The Effect Of Target’s Temperature Changes On The Measurement Of Target’s Emissivity In Natural Conditions

Land surface emissivity (LSE) has already been recognized as a crucialparameter during the measurement of land surface temperature (LST). On one hand,there is a bad-posed retrieval problem about temperature and emissivity insatellite-based measurements of LSE. On the other hand, laboratory-baseddeterminations of LSE are conducted under constant and natural conditions, whichare limited for the use in TIR. This paper takes some typical targets for exampleandstudies the affection of target’stemperature in the process of measuringtarget’semissivity by using the target’semissivity gaining in natural conditions anddiscusses the coupling of LST and LSE to exclude temperature effects and to reducethe error of measuring target’s emissivity. Furthermore, a new approach will beprovided forimprovingthe accuracy of the retrieved LST from passive satellite.An abundant of terrestrial materials in the field includes rocks, man-madematerials and so on. For different materials, the spectral characteristics of emissivityare different and the coupling of LST and LSE is also complex. This paper focuseson studyingvegetation (including shrub and grass) and hard surface (includingconcrete surface,pavement surface,marble suface,brick surface and painted roofsurface). Experiments about synchronic measuring target’s emissivity andtarget’stemperature in natural conditions are conducted and in order to measuremorehigh-accurate target’semissivity, thesuitable temperature-emissivity separationalgorithm is selected. Then the emissivity data is selected and preprocessed.Combined with the measuring target’stemperature, the relationship between target’semissivity or its transforms and target’s temperature are analyzed. In addition, themathematical modelsareset up and made validation. The specific research content asfollows:Firstly, the measurement principle of infrared spectrometer is focused on and aField experiment program is planned. Then, the influence of some environmentalfactors on measuring target’s emissivity is analyzed. Target’semissivityand target’stemperature are selected in ideal conditions.Secondly,by comparative analysis of four kinds of typical temperature and emissivity separation algorithms and the TES algorithm worked on the computer ofFTIR, the ISSTES algorithm is the mostsuitable algorithm in this paper. The waveletanalysis in noise eliminating of spectrum data has remarkable effect, particularly forvegetable.Thirdly, the characteristic wavelength of emissivity is selected as the variable ofthe relational models,according to the correlational analyses of target’s emissivityand target’s temperature. The selection of characteristic wavelength can bedetermined by the coefficients between target’s emissivity or its transformations andtarget’s temperature. The emissivity transformations include: the first derivativetransformation, the second derivative transformation, and the reciprocal oflogarithmic transformation and the first derivative of logarithmic reciprocaltransformation. Shrub, concrete surface, pavement surface, brick surfacecharacteristic wavelengths of emissivity are determining the correlation coefficientsof their second derivative of emissivity and their temperature. Grasscharacteristic wavelengths of emissivity are determining the correlation coefficientsof its first derivative of logarithmic reciprocal of emissivity and their temperature.Concrete surface and painted roof surface characteristic wavelengths of emissivityare determining the correlation coefficients of their emissivity and their temperature.Fourthly, By SPSS, SIMCA-P software, forseven types of targets, thecurve-fitting model, stepwise regression model and partial least squares regressionmodel are established and will be verified. The three types of models show that thechanges of hard surface temperature have greater impact on its emissivity incomparison with vegetation. The relationship between the marble surface emissivityand its temperature is highest.On the contrary, the relationship betweengrassemissivity and its temperature is least. Judging from the effects of models, thepartial least squares regression model and curve fitting model are better than thestepwise regression model in studying the relationship between target’s emissivityand target’stemperature.