An improved Genetic Algorithm for spatial optimization of multi-objective and multi-site land use allocation

• We improved the common Genetic Algorithm for optimal allocation of multi-site land use.
• Additive objectives, spatial objectives, and planning regulatory knowledge are all considered.
• A Goal Programming model with a reference point form was used to manage multi-criteria decisions.
• Two crossover steps and two mutation operations were proposed.
• We demonstrated the use of the Genetic Algorithm for multi-site land use allocation in the Okanagan Valley, British Columbia.

As a result of multiple land use types, spatial heterogeneity, and conflicts of interest among multiple participants, multi-site land use allocation becomes a complex and significant optimization issue. We propose an improved Genetic Algorithm (GA) to deal with multi-site land use allocation, in which maximum economic benefit, maximum ecological benefit, maximum suitability, and maximum compactness were formulated as optimal objectives; and residential space demand and some regulatory knowledge were set as constraints. A Goal Programming model with a reference point form was used to manage trade-offs among multiple objectives. In order to improve the efficiency of the common GA applied to multi-site land use allocation, two crossover steps and two mutation operations were designed. This paper presents an application of the improved GA to the Regional District of Central Okanagan in Canada. Results showed that the proposed GA exhibited good robustness and could generate any optimal land use scenario according to stakeholders' preferred objectives, thus having the potential to provide interactive technical support for land use planning.