Snow Depth Retrieval Based On Satellite-borne Passive Microwave Data In Forest Area Of Northeast China

Snow cover plays an important role in the process of water cycle and energy flow of forest ecosystem in winter,and it is very important to monitor the characteristics and dynamic changes of snow cover in forest areas.Satellite remote sensing technology has become an effective method to monitor snow cover and its temporal and spatial changes because the traditional artificial observation data and automatic fixed-point observation data can not meet the research needs of snow cover parameters at large regional scale.Among them,passive microwave remote sensing is the most effective technique for retrieving snow parameters.However,in forest areas,the accuracy of snow depth retrieval is also reduced due to the influence of vegetation on satellite microwave radiation transmission process.The forest area in Northeast China is mainly distributed in the mountains.The terrain has a distribution effect on the snow cover,and also has an impact on the radiant temperature transmission process.Therefore,it is difficult for the retrieval of snow depth in the forest area in Northeast China.Based on physical model and semi-empirical algorithm,snow depth retrieval algorithms suitable for northeast forest was established in this paper,which provides a new idea for improving the accuracy of snow depth retrieval in forest in Northeast China.The research work is as follows:(1)Optimization algorithm of semi-empirical snow depth retrieval in forest area in Northeast China.Based on the semi-empirical snow depth retrieval algorithm and combined with meteorological observation data,an optimization algorithm of semiempirical snow depth retrieval in forest area in Northeast China was established in this paper.In this algorithm,the permittivity of vegetation varies with temperature and the accuracy of snow depth retrieval in forest is greatly improved.Compared with other semi-empirical algorithms,the Bias of the proposed algorithm is reduced by 3.7 cm and the RMSE reduced by 2.3 cm on average.The proposed algorithm is compared with the commonly used machine learning algorithm and show that the accuracy of the proposed algorithm is higher than the existing machine learning algorithm,in which RMSE decreases by 2.03 cm on average and correlation coefficient R increases by 0.22 on average.(2)The HUT snow microwave radiative transfer model and parameter localization in forest area in Northeast China.Aiming at the forest of northeast China,the snow grain size,density and temperature parameters in the Helsinki University of Technology model(HUT)were localized to obtain the Local-HUT model(LHUT)by combining the effective snow grain size,snow-temperature relationship model and field measurement data of snow.The results show that the simulation of brightness temperature of LHUT model with localized parameters is more accurate.In the 19 GHz band,the RMSE of LHUT model decreases by 5.02 K and the Bias decreases by 5.8 K compared with the original HUT model.(3)Snow depth retrieval algorithm based on Local-HUT model in Northeast China.In this paper,a snow depth retrieval algorithm Improved-LHUT(ILHUT)based on LHUT model is proposed.Considering the influence of topography on snow retrieval,Forest-ILHUT(FILHUT),which is suitable for forest area in Northeast China,is obtained by combining semi-empirical elevation coefficient and ILHUT model.The results show that the accuracy of FILHUT snow depth retrieval is improved compared with the existing algorithms: the Bias is reduced by 5.96 cm on average;RMSE decreased by 3.18 cm on average.