Is It Feasible to Use a Single Remote Sensing Optical Water Index for Rapid Mapping of Water Resources?

Abstract. Water resources are an important component of the earth's system, and the frequent occurrence of floods and droughts in the context of current climate change makes rapid and accurate monitoring of water resources particularly important. The optical water index (OWI) is a commonly used method for extracting water areas on the basis of remote sensing images, often with a high level of automation. However, selecting the right OWI is challenging due to the variety of water types. To quantitatively evaluate the differences in the mapping potential of different OWIs for surface water, we selected 12 commonly used OWIs to conduct comparative experiments among five types of surface water based on Landsat-8 and Sentinel-2 images. The results revealed that the Normalized Difference Water Index (NDWI) was better for turbid water, the Multi-Band Water Index (MBWI) was better for shaded water, the Modified Normalized Difference Water Index (MNDWI) was better for green water, and the Automated Water Extraction Index (AWEIsh)was better for swamp water and saltwater. Sentinel-2 has a higher ability to classify water than Landsat-8. Our work provides prior experience for fast and accurate water resources mapping in case of floods or droughts.