The Research Of Validation Of Remote Sensing Temperature Products And Spatial Scale

With the development of China’s aerospace industry, the successfully launched of the satellite, the corresponding remote sensing products will be widely used. With the FY-3B satellite was successfully launched, the validation of the product surface parameters corresponding to the test need to be resolved. These data provide an important basis both in meteorology, marine, environment, resources and military etc. The land surface temperature which is a important land parameter plays an important role in many places such as urban heat island, forest fire. It is an important index in land parameters, makes its authenticity is very important, so the validation test research of domestic satellites LST products has a great significance.In this paper, the study area is our country mainland off the coast of southern Guangdong province and the northern suburb of Chengdu city, Sichuan province of pengzhou. The experimental data include ground experimental data of ground-satellite synchronous observation experiment, high resolution Landsat8 image data, MODIS LST product with low resolution and domestic environment star HJ-1B CCD and IRS image data with low resolution. The main research contents are as follows:(1) The Landsat8 data and domestic environmental satellite data are pre-processed, including geometric correction, radiometric calibration, atmospheric correction and clipping. Parameters in the Landsat8 formula should be changed by the relevant documents, environmental satellite data need to make the spectral response function. The calibration formula and some constant values use the relevant files.(2) The land surface emissivity is used the method of vegetation index. The visible band has done radiometric calibration, and the calibration model of atmospheric correction on TM4, TM5 band with FLAASH atmosphere. This method divided surface into three major categories including water, vegetation and artificial construction. The result has been able to reach the land surface emissivity required accuracy requirements, so the result is calculated by empirical formula corresponding to the surface type.(3) Described the four kinds of commonly used temperature inversion algorithm, namely the universal single channel algorithm, radiation transmission equation method, single window algorithm and the inversion algorithm based on image for high resolution images for land surface temperature inversion. Then we obtain difference of four kinds of inversion results of the three types and analyzed. The parameters of the radiation transfer equation obtained by the NASA site, the accuracy is higher, so the subsequent follow-up of the FY-3B LST product validation is selected inversion results of this algorithm.(4) In the LST product authenticity verification, because of the scale, there will be a big error if using directly the Landsat8 inversion result to test low resolution products. The effect of the surface heterogeneity and scale transformation will bring about the error.(5) By using wireless sensor network real-time ground survey data obtained, the method of putting nodes need to study. Through the nodes of different distribution considering the research way, using a high precision, high accuracy of the method according to the experimental area. According to the Geo-statistics, we choose a sampling plan to make that the measured value can represent the true value of the pixel. In the comparison of measured value and HJ-1B remote data, the HJ-1B data is auxiliary data. The environment satellite data were up-scaling and compared with the FY-3B temperature product. The error of them can be found about 1.0K and achieve the required accuracy. Conclusion: FY-3B temperature products have a high correlation with high resolution inversion results and MODIS temperature products. But FY-3B has a high missing data rate than MODIS. And it does not include such as rivers and lakes water temperature value, so a lot of edge data make a great impact on the error. FY-3B also has a great correlation with the domestic environmental satellite data. Their error reach three degrees. The time scale has a great effect on the result. The image below 0 degree is basically a deletion status and not available. In a day that the temperature changes little, namely the time scale has little influence on the result, their correlation is high and root mean square error were less than 1. The result meet the accuracy requirements. The overall situation of FY-3B LST products are available, but there is still a certain distance with MODIS temperature products.