Fire regime dynamics in mainland Spain. Part 2: A near-future prospective of fire activity

https://doi.org/10.1016/j.scitotenv.2019.135842Get rights and content

Highlights

  • We conducted a spatial-temporal zoning of fire regime from historical data.

  • The zoning scheme was transferred twenty years into the future using ARIMA.

  • Under current conditions ARIMA models envisage an overall decline in fire incidence.

  • The Mediterranean region experienced the greatest loss of fire activity.

  • Winter fire activity would progress and persist in the northern coast.

Abstract

The current research belongs to a series of two manuscripts aiming at describing spatial-temporal dynamics of fire regime and its drivers in Spain. In this work, we present the first attempt to produce a spatial-temporal delimitation of homogeneous fire regime zones in Spain providing insights into the near future. The analyses were based on historical fire records; leveraging autoregressive models to project fire features into the near future. We evaluated the spatial extent of homogenous fire regime zones in three different periods: past (1974–1994), current (1995–2015) and future (2016–2036). To do so, we applied Principal Component Analysis and Ward's hierarchical clustering to identify zones of fire regime on the basis of the spatial and temporal arrangement of their main fire features: number of fires, burned area, burnt area from natural-caused fires, incidence of large fires (> 100 ha) and seasonality. Clusters of fire regime were trained in the current period, being later projected into the past and future periods using of k-Nearest Neighbor classification.

ARIMA modeling forecasted a shrinkage in all fire features except natural-caused fires that remained stable. Overall, we detected a transition from significant fire incidence in the past towards a situation with moderate impact of fires in the near future. The Mediterranean coast experienced the largest decline in fire activity with few locations maintaining the historical levels of occurrence of large fires. On the other hand, the Northwestern end of Spain depicted a progression towards winter fire activity while still linked to large fires. This pattern persisted in the near future along the northern coast, whereas an intermix of minor fire progression and regression was expected thorough the hinterlands and the Mediterranean.

Introduction

Forest fire mitigation have gained attention over the years, being currently under the spotlight due to the uncertain effects of climate and socioeconomic changes. Expenditures in fire suppression and prevention are increasing globally, especially in fire-prone developed countries where a total fire exclusion policy is often implemented (Stephens et al., 2014). For instance, the US Federal Land Management bureau spent >2 billion $ in fire management during 2015 (Doerr and Santin, 2016). The annual budget in firefighting in the European Union raises to approximately 2.2 billion € (Faivre et al., 2018). In the case of Spain, one of the most fire affected countries within the European Mediterranean region currently ranking second in fire incidence only after Portugal (San-Miguel-Ayanz et al., 2017), fire suppression and prevention have been increasingly funded up to circa 78 million € in 2015 (MAPAMA, 2017). Therefore, it seems clear that firefighting agencies envisage a worsening fire danger scenario in the future, with more hazardous weather conditions increasingly threating human and environmental assets (Alcasena et al., 2019; Badia et al., 2011). For instance, fire incidence seems to be increasing in the Scandinavian countries. In 2018 this region experienced the warmest fire season within the recording period, which undoubtedly contributed to boost fire spread and overcome the extinction capacity (Martin Ruiz de Gordejuela and Puglisi, 2018). However, outside these exceptional cases, the current situation in those regions and countries historically affected by recurrent fires tells otherwise. In the Mediterranean Europe the observed number of fires and burned area is decreasing (Turco et al., 2016). At the same time, a remarkable decline in global fire-related emissions since 1930s is reported reaching the minimum in 2013 (Arora and Melton, 2018; Van Der Werf et al., 2017). One of the main reasons behind this trend relates to the fire exclusion policy, i.e., suppressing all wildfires in a region (Smith, 2000). Such policy considers that wildfires have negative impacts and consequently they must be suppressed by all means. Notwithstanding some authors believe the persistence of such policy will lead to increased large fire activity in the long-run due to substantial fuel accumulation (the so-called ‘fire paradox’, Otero and Nielsen, 2017; Regos et al., 2014; Westerling, 2016) in conjunction with drier and warmer conditions (Chaparro et al., 2016; Ruffault et al., 2017; Turco et al., 2017). At the same time, questions about its sustainability are starting to raise (Curt and Frejaville, 2018).

