Retrieval Method Of Multichannel Ground-based Microwave Radiometer For Atmospheric Parameters Based On BP Neural Network

Microwave remote sensing provides a more continuous,adequate,real-time and convenient way for atmospheric parameter observation,as an important means of atmospheric microwave remote sensing,ground-based microwave radiometer can detect temperature,water vapor density,and liquid water,as well as other atmospheric parameters,which makes the microwave radiometer has been widely used in the meteorological science,radio communications,satellite positioning and navigation,space observation and other fields.Based on the ground-based microwave radiometer,this paper improved BP(Error Back Propagation)neural network algorithm and using it to retrieve atmospheric parameters,thus improving the detection accuracy of atmospheric parameters effectively.Based on brightness temperature detection of ground-based microwave radiometer,this paper focuses on the inversion algorithm of atmospheric parameters such as temperature profile,water vapor density profile,relative humidity,and liquid water content.Main work of this paper are as follows: the historical sounding data of Wuhan area is grouped into cloud free and cloud after the data conversion and interpolation processing.Then,studied the atmospheric microwave transfer model,the microwave brightness temperature set of cloud free conditions or cloud conditions are calculated by using atmospheric microwave transfer equation.The neural network and its improved algorithm for retrieving atmospheric parameters are analyzed,focus on the GABP(genetic algorithm Back Propagation),improving the accuracy of neural network by optimizing the initial connection weight and threshold of network structure,and comparing the performance of different improved neural network algorithms.Finally,using improved neural network to simulate the temperature,water vapor density,relative humidity and liquid water content,and compareing the results with the sounding data,analyzing the retrieve accuracy of the atmospheric parameters at different altitudes.By improving the neural network algorithm,and retrieved the temperature profile,the water vapor density profile,the relative humidity and the liquid water content of the ground microwave radiometer.Compared the retrieval results,and it shows that the neural network is established by Levenberg-Marquardt Back Propagation algorithm or Genetic Algorithm Back Propagation algorithm to retrieve the atmospheric parameters,the accuracy is improved at different altitudes,and we can get good results which provide more accurate atmospheric parameters for atmospheric scientific research and weather observations.