A Neural Network Model for Wildfire Scale Prediction using Meteorological Factors

Abstract

Listen

Translate article

Forest fires are natural hazards defined as movements of fire through unregulated and uncontrolled forested areas. They pose a permanent risk of loss of forest and forest land. The ability to reliably forecast the region that could be involved in a forest fire incident will help to optimize fire prevention efforts. It appears that Portugal may theoretically make better use of the wildfire risk assessment. More than any other region in Europe, it is a country overrun by wildfires. It has a large amount of forest. Forest fires have a long-term impact on the climate because they contribute to deforestation and global warming, which is one of the main causes of the phenomenon. This research employs Back Propagation Neural Network (BPNN) and Recurrent Neural Network (RNN) models with meteorological parameters as inputs to anticipate forest fires as a means of safeguarding forest biodiversity. The results indicate that using meteorological data, it is possible to anticipate the severity of a forest fire at the beginning.