Inversion Of Dry Snow Depth Based On Co-polarized Phase Difference In Typical Area Of Altay

Snow is an important component of the cryosphere and one of the most active natural elements of the Earth's surface.It has active and positive feedback on the climate and affects the surface energy budget,water cycle,and global climate change Snowmelt water is the main water source for domestic water use and industrial production,and is an important source of river water supply.It is particularly important for the arid and semi-arid regions in western China.The spatial distribution and spatio-temporal changes of snow parameters are of great importance for climate change research,ecological environment assessment,regional water resources management and snow disaster monitoring and forecasting,etc.The acquisition of high-precision snow depth for watershed scale climate research and hydrological simulation are of great significance.Compared with optical remote sensing and passive microwave remote sensing,Synthetic Aperture Radar(SAR)has the ability to observe the Earth all-weather,and has the characteristics of high spatial resolution,interferometry,and polarization imaging and is the powerful tool for retrieving snow parameters in high altitude and cold regions.This study is based on the project "Retrieving Snow Depth Using Polarimetric Phase Difference of SAR During Accumulation Period in Typical Area of North Xinjiang Province"(Grant Number:41671344).The snow depth in the middle reaches of Kelan River in north Xinjiang Province is retrieved using CPD model with TerraSAR-X stripmap dual-polarimetric SAR data and measured samples.Some theory and methods about retrieving snow depth using CPD model are discussed based on the analysis of snow microstructure and its polarization characteristics and the sensitive SAR data and snow cover period of the CPD model are selected to build a snow depth inversion model and explore different factors affecting the precision of retrieving snow depth.This research provides a new idea for improving the retrieving snow depth accuracy using CPD model.The main research contents and conclusions of the paper are:(1)Snow microstructure and polarization characteristics analysis and CPD model sensitivity analysis.The snow microstructure mainly refers to the shape of the ice crystals in the snow and its arrangement.It is affected by the self-gravity compaction of snow and delivery of water vapor and energy in vertical direction.The shape of ice particle in snow changes from disk-shaped,near-spherical to vertical-needle shapes and the arrangement of ice particles in snow change from random structure,horizontal arrangement,isotropic random structure to vertical arrangement,and the corresponding phase difference between VV polarization and HH polarization changes from positive to negative.The shape of ice crystals,snow density,angle of incidence of SAR imaging,and wavelength all have an effect on the change of CPD.Selecting SAR data with satisfied parameters is beneficial to improving the accuracy of retrieved snow depth.Sensitivity analysis confirmed that K-band and X-band SAR data are ideal snow depth inversion data sources for CPD models.(2)Construction of CPD model and validation of retrieved snow depth with global window size of filter.In the snow inversion process,in order to suppress the inherent speckle noise of SAR image and improve the correlation between VV polarization and HH polarization data,a certain size of low-pass filter is needed to calculate CPD.The results show that the depth of inversion is related to the window size of the Gaussian low pass filter with full width at half maximum.In general,as the size of the filter window increases,the inversion accuracy rises sharply to the highest value,and then remained stable with some ups and downs,and finally it slowly declined.By traversing the snow depth inversion accuracy of different filter window sizes,the filter window with the highest retrieved snow depth accuracy is selected to calculate the CPD,and the snow depth inversion can obtain the optimal retrieved snow depth results.In the process of global snow depth inversion,the correlation coefficient,water body,forest,shrub land,grassland,and man-made land surface are used to mask the CPD calculation and inversion of snow depth,which can exclude outliers in snow depth inversion and increase retrieved snow depth accuracy.(3)Analysis of applicability of CPD model for retrieving snow depth under different geographical factors.CPD has specific distribution patterns in different land cover types,snow density ranges,and SAR local incidence angle ranges.Changes in any one of these factors can cause changes in CPD of snow depth.Considering the study area as a homogeneous area and using the global optimal window size of filter to calculate CPD for retrieving snow depth,the contribution of different factors to the CPD is mistaken for the CPD caused by snow depth,which is one of the main errors in the retrieving process.Based on the differences in snow density,local incidence angle of SAR system and land types,the study area is divided into different sub-areas,which helps to increase the homogeneity of the sub-areas.The adaptive filter window size in calculating CPD of SAR image is an effective means to improve accuracy of predicted snow depth and enhance the applicability of retrieving snow depth using CPD modelBased on the dual-polarization SAR data,the study analyzes the evolution of the snow structure and polarization characteristics,chooses the best SAR imaging parameters,uses the CPD model to retrival snow depth in snow accumulation period.The influence of window size of filter,land types,snow density and local incidence angle of SAR signature on the accuracy of retrieved snow depth provides a new idea for improving the accuracy of retrieved snow depth using CPD model,and has certain theoretical and application innovation.