Flood disasters account for about a third of all natural catastrophes throughout the world (bynumber and economic losses) and are responsible for more than half of the fatalities. (Gerhard2006) Remote sensing and satellite data are effective tools for hazard related applications like flooding that need a synoptic view when the field is on emergency status and it is hard to make ground surveying. Radar systems are more advantageous in data providing and observation of floodplains than optic systems, since flood occurs mostly on cloudy days.
In this study, the effectiveness of radar images for flood applications is shown with multi-temporal radar data. Also object oriented image processing technique is used, which is a promising technique that enables to use many additional parameters of image objects with their spectral properties. Especially, segmentation and classification in a hierarchical mannerprovides an effective classification. As shown in this study, different regions affected by flooding can easily be extracted and the accuracy obtained (84 %) was satisfactory.