| This paper presents an analysis of temporal and spatial variations of Antarctic ice sheet freezing and thawing based on high-resolution and ultra-wide-band SAR data.Firstly,based on sentinel 1A and 1B satellite data,combined with radar backscattering characteristics and spatial texture characteristics,an ice-sheet freeze-thaw detection algorithm is constructed,and the average melting area map of Antarctic ice sheet is successfully obtained.Then,the temporal and spatial variation trend and regularity of the ice sheet freeze-thaw in the warm seasom of 2015-2018 are analyzed.The SAR image is pre-processed by SNAP software,such as spatial filtering,radiation calibration,geometric correction and so on.The radar backscatter image after eliminating part of the noise and terrain influence is obtained.Super-pixel segmentation algorithm is used to simplify image segmentation while retaining basic features.Considering that radar image is affected by topographic factors in mountain areas,it will produce shadow and overlay.It is difficult to distinguish dry snow area from shadow area,infiltration zone and overlay area only based on radar backscattering characteristics.However,with spatial texture features,each snow area can be separated very well:The texture of the cover area is rougher than that of the infiltration zone,and the texture of the dry snow area is rougher than that of the shadow area.Combining the characteristics of radar backscattering and spatial texture,the corresponding regions of radar images are divided into different snow areas(dry snow area,infiltration zone,wet snow area,shadow area and overlay area).Secondly,the relevant knowledge-based verification rules are summarized and classified to obtain the final freeze-thaw detection results.The results are validated by using the temperature data of automatic weather station and cross-comparison method.The results show that the ice-sheet freeze-thaw detection results in this paper have high accuracy,the overall accuracy can reach 89.91%..According to the characteristics of radar backscattering and texture,this paper presents a frost-thaw detection algorithm for Antarctic ice sheet based on high-resolution and ultra-wide-band SAR data.According to this algorithm,the average frost-thaw region maps for each month of the warm season from 2015 to 2018 are generated,and the spatial and temporal variations are analyzed as follows:(1)The melting of the Antarctic ice sheet varies significantly in time.The experimental results show that the Antarctic ice sheet began to melt in October,and the melting area increased month by month with time.The maximum melting area was reached in December and January next year.After January,the melting area began to change from melting to freezing,and the melting area gradually decreased.The freezing and thawing of the Antarctic ice sheet has a cyclical trend in each year.(2)The melting of the Antarctic ice sheet has obvious spatial differences.The experimental results show that the melting of the Antarctic ice sheet mainly occurs near the coastline,and the melting mainly occurs in the Antarctic Peninsula and the Ross Bay area,while the melting area in the eastern Antarctic is relatively small.The combined effects of land cover type,altitude and geographic location(latitude)make the freeze-thaw of Antarctic ice sheet present a complex spatial distribution.The low latitude area melts more intensely than the high latitude area.The distribution of rocks also makes the surrounding snow area melt faster.Figure[27]Table[6]reference[61]... |
PDF Full Text Request