Snow Depth Variation Monitoring In Mountainous Catchment Based On D-InSAR

Snow depth is a key index to reflect the spatio-temporal variation of snow cover and an indispensable parameter for the study of global and regional climate change and hydrological cycle.Differential interferometric synthetic aperture radar(D-InSAR)measurement techniques have the advantage of quickly obtaining high-precision change information of the surface.The geometric function relationship between the snow depth and differential interferometric phase formed by microwave penetrating snow layer before and after snowfall is established,which has been widely used in regional small-scale snow depth estimation research.However,the estimation accuracy is affected by interferometric coherence,phase unwrapping accuracy,local topography,snow parameters and ascending/descending orbit data.Therefore,on the one hand,how to improve or improve the above important influencing factors based on the single-orbit SAR data of ascending/descending orbit combined with the distribution characteristics of mountain snow? On the other hand,how to make use of the different line-of-sight(LOS)observation ability to the earth and further improve the accuracy of snow depth estimation through the fusion technology of ascending/descending orbit has become a hot issue in the current D-InSAR snow depth research.Based on Sentinel-1 SAR data,this thesis takes Babao River Basin in the northeast of Qinghai-Tibet Plateau as the research area.Aiming at the above two kinds of research problems,the snow depth estimation is studied by improving the single-orbit D-InSAR snow depth estimation method and using the SAR ascending/descending orbit fusion technology.Specific research contents and main conclusions are as follows:(1)Based on the improved single-orbit D-InSAR snow depth estimation method,using Sentinel-1 ascending/descending orbit data,by introducing high correlation coefficient,Sentinel-2 L2 A,Google Earth image and ESA land cover products,ground control points(GCP)were selected accurately to correct the phase error of snow depth unwrapping.Based on the 12.5 m DEM and SAR incidence Angle parameters,the empirical error of the slant-phase relationship model is reduced by using the satellite local incidence angle and measured snow density observation data,so as to improve the single-orbit D-InSAR snow depth estimation method.The quantitative verification results based on 252 surface snow depth measurements show that the improved single-orbit D-InSAR has a high accuracy in estimating the ascending/descending orbit snow depth.Among them,the R~2 of snow depth estimation in the way of ascending orbit can reach 0.84,the RMSE is the lowest 1.38 cm,the MAPE is 16.49 %,and the MBE% is the lowest -09.40 %.The descending orbit snow depth was estimated to be 0.92 R~2,3.71 cm RMSE,20.31 % MAPE and-16.30 % MBE%.Meanwhile,qualitative comparison with Sentinel-2 L2 A optical images shows that the estimation results of ascending/descending orbit snow depth can reasonably reflect the spatial distribution characteristics of snow depth,and the estimated snow depth distribution corresponds to that of snow depth in optical image,which can meet the requirements of regional snow depth monitoring in mountain areas on the whole.(2)Snow depth estimation method based on ascending/descending orbit fusion combines surface deformation monitoring technology of ascending/descending orbit fusion with D-InSAR snow depth estimation method.Firstly,the Sentinel-1 satellite SAR parameters were used to establish a two-dimensional surface deformation measurement model of the ascending/descending orbit.The snow depth value in the LOS direction obtained from the input of the ascending/descending orbit based on the improved D-InSAR snow depth estimation method was used to calculate the two-dimensional surface information after the ascending/descending orbit fusion technology.The vertical snow depth of ascending/descending orbit fusion is obtained.The results of quantitative and qualitative accuracy analysis based on 75 surface snow depth measurements and Sentinel-2 optical images show that the accuracy of snow depth estimation is further improved,with R~2 reaching 0.86,RMSE is 1.95 cm,MAPE is17.90 %,MBE% is-09.40 %.The results of the spatial distribution of snow depth are closer to the actual terrain and snow distribution at different elevations,which significantly supplements the underestimated snow depth distribution in some areas of the single-orbit,and can further reduce the estimation deviation in the monitoring of snow depth in mountainous areas.(3)The improved D-InSAR snow depth estimation method and the snow depth estimation method based on the ascending/descending orbit fusion was used to monitor the snow depth time series of two consecutive snow accumulation periods in the Babao River Basin from 2020 to 2022.The results show that the spatiotemporal distribution of snow depth has high spatiotemporal heterogeneity.Spatially,the snow depth in the mountains with higher elevations is thicker and lasts for a long time,and the snow depth accumulates first and changes less.On the other hand,the distribution of snow depth in the valley and plain at the foot of mountains is unstable,and the snow depth is relatively thin and the spatial change speed is fast.In terms of time,the snow depth changes dramatically,the maximum snow depth difference is 42.80 cm.On the whole,the snow depth is greater in the 2020-2021 snow cover period,while the snow depth decreases in 2021-2022.The variation range of snow depth in accumulation period is relatively small,while the variation range of the ablation period is large and the maximum/light snow depth values of the snow cover period appear in it.Due to the influence of geographical environment,local climate and snow cover characteristics,the temporal variation of snow depth in different regions with different topographic characteristics has great spatial difference.In general,compared with the unimproved D-InSAR snow depth estimation results,the snow depth accuracy of the improved D-InSAR and the ascending/descending orbit fusion technology estimation is significantly improved,which can accurately monitor the quantitative changes of snow depth in the Babao River Basin,and the spatial distribution of snow depth is more consistent with the actual terrain and altitude conditions.But at the same time,77.3 % of the estimated snow depth has underestimated error.Further research shows that the ability of D-InSAR snow depth estimations is not only affected by the interferometric decorrelation factors of D-InSAR system,but also affected by the penetration ability of microwaves in the snow and snow parameters such as snow stratigraphy structure,temperature and snow density.