Nghiên cứu các chỉ số nhận biết nước từ ảnh Sentinel-2 trên nền Google Earth Engine: áp dụng cho thành phố Sa Đéc, tỉnh Đồng Tháp

Abstract

This paper presents the results of using three popular water indices namely NDWI, MNDWI, and WNDWI on detecting water bodies for the Sa Dec city, Dong Thap province based on Google Earth Engine (GEE) cloud platform. The high pass filter is firstly applied for pan-sharpening of the shortwave infrared (SWIR) band from 20 to 10 m in spatial resolution of the domain of interest. The real water and non-water areas in the domain of interest are then digitized directly from Google Earth using the GIS tool in order to create the reference dataset. Next, sensitivity analysis of the threshold value and weighted coefficient are performed when using the WNDWI, resulting in values of -0.11 and 0.50 are obtained for the threshold value and weighted coefficient, respectively. The overall accuracy, kappa coefficient, error of commission and error of omission equal 0.98, 0.85, 0.015 and 0.006, respectively. Among three selected water indices, WNDWI and MNDWI are the most suitable indices for detecting water bodies in the studied area, with the overall accuracy>0.98 and kappa coefficient>0.84. Finally, capacity and time consuming on detecting water bodies using Sentinel-2 images based on GEE are also discussed.