Mapping Wildfires Using Sentinel 2 MSI and Landsat 8 Imagery: Spatial Data Generation for Forestry

Abstract

Forests host diverse ecosystems that involve various habitats. There are many complex interactions between living and non-living things in most forests. It is important to conduct observations and assessments in large forestlands where short-term and long-term direct or indirect negative impacts may occur so that they are known and measured. Scientific studies have been carried out by utilizing the various data offered by today's advanced technology with satellite imagery becoming more readily available. In this study, differenced Normalized Burn Ratio (dNBR=∆NBR) and satellite images with two different resolutions were used to generate pre- and post-wildfire spatial data. An area affected by wildfire in the Mediterranean Region of Turkey was selected as the study area. Google Earth Engine (GEE) and Geographic Information System (GIS) were used to delineate areas affected by wildfire using Sentinel-2 and Landsat 8 multispectral imagery. In order to compare the differences between the two sets of imagery, burn severity levels (low, medium-low, medium-high, and highest) and the effect of water surface were considered. For the most impacted burnt lands, areas detected with Sentinel 2 and Landsat 8 are 31.90% and 32.59%, respectively. However, burn severity classes were also observed in water surface areas likely due to interactions between land cover and water reflectance. The overall results support the use of both satellite platforms and the dNBR for burn severity mapping in medium- and large-scale post-wildfire studies.