Land Use Optimization Using Self‐organizing Algorithms

ABSTRACT.In many spatial systems the interaction between various regions decreases dramatically with distance. This suggests that local trade‐offs may be more important than global ones in land use planning and that a decentralized, parallel optimization of the individual regions may be an attractive supplement to more centralized optimization approaches. In this paper, we solve a forest planning problem using a series of decentralized approaches. The approaches can be characterized as self‐organizing algorithms and are modeled in the framework of a cellular automaton. We compare our results with those obtained by more centralized approaches, viz. a large sample approach, simulated annealing, and a genetic algorithm. We find that the self‐organizing algorithms generally converge much faster to solutions which are at least as good as those obtained by simulated annealing and the genetic algorithm.