Identifying wildland fire ignition factors through sensitivity analysis of a neural network

Abstract

Artificial neural networks (ANNs) show a significant ability to discover patterns in data that are too obscure to go through standard statistical methods. Data of natural phenomena usually exhibit significantly unpredictable non-linearity, but the robust behavior of a neural network makes it perfectly adaptable to environmental models such as a wildland fire danger rating system. These systems have been adopted by many developed countries that have invested in wildland fire prevention, and thus civil protection agencies are able to identify areas with high probabilities of fire ignition and resort to necessary actions. Since one of the drawbacks of ANNs is the interpretation of the final model in terms of the importance of variables, this article presents the results of sensitivity analysis performed in a back-propagation neural network (BPN) to distinguish the influence of each variable in a fire ignition risk scheme developed for Lesvos Island in Greece. Four different methods were utilized to evaluate the three fire danger indices developed within the above scheme; three of the methods are based on network's weights after the training procedure (i.e., the percentage of influence-PI, the weight product-WP, and the partial derivatives-PD methods), and one is based on the logistic regression (LR) model between BPN inputs and observed outputs. Results showed that the occurrence of rainfall, the 10-h fuel moisture content, and the month of the year parameter are the most significant variables of the Fire Weather, Fire Hazard, and Fire Risk Indices, respectively. Relative humidity, elevation, and day of the week have a small contribution to fire ignitions in the study area. The PD method showed the best performance in ranking variables' importance, while performance of the rest of the methods was influenced by the number of input parameters and the magnitude of their importance. The results can be used by local forest managers and other decision makers dealing with wildland fires to take the appropriate preventive measures by emphasizing on the important factors of fire occurrence.

A Review of the Main Driving Factors of Forest Fire Ignition Over Europe

“…shown that a large number of fire ignitions can be explained by the first group of factors (Langhart et al 1998;Mangiavillano 2004;Lampin et al 2005 and2006;Badia-Perpinya and Pallares-Barbera 2006;Catry et al 2007;Romero-Calcerrada et al 2008;Vasilakos et al 2008;Martínez et al 2009;Padilla and Vega-García 2011). On the other hand, socio-economic indicators such as the unemployment rate have also been shown to be clearly linked to fire occurrence in many areas of southern Europe (Ferreira de Almeida et al 1992, Leone 1999, Chuvieco et al 1999, Velez 2000, Sebastian-Lopez 2008.…”

OntoFire: an ontology-based geo-portal for wildfires

“…physical geography), such as vegetative fuels, topography and weather. Wildfires are affected by the vegetative fuels because the quantity, size, density, quality, continuity and moisture content of the vegetation determine the availability of fuel for combustion (Vasilakos et al, 2009). Topography modifies the general climate of a region and thereby affects the availability of fuels and fire behavior.…”

“…Joshi et al (2007) describe a Disaster Mitigation Modeling System that uses ontologies for tying together information from various fields of disaster mitigation. In Ilyas and Ahmed (2010), ontologies are exploited for the automatic data gathering, analysis and decision making in a disaster management system, while Zaharia et al (2009) discuss how intelligent agents can extract knowledge to support local or government decisions in better handling the consequences of disasters. Most of these approaches, however, do not exploit the rich semantics of the underlying data to provide efficient ways for information retrieval.…”

Comparison of the fuzzy AHP method, the spatial correlation method, and the Dong model to predict the fire high-risk areas in Hyrcanian forests of Iran

“…For example, some studies have used Dong models to predict fire high-risk areas in the forests (Dong et al 2005;Erten et al 2005;Eskandari et al 2013a), whereas other researchers have applied analytic hierarchy process (AHP) or fuzzy sets to model forest fire risk (Chuvieco & Congalton 1989;Vadrevu et al 2009;Sowmya & Somashekar 2010;Mahdavi et al 2012;Zarekar et al 2013;Atesoglu, 2014;. Logistic regression has been used to model fire ignition probability (Vasconcelo et al 2001;Rollins et al 2004;Martinez et al 2009;Jurdao et al 2012;Sitanggang et al 2013;Eskandari & Chuvieco 2015), and the artificial neural network has been used to predict fire regimes (Alonso-Betanzos et al 2002;Vakalis et al 2004, Vasilakos et al 2009Satir et al 2016). Finally, the support vector machine approach has also been proposed for fire risk modelling (Cortez & Morais 2007;Sakr & Elhajj 2010).…”