Wildfire Risk Assessment for Strategic Forest Management in the Southern United States: a Bayesian Network Modeling Approach

Abstract

Wildfire occurrence has increased and is projected to continue increasing globally. Strategic, evidence-based planning with diverse stakeholders, making use of diverse ecological and social data, is crucial for confronting and mitigating the associated risks. Prescribed fire, when planned and executed carefully, is a key management tool in this effort. Assessing where prescribed fire can be a particularly effective forest management tool can help prioritize efforts, reduce wildfire risk, and support fire-resilient lands and communities. We collaborated with expert stakeholders to develop a Bayesian network model that integrated a large variety of biophysical, socioecological, and socioeconomic spatial information for the Southeastern United States, to quantify where risk is high, and where prescribed fire would be efficient in mitigating risk. The model first estimated wildfire risk based on landscape-scale interactions among the likelihoods of fire occurrence and severity, and people and resources potentially exposed—accounting for socioeconomic vulnerabilities as well as key ecosystem services. The model then quantified the potential for risk reduction through prescribed fire, given the existing fuel load, climate, and other landscape conditions. Resulting expected risk estimates show high risk concentrated in the coastal plain and interior highland subregions of the southern US, but there was considerable variation among risks to different ecosystem services and populations, including potential exposure to smoke emissions. Capacity to reduce risk through fuel reductions was spatially correlated with risk; where these diverged, the difference was largely explained by fuel load. We suggest that both risk and capacity for risk reduction are important in identifying priorities for management interventions. The model serves as a decision support tool for stakeholders to coordinate large-landscape adaptive management initiatives in the southern US. The model is flexible with regard to both empirical and expert-driven parameterizations and can be updated as new knowledge and data emerge. The resulting spatial information can help connect active management options to forest management goals and make management more efficient through targeted investments in priority landscapes.

Technical info

Files include a CSV data file and a Netica model file that are used for wildfire risk assessment in the southern United States. 

1. Input Data: The variables included in the Input Data CSV file can be used to run the Bayesian network model in Netica. Each row constitutes a single 'case', corresponding to a 1,000-hectare landscape. There are 132,421 cases. For complete data sources information, please refer to our publication and to Gould et al. (2023). 

Variable Name Description

IMP_Water (Watershed Importance):  Importance of HUC-12 watersheds for surface drinking water (index value)

Carbon (Forest Carbon): Total forest carbon stocks (tons/acre)

RHmin (Min humidity): 3-year anomaly of minimum relative humidity in the month with the lowest value in the year, 2019- 2021 (z-score)

SPEI (SPEI drought): 3-year SPEI drought index, leading up to and including December 2021 (z-score)

Temp_max (Max temp): 3-year anomaly of maximum temperature in the month with the highest value in the year, 2019-2021 (z-score)

Fuel_load (Fuel Load): Total available forest fuels in 1000-ha landscape (tons)

Smoke (Smoke/SV): Potential for unsafe smoke emissions to reach human populations if wildfire occurs, given existing fuel load (log of sum population exposed, social vulnerability-weighted)

Smoke_Rx (Smoke/SV): Potential for unsafe smoke emissions to reach human populations if wildfire occurs, given a fuel load reduced through prescribed fire (log of sum population exposed, social vulnerability-weighted)

Burn_Int (Burn Intensity Init): Burn intensity factor* with high loadings for flame length exceedance, forest types, forest tree size classes (index value)

Pop_Str (Structures & Population/SV): Structure/population factor* with high loadings for housing unit density, population density, wildland urban interface (index value)

Burn_Prob (Wildfire Potential Init): Wildfire potential factor* with high loadings for burn probability, risk to potential structures, and wildfire hazard potential (index value)

Soc_Vuln (Social Vulnerability): Social vulnerability factor* with high loadings for socioeconomic status, housing type & transportation, and household composition & disability (index value). 

*For the factor variables, see Gould et al. (2023) for details of the factor analysis that produced these factors.

2. Bayesian network model file: The Bayesian network model is setup in the Netica software platform. In this model, risk assessment involves quantifying the vulnerability of people and ecosystem services to wildfire, and their potential exposure to hazardous fire. Risk is an outcome where these estimates of vulnerability and exposure co-occur. The model provides quantitative, probabilistic estimates of risk and the potential for risk reduction through management intervention, both conditional on the various landscape characteristics represented by the input variables described above. Meter symbols in the model (viewable with free Netica software) indicate the data input variables. Additional nodes represent system variables described in the associated publication. The nodes are connected by arcs/links to show functional relationships between the variables. Conditional Probability Tables (CPTs) defining these dependencies can be viewed at each node. The decision node (blue box) includes two management scenarios: (1) enhanced investment in fuel reduction programs (through prescribed fire), in which existing fuels are reduced by approximately 50%; and (2) 'business as usual', i.e., no enhancement of fuel reduction programs, in which existing fuels are unchanged. The utility node (pink box) represents the value placed on the outcome, given a decision made. To compare the relative utilities of fuel reductions for reducing risk, a data column indicating the decision taken should be included with input data. Running the model twice produces utility values under the two possible decisions. For appropriate smoke exposure potentials under the two decisions, the 'Smoke' variable is used with the 'business as usual' decision, and the 'Smoke_Rx' variable is used with the 'enhanced investment in fuel reduction' decision.