Acta Geodaetica et Cartographica Sinica ›› 2013, Vol. 42 ›› Issue (5): 722-728.

Previous Articles     Next Articles

Soil Spatial Sampling design based on a Multi-objective Micro-neighborhood Particle Swarm Optimization Algorithm

  

  • Received:2012-05-14 Revised:2013-01-06 Online:2013-10-20 Published:2014-01-23

Abstract: The design of a soil spatial sampling network is a complex optimization problem, which must reconcile the conflicts between survey budget, sampling efficiency, sample size and spatial pattern of soil variables. This study presents a soil spatial sampling model on the basis of a multi-objective micro-neighborhood particle swarm optimization algorithm (MM-PSO). The model combines minimum mean kriging variance (MKV) and maximum entropy (ME) as the fitness function of the MM-PSO, and integrates the constraints of sampling barriers, maximum sample size, survey budget and sampling interval as the neighbor operating rules of the particles, in order to improve sampling accuracy and efficiency and to determine sample size and spatial sampling pattern simultaneously. We applied the method to optimizing the sampling networks for soil organic matter in Hengshan County in north-west China. The results indicate that the MM-PSO features a good convergence ability and stability, and can obtain better sampling networks with higher fitness values of the objectives than the single objective and spatial simulated annealing algorithm.

Key words: soil sampling, multi-objective particle swarm optimization, micro neighborhood , geostatistics

CLC Number: