Étude des inondations dans le delta du fleuve Rouge au Viêt-nam à l'aide d'images RSO de RADARSAT et JERS-1

Abstract

The Red River delta is a rice-growing region with well developed hydraulic resources. During the rainy season, the quantity and intensity of precipitation provoke the rising of the Red River waters and its distributaries which can eventually lead to uncontrolled flooding as we can see on the RADARSAT SAR image of the 96-08-27.The objective of this work is to determine to what extent radar images can contribute to the study of floods in the Red River delta plain.The methodological approach consists in collecting ground data at the time of satellite overpass. In this way, we can identify surface states observed on the image in relation to ground observations. Two types of classification are performed on the three sub-images of the flooding area: k-means method and ICM algorithm.The result of the k-means method is about 80% of good classification and the ICM algorithm give 70% of good classification. This algorithm keeps more details but it takes more time to compute. On the other side, the k-means method is a simple and fast method who is included in most of commercial software packages.The three images (JERS-1 95-08-04, RADARSAT 96-08-22 and 96-08-27) used for the classification have permitted to separate five classes: flooding area, open water, rice paddies, gardens and villages. Finally, the interpretation of images taken during winter and summer give the opportunity to locate the regions with high risk of flooding.