arezoo kheyrkhah
Abstract
Correct and consistent uses of natural resources preserve this valuable wealth. Using optimization knowledge can assist us to achieve this object. Thus, this study aims to compare linear programming as a classical method of optimization with genetic evolutionary algorithm for land use optimization of ...
Read More
Correct and consistent uses of natural resources preserve this valuable wealth. Using optimization knowledge can assist us to achieve this object. Thus, this study aims to compare linear programming as a classical method of optimization with genetic evolutionary algorithm for land use optimization of the Bayg watershed. Results showed that linear programming reduced dry farming acreage and increased the acreage of irrigated agriculture. After minimization, surface runoff and sediment yield declined by 1.16, 12.91 percent, respectively. Genetic algorithm led to an increase in rangeland, irrigated agriculture and horticulture acreages, while almond orchard and dry farming acreages were reduced. Furthermore, surface runoff and sediment yield declined by 13.95 and 31.99 percent, respectively. Linear programming acted stronger in satisfying the constraints, as compared with genetic algorithm. The constraint “total acreage” was satisfied by linear programming, while genetic algorithm could not meet this constraint. Sensitivity analysis of linear programming showed that the most critical factor in minimizing runoff and sediment yield function was the coefficient of dry farming with a reduced cost of 67.52. Results also established that the constraints “total acreage and minimum acreage of rangeland” with the shadow prices of 397.40 and 233.28, respectively had the highest negative impact on the optimal solution. Meanwhile, the constraints “maximum acreages of irrigated horticulture and almonds garden” with the shadow prices of -134.97 and -118.44, respectively had the highest positive impact on the optimal solution. As a general conclusion it can be stated that in land use optimization problems with a large number of constraints, genetic algorithm show poorer performance in satisfying constraints, as compared with linear programming.