Projections of fire incidence into the future have been extensively addressed in the literature. They were usually conducted according to climate change scenarios mostly based on General Circulation Models (GCM) coupled to IPCC's emissions scenarios or Regional Climate Models (RCM). Conversely to the observed trend (overall decrease in fire incidence) most works leveraging climate models envisage increased fire activity through the XXI century. Without being exhaustive, an increment in burnt area was reported in Portugal (DaCamara et al., 2014), Canada (Hope et al., 2016), California (Westerling et al., 2011) or the Iberian Peninsula (Sousa et al., 2015); gross fire activity was expected to augment in Canada (Boulanger et al., 2013; Wang et al., 2015), Northeast China (Liu et al., 2012), Finland (Kilpeläinen et al., 2010). Similarly, some works foresee a global (Liu et al., 2010) or regional (Jolly et al., 2015; Moriondo et al., 2006; Wotton et al., 2017) raise in fire weather danger. Some studies point out diverse tendencies depending on the global regions (Krawchuk et al., 2009; Pechony and Shindell, 2010), or even opposite trends with increasing frequency and a stability or slight decrease in burnt area in the Northeast of Spain (Turco et al., 2014). Although the goal of this work is not criticize climate-based approaches, they mainly address long-term trends while often disregard suppression efforts in their prediction. Moreover, several authors have pointed out the bias found between GCM's simulations and observations (Maraun, 2012). Differences in precipitation and surface temperature between the present and future climates indicate that present-climate biases are systematically propagated into future-climate projections at regional scales (Liang et al., 2008). In the particular case of Spain, the correlation between fire weather danger and fire incidence has been found to be rather weak, with weather controlling the seasonal patterns but exerting limited influence in the observed trends (Jiménez-Ruano et al., 2019). In this sense, suppression-related features such as the time elapsed until fire brigades reach the fire site or the scattering of suppression media during simultaneous fire events control the success of the initial attack whereas fire weather relates to sporadic large fire events (Connor et al., 2017; Duane and Brotons, 2018; Rodrigues et al., 2019). Accordingly, we propose decoupling the temporal behavior of fire activity from other covariates (either climate or human related) to explore the near future evolution of wildfire features under the premise that their temporal behavior already integrates the influence of their underlying drivers. By doing so, we assume that weather conditions and human influence in wildfire activity would remain ‘stable’, i.e., they follow the same evolving trajectory and exert the same influence observed from past to current conditions as described in Rodrigues et al. (2020).

Among the few modeling techniques that allow to forecast time series of data, the most well-known and widespread are the Auto-Regressive Integrated Moving Average (ARIMA) models. ARIMA only requires a univariate time series to forecast its future evolution, although versions that are more sophisticated allow incorporating additional covariates. ARIMA models are best known for its performance in economics and marketing (Loi and Ng, 2018; Matyjaszek et al., 2019), but also, environmental studies, such as vegetation dynamics (Jiang et al., 2010) or climate change (Afrifa-Yamoah, 2015). There are also experiences of ARIMA modeling in wildfire science. In North-America, Preisler and Westerling (2007) employed ARIMA to forecast temperature 1-month ahead to evaluate fire danger whereas Miller and Safford (2012) explored trends in large high severity fires. In Spain, Boubeta et al. (2016) applied ARMA (ARIMA without the integrated component) to predict weekly burnt area in Galicia. However, to the best of our knowledge there was no experience assessing the mid-to-long term evolution of fire regime features using ARIMA or any other autoregressive technique, at least in Spain.

In this work, we developed and exemplified a framework to identify and outline fire regime regions over time. The proposed approach included for the first time a near-future prospective based on the ongoing evolution of fire regime features. In the case of Spain, fire regime zoning experiences are scarce, finding some examples in Jiménez-Ruano and et al. (2018), Montiel Molina and Galiana-Martín (2016) and Moreno and Chuvieco (2013). Nevertheless, these works provide a stationary picture of fire regimes without taking into account their temporal evolution. But fire incidence is non-stationary (Jiménez-Ruano et al., 2017; Silva et al., 2019), a feature that encourages embracing a temporal perspective in fire regime assessments. Our core methodology allowed to identify homogenous fire regime zones in three different time periods. Past (1974–1994) and current (1995–2015) situations were addressed using historical fire records from the Spanish fire database. A third period was set in the near-future (2016–2036) forecasted by means of ARIMA extending from the current period. We hypothesized that the immediate evolution of fire regime would follow the ongoing pathway, thus assuming that climatic and human drivers of fire activity (both related to ignition and suppression) remain stable towards the near future. Our main goals were to a) outline current and past fire regime zones, b) forecast their immediate future evolution, c) describe the most representative spatial and temporal trajectories of fire regimes and (e) evaluate the disruptive effects of the current fire suppression policy.

