Waterbody Extraction from Small and Medium-Sized Streams Using Sentinel-1 Images Based on Shortest Plane Distance

In this study, SAR (Synthetic Aperture Radar) images are used to validate a method for extracting the waterbody from small and medium-sized streams based on the shortest plane distance (DST). The research area is the Hantangang River and Jungnangcheon in the Hangang River Basin of South Korea. To validate the water body extraction method, SAR satellite image data (Sentinel-1) and high-resolution optical satellite PlanetScope are employed concurrently. After preprocessing, the brightness distribution of the Sentinel-1 photos is equalized using histogram matching. To achieve an efficient stream extraction, a weight is applied that is the DST from the stream centerline. The optimal parameter value is obtained using the k-means method after combining this weight value with Sentinel-1's VH and VV polarizations. Depending on the resolution limit, this value allows the waterbody to be extracted with maximum accuracy from Sentinel-1 images. The waterbody extraction can be calculated using an elliptic equation based on the correlation between the VV, VH, and DST. Results show that the average accuracy is 0.45-0.75, and the average Kappa coefficient is 0.60-0.85. This study demonstrates that the DST can be used to estimate the area of a waterbody. Furthermore, the proposed method extracts the waterbody more easily and quickly than the existing method.

This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (No. NRF-2021R1A2C2003471).