Keywords
multifunctional agriculture, stakeholder discussions, optimization, scale
Start Date
1-7-2006 12:00 AM
Abstract
Intensively managed agricultural areas in North-Western Europe are undergoing a shift from solely production oriented use to provision of multiple services and functions. Design of multifunctional agricultural landscapes can be supported by exploration of the potential to effectively combine economic performance, landscape identity, nature conservation and environmental quality. The Landscape IMAGES methodology enables spatially explicit exploration of options by multi-objective optimization, for multifunctional agriculture in landscapes at a scale of a few km². The framework has been developed to support stakeholder discussions and informed decision making. For simultaneous optimization of multiple objectives the evolutionary algorithm of Differential Evolution is employed. Selection pressure normal to the trade-off surface is exerted by Pareto-based ranking, while a crowding metric is used to provide tangential selection pressure. A large range of alternative configurations of a landscape representing the trade-off surface between the objectives was generated and explicit insight in the trade-off between the objectives was provided. Enriching the initial population of the optimization with extremes obtained from single objective optimizations resulted in an improvement of the quality of the obtained non-dominated solution set. A landscape visualization module enables rapid assessment of alternative landscape and land-use designs. In this paper, the methodology is elaborated and its functioning is illustrated with a hypothetical example of a grassland-based landscape with hedge rows bordering the fields.
Landscape design and agricultural land-use allocation using Pareto-based multi-objective Differential Evolution
Intensively managed agricultural areas in North-Western Europe are undergoing a shift from solely production oriented use to provision of multiple services and functions. Design of multifunctional agricultural landscapes can be supported by exploration of the potential to effectively combine economic performance, landscape identity, nature conservation and environmental quality. The Landscape IMAGES methodology enables spatially explicit exploration of options by multi-objective optimization, for multifunctional agriculture in landscapes at a scale of a few km². The framework has been developed to support stakeholder discussions and informed decision making. For simultaneous optimization of multiple objectives the evolutionary algorithm of Differential Evolution is employed. Selection pressure normal to the trade-off surface is exerted by Pareto-based ranking, while a crowding metric is used to provide tangential selection pressure. A large range of alternative configurations of a landscape representing the trade-off surface between the objectives was generated and explicit insight in the trade-off between the objectives was provided. Enriching the initial population of the optimization with extremes obtained from single objective optimizations resulted in an improvement of the quality of the obtained non-dominated solution set. A landscape visualization module enables rapid assessment of alternative landscape and land-use designs. In this paper, the methodology is elaborated and its functioning is illustrated with a hypothetical example of a grassland-based landscape with hedge rows bordering the fields.
