Study On Channel Selection Of Ground-based Hyperspectral Microwave Radiometer For Atmospheric Temperature And Humidity Profiles Detection

The development of hyperspectral microwave radiometers is of great significance for improving the accuracy of atmospheric parameter inversion.The primary work to carry out research on hyperspectral microwave radiometer is channel location determination through simulation research.The research work of calibration,inversion and hardware design needs to be carried out at selected channel locations.Therefore,this paper introduces the concept of information content,and uses it as an indicator to quantitatively describe the channel detection performance.The problem of channel selection for inversion of atmospheric temperature and wet profile by ground-based hyperspectral microwave radiometer is studied.This paper mainly contains the following research contents:1.Introduce the concept of information content into the inversion problem of atmospheric parameters,and find the DFS(signal degrees of freedom)which can represent the information content as the basis for measuring the channel detection ability.In this paper,considering the feasibility of the calculation,the information content before and after the inversion of each channel is obtained by the method of channel-by-channel selection,and the channels are arranged according to the information content.When inverting the atmospheric temperature profile,the background error and the observation error affecting the information content of each channel are analyzed.When inverting the relative humidity profile of the atmosphere,sensitivity analysis was carried out on the temperature error,background field error and observation error that affect the information content of each channel.2.The microwave radiation transmission model pwr is used to calculate the atmospheric downward brightness radiation temperature,and the temperature Jacobian matrix and water vapor Jacobian matrix of each channel are calculated.3.Inversion models of inversion of atmospheric temperature and relative humidity profiles were constructed using BP neural network,and multiple inversion experiments were performed using inversion models.200 channels with higher information content were selected as channels for temperature profile inversion.120 channels with higher information content are used as channels for inversion of relative humidity profiles.