Determinación del riesgo de incendios forestales mediante el procesamiento de imágenes multiespectrales en zonas con presencia de vegetación.

Abstract

This degree work was aimed to determine the risk of forest fires by processing multispectral images in areas with vegetation. An image acquisition system based on an Unmanned Aerial Vehicle platform and a five-band multispectral sensor was used to obtain the images. Aerial images were taken of each vegetation stratum (arboreal, shrub, and herbaceous) surrounding the ESPOCH-LIZARZABURU health center. Afterward, the radiometric corrections were carried out in each image using the Agisoft software, and through MATLAB, the algorithm was developed to calculate the normalized difference vegetation index (NDVI). Based on this index, the separation between uncovered soils and vegetation was carried out by classifying unsupervised images and segmentation of the images. Subsequently, the results obtained (percentage of vegetation with NDVI values between 0.1 and 0.3) and the susceptibility of the vegetable fuel allowed to establish the fire risk of the vegetation cover of the place. They were helpful to establish prevention and mitigation measures in areas identified as fire-risk zones. Finally, it was determined that the vegetation of the place corresponding to the arboreal stratum has a low risk of fire due to a small amount of fuel capable of burning. The shrub stratum presents a moderate fire risk due to the discontinuous distribution of vegetable fuel, and the herbaceous stratum presents a high fire risk since, almost in its entirety, it corresponds to dry and dead vegetation that burns quickly. The project showed that it is possible to determine the fire risk of vegetated areas from multispectral image processing in cooperation with other parameters obtained from the intrinsic characteristics of the vegetation.