Chang, CH; Lee, H; Kim, D; Hwang, E; Hossain, F; Chishtie, F; Jayasinghe, S; Basnayake, S (2020). Hindcast and forecast of daily inundation extents using satellite SAR and altimetry data with rotated empirical orthogonal function analysis: Case study in Tonle Sap Lake Floodplain. REMOTE SENSING OF ENVIRONMENT, 241, 111732.

The Tonle Sap Lake (TSL) is the largest natural freshwater lake in Southeast Asia and is called the "heart of the lower Mekong" due to its high aquatic biodiversity and is considered as one of the most productive freshwater ecosystems of the world. Its floodplain eco-system, which is strongly tied to seasonal flood pulse, is extremely important for food security, trade and economy of Cambodia, supporting the livelihoods of about 1.7 million people. On the other hand, flood can also be extremely devastating in the region along the TSL. In recent years, studies have pointed out that rapid growing number of water infrastructures as well as future climate changes may alter the hydrological cycle of the Mekong River Basin (MRB) and are expected to influence the flood pulse of the TSL and surrounding TSL floodplain. Therefore, it is timely to understand historical inundation extent and predict its likely future state. In this study, we proposed a Rotated Empirical Orthogonal Function (REOF) analysis-based daily inundation extent estimation framework, integrating multi-temporal stack of Sentinel-1A Synthetic Aperture Radar (SAR) imagery and Jason-series satellite altimetry data. The framework can generate daily, cloud-free and gap-free inundation extents for any given time depending on the altimetry data provided. A long-term El Nino and Southern Oscillation (ENSO) index-based daily TSL level forecasting method with months of lead time was also proposed to fulfill the framework's forecasting capacity. In this study, the framework was adopted in the TSL floodplain area for hindcast (2003 to 2015) and forecast (January to July 2019) of daily inundation extents. Estimated inundation extents were cross-compared with MODIS-derived and Sentinel-1-derived inundation maps, resulting in up to higher than 90% of Critical Success Index (CSI). The proposed framework has (1) innovative capacity of estimation of future daily areal inundation extents and is (2) a fully remote sensing-based framework which can empower local authorities tasked with water resource management decisions without relying on upstream countries. The framework has potential to be implemented in other major river basins or wetlands (e.g., Amazon River Basin, and Congo River Basin). The implementation on SAR imagery from other satellites with different bands of electromagnetic wave is also possible but requires more investigation.