Bayes Nets as a method for analysing the influence of management actions in fire planning.

摘要

Wildfire can result in significant economic costs with inquiries following such events often recommending an increase in management effort to reduce the risk of future losses. Currently, there are no objective frameworks in which to assess the relative merits of management actions or the synergistic way in which the various combinations may act. We examine the value of Bayes Nets as a method for assessing the risk reduction from fire management practices using a case study from a forested landscape. Specifically, we consider the relative reduction in wildfire risk from investing in prescribed burning, initial or rapid attack and suppression. The Bayes Net was developed using existing datasets, a process model and expert opinion. We compared the results of the models with the recorded fire data for an 11-year period from 1997 to 2000 with the model successfully duplicating these data. Initial attack and suppression effort had the greatest effect on the distribution of the fire sizes for a season. Bayes Nets provide a holistic model for considering the effect of multiple fire management methods on the risk of wildfires. The methods could be further advanced by including the costs of management and conducting a formal decision analysis.Digital Object Identifier http://dx.doi.org/10.1071/WF10076