Detecting and modeling multi-scale space-time structures: the case of wildfire occurrences

Gabriel, Edith; Opitz, Thomas; Bonneu, Florent

Journal de la société française de statistique, Tome 158 (2017) no. 3, pp. 86-105.

Les feux de forêt, qu’ils soient d’origine naturelle ou malveillante, représentent d’importants risques économiques et écologiques. La structure relative aux éclosions de feux est complexe, avec de fortes variations spatiales et temporelles, souvent dirigées par de nombreuses covariables telles que l’occupation des sols, le climat et la météo. Dans cet article, nous nous intéressons à l’éclosion de feux recensés quotidiennement depuis 1981 dans les Bouches-du-Rhône, région très touristique marquée par un climat méditérannéen, et pour lesquels nous connaissons la surface brûlée associée. Nous adoptons une approche basée sur les processus ponctuels afin d’étudier les variations spatio-temporelles à grande échelle et les interactions à petite échelle. Après un aperçu sur la litterature existante, nous explorons l’influence des covariables climatiques (comme la température, les précipitations) et de l’occupation des sols sur la probabilité d’occurrence d’un feu. Les questions statistiques sont soulevées par la structure multi-échelles des données, qui sont de plus définies sur des supports variés, comme des grilles très fines pour l’occupation des sols, des grilles grossières pour la localisation des feux qui engendrent de l’incertitude de positionnement, et des séries météorologiques observées en des sites irrégulièrement répartis. Nous ajustons à ces données un modèle de Cox log-gaussien incluant l’information des covariables et des effets spatio-temporels via la méthode INLA (Integrated Nested Laplace Approximation) et nous analysons la structure d’interaction résiduelle via la fonction $K$ spatio-temporelle inhomogène. Nous mettons en évidence et modélisons en particulier les effets inhibiteurs localement présents dans l’espace et dans le temps induits par de très grandes surfaces brûlées.

Wildfires due to natural origin or arson come along with important economic and ecological risks. Typically, the structure of relative risk of fire events is highly complex, shows strong variation over space and time and is driven by numerous covariates like land use, climate and weather. We here adopt a point process approach to study space-time large-scale variations and local interaction behavior in a data set reporting geolocalized fire events and burnt surfaces on a daily basis in the Bouches-du-Rhône county in Southern France, marked by a Mediterranean climate and high touristic activity. After a review of the existing literature, we explore the influence of land use and climatic covariates like temperature and precipitation on the probability of event occurrence. Statistical challenges arise from the multi-scale structure of data defined over various supports like fine grids for land use covariates, coarse grids for fire position leading to positional uncertainty, and meteorogical series observed at irregularly spaced measurement sites. We fit a log-Gaussian Cox process including covariate information and nonparametric spatial and temporal effects to our data through the technique of Integrated Nested Laplace Approximation, and we analyze the residual interaction structure through the space-time $K$-function adapted to the setting of second-order intensity reweighted stationarity. Specifically, we also study inhibitive effects that arise locally in time and space after fire events with relatively large burnt surfaces.

Keywords: INLA, multi-scale data, K-function, point process, space-time dependence, wildfire
Mot clés : dépendance spatio-temporelle, données multi-échelles, feux de forêt, processus ponctuel

[1] Baddeley, AJ; Møller, J; Waagepetersen, R Non- and semi-parametric estimation of interaction in inhomogeneous point patterns, Stat Neerl, Volume 54 (2000) no. 3, pp. 329-350 | MR | Zbl

[2] Bonneu, F. Exploring and modeling fire department emergencies with a spatio-temporal marked point process, Case Studies in Business, Industry and Governement Statistics, Volume 1 (2007), pp. 139-152

[3] Comas, C; Conde, J; J, Mateu A second-order test to detect spatio-temporal anisotropic effects in point patterns, Statistics (2017) (To appear) | MR | Zbl

[4] Diggle, PJ; Chetwynd, AG; Häggkvist, R; Gooding, S Second-order analysis of space-time clustering, Stat Methods Med Res, Volume 4 (1995), pp. 124-136

[5] Gabriel, E Estimating second-order characteristics of inhomogeneous spatio-temporal point processes: influence of edge correction methods and intensity estimates, Methodololy and Computing in Applied Probability, Volume 16 (2014) no. 2, pp. 411-431 | DOI | MR | Zbl

[6] Gabriel, E. Spatio-temporal point pattern analysis and modelling, Encyclopedia of GIS, 2nd Edition (2016) | DOI

[7] Genton, M.; Butry, D.; Gumpertz, M.; Prestemon, J. Spatio-temporal analysis of wildfire ignitions in the St Johns River water management district, Florida, International Journal of Wildland Fire, Volume 15 (2006), pp. 87-97

