Application of remote sensing for monitoring of flood areas

Abstract

Traditional measurement techniques “in situ” sometimes fail to magnify the spatial distribution of floods. For these cases, the remote sensors provide methodologies of very low economic cost and high reliability when mapping flooded areas and quantifying the damages. Due to the dynamic nature of these phenomena, it is necessary to use satellite images of high temporal resolution, however this type of images usually have a low spatial resolution. In relation to this problem, traditional classification techniques are not reliable enough for flood delineation and monitoring since they use “hard methods” of classification, where the coarse pixel is assigned a single type of coverage. On the other hand, “smoothed methods” have the ability to assign different kinds of coverage to the interior of the thick pixel. The present investigation makes the application of a sub-pixel analysis methodology (sub-pixel analysis - SA) for the monitoring of flooded areas. The improvement of the delimitation is achieved with the use of topographic attributes provided by a digital terrain model (DTM). The methodology was applied to the monitoring in the Great Depression Momposina, specifically to delineate the swamp of Zapatosa.