Analysis of Flood Conditions in Mekong Delta in Vietnam, Based on Sentinel Data
- Author(s): Feng, Chufan
- Advisor(s): Nguyen, Phu PDN
- et al.
Mekong River is a trans-boundary river in east Asia and southeast Asia, flows through six countries: China, Myanmar, Laos, Thailand, Cambodia, and Vietnam. Due to the abundant water resources and diverse biological resources, the Mekong River Basin has been always significant. However, there have been natural problems in the Mekong River Basin for a long time, such as the extreme droughts in the Upper Basin and floods in the Lower Basin, especially in Mekong Delta in Vietnam. Through literature review, there are few studies on flood conditions in the Mekong Delta in recent years, and it is particularly rare to find researches applying the Sentinel Data. Therefore, in this paper, Sentinel-1 SAR and Sentinel-2 MSI Imageries are used to conduct the analysis of flood conditions in the Mekong Delta, based on the Google Earth Engine (GEE) Platform.In the first part of this study, different surface water extraction methods are assessed. The Single-band Threshold Method is used for Sentinel-1 SAR data, and the surface water is extracted based on two different polarization modes (“VV” and “VH”). At the same time, Multi-Band Threshold Method is used for Sentinel-2 MSI data, and surface water is extracted based on different water indices, considering NDWI, MNDWI, LSWI, AWEI, and WI indices. After analyzing the accuracy of different methods, the Sentinel-1 data with the “VV” polarization mode performs the best, with the Overall Accuracy of 0.8618 and 0.9204 in two scenarios. In the second part, all data can be processed to extract the water surface from 2016 to 2020, using the Sentinel-1 data with the “VV” polarization mode. Then, permanent water and non-permanent water can be classified. More specifically, the pixels that recognized as water with a frequency greater than 0.7 are classified as permanent water. Accordingly, under the time series, the trends of total surface water and seasonal surface water can be generated. In the final part, a simple damage assessment is conducted, to compare the annual extreme flood events. As a result, in the past five years from 2016 to 2020, the most extreme flood event occurred in 2018, with the affected area of 695,117 hectares and an exposed population of 1,500,517.