SEMI-AUTOMATIC APPROACH OF BURNT AREA EXTRACTION BASED ON DIFFERENT INDICES

The following study is focused on the analysis of burnt areas extracted by the use of an automatic tool. The burnt areas are obtained by applying different vegetation and burnt areas indices available in the literature (NDVI, NDB, SAVI, BAIM and BAIS2), calculated as combinations of two or more bands of the satellite images to synthesize the information carried by multispectral sensors and frequently used for the mapping fires. The research aims to find strong and alternative methods to combine with survey field activities. Traditional methods can often be subject to sampling errors or even to the non-perimeter of the event where the area is inaccessible. Satellite images allow to perform these activities simply and automatically, reducing costs and significantly increasing the accuracy and completeness of the data. Some areas analysed in the Copernicus Rapid Mapping and Risk and Recovery Mapping service were taken as case studies. A validation test has been made on all the selected indices, to evaluate the limits and potential of the different approaches and provide a burnt area closer to reality. Not having ground surveys data, the reference information for validation is given by the photointerpretation by experts in the sector.