Optimal sampling design for reclaimed land management in mining area: An improved simulated annealing approach

Abstract

Soil properties in reclaimed land altered inevitably due to human activities. To realize sustainable land use and management in mining area, the long-term dynamic monitoring of reclaimed soil properties is required. Traditional dense sampling network is time consuming and expensive. Therefore, it is necessary to devise a scientific sampling design that maximizing the accuracy of results while minimizing the costs. In this study, we developed an improved simulated annealing method considering the spatial correlation to optimize the sampling design for soil properties, to enhance the efficiency of sampling plans for the prediction of space variables. A separate optimization for each soil property and a sampling optimization integrating various soil properties were conducted by combining simulated annealing and artificial neural networks. The number of sampling points from present improved simulated annealing method was markedly reduced, the spatial correlation was increased and the sampling points obtained showed more accurate predictions for soil properties. The sampling points for each soil property were mostly distributed in the marginal areas and appropriately added in the middle of the dumps. There were more sampling points distribution in these areas with large changes in topography for each soil property after improved simulated annealing optimization and more sampling points were needed to reflect soil information in southern dump in Antaibao opencast coal mine. When operating optimization integrating all the properties, some sampling points of various properties were added to the sampling pattern, showing a uniform distribution of sampling points. This study provides a theoretical basis for optimal design for soil property monitoring and sustainable land management.