Design And Implementation Of SAR Pixel Tracking Parallel Algorithms For Mountain Glacier Motion Monitoring

Glaciers are sensitive indicators to global climate change and are also valuable freshwater resources on the earth.As a direct reflection of glacier movement,surface velocity has a vital role and significance for glacier dynamics investigation,disaster prediction and risk assessment,and deep understanding of climate response mechanism.Under the background of the increasing amount of remote sensing image data,it has become an inevitable requirement and trend to carry out glacier flow velocity time-series analysis.Pixel-Tracking technology is based on SAR image intensity information to achieve sub-pixel-level accuracy of azimuth and range to two-dimensional surface deformation offset information through multi-level registration,which is one of commonly used means to extract glacier surface velocity.However,with the continuous enrichment and accumulation of spaceborne SAR data,the traditional PT method has been difficult to meet the needs of large data volume and long-term glacier monitoring.In this paper,we will improve the performance of traditional PT algorithm by parallel programming,and make up for the defects of traditional PT algorithm,so as to realize the long-term monitoring and time-series analysis of glacier surface motion based on spaceborne SAR data.Based on the parallelized pixel-tracking algorithm,this paper uses small baseline time series analysis technology to carry out research on two representative large mountain glaciers in the middle and low latitudes of Asia.The main work involves the following three aspects:1.Parallelization research is carried out for the pixel-tracking algorithm commonly used in the extraction of glacier surface velocity.A parallel program based on CPU multi-threading and GPU multi-computing core is developed.The CPU uses good logic operation function and GPU data processing multi-core characteristics.The parallel algorithm of pixel-tracking is implemented,and the advantages and performance comparison of CPU multi-threading and GPU multi-core are tested and analyzed through experimental data.The experimental results show that the parallel program greatly improves the efficiency of the pixel-tracking algorithm on the premise of the same accuracy.2.Using ALOS/PALSAR data to study the movement of Fedchenko and its surrounding glaciers,based on the parallelized pixel-tracking algorithm to extract the surface motion of the glacier,through geometric transformation model fitting,terrain effect compensation and filtering,etc.The spatial and temporal distribution of glacier velocity is estimated,and the spatial-temporal characteristics of the surface motion of the Fedchenko Glacier and the Grumm-Grzhimaylo glacier are analyzed.3.With the help of the small baseline subset strategy,the parallelized SBAS-PT algorithm is implemented.By selecting the Sentinel-1A satellite IW mode SAR data of the South Inylchek glacier research area,the GPU-accelerated parallelized pixeltracking algorithm is used to analyze the surface motion of the glacier,which proves the feasibility and reliability of the parallelized SBAS-PT algorithm.