Comparisons of three different methods used to generate forest landscapes for spatial harvest scheduling problems with adjacency restrictions

Abstract

Dealing with adjacency constraints is currently one of the main research focuses of spatial harvest scheduling problems. In many situations, hypothetical landscapes are generated for research purposes. The simulation methods used to produce hypothetical landscapes are important and may affect the planning solutions. In this study, we examined three landscape simulation methods (grids, Voronoi diagrams, and random graphs) and compared them to a forest developed using irregularly-shaped polygons within a geographic information system. The effects on the optimal planning solutions due to different spatial layouts were investigated as well. The results show that landscape simulation methods have a significant impact on the mean solution values and CPU running time due to different adjacency patterns in each simulated landscape. Advantages and disadvantages using each of the three methods are discussed and useful suggestions are provided regarding the selection of one particular method for research purposes.