Time series analysis of the Alos PALSAR radar data using an ecosystem approach for the detection and mapping of flooded areas in Continental Colombia

Abstract

This article describes a methodology for the interpretation and the processing of wide beam Alos PALSAR Images (WB) of HH polarization, for the detection of flooding areas at 100m resolution. Speckle has to be filter as well as other typical radar geometrical distortions need to be corrected before any classification procedures. The physical interactions between the L band radar waves and the land surface allows the detection of open sky flooding and floods under the canopy. The analysis of a time series of 29 radar (WB) mosaics, covering all the continental Colombian territory, allowed the classification of 29 flooding maps, along an observation window of 5 years from 2007 to 2011, comprising years of extreme climatic variations and consecutive El Niño la Niña events, making of this time a good baseline for flooding studies. The interpretation and the processing of the WB radar images is done following and ecosystemic approach that includes the use of a forest structural map for the correction of the flooding detections. The resultant flooding maps are compile in a single flood frequency map, in which every pixel indicates the number of times that the pixel was detected as flooded within the 29 flooding maps. This flood frequency map was used as an input for the evaluation of the wetland map at national level, and for the delineation of wetland ecosystems. The vegetation structural map used for the flooding analysis was also created with radar images but of the fine beam, Alos PALSAR-FB, polarizations (HV) and (HH). The LCCS (Land Cover Classification system) of the FAO was used to describe the structural vegetation types. A validation procedure was applied to both maps following a random stratification method over 19 windows distributed along the National territory.

Keywords

• Inundación bajo dosel
• Mapa de frecuencias de inundación
• Mapas de inundación
• Mapeo de tipos de humedales
• Series de tiempo