Abstract: In the processes of designing urban green land soil monitoring system, there is a technical problem with respect to optimize the existing monitoring sites through an easy to operate, low-cost and representative optimization method which considers multiple soil quality indices. Aiming at the optimization of soil quality monitoring of urban green space in Shanghai, we proposed a technical framework for the optimization of the secondorder forward sampling. The Sandwich spatial sampling model and Kriging interpolation method were used to evaluate the accuracy of existing point positions. Spatial simulated annealing was used to add extra points on the existing sampling system, considering the spatial characteristics of spatial autocorrelation and heterogeneity of various soil indicators. Then, the final scheme was used to test whether the optimal design number could meet the minimum number requirement of the stratified sampling method. The results showed that the sampling variance of the existing monitoring sites was lower than 10% of the mean values of nutrients and heavy metals concentrations in Shanghai green space soils. However, the representativeness of certain elements in Pudong, Jing’an, Hongkou and Xuhui Districts needed to be improved. The mean Kriging variance could be reduced by 64.5% by adding 350 monitoring sites. This multiobjective optimization method balances the equity between the accuracy and the cost, and helps city managers to build a reasonable monitoring program for urban green space.

Key words: multi-objective optimization, monitoring network assessment, spatial simulation annealing, Kriging, Sandwich spatial model.