SAR AND OPTICAL IMAGERY ANALYSES FOR THE 2017 LANDSLIDE DISASTERS IN SON LA DISTRICT, NORTH VIETNAM

In the world, 17% of fatalities attributed to various hazards were caused by landslides during a decade from 1993. Various damages due to landslides increase dramatically under the influences of global warming, particularly in south east Asian countries. For example, north Vietnam is regarded as one of the most prone regions to landslides in the world. Thus, it is an urgent need to reveal the characteristics of hazards. In recent years, European, US and Japanese space agencies have released satellite data in order to promote private use. Two types of satellite data can be used for the land surface motion analyses in this study: Sentinel-2 and -1 as optical and SAR data. The study area is It-Ong region, Son La District, north Vietnam where many landslides have occurred in 3^rd-4^th August 2017 due to heavy rains. We analyzed two types of data in this region as follows. First, we selected fifteen Sentinel-2 datasets without cloud covers from 2016 to 2019. As results, we extracted 9 landslides in an area of 2 km² on the basis of the Normal Difference Vegetation Index (NDVI). These extracted landslides correspond to the landslides observed with drone system from ground survey. Second, we selected the SAR datasets of August 1^st, 4^th, 13^th and 16^th 2017 in order to extract abrupt changes of topography and land cover. The extracted changes must be determined by ground proof what they mean. A SAR analysis generally requires two datasets which are acquired before and after an incident that you attempt to analyze. In addition, the datasets used for an analysis need to have same orbital data. Thus, we need two datasets for each orbit to complete the analysis on an area. In our case, the datasets August 1^st and 13^th were acquired in descending, and 4^th and 16^th were acquired in ascending. As results of SAR data analyzing, we detected several 300-m-sized areas that abrupt topographic change has occurred during the 2017 disaster. The topographic change areas were divided into subsidence and uplifting which correspond to be landslides and sedimentation, respectively. The SAR results can be explained by the NDVI results. Thus, we identified landslides in a mountainous region with the two satellite technologies. We need further combined studies with other significant datasets (e.g. any aerial photos, geological surveys and so on) and other monitoring systems.