Flood Mapping From Sentinel-1 Imagery Using Index Composition and HSI Transformation

Abstract

Flood is one of the most influential natural disasters. An accurate and rapid flood mapping is critical for an effective flood emergency response. Synthetic aperture radar (SAR) is important for flood mapping in terms of its ability to image the Earth's surface regardless of weather conditions and time. Sentinel-l is currently the only free SAR constellation in operation. Its high spatial-temporal resolution provides important support for SAR-based flood monitoring. However, previous approaches for SAR-based flood mapping have limited applicability to detect floods in mixed pixels. To overcome this problem, we apply index composition and HSI (hue, saturation, and intensity) transformation method to flood mapping via Sentinel-1 imagery. First, three different indexes are used to create a false-color image, i.e., the Height Above Nearest Drainage (HAND), the Normalized Difference Flood Index (NDFI), and the Normalized Difference Flood in Vegetated areas Index (NDFVI). Then, an HSI transformation is employed to extract flood areas. This method is implemented on Google Earth Engine (GEE) to support the near real-time flood monitoring. Three study areas are selected to verify the accuracy and robustness. The average overall accuracy of the three study areas is 96%, and kappa coefficients vary from 0.6234 to 0.8230, showing promising applicability to detect floods.