|
Featured Paper by Barnali Dixon, Nivedita Candade, Robert Stetson Kriging is a statistical technique that uses semi-variograms to produce a continuous surface from point data and provide an estimate for unsampled locations. This study aims to predict contamination wells using different kriging methods such as universal, ordinary, disjunctive, probability, indicator, and simple kriging to determine which method is optimal in predicting contamination of wells. Each method's performance was assessed by calculating the proportion of sites that were misclassified as non-contaminated when they were actually contaminated and vice-versa. The methods were validated with a subset of wells not included in the original simulation. The study looked at nine organic and inorganic contaminants. Comparisons of the various methods were made using ArcGIS Geostatistical Analyst and MAPCALC. Results show that there are considerable differences in prediction of contaminate potential when different kriging methods were used. This study will provide a basis for future comparisons of vulnerability maps to the well contamination.
|
|
|||||||
|
|
||||||||
|
||||||||
