Use of fractional forest cover maps for detection of forest fires in the Brazilian Amazon

Abstract

The Brazilian Amazon is affected by large forest fires resulting from the combination between stronger droughts and human activities. The monitoring of forest fire distribution and its mapping on a large regional scale, especially with remote sensing remains a challenge because the impacts of forest fires are less visible than lands changing and because its spectral signature changes rapidly because of the new growth, until if damages to the forest has outcomes for the structure of forests. We propose a new approach based on the use of the fractional forest cover maps from CLASlite with Landsat data for detected forest fire impacts. The method applied in Paragominas, is a county in the agricultural frontier of the Amazon in Brazil, covering an area of 19,395 km², was impacted by a large fire during the 2015 dryness. The degradation forester by fire had detected by using 2016 Landsat images. A method has been validated with observation points and aerial photo interpretation. The results showed the Non-Photosynthetic Vegetation band increased clearly in the burned area. This variable is useful to map the extent of the burned area.