Simulating Spatial Distributions, Variability and Uncertainty of Soil Arsenic by Geostatistical Simulations in Geographic Information Systems

This study quantifies and delineates the spatial distributions, variability and uncertainties of soil arsenic (As) in the northern part of Changhua County in central Taiwan by using kriging, sequential Gaussian simulation (SGS) and simulated annealing simulation (SAS) in geographic information systems. Thousand realizations of soil As are simulated by using SGS and SAS. The impacts of the number of generated realizations on the standard deviation of the soil arsenic distributions simulated by SGS and SAS were performed for assessing and mapping spatial variability and distributions of soil arsenic. The semivariogram results show that As data exhibited small scale variation in the study area. Kriging captures spatial distribution of soil As, but underestimate high As concentration area. However, both SGS and SAS well capture spatial distributions and variability of soil arsenic in the study area, but SGS results in higher average standard deviation than SAS with the same number of realizations. 40 realizations of SAS are reliable to simulate and map the distributions, variability and uncertainty of soil arsenic, but 100 realizations are needed by using SGS. Both estimates and simulates demonstrated that the high As concentration area distributed around the irrigation ditch systems and industrial plants in the study site. Finally, the cumulative distribution of soil As of 100 SGS realizations is obtained and can be used for further risk assessment.