Modelling wildfire occurrence at regional scale from land use/cover and climate change scenarios
Introduction
Nowadays wildfires are one of the significant threats to forested areas worldwide, and the European Mediterranean Basin is one of the more susceptible areas to fire episodes, reporting more than 85% of the total burned area in Europe (San-Miguel-Ayanz et al., 2012). Wildfires, however, are complex phenomena involving multiple factors mediate, e.g. fuel availability, moisture conditions, natural and human ignitions, meteorological/climate drivers (Gudmundsson et al., 2014) and management decisions (Hély et al., 2001), which also operate at different spatial and temporal scales (Bedia et al., 2015).
Reported changes in the Mediterranean region regarding the use of land and climate (Spinoni et al., 2020) are affecting the fire cycle (Pausas and Fernández-Muñoz, 2012), increasing the frequency and severity of wildland fires (Moreno et al., 2013), and threatening ecosystem stability, the provision of services, habitat and biodiversity conservation, landscape value and aesthetics, as well as property and human lives. These changes are expected to become more intense in the coming century (Syphard et al., 2019), increasing their effects on fires and the consequential impacts on human communities worldwide. Spain is representative of these changes within Mediterranean Europe (Stellmes et al., 2013), as lengthened, and longer fire weather seasons have become more frequent (Jolly et al., 2015). Understanding the past relationships among land use land cover (LULC) changes, climate and wildfire occurrence will allow for the prediction of future impacts and the evaluation of vulnerabilities, which will serve as input for management and policy actions (Gallardo et al., 2015). Resulting relationships will depend on the type and force of driving factors and their importance in different regions.
The European Mediterranean Basin has experienced profound LULC changes derived from human activities (Geri et al., 2010). In the last century, LULC changes in rural areas of Southern European countries were first linked to the intense abandonment of the countryside (1950s–1960s). Then, there was a shift to a new agricultural and mechanized system (1980s), followed by urban expansion in forested areas and increased recreational activities and the consequential intensification of the pressure on natural zones (2000s) (Vilar et al., 2016a). Such abandonment led to exceptional fuel accumulation due to natural reforestation processes (Geri et al., 2010) and urban pressure due to the increase in contact areas between forest and urban constructions (the so-called Wildland Urban Interface or WUI), all of which triggered an increase in wildfire risk. Besides, LULC also creates an impact through the ignition of fires, with more than 80% of fires in this area being linked to human activities and the result of negligence, accidents or acts of arson (Ganteaume et al., 2013), e.g., the use of fire to control herbaceous vegetation for cattle grazing or to clean brushwood in crops (lightning causes ∼5% in average of the known fires in this basin). Given the wide range of the human activities effects, several studies include the contact areas between forest and other covers, the so-called interfaces, such as WUI (Vilar et al., 2016a; Chas-Amil et al., 2013; Modugno et al., 2016; Lampin-Maillet et al., 2011), agricultural-forest interface (Gallardo et al., 2015; Martínez et al., 2009; Rodrigues et al., 2016) and grassland-forest interface (Rodrigues, 2014; Vilar et al., 2019), among others as human drivers of wildfire occurrence. Assuming that socioeconomic changes are likely to continue, further changes in LULC are also expected. Projecting the amount of LULC change and the location thereof will allow for future LULC derived interfaces to be obtained and for the human factor of the wildfire occurrence be represented (Gallardo et al., 2015).
On the other hand, the climate role in wildfires is mainly linked to the control of vegetation characteristics and status (Westerling et al., 2011). Pre-fire-season weather conditions have been proved to have a strong influence on the ignition and propagation of large fires because of their effect on fuel load and flammability (Urbieta et al., 2015). Moreover, the fuel moisture content (FMC), defined as the mass of water contained within vegetation per dry mass, is a critical variable affecting fire interactions with fuel (Yebra et al., 2013), and might affect both fuel ignition and fire spread rate (Viegas et al., 1992; Rossa et al., 2017). As FMC increases, the flammability of fuels tends to decrease, as more energy is needed to evaporate water before burning organic tissues (Argañaraz et al., 2018). FMC is usually separated into live (LFMC) and dead fuels (DFMC) (Chuvieco et al., 2004). Most operational fire danger rating systems include the estimation of DFMC, those lying on the forest floor (leaves, branches and debris) (Camia et al., 2003). Still, the estimation of LFMC is included less often. Less significant relations between fire spread or intensity were found in experimental data field analysis for a shrub or conifer forest (Fernandes and Cruz, 2012). Among other reasons, this is because LFMC is the result of complex interactions between previous and concurrent weather and the varied biological mechanisms that influence water content and dry matter accumulation (Jolly et al., 2014; Turner, 1981).
