Characterization of soil lead by comparing sequential Gaussian simulation, simulated annealing simulation and kriging methods

Abstract.

This study attempted to characterize the spatial patterns of lead (Pb) for further soil monitoring and remediation by comparing the sequential Gaussian simulation, simulated annealing techniques and ordinary kriging methods to delineate soil lead in a rice paddy field in the north of Changhua County, Taiwan. For reproducing the statistics of Pb and natural log Pb (ln(Pb)), simulation techniques yielded better results than ordinary kriging. Meanwhile, sequential Gaussian simulation and simulated annealing reproduced the spatial variation of the measured Pb and ln(Pb), as well as identified the global spatial continuity and discontinuity patterns. Furthermore, the simulated annealing method equaled the global measurement statistics and spatial patterns of Pb and ln(Pb) more so than sequential Gaussian simulation and kriging. Finally, the realizations generated by sequential Gaussian simulation displayed significantly higher local heterogeneity than those generated by simulated annealing. The realizations of simulated annealing simulation are consistent in presenting the spatial patterns of soil Pb.