Analysis on recent changes of water area in China's Poyang lake region using Sentinel-1 data

Abstract

Poyang Lake is the largest fresh water lake in China. Monitoring water area changes of Poyang Lake is essential for flood mitigation and control in the middle and lower part of Yangtze River. It helps to better conserve wetland and ecological resources, and plays an important role for sustainable water use and management. Landsat and MODIS sensor images are widely used for mapping water body with their sensitivity for spectral reflectance of water bodies and relatively high spatial resolution and temporal resolution. Studies using Landsat images mainly investigate changes of Poyang Lake in dry season because of its 16 days revisit cycle and effect of cloud cover. It builds on the assumption that Poyang Lake undergone relatively small changes in dry season. However, this assumption may not be reliable. MODIS images with very short revisit period, on the other hand, has been proven not be able to assess water area of Poyang Lake accurately in dry season. The contribution of this study is to investigate recent Poyang Lake water area changes with unforeseen temporal resolution in high water period and spatial resolution in low water period using Sentinel-1 satellite, with the specific objective of investigating recent Poyang Lake water area changes in intra-month scales. After preprocessing, calibration, speckle filtering and terrain correction, water bodies were extracted using the histogram-based threshold method. Binarization results were imported into Google Earth for visualization. Shape and area information of Poyang Lake were extracted in ArcGIS. Results show that significant intra-month variability exists, even in dry season. During the observation period from October 2014 to March 2016, October 2014 was the month with the largest max/min water coverage ratio. Water coverage of winter in 2014 and 2015 was completely different, with severe drought happened in 2014 and unusual winter flood in 2015. Parts covered by water in the whole year were identified. Sequence exists for marshland to emerge during water retreat. This study demonstrated the value of Sentinel-1 SAR data to reveal intra-month variation with its regular observation independent of weather conditions. Sentinel-1 SAR data, with near real-time and free of charge distributing policy, relatively high spatial and temporal resolution, is bound to be an important data source for early warning, monitoring, assessment of drought disaster, and will plays an important role for mapping flood extent during the flood, enabling better disaster management. The approach used here can also be applied to other large lakes or wetlands research.