Section snippets

Study area

The study area covers the mainland Spain. See the companion work by Rodrigues et al. (2020) for further description of the region.

Data and methods

The proposed methodology was sequenced in three stages. First, we retrieved historical fire records in the period 1974–2015 and organized them in two separate datasets (1974–1994 and 1995–2015). Then, we forecasted the evolution of fire incidence in the near future (2016–2036) using ARIMA models. Finally, we identified fire regime typologies in the current period

Evolution of the fire regime features

Taking the historical period as baseline, future fire features showed a general decrease in their total values (Fig. 2). According to, Pearson's R2 coefficient between forecasted and observed fire features (Fig. 3), ARIMA predictions were reliable, capturing at least the 46% of variance (fire frequency) up to 76% in burned area. The coefficient of determination was consistently higher in the 30 × 30, compared to the original 10 × 10 km grid.

The overall decreasing trend in fire activity was also

Discussion

In this work, we conducted the multitemporal outline of homogeneous fire regime typologies and zones in mainland Spain to a) outline fire regime zones for past and current periods, b) predict their immediate future evolution, and c) analyze the main transitions in terms of spatial and temporal patterns. We aimed at improving the identification and definition of fire regimes in Spain, as well as providing insights into the effects the current of fire suppression policy. A better knowledge of the

Conclusions

The current research belongs to a series of two manuscripts aiming at describing spatial-temporal dynamics of fire regime and its drivers in Spain. In this work, we proposed the first attempt to outline fire regime zones that incorporates a temporal perspective towards the near future. We investigated three different temporal spans. Two historical periods, i.e., past (1974–1994) and current (1995–2015), which were built from historical fires (>1 ha), and a third located in the future

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

Adrián Jiménez-Ruano and Marcos Rodrigues contributed equally to the development of this work. This work has been financed by the Spanish Ministry of Education, Culture and Sports (FPU grant 13/06618) and partially by the Regional Government of Aragon (Geoforest research group S51_17R) co-financed with FEDER 2014-2020 “Construyendo Europa desde Aragon”. Marcos Rodrigues is a postdoctoral research fellow in the Juan de la Cierva program supported by the Spanish Ministry of Economy and Finance (

References (76)

  • M.V. Moreno et al.

    Fire regime changes and major driving forces in Spain from 1968 to 2010

    Environ. Sci. Pol.

    (2014)
  • I. Otero et al.

    Coexisting with wildfire? Achievements and challenges for a radical social-ecological transformation in Catalonia (Spain)

    Geoforum

    (2017)
  • J. Piñol et al.

    Modelling the Effect of Fire-Exclusion and Prescribed Fire on Wildfire Size in Mediterranean Ecosystems

    (2005)
  • M. Rodrigues et al.

    A comprehensive spatial-temporal analysis of driving factors of human-caused wildfires in Spain using geographically weighted logistic regression

    J. Environ. Manag.

    (2018)
  • M. Rodrigues et al.

    Modeling initial attack success of wildfire suppression in Catalonia, Spain

    Sci. Total Environ.

    (2019)
  • J. San-Miguel-Ayanz et al.

    Analysis of large fires in European Mediterranean landscapes: lessons learned and perspectives

    For. Ecol. Manag.

    (2013)
  • P.M. Sousa et al.

    Different approaches to model future burnt area in the Iberian Peninsula

    Agric. For. Meteorol.

    (2015)
  • E. Afrifa-Yamoah

    Application of ARIMA models in forecasting monthly average surface temperature of Brong Ahafo region of Ghana

    Int. J. Stat. Appl.

    (2015)
  • H. Akaike

    A new look at the statistical model identification

    IEEE Trans. Automat. Contr.

    (1974)
  • D. Altman

    Practical Statistics for Medical Research

    (1991)
  • V.K. Arora et al.

    Reduction in global area burned and wildfire emissions since 1930s enhances carbon uptake by land

    Nat. Commun.

    (2018)
  • M. Boubeta et al.

    Burned area prediction with semiparametric models

    Int. J. Wildl. Fire

    (2016)
  • Y. Boulanger et al.

    Fire regime zonation under current and future climate over eastern Canada

    Ecol. Appl.

    (2013)
  • D. Chaparro et al.

    Surface moisture and temperature trends anticipate drought conditions linked to wildfire activity in the Iberian Peninsula

    Eur. J. Remote Sens.

