Inversion Of Atmospheric Temperature And Humidity Profiles Based On CrIS Infrared Hyperspectral Satellite Data

The atmospheric temperature and humidity profiles are indispensable basic data for scientific research such as numerical weather prediction and climate change assessment.It is of great importance to accurately retrieve atmospheric temperature and humidity profiles from hyperspectral satellite data for improving weather forecast and climate prediction ability.This paper takes East China as the research area,and the D-R(Dual-Regression)inversion algorithm is used to study the atmospheric temperature and humidity profiles of the high spectral infrared radiation data of a new generation of Cr IS(Cross-track Infrared Sounder)mounted on Suomi-NPP(National Polar-orbiting Partnership)satellite.The main contents and conclusions of this paper are as follows:(1)In the statistical regression inversion method of atmospheric temperature and humidity profile,there are often insufficient data and limited representativeness of the matched samples.A combined ECMWF's(European Centre for Medium-Range Weather Forecasts)atmospheric temperature and humidity profile reanalysis data is proposed,and a number of representative samples are selected as the background field,and the inversion accuracy of the atmospheric temperature and humidity can be significantly improved.(2)Based on the above method,the regression coefficient is calculated from the atmospheric temperature and humidity profile data of ECMWF in East China from from September to September in 2012 and 2013.In this paper,the atmospheric temperature and humidity profiles of four meteorological stations which are Baoshan,Hangzhou,Nanjing and Qingdao were retrieved from June to September in 2014,2015 and 2016.The results showed that when the temperature profile was in the height layer above 800 h Pa,the RMSE(root mean square error)remained basically around 1K,and the RMSE increased slightly at the height layer below 800 h Pa,but it was still within 2K.The relative humidity profiles in the whole layer of RMSE are relatively small,which remain within 20%,and the inversion accuracy is relatively high.In addition,the BIAS and standard deviation of the retrieved atmospheric temperature and humidity profiles are very small,indicating that the dispersion degree of the deviation is very small,the inversion effect is stable,and there will be no outliers.(3)In order to further verify the accuracy of using the background field selected in this paper and the D-R algorithm to inverse the atmospheric temperature and humidity profiles with Cr IS data,this paper compares and verifies the atmospheric temperature and humidity profiles measured by the encrypted sounding of Baoshan meteorological station in Shanghai and the NOAA's(National Oceanic and Atmospheric Administration)atmospheric temperature and humidity products based on the NUCAPS(NOAA Unique Combined Atmospheric Processing System)inversion.The results show that the overall BIAS of the atmospheric temperature profile obtained by D-R algorithm based on ECMWF's temperature and humidity reanalysis is almost within 1K and the RMSE is around 1.5K,and the inversion accuracy is equivalent to that of the NUCAPS algorithm.In the near surface layer,the RMSE of the D-R algorithm is still within 2K,which is better than the NUCAPS algorithm.The RMSE of the relative humidity profile is basically less than 20% under the pressure 350 h Pa,and the BIAS is basically within the range of 15%,which is equivalent to the NUCAPS algorithm's inversion accuracy.However,when the air pressure is 100 h Pa to 350 h Pa,the relative humidity RMSE and BIAS retrieved by D-R algorithm increase slightly,and the inversion accuracy decreases.(4)Finally,this paper analyzes and discusses the strategy of optimizing the inversion of atmospheric temperature and humidity,and studies the influence of the two aspects of the selection range of the background field and the selection of the sensitive channel.The results show that the narrowing of the selection range of the background field does not necessarily improve the inversion accuracy of the statistical algorithm,and may even reduce the inversion accuracy due to the reduction of the feature sample profiles under different weather conditions.The RMSE of retrieving the temperature profile after the channel optimization is further reduced,which indicates that the accuracy of the Cr IS temperature profile retrieved by the sensitive channel is improved.