Passive microwave surface temperature and emissivity joint inversion method based on deep learning and optimization algorithm

Abstract

The invention discloses a passive microwave surface temperature and emissivity joint inversion method based on deep learning and an optimization algorithm, and belongs to the technical field of satellite remote sensing quantitative inversion of surface temperature. The method comprises the following steps: carrying out gridding sampling on atmosphere data of a target area to obtain a representative cloud / clear sky atmosphere parameter combination; constructing representative earth surface emissivity data based on the earth surface classification system; a multi-dimensional radiation transmission simulation data set is constructed through radiation transmission mode simulation; establishing a high-precision quantitative regression model of atmospheric components and radiation transfer key parameters; and a trust region constraint algorithm is introduced, and physical consistency correction and verification of the earth surface parameters and the atmospheric parameters are realized through iterative optimization. The precision and the physical interpretability of passive microwave inversion of the surface temperature are improved, and the method is particularly suitable for surface temperature monitoring under cloudy and foggy weather conditions.

Images