The development of a Bayesian Belief Network as a decision support tool in feral camel removal operations
Piantadosi, J., Anderssen, R.S. and Boland J. (eds) MODSIM2013, 20th International Congress on Modelling and Simulation · 2013
The removal of feral camels in Australia is complicated by the vast area over which they range, their remoteness and the changing weather conditions that constantly affect their distribution. Decision Support Systems (DSS) provide a framework in which program managers can undertake a more formal assessment of pest removal actions under different conditions, using past data and expert knowledge. The objective of a DSS in pest management is to minimise costs and optimise on-ground effectiveness. In this study we develop a Bayesian Belief Network (BBN) as a component of a camel DSS. BBNs provide a transparent visualisation of the components of the problem, underpinned by probability tables consisting of likelihoods and states in an uncertain environment. They enable managers to interrogate different scenarios, often consisting of incomplete intelligence data, and help seek the best course of action. We describe a novel approach of eliciting data from past camel culling operations into a BBN using a simulation algorithm. The algorithm simulates all aspects of the operation including search patterns, sightability, the time it takes to undertake the operation, fuel costs and camel densities. We verified the output of a range of scenarios from these simulations interactively with a group of experts and then using a wide range of environmental conditions we populated the states and dependencies of the final BBN. Using some hypothetical scenarios we demonstrate the BBN outputs including probabilities associated with a different number of camels removed and the associated costs.