Reducing the risk of house loss due to wildfires

https://doi.org/10.1016/j.envsoft.2014.12.020Get rights and content

Highlights

  • Bayesian Network models provide a powerful tool for analysing fire risk scenarios.

  • Combining multiple data sources allow for the exploration of a range of scenarios that cannot be empirically tested.

  • Fire risk reduction strategies can reduce risk, but other factors limit their implementation.

  • Approaches used here are broadly applicable to other natural hazards.

Abstract

Wildfires will continue to reach people and property regardless of management effort in the landscape. House-based strategies are therefore required to complement the landscape strategies in order to reduce the extent of house loss. Here we use a Bayesian Network approach to quantify the relative influence of preventative and suppressive management strategies on the probability of house loss in Australia. Community education had a limited effect on the extent to which residents prepared their property hence a limited effect on the reduction in risk of house loss, however hypothetically improving property preparedness did reduce the risk of house loss. Increasing expenditure on suppression resources resulted in a greater reduction in the risk of loss than preparedness. This increase had an interaction effect with increasing the distance between vegetation and the houses. The extent to which any one action can be implemented is limited by social, environmental and economic factors.

Introduction

Wildfires can cause considerable damage to people and property with the effects on communities and individuals lasting for many years after the event. The Black Saturday fires in Victoria, Australia, resulted in the damage or destruction of over 2000 houses and the loss of 173 lives (Gibbons et al., 2012, Leonard et al., 2009b, Price and Bradstock, 2012). Similarly, the 2007 wildfires in California resulted in the evacuation of 300 000 people and the loss of 2223 houses (McCaffrey and Rhodes, 2009). In the following years, wildfires have considerable economic impacts on communities, local business and production (e.g. agriculture and forestry) (Ganewatta, 2008). Societal impacts continue for decades as many residents suffer post-traumatic stress as a result of the wildfire (Langley and Jones, 2005, McFarlane et al., 1997, Papadatou et al., 2012). Minimising the damage of wildfires to people and property will therefore have a range of economic and social benefits.

Fire management agencies have large budgets devoted to landscape fire management in an attempt to reduce the risk of fires reaching property (Berry et al., 2006, Calkin et al., 2005). These are primarily fuel treatment (e.g. thinning, clearing, prescribed burning) and fire suppression (i.e. the coordinated use of fire-fighting resources such as trucks, helicopters and aircraft, in an attempt to contain or extinguish the fire). Optimised placement of fuel treatments and resources can reduce the risk to the interface, i.e. those houses which form the boundary between native vegetation and urban areas (Bradstock et al., 2012, Finney et al., 2007, Penman et al., 2014, Plucinski, 2012, Wilson and Wiitala, 2005). However, these actions are not expected to contain all wildfires, particularly under more severe fire weather conditions (Cary et al., 2009, LaCroix et al., 2006, Penman et al., 2011a, Price and Bradstock, 2010). Given that wildfires under severe fire weather conditions are generally responsible for the majority of area burned and greatest loss of houses (Blanchi et al., 2010, Bradstock et al., 2009, Mees and Strauss, 1992, Podur and Martell, 2007), wildfires will continue to reach houses regardless of the extent of management intervention in the landscape (Bradstock et al., 2012, Cary et al., 2009, Penman et al., 2014, Syphard et al., 2011). The frequency with which fire impacts upon the interface is predicted to increase due to the expansion of populations into native vegetation and the severity of fire weather increases (Clarke et al., 2013, Penman et al., 2013a, Syphard et al., 2007). Therefore house-based strategies are required to complement the landscape strategies in order to minimise house loss.

Management strategies that may reduce the risk of individual property loss can be considered to be preventative or defensive, because they are predicated on the assumption that fires will reach the vicinity of houses. Considered decisions about placement of property relative to flammable vegetation and building construction (Blanchi and Leonard, 2008, Cohen, 2000, Radeloff et al., 2005, Ramsay et al., 1987) will affect the level of exposure to fire, hence the probability of loss. In the short term, the primary preventative option is educating land owners to prepare their property for wildfire by reducing or removing fuels within their property (Blanchi and Leonard, 2008, Gibbons et al., 2012, Gill, 2005, McGee, 2011) to reduce both the risk of ignition within the property and the severity of subsequent fire(s). Other defensive actions include fire suppression in and around houses, although the level of suppression can vary from work carried out by individual residents through to volunteer or professional fire agency resources, e.g. fire trucks, helicopters etc.