Climate change trends in Southern Europe are expected to lead to increased temperature, a greater number of heatwaves and dryer days (Cramer et al., 2008) and a decreased summer precipitation (Kovats et al., 2014). Modelling studies predict that this will lead to an increase in fire activity (Sousa et al., 2015), the number of large fires (Vázquez de la Cueva et al., 2012) and the burnt area (Amatulli et al., 2013; Turco et al., 2018). Dupuy et al. (2020) recently reviewed 23 studies that projected fire danger indices or fire activity (number of fires, size and burnt area) and at modelled climate-fire relationships in the European context at local, regional or continental scale. Results showed a relative increase (2–4% per decade) in mean seasonal fire danger under pessimistic climate change scenarios in the Mediterranean regions. Burnt areas are projected to increase everywhere in Southern Europe at a rate of 15–25% per decade (Dupuy et al., 2020).
Several simulation scenario studies have combined factors on humans, topography, vegetation and FMC along with climate change conditions as drivers of future wildfires in fire-prone areas worldwide. Examples of variables included in the modelling processes include, housing density (Westerling et al., 2011), distance to populated places (Liu et al., 2012), land use effects (Syphard et al., 2018), distance to roads (Syphard et al., 2019), road density (Liu et al., 2012), WUI and other land cover interfaces (Gallardo et al., 2015), aspect and slope (Westerling et al., 2011), vegetation type (Syphard et al., 2018) and fine fuel moisture content (Liu et al., 2012). However, in the European context, Dupuy et al. (2020) claimed that one of the sources of uncertainty in future estimations was the lack of information on the influence of human factors on climate-fire relationships. Moreover they affirmed that the influence of FMC should be taken into account and constitutes a possible source of bias for future predictions.
The main aim of this work was to develop an integrated modelling framework at 1 km2 target resolution to better understand the importance of climate, LFMC and human factors on future spatial and temporal characteristics of wildfire occurrence in different Southern European regions. To that end, the variable effect and distribution of current and future projected probability of wildfires were analyzed in four areas of Spain. These regions were selected as representative of LULC changes and climate conditions in Southern Europe and have different socioeconomic, biophysical and wildfire characteristics. The specific objectives were first, to calibrate statistical-based regression models for wildfire occurrence combining climate, LFMC, topography and LULC interfaces. Secondly, the regression models were applied to the projected LULC and climate variables obtaining the future wildfire probability. The three main research questions addressed are, (1) would wildfire probability increase or decrease in relation to LULC and climate changes? and how?; (2) which variables would be more influential in the different regions?; and (3) could this modelling framework be applied to other scales and areas for planning and management actions?
Section snippets
Study sites
This paper covers four Spanish regions: Ourense, Zamora, Madrid and Valencia (Fig. 1, Table 1). These regions are representative of the landscape types derived from the different socioeconomic processes affecting Southern Europe in the last decades and the relation thereof to historical fire events and trends, as explained in the Introduction.
More specifically, two of the sites are rural-oriented (Ourense and Zamora in the northwest of Spain), and the other two are representative of urban
Methods
The methodological steps followed to obtain the future wildfire occurrence in the study sites for the 20-year time period (2041–2060, centered on 2050) include four main phases: (1) modelling the baseline wildfire occurrence, (2) simulating LULC change scenarios, (3) building climate projections and (4) modelling future wildfire occurrence (Fig. 2). This integrated modelling framework was proposed at 1 km2 target resolution as appropriate for regional scale in Spain due to the average size of
Explanatory and response variables
Fig. 4 shows the LULC interfaces derived from CCI-LC 2005. The extent and spatial distribution differed among sites. FAI dominated in all regions. FGI presence was also notable in Ourense (Fig. 4a). Cells covered by WUI were also crucial in all study sites except Zamora (Fig. 4b).