    (2016)
  • C.D.O. Connor et al.

    An empirical machine learning method for predicting potential fire control locations for pre-fire planning and operational fire management

    Int. J. Wildl. Fire

    (2017)
  • P. Costa et al.

    Prevention of Large Fires Using the Fire Type Concept. Barcelona

    (2011)
  • T. Curt et al.

    Wildfire policy in Mediterranean France: how far is it efficient and sustainable?

    Risk Anal.

    (2018)
  • C.C. DaCamara et al.

    Impacts of climate change on the fire regime in Portugal

  • S.H. Doerr et al.

    Global trends in wildfire and its impacts: perceptions versus realities in a changing world

    Philos. Trans. R. Soc. London

    (2016)
  • N. Faivre et al.

    Forest Fires - Sparking Firesmart Policies in the EU

    (2018)
  • T. Fréjaville et al.

    Seasonal changes in the human alteration of fire regimes beyond the climate forcing

    Environ. Res. Lett.

    (2017)
  • J.R. González-Olabarria et al.

    Different factors for different causes: analysis of the spatial aggregations of fire ignitions in Catalonia (Spain)

    Risk Anal.

    (2015)
  • E.S. Hope et al.

    Wildfire suppression costs for Canada under a changing climate

    PLoS One

    (2016)
  • R. Hyndman et al.

    Automatic time series forecasting: the forecast package for R

    J. Stat. Softw.

    (2008)
  • A. Jiménez-Ruano

    Identifying pyroregions by means of self organizing maps and hierarchical clustering algorithms in mainland Spain

  • A. Jiménez-Ruano et al.

    Exploring spatial – temporal dynamics of fire regime features in mainland Spain

    Nat. Hazards Earth Syst. Sci.

    (2017)
  • W.M. Jolly et al.

    Climate-induced variations in global wildfire danger from 1979 to 2013

    Nat. Commun.

    (2015)
  • A. Kilpeläinen et al.

    Climate change impacts on forest fire potential in boreal conditions in Finland

    Clim. Chang.

    (2010)
  • Cited by (12)

    • Wildfire mitigation and adaptation: Two locally independent actions supported by different policy domains

      2023, Land Use Policy
      Citation Excerpt :

      An associative capacity in place (e.g., FOAs), promoted by public policies during the 1990 s (Canadas et al., 2014), and the existence of forestland managed by pulp and paper companies also translates into the ability to compete for public funds allocated to wildfire mitigation. The adaptation domain is more recent and less consolidated, contrary to Anglo-Saxon countries (Everett and Fuller, 2011; Haynes et al., 2020; McLennan and Birch, 2005). Until 2017, human casualties were mainly firefighters, and the success of wildfire policies was mostly assessed by the reduction in burnt areas (Moreira et al., 2020; Stephens et al., 2013).

    • Economic drivers of global fire activity: A critical review using the DPSIR framework

      2021, Forest Policy and Economics
      Citation Excerpt :

      Changes in fire regimes in North America have been well-documented through a number of empirical and modeling studies, in terms of increased frequencies and extents of large fires with high severity, lengthening durations of wildfires and fire-season, and their worsening trends with warmer and drier climate affecting regional water-balance (e.g. Abatzoglou and Williams, 2016; Dennison et al., 2014; Littell et al., 2016; Mueller et al., 2020; Singleton et al., 2019; Westerling et al., 2006; Whitehair et al., 2018). The trend of changing fire regime is similar in southern Europe (Curt and Frejaville, 2017; DaCamara et al., 2014; Fréjaville and Curt, 2017; Jiménez-Ruano et al., 2020; Rodrigues et al., 2020). Wildfires became one of the most important agents of land-use/land-cover change in the tropics (Lavorel et al., 2007).

    • Fire regime dynamics in mainland Spain. Part 1: Drivers of change

      2020, Science of the Total Environment
      Citation Excerpt :

      In this work, we presented and applied a methodology to identify and spatialize fire regime typologies that enabled further insights into the underlying drivers of its spatial-temporal dynamics. To the best of our knowledge, this and the companion work by Jiménez-Ruano et al. (2020) were the first attempts to incorporate the temporal perspective into fire regime zoning in Spain, complementing the findings by Curt and Frejaville (2017) in southern France. The main novelty of our proposal lies not only in applying the zoning scheme in two historical periods but in completing them with regression models, deepening into the traits behind the spatial-temporal behavior of fire regime.

    View all citing articles on Scopus
    View full text