All these strategies are considered to reduce the risk of house loss however there has been no quantification of the individual or interactive effects. Fire management agencies require this information to determine how to invest limited budgets in order to reduce the risk of house loss. There are limited data available to address the issue, primarily because houses are lost during emergency situations where the focus is on protecting life and property, rather than data collection. Generating such a data set after an event relies on methods such as detailed structured interviews of a large number of individuals in an attempt to reconstruct the range of actions and responses. Furthermore, generation of suitable data that covers sufficient events for a quantitative analysis has generally been considered too difficult and expensive (Gill et al., 2013). An alternative to reconstruction is to use a formal elicitation process to generate meaningful values for quantitative analysis (Burgman et al., 2011, Martin et al., 2012, Wintle et al., 2013).

Here we use a process based model combined with a formal elicitation process to quantify the relative influence of preventative and suppressive management strategies on the probability of house loss. Specifically, we ask the questions:

  • 1.

    What is the optimal strategy or strategies at the interface to reduce the risk of house loss in the event of a fire?

  • 2.

    Does this capacity differ between urban interface and intermix communities (low density housing within extensive native vegetation)? (Radeloff et al., 2005)

Section snippets

Study area

The study was conducted in the Sydney Basin Bioregion (Environment Australia, 2000), the most populated area of Australia. Within the Sydney Basin Bioregion are three large urban centres (Sydney, Newcastle and Wollongong) which support a combined population of 5.5 million people (www.abs.gov.au, Accessed March 2011). Fire prone native vegetation, predominantly dry Eucalypt forest (Keith, 2004), surrounds all three urban centres creating a large and complex urban interface (Fig. 1). Between 2000

Elicitation

Opinions of the participants suggested that the education campaigns considered were likely to be ineffective in influencing preparedness by residents (Fig. 4). The greatest change in the distribution was predicted to occur if an education campaign included a ‘street walk’, with much smaller changes predicted when either the letterbox drop or television advertising were included in the mix. Regardless of the advertising campaign very few houses were considered to be in the “Good” category of

Discussion

There are a range of strategies in and around properties that will reduce the risk of house loss in the event of a fire. Increasing the distance between vegetation and structures had the strongest influence as they reduce the exposure and increase the ability of fire suppression to address fires when they occur. Residents have the potential to reduce their own risk if they prepare property for wildfire, yet they often fail to do so. While there were no differences in the relative role of

Community education

Community education campaigns were predicted to be relatively unsuccessful in altering the extent to which property preparedness and thereby reducing the risk of house loss. Participants in the elicitation exercise believed that few residents were likely to respond to any of the proposed education campaigns which is consistent with empirical studies elsewhere (McLennan et al., 2012). The idealised education campaign which hypothetically improved the overall community standard of house condition

Model limitations

The model presented here attempted to quantify the risk of loss that can be attributed to management actions of various spatial and temporal time scales. There is a lack of empirical data that can be used for such a model and this resulted in the need to use expert elicitation for some of the key nodes in the model. However, this is not necessarily a limitation as Bayesian Networks built purely on empirical data tend overfit the data to the situation in which data were collected and as result

Conclusion

In the study, we have brought together spatial data, a process model and expert opinion in order to undertake a risk assessment of various fire management strategies. BNs provide the ideal framework for such a task and have been increasingly used in the field of risk assessment (Borsuk et al., 2004, Burgman et al., 2010, Chen and Pollino, 2012, Dlamini, 2010, Ejsing et al., 2008, Johnson et al., 2010, Jolma et al., 2014, Lucas, 2004, Oatley and Ewart, 2003, Punt and Hilborn, 1997). The approach

Acknowledgements

The work was funded by the NSW Rural Fire Service. The probability elicitation exercise was approved by the University of Wollongong of Human Research Ethics Committee approval number HE12/149. All participants in the elicitation provided written consent prior to the exercise. Anonymity of participants has been guaranteed.

References (111)

  • S. Johnson et al.

