Analysis of dense time-series Sentinel-1 images for risks management in the drinking water resource

摘要

This paper analyzes the potential of dense time-series Sentinel-1 images for risks management in the drinking water resource. Totally 50 SAR images ranging from March 8, 2015 to August 18, 2017 were employed to monitor land cover/use changes around the Yuqiao reservoir in Tianjin Province, China. Firstly, 49 change maps were obtained by comparing the image acquired on March 8, 2015 with all other 49 images. Then with the help of statistical procedures and very high spatial optical images from Google Earth, 258 intensive local changes in various sizes were collected and analyzed in both a visual and quantitative way. Results indicate that land cover/use changes a lot in the study area in the past three years. Among the selected changes, some are due to phenology or cultivation; a few are sudden changes on the time of imaging, while most are human induced land cover/use changes with specific purposes. Meanwhile, no human induced changes were found within the core protection region of the Yuqiao reservoir, except a few buildings were dismantled indicating the effectiveness of the environmental protection regulation. Overall, the dense time-series Sentinel-1 images can accurately capture high spatial and temporal variability of land cover/use which indicate their great potential in risks management in the drinking water resource. In the future, we will work on more features and procedures to detect changes and discriminate change types in a more intelligent way.