[8] Gabriel, E.; Diggle, P.J. Second-order analysis of inhomogeneous spatio-temporal point process data, Statistica Neerlandica, Volume 63 (2009), pp. 43-51 | MR

[9] Gonzalez-Monsalve, J.; Rodriguez-Cortes, F.; Cronie, O.; Mateu, J. Spatio-temporal point process statistics: a review, Spatial Statistics (2016) | DOI | MR

[10] Gabriel, E; Rowlingson, B; Diggle, PJ stpp: An R package for plotting, simulating and analysing spatio-temporal point patterns., J Stat Softw, Volume 53 (2013) no. 2, pp. 1-29 | DOI

[11] Gabriel, E; Wilson, DJ; Leatherbarrow, H; Cheesbrough, J; Gee, S; Bolton, E; Fox, A; Fearnhead, P; Hart, A; Diggle, PJ Spatio-temporal epidemiology of Campylobacter jejuni enteritis, in an area of Northwest England, 2000-2002, Epidemiol Infect, Volume 138 (2010), pp. 1384-1390

[12] Juan, P.; Mateu, J.; Saez, M. Pinpointing spatio-temporal interactions in wildfire patterns, Stochastic Environmental Research and Risk Assessment, Volume 26 (2012) no. 8, pp. 1131-1150

[13] Lindgren, Finn; Rue, Håvard Bayesian spatial modelling with R-INLA, Journal of Statistical Software, Volume 63 (2015) no. 19

[14] Lindgren, Finn; Rue, Håvard; Lindström, Johan An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach, Journal of the Royal Statistical Society: Series B (Statistical Methodology), Volume 73 (2011) no. 4, pp. 423-498 | MR | Zbl

[15] Møller, J.; Diaz-Avalos, C. Structured Spatio-Temporal Shot-Noise Cox Point Process Models, with a View to Modelling Forest Fires, Scandinavian Journal of Statistics, Volume 37 (2010) no. 1, pp. 2-25 | MR | Zbl

[16] Møller, J; Ghorbani, M Aspects of second-order analysis of structured inhomogeneous spatio-temporal point processes, Statistica Neerlandica, Volume 66 (2012), pp. 472-491 | MR

[17] Opitz, Thomas Latent Gaussian modeling and INLA: A review with focus on space-time applications (2016) (Technical report Submitted to Journal de la Société Française de Statistique)

[18] Pereira, P.; Turkman, K.; Amaral-Turkman, M.; Sa, A.; Pereira, J. Quantification of annual wildfire risk; A spatio-temporal point process approach, Statistica, Volume 73 (2013) no. 1, pp. 55-68

[19] RESSTE Network Analyzing spatio-temporal data with R: Everything you always wanted to know – but were afraid to ask (2016) (Submitted to Journal de la Société Française de Statistique) | Numdam | MR | Zbl

[20] R Core Team R: A Language and Environment for Statistical Computing (2016) https://www.R-project.org/

[21] Rue, Håvard; Martino, Sara; Chopin, Nicolas Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations, Journal of the royal statistical society: Series b (statistical methodology), Volume 71 (2009) no. 2, pp. 319-392 | MR | Zbl

[22] Siino, M.; Adelfio, G.; Gabriel, E.; Mateu, J. Classes of anisotropic spatio-temporal log-Gaussian Cox processes to model earthquake events (2017) (Technical report)

[23] Simpson, Daniel; Illian, Janine; Lindgren, Finn; Sørbye, Sigrunn; Rue, Håvard Going off grid: Computationally efficient inference for log-Gaussian Cox processes, Biometrika (2016) (In press) | MR | Zbl

[24] Serra, L.; Saez, M.; Mateu, J.; Varga, D.; Juan, P.; Diaz-Avalos, C.; Rue, H. Spatio-temporal log-Gaussian Cox processes for modelling wildfire occurrence: the case of Catalonia, 1994–2008, Environmental and Ecological Statistics, Volume 21 (2014) no. 3, pp. 531-563 | MR

[25] Tamayo-Uria, I.; Mateu, J.; Diggle, P.J. Modelling of the spatio-temporal distribution of rat sightings in an urban environment, Spatial Statistics (2014) | MR

[26] Turner, R. Point patterns of forest fire locations, Environmental and Ecological Statistics, Volume 16 (2009), pp. 197-223 | MR

[27] Xu, H.; Schoenberg, F. Point process modelling of wildfire hazard in Los Angeles county, California, The Annals of Applied Statistics, Volume 5 (2011), pp. 684-704 | MR | Zbl