Flood identification using Sentinel-1 SAR Imagery in Tallo Watershed

Abstract

Indonesia is a country with a high level of vulnerability to disasters, especially floods with an average flood of disasters occurring annually around 845 incidents in Indonesia. Tallo Watershed is an urban Watershed, which is prone to flooding, especially in the downstream of the river. Remote Sensing Data with RADAR sensor types (Radio Detection and Ranging) can help identify and classify floods that occur in Tallo Watershed. This Study aims to quickly identify flood disasters in Tallo Watershed by using Sentinel-1 SAR imagery data before and after flood incidents. Flood identification using 2 methods, with manually determining threshold values and visual interpretation. 102 sample survey results were used as reference data. Out of the Bag (OOB) sampling technique is used to separate reference data (32 samples) that are used to determine threshold values and validation of identification results (70 samples). This analysis showed that the accuracy of the manual threshold method was 43.13% with the results of visual interpretation reaching 88.2%. The influence of settlements in Tallo watershed can also affect the results interpretations of Sentinel-1 SAR image because it is very sensitive to surface structures.