Rapid Mapping of Forest Fires in the European Mediterranean Region - a Change Detection Approach Using X-Band SAR-Data

Usually, optical data are used for detecting burned areas in a rush mode, but the analysis of these data is often limited due to persistent cloud cover or haze. This paper focuses on the potential of X-band SAR data for mapping forest fires in the European Mediterranean region. The goal of the study was the development of an object-based, semi-automatic, robust and fast but at the same time accurate and transferable algorithm for the detection of burned areas in case of rapid mapping. For this reason change detection techniques based on image differencing, rationing and index calculation were applied. Prior to the SAR data analysis, a number of pre-processing procedures were carried out, including the elimination of speckle noise by Gamma-DE-MAP filtering, radiometric and geometric calibration as well as computation of sigma nought, a scattering coefficient. In the course of devastating fires in Greece and La Palma in 2009, a multi-temporal backscatter coefficient analysis of fire affected and unburned areas were performed. Within this comparison, the detected signal of VV and HH polarisations showed significantly higher values in images acquired after the fires than before the fires. However, VV polarised data showed a higher amplitude between pre- and post-disaster backscatter levels than HH polarised data. The classification result has been validated by SPOT 5 data and achieved an overall accuracy of 77.9%. Thus, burned area detection with SAR data could potentially play a significant role in forest fire detection in Europe.