    Modelling cheetah relocation success in southern Africa using an Iterative Bayesian Network Development Cycle

    Ecol. Model.

    (2010)
  • A. Jolma et al.

    A software system for assessing the spatially distributed ecological risk posed by oil shipping

    Environ. Model. Softw.

    (2014)
  • T. Konishi et al.

    Aerial firefighting against urban fire: mock-up house experiments of fire suppression by helicopters

    Fire Saf. J.

    (2008)
  • T. Krueger et al.

    The role of expert opinion in environmental modelling

    Environ. Model. Softw.

    (2012)
  • B.G. Marcot et al.

    Using Bayesian belief networks to evaluate fish and wildlife population viability under land management alternatives from an environmental impact statement

    For. Ecol. Manag.

    (2001)
  • T.K. McGee

    Public engagement in neighbourhood level wildfire mitigation and preparedness: case studies from Canada, the US and Australia

    J. Environ. Manag.

    (2011)
  • G.C. Oatley et al.

    Crimes analysis software: ‘pins in maps’, clustering and Bayes net prediction

    Expert Syst. Appl.

    (2003)
  • J. Pearl

    Fusion, propagation, and structuring in belief networks

    Artif. Intell.

    (1986)
  • T.D. Penman et al.

    Reducing wildfire risk to urban developments: simulation of cost-effective fuel treatment solutions in south eastern Australia

    Environ. Model. Softw.

    (2014)
  • T.D. Penman et al.

    Defining adequate means of residents to prepare property for protection from wildfire

    Int. J. Disaster Risk Reduct.

    (2013)
  • C.A. Pollino et al.

    Parameterisation and evaluation of a Bayesian network for use in an ecological risk assessment

    Environ. Model. Softw.

    (2007)
  • O.F. Price et al.

    The efficacy of fuel treatment in mitigating property loss during wildfires: insights from analysis of the severity of the catastrophic fires in 2009 in Victoria, Australia

    J. Environ. Manag.

    (2012)
  • H.M. Regan et al.

    The effects of fire and predators on the long-term persistence of an endangered shrub, Grevillea caleyi

    Biol. Conserv.

    (2003)
  • M.C. Runge et al.

    Which uncertainty? Using expert elicitation and expected value of information to design an adaptive program

    Biol. Conserv.

    (2011)
  • A. Ahern et al.

    How Far Do Bushfires Penetrate Urban Areas?

    (1999)
  • A.H. Berry et al.

    Prescribed burning costs and the WUI: economic effects in the Pacific Northwest

    West J Appl. For.

    (2006)
  • R. Blanchi et al.

    Property safety: judging structural safety

  • R. Blanchi et al.

    Investigation of Bushfire Attack Mechanisms Resulting in House Loss in the ACT Bushfire 2003

    (2005)
  • R. Blanchi et al.

    Meteorological conditions and wildfire-related houseloss in Australia

    Int. J. Wildland Fire

    (2010)
  • R.A. Bradstock et al.

    Prediction of the probability of large fires in the Sydney region of south-eastern Australia using fire weather

    Int. J. Wildland Fire

    (2009)
  • G.M. Budd et al.

    Project Aquarius 4. Experimental bushfires, suppression procedures, and measurements

    Int. J. Wildland Fire

    (1997)
  • M.A. Burgman et al.

    Expert status and performance

    PLoS ONE

    (2011)
  • M.A. Burgman et al.

    Reconciling uncertain costs and benefits in Bayes Nets for invasive species management

    Risk Anal.

    (2010)
  • D.E. Calkin et al.

    Forest service large fire area burned and suppression expenditure trends, 1970-2002

    J. For.

    (2005)
  • G.J. Cary et al.

    Relative importance of fuel management, ignition management and weather for area burned: evidence from five landscape-fire-succession models

    Int. J. Wildland Fire

    (2009)
  • K.P. Chen et al.

    Quantifying bushfire penetration into urban areas in Australia

    Geophys. Res. Lett.

    (2004)
  • H. Clarke et al.

    Changes in Australian fire weather between 1973 and 2010

    Int. J. Climatol.

    (2013)
  • M.J. Clayton

    Delphi: a technique to harness expert opinion for critical decision-making tasks in education

    Educ. Psychol.

