Multivariate analysis of landscape wildfire dynamics in a Mediterranean ecosystem of Greece

Abstract

This paper focuses on spatial distribution of long-term fire patterns versus physical and anthropogenic elements of the environment that determine wildfire dynamics in Greece. Logistic regression and correspondence analysis were applied in a spatial database that had been developed and managed within a Geographic Information System. Cartographic fire data were statistically correlated with basic physical and human geography factors (geomorphology, climate, land use and human activities) to estimate the degree of their influence at landscape scale. Land cover types of natural and agricultural vegetation were the most influential factors for explaining landscape wildfire dynamics in conjunction with topography and grazing.

Probabilistic prediction of daily fire occurrence in the Mediterranean with readily available spatio-temporal data

“…In agreement with previous studies (Cardille et al 2001, Pew & Larsen 2001, Amatulli et al 2006, Kalabokidis et al 2007, Syphard et al 2008, Vilar et al 2010, Padilla & Vega-García 2011, Miranda et al 2012, Oliveira et al 2012, Martínez-Fernández et al 2013, we found that including anthropogenic factors as explanatory variables can significantly improve the prediction of fire occurrence. The comparison of different models showed that a model with land cover types, population and road density has a significantly better predictive ability than one based on FWI alone.…”

“…Various studies have looked into the combined effect of weather and anthropogenic factors (Cardille et al 2001, Pew & Larsen 2001, Amatulli et al 2006, Kalabokidis et al 2007, Syphard et al 2008, Vilar et al 2010, Padilla & Vega-García 2011, Miranda et al 2012, Oliveira et al 2012, Martínez-Fernán-dez et al 2013. The temporal resolution of these studies is seasonal or yearly, and thus the weather factors include mean, minimum and maximum temperatures, as well as cumulative precipitation.…”

“…Due to the uncertainties in the influencing factors, as well as to random effects in the fire process, such a prediction must necessarily be probabilistic. Various probabilistic models are proposed in the literature, including Poisson models (Cunningham & Martell 1973, Mandallaz & Ye 1997, Syphard et al 2008, logistic regression (Vasconcelos et al 2001, Preisler et al 2004, Kalabokidis et al 2007, Syphard et al 2008, Chuvieco et al 2009, Arndt et al 2013, multiple regression (Sebastián-López et al 2008, Oliveira et al 2012, neural networks (Vasconcelos et al 2001, Vasilakos et al 2009) and Bayesian networks (Dlamini 2009). Recently, machine learning algorithms have been found to be well suited to modeling and predicting fire occurrences, due to their greater flexibility compared to classical regression analysis.…”

Human-caused fire occurrence modelling in perspective: a review

“…Usually, lower elevation (Kalabokidis et al 2007;Sebastián-López et al 2008;Kwak et al 2012;Narayanaraj and Wimberly 2012;Liu and Wimberly 2015) and smaller slope gradient (Preisler et al 2004;Syphard et al 2008;Dondo Bühler et al 2013;Oliveira et al 2014;Argañaraz et al 2015) increase HCF occurrence. Since surface temperature and humidity are affected by terrain, these may be reflecting climatic conditions.…”

Integration of Lightning- and Human-Caused Wildfire Occurrence Models

“…Logistic regression analysis has been used widely both to predict and also to explain human-and/or lightning-caused fires by integrating geophysical, environmental, or socioeconomic variables (e.g., related to topography, vegetation, land uses, climate and meteorological conditions, environmental parameters, fire danger indices, human factors) with observed fire occurrence (Martell et al 1987;Vega-García et al 1995;Lin 1999;Pew and Larsen 2001;Vasconcelos et al 2001;Martínez et al 2004;Wotton and Martell 2005;Kalabokidis et al 2007;Prasad et al 2008;Martínez et al 2009;Modugno et al 2008;Vilar et al 2008;Nieto et al 2006). Other statistical methods such as linear regression, classification regression trees, neural networks, generalized additive models, or Bayesian probability have also been used in fire risk mapping to generate risk models (Chuvieco et al 1999;McKenzie et al 2000;Sebastián et al 2001;Chao-Chin 2002;Koutsias et al 2004;Preisler et al 2004;Robin et al 2006;Amatulli et al 2006Amatulli and Camia 2007;Syphard et al 2007;Vega-García 2007;Yang et al 2007;Romero-Calcerrada et al 2008).…”