Tropospheric Wind Field Measurement Based On Infrared Hyperspectral Observations

The global three-dimensional wind field has extensive application value and significant research significance in the aspects of climate change,extreme weather disasters,numerical weather forecast prediction,atmospheric dynamics,etc.The detection methods for wind fields in every layer of the atmosphere are constantly strengthened and improved,among which the tropospheric atmospheric moves frequently and affects human life closely.Tropospheric wind field is one of the observation elements that fail to meet the requirements of improving weather forecast and the accuracy of troposphere wind field data is also an important basis for determining whether it can enter the assimilation model of numerical weather forecast prediction.At present,the main detection methods of tropospheric wind field include satellite-borne wind lidar monopulse measurement of zonal wind and retrieval of atmospheric motion vector by meteorological satellite imager.Although the wind field information has improved numerical weather prediction on regional and global scales,the vertical resolution of wind field is affected by the fuzzy distribution of vector height,resulting in errors,the height distribution of wind field is the main error source of wind field information.Since the conception of infrared hyperspectral vertical detection of atmospheric parameters was proposed,with the launch of the satellite-borne infrared hyperspectral sounder into space,the method of infrared hyperspectral sounder high-precision vertically detection of wind field retrieving has gradually become the new research direction of wind field detection methods.In this paper,we develop research on the measurement method of the troposphere wind field,based on the radiation spectrum data of FY3D/HIRAS,an infrared hyperspectral sounder onboard polar meteorological satellite in our country.According to HIRAS infrared hyperspectral data,we carried out experiments on the evalution of radiation accuracy consistency,retrieval of clear-sky atmospheric temperature and humidity profile and wind field measurement of vapor channel respectively.The main results of this paper are as follows:(1)Aiming at the consistency of radiation accuracy of HIRAS radiation spectrum data,the HIRAS measurements are compared with other international high-precision similar instruments through simultaneous nadir overpass observations,including IASI on European meteorological operational satellite Met Op-A/B and CrIS on American meteorological satellite.The inter-comparison samples are paired through strict spatial,temporal and geometrical collocation,and the observation objects are satisfied the uniform scenes,this developed inter-comparison algorithm is applied in operational data processing.(2)Aiming at the radiation spectrum samples collocated with HIRAS and similar instruments,we compared the mean deviation and standard deviation of bright temperature,analyzed the temperature dependence between the deviation and the observed object,and monitored the variation trend of the deviation.The matching samples of HIRAS-IASI are distributed in the North and South Arctic regions with lower target temperatures,HIRAS agrees well with IASI at the long-wave IR and middle-wave IR with less than 0.5K in most channels,there are no apparent scene-dependent features for individual spectral channels,the standard deviation is less than 2K.The spectral BT of HIRAS is lower than IASI overall with 1.5K mean deviation in most channels at short-wave IR,the deviation shows scene-dependent features,and the standard deviation decreases as the object temperature increases.The trend of BT deviation on the monitoring of HIRAS-IASI,HIRAS-CrIS at early on-orbit test(April 2018)and HIRAS-IASI long-term test(April-December 2018)shows that consistency of HIRAS is stable,and the deviation at short-wave IR is slightly larger in the lower object temperature conditions.(3)Aiming at retrieval of clear-sky atmospheric temperature and humidity profile based on HIRAS infrared hyperspectral data,we constructed a one-dimensional variational retrieval system.Through the retrieval channels are selected by information iteration method and weight function covering method,we compared the influence of channel number on retrieval deviation,carried out clear-sky pixel selection and data quality control of spectral samples,the observation is strictly collocated with data of forecast field and reanalysis field.We calculated the root mean square error,mean absolute error and mean error between the retrieval results and the true values,and then analyzed the accuracy of retrieval results.The retrieval pressure level of the temperature profile is 1000 h Pa-10 h Pa,and the root mean square error compared with the true value is 0.5K-3.5K,the retrieval accuracy is better than the predicted value overall.The retrieval pressure level of humidity profile is 1000 h Pa-200 h Pa,the retrieval accuracy of lower layer is better than 20% and that of higher layer is better than that of predicted value,the reason for large deviation position is probably affected by initial guss error.(4)Aiming at HIRAS infrared hyperspectral data to measure tropospheric three-dimensional wind field,we developed a cross-platform clear-sky tropospheric three-dimensional wind field measurement algorithm.We adopted an inter-comparison matching method to obtain revisited data and allocated the height of the wind field according to the spectral channel weight function covering method.We used a dense optical flow method to track wind vector,combined with reanalysis data,we quantitatively calculated the root mean square error of wind speed,the mean absolute error of wind speed and the mean absolute error of wind direction,and compared the measurement accuracy of wind field.The feasibility of the algorithm is tested by cross-platform matching data of FY3D/HIRAS and Suomi NPP/CrIS,and the wind field measurement experiment is carried out by cross-platform matching data of FY3D/HIRAS and NOAA-20/CrIS based on the improved algorithm.The quantitative comparison results of wind field accuracy show that,the variation trend of wind speed range is consistent with the reanalysis data,and the wind speed range decreases with the decrease of pressure level,the mean absolute error of wind speed is not exceeding 4m/s and mostly less than 3m/s,the root mean square error of wind speed is not exceeding 4.5m/s and mostly less than 3.5m/s,the mean absolute deviation of wind direction is not exceeding the degree of 40 and mostly less than the degree of 30.(5)For the results of the cross-platform infrared hyperspectral wind measurement,the major error sources of wind field information are including observation errors caused by different design parameters of infrared hyperspectral instruments on different satellite platforms and positioning errors caused by spatial resolution changes caused by reprojection process.We proposed to analyse parameters of a wind measurement system of the same satellite platform,put forward the design requirements for satellite number,revisit time and global coverage observation by theoretically calculating,analyzed mathematical relationship between spatial resolution,temporal resolution and wind speed accuracy resolution of infrared hyperspectral wind measurement instrument,the structure of small-size integration of the instrument and the expected requirements for key parameters are designed.