Seasonal climate-related variables within the baseline modelling for the 2001–2010 period included for the factor analysis showed differences among the years of the study period, illustrating then the expected
Discussion
This paper concludes that both LULC and climate changes will have an effect and drive future wildfire probability of occurrence. Drivers of change and resulting probability will vary across and within the analyzed sites. Projected wildfire probability will increase by target 1 km2 grid cells mostly in Zamora site, where the percentage of cells with an increase will be larger than the percentage of cells with a decrease. Also, in this site and in Madrid higher projected values occurred where it
Conclusions
Future wildfire projections will result from the complex interactions among diverse factors related to human activities (LULC interfaces), climate and fuel moisture content (LFMC). Expected changes will produce an increase in wildfire occurrence in three out of four analyzed Spanish sites, indicating the existing variation in fire-climate and land-use effects by site. LULC change-projected scenarios properly simulated the conversion to natural vegetation and urban development, resulting in LULC
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This paper was funded by the LUC4FIRE project (CSO2015-73407-JIN), supported by the Spanish Ministry of Economy and Finance (MINECO) and the Environmental Remote Sensing and Spectroscopy Laboratory (Speclab) at the Spanish National Research Council (CSIC). E.T-G was supported by the 2017 Youth Guarantee initiative from Madrid region (contract number CAMPD17_IEGD_001). We want to thank three anonymous rewievers for useful comments to improve the manuscript. We also acknowledge for the provision
References (96)
- et al.
The case of urban sprawl in Spain as an active and irreversible driving force for desertification
J. Arid Environ.
(2013) - et al.
Dangers of using global bioclimatic datasets for ecological niche modeling. Limitations for future climate projections
Global Planet. Change
(2013) - et al.
What's new about old fields? Land abandonment and ecosystem assembly
Trends Ecol. Evol.
(2008) - et al.
Human-ignited wildfire patterns and responses to policy shifts
Appl. Geogr.
(2015) - et al.
Forest fires in the wildland-urban interface: a spatial analysis of forest fragmentation and human impacts
Appl. Geogr.
(2013) - et al.
Development of a framework for fire risk assessment using remote sensing and geographic information system technologies
Ecol. Model.
(2010) - et al.
Forest fire spatial pattern analysis in Galicia (NW Spain)
J. Environ. Manag.
(2013) - et al.
Sensitivity of a common land use cover change (LUCC) model to the minimum mapping unit (MMU) and minimum mapping width (MMW) of input maps
Comput. Environ. Urban Syst.
(2019) - et al.
Human activity impact on the heterogeneity of a Mediterranean landscape
Appl. Geogr.
(2010) - et al.
Generalized linear and generalized additive models in studies of species distributions: setting the scene
Ecol. Model.
(2002)
About regression-kriging: from equations to case studies
Comput. Geosci.
Regional estimation of woodland moisture content by inverting Radiative Transfer Models
Rem. Sens. Environ.
Land cover analysis in wildland–urban interfaces according to wildfire risk: a case study in the South of France
For. Ecol. Manag.
Space–time process and drivers of land abandonment in Europe
Catena
Human-caused wildfire risk rating for prevention planning in Spain
J. Environ. Manag.
Mapping regional patterns of large forest fires in Wildland–Urban Interface areas in Europe
J. Environ. Manag.
Fire regime changes and major driving forces in Spain from 1968 to 2010
Environ. Sci. Pol.
Design of a GIS and multi-criteria based land evaluation procedure for sustainable land-use planning at the regional level
Agric. Ecosyst. Environ.
European Soil Data Centre: response to European policy support and public data requirements
Land Use Pol.
