Quantification and optimization of spatial contiguity in land use planning

Contiguous land is desired in various planning domains, such as forest management and reserve design. Landscape contiguity is a necessity for species everyday dispersal. Much research shows that landscape fragmentation is a serious threat to species survivability. Promoting contiguity/controlling fragmentation thus has been a common planning goal. Though the objective of encouraging land contiguity is apparent, there exists no accurate measure of spatial contiguity. As a result, current land acquisition models employing proxy measures generate biased land use plans. This research focuses on controlling spatial contiguity optimally in order to reduce fragmentation using spatial analysis methods and mathematical operational research. This research systematically reviews and clarifies the measurement issues of the existing methods of evaluating spatial contiguity. Based upon graph theory and spatial interaction, a direct and unbiased measure of contiguity is constructed. Empirical analysis shows that the new measure is reliable and consistent and does not have the measurement issues of the existing approaches. Incorporating this measure, a new land acquisition model is proposed in order to explicitly optimize spatial contiguity. As this model is nonlinear, efficient local search heuristics are explored and developed. Application results prove that the new model has superiority of optimizing contiguity over existing alternatives. In order to solve large planning problems and generate a large number of diverse alternative solutions, a new evolutionary algorithm is proposed, providing flexibility in handling geographical constraints and efficiently converges to good solutions. Application results demonstrate that the developed evolutionary algorithm (EA) finds numerous non-dominated alternative solutions that are evenly distributed on the tradeoff curve. Compared to classical multiobjective optimization approaches, the EA does not need auxiliary information in order to approximate the Pareto front. This research proposes theoretical framework and spatial optimization techniques for optimizing spatial contiguity, which could be beneficial for studies of addressing other spatial objectives in land use planning.