Inversion And Validation Of HJ Thermal Infrared Image Sea Surface Temperature

Sea Surface Temperature(SST)is a main parameter to describe the interaction between sea-air and the structural state of the ocean.The use of thermal infrared remote sensing technology can quickly obtain large-scale real-time sea surface temperature data around the world.At present,the most mature and widely used SST products are MODIS,ASTER,and AVHRR sea temperature products.The HJ-1B satellite launched in 2008 can obtain thermal infrared images with high spatial and temporal resolution on a global scale,providing basic data for us to obtain high-quality SST products.However,the existing single-channel algorithm is for Landsat TM,which is not suitable for the inversion of the HJ-1B satellite SST.At the same time,the authenticity check of the inversion products is more based on the comparative analysis of other satellite SST products of known precision,but not quantitative analysis.Therefore,the application of HJ-1B satellite SST products is limited.This paper takes the South China Sea as the research area,selects the HJ-1B IRS images from 2010 to 2019 as the inversion data source,studies and analyzes the SST inversion algorithm suitable for the HJ-1B satellite,and obtains a high-precision HJ-1B IRS SST product.Then,using Argo buoy data and MODIS SST products as verification data sources,the authenticity of the inversion HJ-1B SST product is verified,which provides convenience for the application in the sea area.The research work and results of the dissertation have the following four aspects:(1)Using the general single-channel algorithm proposed by Jiménez-Mun(?)oz and Sobrino for SST inversion.Firstly,according to the temporal discontinuity and spatial dispersion of HJ-1B satellite data and Argo buoy data,seven sub-study areas in the South China Sea are proposed to be selected for the convenience of HJ-1B IRS and CCD,as well as Argo buoy profile points,MODIS SST Spatial and temporal matching of data.Then,the MODTRAN model was used to re-simulate the functional relationship between atmospheric transmittance,atmospheric up-down radiation and atmospheric water vapor content in the JM&S inversion algorithm.A total of 112 time-phase IRS SST products in 7 sub-study areas were inverted using the revised general single-channel algorithm.(2)The inversion HJ-1B IRS SST product was directly checked based on the measured data of the international Argo buoy.The verification results show that the best inversion accuracy is the sub-study area with the lowest geostrophic deflection force at the lowest latitude.The R~2between the inversion temperature and the verification temperature is 0.9954,and AAD of the inversion is 0.3K.The second highest inversion accuracy is the sub-study area with the highest latitude and small solar radiation,with an R~2of 0.9903 and an AAD of0.5K.The worst inversion accuracy is in the sub-study area that is very vulnerable to typhoons,with R~2of 0.9584 and AAD of the inversion greater than 1K.The deviation between the inversion temperature and the buoy temperature in all phases was larger from April to September,and the maximum deviation reached±1.5K.(3)In order to further verify the accuracy of the inverse SST product,an indirect verification based on the MODIS SST product was carried out.The validation results show that the inversion accuracy remains basically consistent with the direct test accuracy.The sub-study area with the best inversion accuracy based on point-to-point verification has an AAD value of 0.3 K.The sub-study area in the origin of typhoons,which is highly susceptible to typhoon attacks and susceptible to the influence of Kuroshio from the East China Sea,has the worst inversion accuracy with an AAD value of more than 1.1 K.The SSIM values of the inversion and verification images are both greater than 0.8 K for the image-based indirect verification.The absolute deviation between the inversion image and the validation image at all moments is larger in May-September,with a maximum of 2 K.The overall accuracy of the inversion of the sea surface temperature is higher due to the small spatial heterogeneity of the water surface.(4)Seasonal variation analysis on the time scale of 112 time phases was conducted for the HJ-1B SST inversion products.The validation results show that in spring,summer,autumn and winter,the inversion accuracy is high in spring and winter with correlation coefficients of 0.9929 and 0.9987,respectively,and the inversion temperature is concentrated with MODIS temperature distribution.The worst inversion accuracy is in summer,the correlation coefficient is only 0.9649,and the values of AAD,RMSE,and STD are greater than 1 K.The deviation of inversion temperature from MODIS temperature at all times is concentrated between±0.5 K in spring and winter,and the deviation in summer and autumn is discrete,especially in summer,which is basically distributed between±1.5 K.