Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/127067
Title: Creación de un modelo de predicción de riesgos de incendios forestales usando una red neuronal convolucional sobre datos históricos de meteorología de California
Author: Ricci Voltas, Xavier
Director: Solé-Ribalta, Albert  
Tutor: Muñoz Bollas, Anna
Abstract: The main aim of this project was to create a model able to predict the chances of fire ignition in a specific geographical point under specific weather conditions. In order to create this prediction model, first it was needed to create a program to collect the available information about the relevant elements involved on the fire ignition, as topography, vegetation and meteorology. Then, using the geographical positions of the ignition of the fires in California from 2000 to 2020, information matrixes of the area around these points were created. Simultaneously, for every ignition point, on the basis of the same day weather information, random points of areas ¿not on fire¿ from the California area were selected, and for each point a matrix was also created. The two groups of matrixes where labelled and used to train a Convolutional Neurological Network to differentiate between the conditions of the areas with fire and the ones with no fire. After trying different models based on several structures of the CNN, and checked with the prediction from historical fires and random ¿no fire¿ points, eventually, the best model got an accuracy of 95%. Using this model, a system to create a fire ignition probability heatmap of any area in California with 500m terrain definition was made.
Keywords: geographic information system
artificial neural network
wildfire management
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 3-Jan-2021
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 

M2.878 - TFM_Xavier_Ricci.m4v

84,32 MBMP4View/Open
xricciTFM0121memoria.pdfMemoria del TFM11,32 MBAdobe PDFThumbnail
View/Open