The Research on a Novel Method of Retrieving Atmospheric Parameters Based on GABP

Abstract

In this article a new and novel method based on the genetic algorithm and the BP neural network(GABP) was proposed to improve the precision and speed of retrieving atmospheric parameters(temperature, humidity and liquid water content) from a microwave radiometer. Firstly, the genetic algorithm(GA) improves the BP neural network weights and thresholds and obtains an ideal range of those weights and thresholds. Secondly, the BP neural network is trained from these optimized weights and thresholds, and an ideal BP neural network is obtained in a shorter time. Finally, the model is tested by the measured data from a multi-channel microwave radiometer. The results by GABP are in a good agreement with the sounding data and those by other methods. And also GABP is a new algorithm for high precision and rapid inversion of atmospheric parameters.