| Spatio-temporal distribution of surface water is an important component for drought and flood management, hydrology and agriculture, especially in high poplulated catchments and in regions under climate change presure. With an increased number of Earth-observation satellites providing a large diversity of remote sensing data, there is the great potential to monitoring the surface water from local to regional as well as global scale. This is a main reason why there is extremly attended to use the satellite imagery in general and Sentinel-2 in particular. The later is often used because of (i) the Sentinel-2 mission is a land monitoring constellation of two satellites that provide high resolution optical imagery and provide continuity for the current SPOT and Landsat mission, (ii) providing a global coverage of the Earth\'s land surface every 10 days with one satellite and 5 days with 2 satellites, making the data of great use in on-going studies and (iii) the satellites are equipped with the state-of-the-art Multispectral Imager instrument, that offers high-resolution optical imagery.<br>
In this context, Artifical Neural Network (ANN) is applied to detect the water and non-water area using Sentinel-2 images for An Giang province within the Vietnamese Mekong Delta. Firstly, Sentinel-2 images, in which cloud and cloud shadow areas are less than 10% of the total image area, is collected from 2015 to 2020, resulting 39 images. Secondly, Artifical Neural Network is used to train on selected images. Sensitivity tests are performed and carried out to optimize the performance of the classification and assess the retrieval accuracy. The results show that estimated surface water maps from the Sentinel-2 images is well in line with the observed water elevation at Tan Chau location. The correlation coefficient between estimated surface water area and water elevation is closely to unity. Finally, the spatio-temporal surface water in the studied area in the period from 2015 to 2020 is presented and discussed.
|
