Abstract:

he aims of the study were to determine the vegetation change due to the effect of the flood in Tra Vinh and Vinh Long provinces during the annual flood pulse. For performing this purpose, the flooded maps of vegetation were constructed from classified results of Sentinel-1A data, Digital Elevation Models (DEM), and water level data in the study area. In this paper, we used an object-oriented classification method for multi-temporal Sentinel-1A data with 24 days interval based on the association with DEM and the water level data in the year (from October 2014 to November 2015) with overall accuracy and kappa coefficient (0.81 and 0.78, respectively). The advantage of this study was that land cover information can be observed from radar data in all weather condition, and able to detect the surface condition changes beneath the canopy of vegetation. The areal variations of vegetated types were estimated based on the backscattering coefficient values change because these value changes were affected by varying the floodwater level for vegetation. The results indicated that a large portion of the lowland vegetation (about 10.3% of the total study area) was covered with the water surface at the peak of the flooding. A backscattering coefficient changes from -7.6 dB to -20.6 dB in the flood development stage showed that the areal vegetation was completely inundated about 15.38% of the total study area.