The Research Of Spatial Downscaling Retrieved From Thermal Infrared Remote Sensing Based On IDL

The land surface temperature(LST) is an important physical parameter of the earth’s surface, is an important index to study the surface, atmospheric related parameters of land surface. With the development of thermal infrared remote sensing technology, the land surface temperature inversion from thermal infrared remote sensing data has become a research hotspot. With the maturing of the temperature inversion methods, how to improve the spatial resolution of the temperature inversion results became the thermal infrared remote sensing research focus. The process of improving the spatial resolution also called spatial downscaling.In this paper, we using ETM + remote sensing image to realize the inversion of surface temperature in Beijing with single window algorithm. By studying the correlation relationship between the surface temperature and surface characteristic parameters with the low spatial resolution, then applying this relationship to high spatial resolution to improve the spatial resolution of the surface temperature results. The surface characteristic parameters including vegetation index, vegetation coverage, the surface albedo and impermeable layer coverage(ISP). By studying the relationship between the surface temperature and surface characteristic parameters and using Ts HARP method and HUTS method to downscaling for the surface temperature, on the basis of relationship between ISP and LST in urban areas, we use the ISP to improve the Ts HARP method(MTs HARP). We set two groups of downscaling in two spatial resolution of 120 m and 60 m, then respectively get two temperature image of 60 m and 30 m spatial resolution. Through comparing the 60 m spatial resolution downscaling images and inversion temperature of 60 m spatial resolution images, we can evaluate the effect of different methods for temperature spatial downscaling. All the downscaling research process is based on the IDL language and ENVI secondary development technology.The study shows that the NDVI and the LST have the strongest correlation,their two fitting equation fitting degree is the highest, the LST and ISP have a positive correlation in urban areas. All the downscaling image inverted by three methods retain the original temperature of the image spatial characteristics and tallies with the characteristics and the type of land surface. The temperature results which is inverted by MTs HARP method has a better precision in urban areas than Ts HARP and HUTS method. The results show that using the correlativity between LST and surface characteristic parameters to downscaling can improve the spatial resolution of the land surface temperature, and has a certain accuracy.