GIS-based statistical, geostatistical, and stochastic analyses of nitrate contamination in the Cimarron Terrace aquifer in Oklahoma.

摘要

Ordinary kriging estimates of groundwater quality monitoring data were integrated into GIS to provide a quantitative, statistical, and weighted means of nitrates defining the statistical significance of spatial changes. The tendency of probability to remain geographically stationary or to change with time was also evaluated as chronic exceedance frequencies of nitrates by using indicator kriging. At threshold 10 mg/l, nitrates were frequently observed in the central portion of the aquifer. From logistic regressions, the best radius of well influence so that the land use could show significant effects on nitrate concentration in wells was determined to be 1,000 meters.; Developed land, Fertilizer N, and Depth to water table were significant predictors of groundwater nitrate concentration in excess of 4 mg/l while percent of Clay, Fertilizer N, and Depth to water table were significant predictors in excess of 10 mg/l. The probability curves of Monte Carlo simulation indicated that mean probabilities of occurrence of nitrate in excess of 4 mg/l and 10 mg/l were 0.8214 and 0.2581 respectively. Fertilizer N was found to be the most sensitive parameter for occurrence of both levels of nitrate.