Modeling the spatial variation of the explanatory factors of human-caused wildfires in Spain using geographically weighted logistic regression
Appl. Geogr.
Estimating urban areas: new insights from very high-resolution human settlement data
Remote Sensing Applications: Society and Environment
Different approaches to model future burnt area in the Iberian Peninsula
Agric. For. Meteorol.
Mapping syndromes of land change in Spain with remote sensing time series, demographic and climatic data
Land Use Pol.
The relative influence of climate and housing development on current and projected future fire patterns and structure loss across three California landscapes
Global Environ. Change
Modeling temporal changes in human-caused wildfires in Mediterranean Europe based on Land Use-Land Cover interfaces
For. Ecol. Manag.
Comparative analysis of CORINE and climate change initiative land cover maps in Europe: implications for wildfire occurrence estimation at regional and local scales
Int. J. Appl. Earth Obs. Geoinf.
Estimation of live fuel moisture content from MODIS images for fire risk assessment
Agric. For. Meteorol.
A global review of remote sensing of live fuel moisture content for fire danger assessment: moving towards operational products
Rem. Sens. Environ.
A fuel moisture content and flammability monitoring methodology for continental Australia based on optical remote sensing
Rem. Sens. Environ.
Estimating future burned areas under changing climate in the EU-Mediterranean countries
Sci. Total Environ.
Los incendios forestales en España. Decenio 2006-2015
Determining fuel moisture thresholds to assess wildfire hazard: a contribution to an operational early warning system
PloS One
Socio-geographic analysis of the causes of the 2006's wildfires in Galicia (Spain)
Global patterns in the sensitivity of burned area to fire-weather: implications for climate change
Agric. For. Meteorol.
Scenarios for future wildfire risk in California: links between changing demography, land use, climate, and wildfire
Environmetrics. Special Issue Paper
Methods Used to Estimate Moisture Content of Dead Wildland Fuels, Wildland Fire Danger Estimation and Mapping
Mathematical Methods of Statistics
Socioeconomic factors drive fire-regime variability in the Mediterranean Basin
Ecosystems
Conversion of fuel moisture content values to ignition potential for integrated fire danger assessment
Can. J. For. Res.
Critical live fuel moisture in chaparral ecosystems: a threshold for fire activity and its relationship to antecedent precipitation
Int. J. Wildland Fire
Land Cover Classification System
Predictive modelling of fire occurrences from different fire spread patterns in Mediterranean landscapes
Int. J. Wildland Fire
Climate change impact on future wildfire danger and activity in southern Europe: a review
Ann. For. Sci.
A meteorological estimation of relevant parame-ters for snow models
Ann. Glaciol.
A computer-based system simulating snowpack structures as atool for regional avalanche forecasting
J. Glaciol.
A short presentation of the land change modeler (LCM)
Plant flammability experiments offer limited insight into vegetation–fire dynamics interactions
New Phytol.
Impacts of future land use/land cover on wildfire occurrence in the Madrid region (Spain)
Reg. Environ. Change
Cited by (22)
Validating the effect of fuel moisture content by a multivalued operator in a simplified physical fire spread model
2023, Environmental Modelling and SoftwareWildfire prediction using zero-inflated negative binomial mixed models: Application to Spain
2023, Journal of Environmental ManagementCitation Excerpt :The use of the provincial scale has many advantages in forecasting models. On the one hand, aggregated information data is readily available (Ordóñez et al., 2012) and it is a sufficiently disaggregated scale to incorporate socio-economic heterogeneity (Juan, 2019; Vilar et al., 2021). On the other hand, a certain territorial homogeneity is achieved, which allows the methodology to be applied to other datasets (Díaz-Avalos et al., 2016; Pham et al., 2020; Pérez-Porras et al., 2021).
Prediction of wildfire rate of spread in grasslands using machine learning methods
2022, Environmental Modelling and SoftwareCitation Excerpt :Wildfires are a natural hazard in remote regions and of worldwide significance (Vilar et al., 2021), with their occurrence and adverse impacts anticipated to rise in the future (Ellis et al., 2022).