Analysis of Different Methods for Burnt Area Estimation using Remote Sensing and Ground Truth Data

Abstract

Mapping of burnt areas and burnt severity is a key parameter to analyze the medium and long-term effects of forest fires in ecosystems. In this work different remote sensing techniques have been tested to check its suitability over a steepness burnt area in Tenerife and Gran Canaria Islands (Canary Islands-Spain) affected by two important fires during 2007 fire season. The aim of this paper is to show the results of a comparative analysis of some of the most commonly used spectral indexes in burnt land mapping applications. In this way SVI, NDVI, TVI and SAVI were used and its operative consistency with ASTER data was assessed, establishing the discrimination ability of each index between the recently burned zones and other land covers using a post-fire image. We have used supervised classification SVM (Support Vector Machines) algorithm. The results have been compared with the burnt area perimeter provided by a SPOT multitemporal image study in Gran Canaria Island and ground truth GPS perimeter provided by the regional forest service in Tenerife Island.