基于FY-4B/GIIRS的华东区域大气温湿廓线反演与融合OA

In order to improve the accuracy of retrieving the atmospheric temperature and humidity profiles from FY-4B/GIIRS data and determine the optimal method for obtaining the atmospheric temperature and humidity profiles,this study used the BP neural network algorithm to retrieve the atmospheric temperature and humidity profiles,based on the FY-4B/GIIRS Level 1 brightness temperature data and the ERA5 reanalysis data in the clear sky in East China in July 2022.Moreover,the one-dimensional variational and optimal interpolation methods were respectively used to fuse the retrieval results with numerical forecast products to obtain atmospheric temperature and humidity profiles with higher accuracy.Finally,the accuracy of the retrieval and fusion results was evaluated with the ERA5 data and sounding data.Results show that:(1)for the clear-sky atmospheric temperature profile,when compared with the ERA5 data,the error of the fusion result of the retrieval and the forecast data by the optimal interpolation method is the smallest,with an RMSE of 0.56 K;when compared with the sounding data,the error of the fusion result of the retrieval and the forecast data by the one-dimensional variational method is the smallest,with an RMSE of 0.87 K.(2)For the clear-sky atmospheric humidity profile,when compared with the ERA5 data,the error of the retrieval result is the smallest,with an RMSE of approximately 7.5%;when compared with the sounding data,the error of the fusion result of the retrieval and the forecast data by the one-dimensional variational method is the smallest,with an RMSE of approximately 13%.All in all,in the clear sky,based on the FY-4B/GIIRS data,the results of retrieving the atmospheric temperature and humidity profiles using the BP neural network are better than the current FY-4B/GIIRS Level 2 products.The fusion model further improves the accuracy of the retrieval results,with the optimal temperature error being less than 1 K and the humidity error being less than 15%.