High performance genetic algorithm for land use planning

This study uses genetic algorithms to formulate and develop land use plans. The restrictions to be imposed and the variables to be optimized are selected based on current local and national legal rules and experts’ criteria. Other considerations can easily be incorporated in this approach. Two optimization criteria are applied: land suitability and the shape-regularity of the resulting land use patches. We consider the existing plots as the minimum units for land use allocation. As the number of affected plots can be large, the algorithm execution time is potentially high. The work thus focuses on implementing and analyzing different parallel paradigms: multi-core parallelism, cluster parallelism and the combination of both. Some tests were performed that show the suitability of genetic algorithms to land use planning problems.

► A parallel genetic algorithm for land-use planning problem is proposed. ► The land suitability and the shape-regularity of the land-use patches are optimized. ► Three parallel versions were developed for different types of memory architectures. ► Tests showed a good performance in multi-core and cluster machines.