、ran4 for 1761 atmospheres12/11.2.3.4.5.6.7.8.91.0Water Vapor Content (g/cm2)012345678 = 0 = 64.9 Slope = -13.688, Offset = 13.905, r2 = 0.984Slope = -8.004Offset = 8.201r2 = 0.983NOAA 14Relationship between transmittance ratio and W Simulated channel transmittances: atmospherictransmittance/radiance

48、computercode-MODTRAN4.0NOAA14filterresponsefunctions1761atmosphericprofiles:Wupto7.6g/cm2cos .3.4.5.6.7.8.91.01.1Slope or Offset-20-15-10-505101520SlopeOffsetOffset = 4.062 + 9.891 cos r2=0.9994Slope = -3.876 - 9.858 cos r2=0.9995Angular dependence of regression coefficients For NOAA 141112)cos86. 9

49、88. 3(cos89. 906. 4W2/13. 0cmgwithWvaluesretrievedfromsatellitedataarecomparableinmagnitudewithWobservedbyradiosoundings,g/cm2,g/cm2W measured by radiosonde (g/cm2)0.51.01.52.02.53.03.54.0W derived from ATSR data (g/cm2)0.51.01.52.02.53.03.54.0SGP97BarraxCabauw1:1 lineComparison with Radiosonde Meas

50、urements 04. 0W22. 0W第三节第三节陆面温度遥感陆面温度遥感6.3.1 陆面温度遥感反演概述陆面温度遥感反演概述6.3.2 陆面地表温度的分裂窗反演算法陆面地表温度的分裂窗反演算法6.3.3 地表比辐射率的遥感反演方法地表比辐射率的遥感反演方法1. Why Measure Surface Emissivity Surface emissivity isAmeasureoftheinherentefficiencyofthesurfacetoconvertheatenergyintoradiantenergyoutsidethesurface.DependsonthevComp

51、ositionofsurfacevRoughnessofsurfacevOtherphysicalparameters(cf.soilmoisture)vwavelengthKnowledge of permits:DiscriminationandsometimesidentificationofdifferenttypesofsurfacesTogetsurfacetemperatureTs:)6030(sTRemarks:hasveryinterestingspectralproperties,isverycomplementarytoVIS,NIRandMicrowave How to

52、 Get Surface Emissivity 1 1From Field or Laboratory MeasurementsSpectralradiometerRadiometercombinedwithCO2laserBoxNeed some assumptions:TsanddonotchangeduringthemeasurementsSurfaceisLambertianoraspecularreflectingunlessacompletesetofisalsomeasuredbHow to Get Surface Emissivity 2 2From space measure

53、ments(1)onthebasisofthestatisticalrelationshipsbetweenandthespectral informationintheVIS/NIR(cf.NDVI/). =a+b log(NDVI)Coefficientsaandbweredeterminedonlyforthe8-14mbandaandbaresurfacedependentVandeGriendandOwe(1993)Using emissivity databaseNeeds:a)ImagesclassificationfromVIS/NIRandTIRimagesb)Fieldan

54、dLaboratoryemissivitymeasurementsforeachclassareNDVIafteratmosphericcorrectionsforforbaresoilandforinfiniteLAIdependentonstudiedimageEmissivity Determination From SpaceUsing NDVI thresholds methodFor,1110565. 09795. 01120275. 09815. 0Reflectivity in channel 1 of AVHRRFormaxminNDVINDVINDVIvf021. 0968

55、. 011vf015. 0974. 012Fraction of Vegetation2minmaxminNDVINDVINDVINDVIfvminNDVINDVI For,maxNDVINDVI 989. 01211maxmin,NDVINDVIUsing TISI Method from a couple of day/night dataFrom space measurements(2)onthebasisofthestatisticalrelationshipbetweentheminimumandthemaximumoftherelativeemissivitydifference

56、inthemultispectralradiometerHow to Get Surface Emissivity 3 3 (Geillespie et al. 1998)The success of this method depends on the number of channels and their placementNot appropriate for AVHRR sensors with only two channels in IRTHow to Get Surface Emissivity 4 4 From space measurements(3)usingmulti-

57、temporalthermalinfraredmeasurementsNeed some assumptions on emissivityDay=night(Watsonetal.,1990;WanandLi1997)day1=day2(GoitaandRayer1999)DayTISIE=NightTISIE(LiandBecker1993)TISIE=appropriateratioofemissivitiesEmissivity Determination From Space (TISI Method) (1)Definition of TISITheradiancemeasured

58、atgroundlevelisgivenbyiatiSiigiiiRTBTBR)1 ()()(UsingthepowerlawapproximationofthePlancksfunctioninsiingiiiCTTRiiWhereCiisacorrectivetermthattakesintoaccounttheatmosphericreflectedradiance:iatiiRRbCi11)(siatiTBRbiTakingtheproductofNwithpowerak(k=1,.N)gkTNiNiaiainasnagiiiNiiiiiCTT111Choosingtheakssuch

59、that01Niiina11NiiinasTthenEmissivity Determination From Space (TISI Method) (2)Definition of TISI (continuation)LeadingtoNiNiainagiiiiCT11with11NiaiiCCDefiningTISIasNinagiiiTCTISI11DefiningTISIEasNiaiiTISIE1TISIETISI Properties of TISITISIisindependentofsurfacetemperatureTISIisdirectlyrelatedtoemiss

60、ivityratioTISIisdirectlymeasurablefromspaceafteratmosphericcorrectionsTISIcanbetailoredtoweightcertainbandsmoreheavilythanothersNkkkkaTISITISI1with01NiiinaCmaybecomputedbyBi(Tgi)insteadofBi(Ts)whereTgiisthehighestTgamongallchannelsusedforagivenpixel.TISI=TISIEisonlytruewhennosolarreflectionoccursTIS