Landsat and MODIS Images for Burned Areas Mapping in Galicia, Spain

The extent, frequency and intensity of forest fires in Mediterranean regions have become an important problem in recent decades. Nowadays, remote sensing is an essential tool for the planning and management of the land at different scales. In the field of forest fires remote sensing images have been used in many different types of studies and currently applied to detect burned areas by means of images, providing quickly, easily and affordable the limits of burned areas immediately during or after the fire season. The importance of these products lies in the possibility to obtain perimeter, area and damage level caused by wildfires. The objective of this study was the evaluation of multi-scale remotely sensed images and various mapping methods for the identification and estimation of burned areas. The area of the study was situated in Galicia, a region of Spain punished year after year by important wildfires. By employing 7 images before, during and after the occurrence of forest fires, and working with different methods it was possible the collection of several products and results. The satellite imagery used was Landsat TM5 and MODIS, and the methods carried out were mainly spectral indices such as Normalized Burnt Ratio (NBR), Short Wave InfraRed Index (SWIR), Burnt Area Index (BAI), Burnt Area Index for MODIS (BAIM) and supervised classifications. Based on a wide literature review there were selected as suitable techniques for assess, localize and quantify burned areas. The work was separated in two sections, being differenced monotemporal and multitemporal analyses, depending on the images involved in each part. The results showed that which indices can distinguish burned areas with the high precision. There were found common problems of all indices as the classification of burned areas in shaded regions as unburned areas. Landsat images proved to be the most accurate images to perform studies with burned areas due to its high spatial resolution comparing with MODIS images. As a final products were obtained with precision the total burned area, the perimeter, the localization and the burn severity of the regions affected by wildfires. The data obtained could be used to create a database of burned areas, or based in the repetitive patterns, as useful information in order to prevent future forest fires