Citation

Bispo R, Vieira FG, Bachir N, Espadinha-Cruz P, Lopes JP, Penha A, Marques FJ, Grilo A. Fire Safety J. 2023; 138: e103802.

Copyright

(Copyright © 2023, Elsevier Publishing)

DOI

10.1016/j.firesaf.2023.103802

PMID

unavailable

Abstract

Fires in urban areas typically carry severe consequences. High population density together with the complexity of urban network potentially imply significant impacts in property loss, physical damage and life losses. However, despite the impact that fires may have in urban areas, research in urban fire prediction remains limited. In this study, we modelled urban fire occurrences while making a comparative analysis of different strategies to account for spatial autocorrelation. Considering space dependence in addition to a range of social-economic explanatory variables has proven to strengthen the validity of the fitted models. The spatial Durbin error model, including population density, degraded buildings density and buying power, was selected as having the best fit. This model allowed to map the estimated probability of fire occurrence across Portugal, revealing a spatial pattern with clusters centred on the two main Portuguese city districts (Lisboa and Porto). Ultimately, the analysis of the relation between the observed urban fire incidence and the actual number of fire stations in each municipality allowed to underline the need for planning the spatial configuration of fire stations, both in number and location, at a regional scale.

Language: en

Keywords

Hotspot analysis; Spatial autocorrelation; Spatial modelling; Urban fires