    (1997)
  • J.D. Cohen

    Preventing disaster: home ignitability in the wildland-urban interface

    J. For.

    (2000)
  • J.D. Cohen

    Relating flame radiation to home ignition using modeling and experimental crown fires

    Can. J. For. Res.

    (2004)
  • J.D. Cohen et al.

    Modelling potential structure ignitions from flame radiation exposure with implications for wildland/urban interface fire management

  • J.D. Cohen et al.

    Home destruction Examination, Grass Valley Fire, Lake Arrowhead, California

    (2008)
  • T.W. Collins

    Influences on wildfire hazard exposure in Arizona's high country

    Soc. Nat. Resour.

    (2009)
  • A. Cottrell et al.

    Community perceptions of bushfire risk

  • S.G. Cumming

    Effective fire suppression in boreal forests

    Can. J. For. Res.

    (2005)
  • T. Daniel et al.

    Assessing public tradeoffs between fire hazard and scenic beauty in the wildland–urban interface

  • E. Ejsing et al.

    Predicting probability of default for large corporates

  • Environment Australia

    Revision of the Interim Biogeographic Regionalisation of Australia (IBRA) and the Development of Version 5.1. - Summary Report

    (2000)
  • R.S. Etienne et al.

    Ecological impact assessment in data-poor systems: a case study on metapopulation persistence

    Environ. Manag.

    (2003)
  • M. Finney et al.

    Modeling containment of large wildfires using generalized linear mixed-model analysis

    For. Sci.

    (2009)
  • Cited by (39)

    • Framework for spatial incident-level wildfire risk modelling to residential structures at the wildland urban interface

      2022, Fire Safety Journal
      Citation Excerpt :

      Modelling of fire spread between non-burnable pixels based on spatial patterns is something that should also be considered for future developments. Suppression and defensive actions: the role of fire fighting activities was identified as being significant in reducing structure losses [49], but the role of defensive actions and the extent of suppression effectiveness in structure survival has been historically difficult to quantify, mostly because of data rarity [48]. Most wildfire suppression research focuses on the containment or control of fires in the wildland landscape rather than their role in the protection of exposed structures [49].

    • Integrating geospatial wildfire models to delineate landscape management zones and inform decision-making in Mediterranean areas

      2022, Safety Science
      Citation Excerpt :

      Specifically, we first assembled modeling outcomes for fire IP (Rodrigues and de la Riva, 2014), IA escape probability (Rodrigues et al., 2019), and fire transmission to communities (Alcasena et al., 2018a) to delineate different wildfire management zones (WMZs). Then, we prioritized and summarized risk reduction strategies within WMZs based on four main archetypes (Curt and Frejaville, 2018; Penman et al., 2015; Wunder et al., 2021). Ultimately, we advise a comprehensive solution where (i) human ignition prevention, (ii) fire reintroduction, (iii) fuel treatments, (iv) fire suppression, and (v) human community adaptation efforts are combined and strategically designated within a vast fire-prone cultural landscape.

    • Predicting Paradise: Modeling future wildfire disasters in the western US

      2021, Science of the Total Environment
      Citation Excerpt :

      Given predicted climate change effects on the global fire footprint (Abatzoglou and Williams, 2016; Abatzoglou et al., 2019; Liu et al., 2010), analytical templates need to begin incorporating the assessment of wildfire exposure from extreme wildfire events as with other natural disasters. As noted in the introduction, despite the potential contributions from both predictive and empirical reconstruction studies to understanding the WUI wildfire problem, most of the recent empirical studies (Alexandre et al., 2016a; Kramer et al., 2019) have not compared their findings relative to the numerous wildfire simulation studies to understand exposure and loss (e.g., Bar Massada et al., 2009; Oliveira et al., 2020; Penman et al., 2014; Calkin et al., 2010; Haas et al., 2013) and the effects of fuel treatments (Ager et al., 2007; Penman et al., 2015). Simulation models are widely used in disaster and hazard prediction, including hurricanes and earthquakes (Irikura and Miyake, 2011; Vickery et al., 2000), and when carefully interpreted on a background of empirical data can provide predictive capabilities not afforded in empirical studies of past events.

    View all citing articles on Scopus
    View full text