Genetic Programming for Computationally Efficient Land Use Allocation Optimization

Land use allocation optimization is essential to identify ideal landscape compositions for the future. However, due to the solution encoding, standard land use allocation algorithms cannot cope with large land use allocation problems. Solutions are encoded as sequences of elements, in which each element represents a land unit or a group of land units. As a consequence, computation times increase with every additional land unit. We present an alternative solution encoding: functions describing a variable in space. Function encoding yields the potential to evolve solutions detached from individual land units and evolve fields representing the landscape as a single object. In this study, we use a genetic programming algorithm to evolve functions representing continuous fields, which we then map to nominal land use maps. We compare the scalability of the new approach with the scalability of two state-of-the-art algorithms with standard encoding. We perform the benchmark on one raster and one vector land use allocation problem with multiple objectives and constraints, with ten problem sizes each. The results prove that the run times increase exponentially with the problem size for standard encoding schemes, while the increase is linear with genetic programming. Genetic programming was up to 722 times faster than the benchmark algorithm. The improvement in computation time does not reduce the algorithm performance in finding optimal solutions; often, it even increases. We conclude that evolving functions enables more efficient land use allocation planning and yields much potential for other spatial optimization applications.