Premium
Multivariate Geostatistïcal Analysis of Ground‐Water Contamination: A Case History
Author(s) -
Istok Jonathan D.,
Smyth Jeffrey D.,
Flint Alan L.
Publication year - 1993
Publication title -
groundwater
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.84
H-Index - 94
eISSN - 1745-6584
pISSN - 0017-467X
DOI - 10.1111/j.1745-6584.1993.tb00829.x
Subject(s) - nitrate , environmental science , variogram , multivariate statistics , groundwater , aquifer , sampling (signal processing) , hydrology (agriculture) , contamination , soil science , pesticide , environmental chemistry , kriging , geology , statistics , mathematics , chemistry , ecology , geotechnical engineering , organic chemistry , filter (signal processing) , computer science , computer vision , biology
A case history is presented for the application of multivariate geostatistical methods to the problem of estimating pesticide concentrations in ground water from measured concentrations of nitrate and pesticide, when pesticide is under‐sampled. The shallow, poorly confined, sand and gravel aquifer underlying the lower Malheur River basin near Ontario, Oregon is contaminated by nitrate and metabolites of the herbicide Dacthal (dimethyl tetrachloroterephthalate) or DCPA. The results of extensive ground‐water sampling indicate that a significant positive correlation exists between measured nitrate and DCPA concentrations in the aquifer. This suggests that future sampling should include a large number of the less‐expensive nitrate analyses, and these data should be used to support the interpretation of fewer, more expensive DCPA analyses. Sample variograms were computed for nitrate and DCPA concentrations and were fit with isotropic, spherical variogram models with correlation ranges of 4 km. Incorporating measured nitrate concentrations in the DCPA estimates obtained by cokriging reduced estimation variances from 14 to 34%. A simple economic analysis demonstrated that for this aquifer, acquiring additional nitrate samples is a cost‐effective way to reduce estimation variances for DCPA.