The Error Source Analysis Of The Forward Modeling Brightness Temperature Precision Of FY Satellite Infrared Water Vapor Channels

Forward modeling of satellite atmospheric sounding instruments is the foundation of assimilation and inversion.In this paper, we reduce the uncertainties on the accuracy of infrared water vapor channel to three factors:the bias caused by the difference of water vapor line for the line-by-line transmittance calculation which are the base of the fast radiative coefficients calculation, the bias caused by the fast radiative coefficients calculation arithmetic and the bias caused by the profile water vapor content which is the input for further fast radiative transmittance calculation.The study takes the bias of channel brightness temperature as the standard for measure the influence of the three possible factors mentioned above.To achieve this purpose, the study contains three sections.Firstly,different water vapor lines were input into the line-by-line model(here using LBLRTM) to get the brightness temperature by spectrum,then we got the bias caused by the different water vapor lines through the convolution with FY-3 IRAS channel spectral response function.Secondly,using TIGR 43 profile library as the training sample and NESDIS 35 profile library as the independent sample, group training was done in this section with the profiles classified by the 0.045kg/m~2 water content of the air column. Different coefficients got from the group training are used to establish RTTOV forward model and calculate the brightness temperature of FY-3 IRAS to get the bias caused by the fast radiative transmittance calculation change. Finally,a linear change of water vapor content of the American standard atmospheric profile USSA-1976 was made before it was input into the RTTOV to get the bias caused by the uncertainty of the water vapor content of input profile for fast radiative transmittance calculation.Research results show that the bias caused by the difference of water vapor lines which were offered with the update of LBLRTM is less than 0.04 K,which means less influence on the final result of the calculated the brightness temperature of infrared water vapor channel. The fast radiative coefficients generated from the group training based on the water content of air column show improvement in the forward modeling calculation of the brightness temperature of infrared water vapor channel,the evident improvement was found in the high water content case,which up to 0.2K. The bias caused by the linear change of the water vapor content of input profile shows an obvious influence,and the bias was more evident in the linear decrease of water vapor content case.The bias was more than 0.3K when there were 5% linear change in the water vapor content of the input profile,and the bias exceeded 1K when the linear change reached 30%,in the 30% linear decrease case the bias could even exceeded 2.5K.The comparison of the studies mentioned above shows clearly that the uncertainty of thewater vapor content of input profile for fast radiative transmittance calculation has significant influence on the forward modeling result of the brightness temperature of FY-3 IRAS water vapor channel,which also shows the importance of the water vapor observation accuracy.Besides, concerning the bias of simulated water vapor brightness temperature, the study suggests a method based on the group training of atmospheric water content that could improve the forward modeling calculation accuracy of the brightness temperature of infrared water vapor channel,which up to 0.2K.The study systematically analyzed the error source of infrared water vapor brightness temperature simulation and put forward an effective method to improve the accuracy of the brightness temperature of infrared water vapor channel, it shows strong application value on the retrieval of satellite water vapor channel, the assimilation of satellite data and so on.