Research On The Algorithm Of Ground-based Infrared Radiometer For Remote Sensing Of Atmospheric Precipitation

In this paper,the training database of global profiles(SeeBor V5.0)and the radiation transfer model(MODTRAN 5.0)are used to simulate the radiation transfer,and the sky brightness temperature(Tsky)under different observations and different atmospheric conditions is obtained.In this paper,feasibility of retrieving the precipitable water vapor(PWV)from the grunod-based infrared measurements is analyzed,and an algorithm for retrieving the PWV under the clear sky from the ground-based infrared measurement is established.This study is of great significance for the selection of channels and the development of retrievive algorithm of ground-based infrared radiometer.Main research contents of this article are as follows:(1)The correlation between Tsky and PWV is analyzed.The results showed that spectral response function and aerosol have a great influence on Tsky.In addition,there is a good exponential function relationship between Tsky and PWV in low aerosol loading conditions.With the increase of aerosol loading,the correlation between Tsky and PWV becomes worse.A simple exponential model for PWV retrieval based on Tsky in low aerosol loading conditions,the retrieval accuracy of PWV is better than 0.31 cm,and the correlation between PWV and Tsky of instrument spectral response functions in the infrared window(10-13 ?m)is better than that instrument spectral response functions including water vapor(i.e.,6.3 ?m)and ozone(i.e.,9.6 ?m)absorption bands.However,the Tsky in the infrared window is low(120-290 K),that means the infrared mensurements working in the infrared window is needed to detect the lower temperature(<150 K).(2)The impact on Tsky is analyzed,which is made by aerosol,temperature,view zenith angle(VZA)and cloud.Tsky has a positive correlation with aerosol loading,temperature and VZA.Because the presence of clouds will make Tsky increase,it is impossible to detect PWV when there are clouds.(3)The neural network model of PWV retrieval was built by MATLAB.The results showed that the introduction of multi-angle(0° and 30°)Tsky can reduce the impact of aerosols on PWV retrieval and improve the accuracy of the PWV retrieval model.Meanwhile,the introduction of near-surface air temperature(Tair)and surface pressure(SP)can effectively improve the accuracy of PWV retrieval determination coefficient,root-mean-square error,and bias of the neural network model to retrieve PWV are 0.984,0.232 cm and 0.010 cm,respectively.