Modeling the spatial variation of the explanatory factors of human-caused wildfires in Spain using geographically weightedlogistic regression

M. Rodrigues, J. de la Riva, Stewart Fotheringham

Research output: Contribution to journalArticle

81 Citations (Scopus)

Abstract

Forest fires are one of the main factors transforming landscapes and natural environments in a wide variety of ecosystems. The impacts of fire occur both on a global scale, with increasing emissions of greenhouse gases, and on a local scale, with land degradation, biodiversity loss, property damage, and loss of human lives. Improvements and innovations in fire risk assessment contribute to reducing these impacts. This study analyzes the spatial variation in the explanatory factors of human-caused wildfires in continental Spain using logistic regression techniques within the framework of geographically weighted regression models (GWR). GWR methods are used to model the varying spatial relationships between human-caused wildfires and their explanatory variables. Our results suggest that high fire occurrence rates are mainly linked to wildland-agricultural interfaces and wildland-urban interfaces. The mapping of explanatory factors also evidences the importance of other variables of linear deployment such as power lines, railroads, and forestry tracks. Finally, the GWLR model gives an improved calculation of the probabilities of wildfire occurrence, both in terms of accuracy and goodness of fit, compared to global regression models.

Original languageEnglish (US)
Pages (from-to)52-63
Number of pages12
JournalApplied Geography
Volume48
DOIs
StatePublished - Mar 2014
Externally publishedYes

Fingerprint

wildfires
wildfire
spatial variation
Spain
regression
modeling
wildland-urban interface
power line
land degradation
railroads
forest fires
greenhouse gas emissions
railroad
forest fire
forestry
risk assessment
biodiversity
logistics
damages
greenhouse gas

Keywords

  • Fire risk
  • Forest fires
  • GIS modeling
  • GWR
  • Human causality
  • Logistic regression

ASJC Scopus subject areas

  • Forestry
  • Tourism, Leisure and Hospitality Management
  • Environmental Science(all)
  • Geography, Planning and Development

Cite this

@article{1c992d36bd0c425091b2e47e37317601,
title = "Modeling the spatial variation of the explanatory factors of human-caused wildfires in Spain using geographically weightedlogistic regression",
abstract = "Forest fires are one of the main factors transforming landscapes and natural environments in a wide variety of ecosystems. The impacts of fire occur both on a global scale, with increasing emissions of greenhouse gases, and on a local scale, with land degradation, biodiversity loss, property damage, and loss of human lives. Improvements and innovations in fire risk assessment contribute to reducing these impacts. This study analyzes the spatial variation in the explanatory factors of human-caused wildfires in continental Spain using logistic regression techniques within the framework of geographically weighted regression models (GWR). GWR methods are used to model the varying spatial relationships between human-caused wildfires and their explanatory variables. Our results suggest that high fire occurrence rates are mainly linked to wildland-agricultural interfaces and wildland-urban interfaces. The mapping of explanatory factors also evidences the importance of other variables of linear deployment such as power lines, railroads, and forestry tracks. Finally, the GWLR model gives an improved calculation of the probabilities of wildfire occurrence, both in terms of accuracy and goodness of fit, compared to global regression models.",
keywords = "Fire risk, Forest fires, GIS modeling, GWR, Human causality, Logistic regression",
author = "M. Rodrigues and {de la Riva}, J. and Stewart Fotheringham",
year = "2014",
month = "3",
doi = "10.1016/j.apgeog.2014.01.011",
language = "English (US)",
volume = "48",
pages = "52--63",
journal = "Applied Geography",
issn = "0143-6228",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Modeling the spatial variation of the explanatory factors of human-caused wildfires in Spain using geographically weightedlogistic regression

AU - Rodrigues, M.

AU - de la Riva, J.

AU - Fotheringham, Stewart

PY - 2014/3

Y1 - 2014/3

N2 - Forest fires are one of the main factors transforming landscapes and natural environments in a wide variety of ecosystems. The impacts of fire occur both on a global scale, with increasing emissions of greenhouse gases, and on a local scale, with land degradation, biodiversity loss, property damage, and loss of human lives. Improvements and innovations in fire risk assessment contribute to reducing these impacts. This study analyzes the spatial variation in the explanatory factors of human-caused wildfires in continental Spain using logistic regression techniques within the framework of geographically weighted regression models (GWR). GWR methods are used to model the varying spatial relationships between human-caused wildfires and their explanatory variables. Our results suggest that high fire occurrence rates are mainly linked to wildland-agricultural interfaces and wildland-urban interfaces. The mapping of explanatory factors also evidences the importance of other variables of linear deployment such as power lines, railroads, and forestry tracks. Finally, the GWLR model gives an improved calculation of the probabilities of wildfire occurrence, both in terms of accuracy and goodness of fit, compared to global regression models.

AB - Forest fires are one of the main factors transforming landscapes and natural environments in a wide variety of ecosystems. The impacts of fire occur both on a global scale, with increasing emissions of greenhouse gases, and on a local scale, with land degradation, biodiversity loss, property damage, and loss of human lives. Improvements and innovations in fire risk assessment contribute to reducing these impacts. This study analyzes the spatial variation in the explanatory factors of human-caused wildfires in continental Spain using logistic regression techniques within the framework of geographically weighted regression models (GWR). GWR methods are used to model the varying spatial relationships between human-caused wildfires and their explanatory variables. Our results suggest that high fire occurrence rates are mainly linked to wildland-agricultural interfaces and wildland-urban interfaces. The mapping of explanatory factors also evidences the importance of other variables of linear deployment such as power lines, railroads, and forestry tracks. Finally, the GWLR model gives an improved calculation of the probabilities of wildfire occurrence, both in terms of accuracy and goodness of fit, compared to global regression models.

KW - Fire risk

KW - Forest fires

KW - GIS modeling

KW - GWR

KW - Human causality

KW - Logistic regression

UR - http://www.scopus.com/inward/record.url?scp=84894160311&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84894160311&partnerID=8YFLogxK

U2 - 10.1016/j.apgeog.2014.01.011

DO - 10.1016/j.apgeog.2014.01.011

M3 - Article

AN - SCOPUS:84894160311

VL - 48

SP - 52

EP - 63

JO - Applied Geography

JF - Applied Geography

SN - 0143-6228

ER -