Under-the-cloud pixel LST estimation method based on microwave remote sensing and space-time information

Abstract

The invention discloses an under-the-cloud pixel LST estimation method based on microwave remote sensing and space-time information. The under-the-cloud pixel LST estimation method is characterized in that by considering land surface temperature space distribution consistency and periodicity of a time sequence, a statistical model and time sequence filtering are effectively combined together. A multi-channel statistical model based on passive microwave remote sensing is used for initial estimation of an under-the-cloud pixel LST value, and during an initial estimation process, an NDVI is used as classification basis for land surface coverage classification, and in addition, a conventional land surface type remote sensing product is abandoned to improve classification precision. An estimation value based on the statistical model is used as a background value to be filled in an LST time sequence, and by considering the influences of the former period and the latter period on a period without LST, a moving weighted filter is used to modify the estimation value to acquire an under-the-cloud pixel LST reconstruction result.

Description

technical field [0001] The invention relates to the technical field of MODIS under-cloud pixel LST estimation method, in particular to an under-cloud pixel LST estimation method based on microwave remote sensing and spatio-temporal information. Background technique [0002] Land Surface Temperature (LST) is a key parameter to measure the balance of water and heat on the earth's surface, and it is of great significance for related research in climate, hydrology, geophysics and other scientific fields. Since the 1970s, domestic and foreign scholars have carried out a lot of research on how to use thermal infrared remote sensing to obtain land surface temperature, and proposed a variety of inversion algorithms including the split window algorithm. Although thermal infrared remote sensing has a wide coverage, strong real-time performance, and can quickly obtain surface emissivity and temperature information, it is easily affected by atmospheric water vapor and cannot penetrate c...

AI Technical Summary

Problems solved by technology

However, the algorithm is based on the existing surface type products (such as MOD12) in practical applications, and the inversion accuracy is low, with an average error of about 3K; the inversion results are obviously limited by environmental and meteorological conditions, and the higher the missing value rate, the higher the missing value rate. The larger the inversion error is; the inversion area is based on a pixel, and the surface temperature reconstruction of a large area cannot be realized, and the practicability of the algorithm is low

Images