Determining Optimal Policies for Ecosystems
STANFORD UNIV CA SYSTEMS OPTIMIZATION LAB
Pagination or Media Count:
The paper is a review of research done primarily at IIASA. The problem of finding the optimal policy for controlling the spruce budworm -- an insect whose outbreaks from time to time do great damage to the fir forests of New Brunswick, Canada -- represents a rare opportunity to develop and to successfully apply the methodology of optimization. The two interacting populations, the tree and the insect, constitute about the simplest ecosystem of practical importance. A very detailed computer simulation model is used to evaluate and to compare proposed policies regarding when to apply insecticides and when to cut down trees. The model is considered by biologists to be sufficiently representative that its simulation on the computer can be viewed as one way to bring the real world into the laboratory. The effectiveness of different policies can then be determined by trying them out on the simulation model. In this paper the authors discuss how the simulator can be supplemented with optimization methods to determine an optimal policy, in particular how a Markov model is appropriate.