Agrichemicals (herbicides and nitrate) are significant sources of diffuse p
ollution to groundwater. Indirect methods are needed to assess the potentia
l for groundwater contamination by diffuse sources because groundwater moni
toring is too costly to adequately define the geographic extent of contamin
ation at a regional or national scale. This paper presents examples of the
application of statistical, overlay and index, and process-based modeling m
ethods for groundwater vulnerability assessments to a variety of data from
the Midwest U.S. The principles for vulnerability assessment include both i
ntrinsic (pedologic, climatologic, and hydrogeologic factors) and specific
(contaminant and other anthropogenic factors) vulnerability of a location.
Statistical methods use the frequency of contaminant occurrence, contaminan
t concentration, or contamination probability as a response variable. Stati
stical assessments are useful for defining the relations among explanatory
and response variables whether they define intrinsic or specific vulnerabil
ity. Multivariate statistical analyses are useful for ranking variables cri
tical to estimating water quality responses of interest. Overlay and index
methods involve intersecting maps of intrinsic and specific vulnerability p
roperties and indexing the variables by applying appropriate weights. Deter
ministic models use process-based equations to simulate contaminant transpo
rt and are distinguished from the other methods in their potential to predi
ct contaminant transport in both space and time. An example of a one-dimens
ional leaching model linked to a geographic information system (GIS) to def
ine a regional metamodel for contamination in the Midwest is included. (C)
1999 IAWQ Published by Elsevier Science Ltd. All rights reserved.