Groundwater sensitivity mapping in Kentucky using GIS and digitally vectorized geologic quadrangles

ORIGINAL PAPER

Groundwater sensitivity mapping in Kentucky using GIS

and digitally vectorized geologic quadrangles

Andrea Croskrey Æ Chris Groves

Received: 31 December 2005 / Accepted: 23 March 2007 / Published online: 21 August 2007

� Springer-Verlag 2007

Abstract Groundwater sensitivity (Ray and O’dell in

Environ Geol 22:345–352, 1993a) refers to the inherent

ease with which groundwater can be contaminated based

on hydrogeologic characteristics. We have developed dig-

ital methods for identifying areas of varying groundwater

sensitivity for a ten county area of south central Kentucky

at a scale of 1:100,000. The study area includes extensive

limestone karst sinkhole plains, with groundwater extre-

mely sensitive to contamination. Digitally vectorized

geologic quadrangles (DVGQs) were combined with ele-

vation data to identify both hydrogeologic groundwater

sensitivity regions and zones of ‘‘high risk runoff’’ where

contaminants could be transported in runoff from less

sensitive to higher sensitivity (particularly karst) areas.

While future work will fine-tune these maps with addi-

tional layers of data (soils for example) as digital data have

become available, using DVGQs allows a relatively rapid

assessment of groundwater sensitivity for Kentucky at a

more useful scale than previously available assessment

methods, such as DRASTIC and DIVERSITY.

Keywords Groundwater � Karst � DVGQs � Runoff �

Contamination

Introduction

As communities expand to accommodate increased popu-

lation, the demand for and threat to potable water supplies

increases. Protection of groundwater resources is both

critical and problematic in the state of Kentucky since over

a million of its citizens depend on public water supplies

that use groundwater. In addition, over half a million

people use groundwater as a private water source, of which

half are estimated to be contaminated by bacteria (Ken-

tucky Division of Water 2004). This number does not take

into consideration the thousands that use surface water that

is often fed from groundwater sources. Complicating

matters is that about half of Kentucky is underlain by

limestone karst aquifers formed from extensive dissolution

of soluble bedrock resulting in areas with caves, under-

ground rivers and large springs. Typically, in these areas

groundwater is very susceptible to contamination due to

high aquifer permeability that results in minimal attenua-

tion of contamination along with rapid groundwater

transport velocities.

Groundwater sensitivity describes the intrinsic suscep-

tibility of groundwater to contamination, which is closely

related to the mechanisms and rates of storage and trans-

port of water and contaminants within bedrock zones. It is

a comparison of how well the vadose zone system protects

the underlying aquifer from the pollution released on the

surface and within the bedrock. While there are a number

of ways the associated concepts have been labeled (e.g.

Johnson and Van Driel 1978; Vrba and Zaporozec 1994)

in this work we are consistent with the terminology of

Geographic definitions: United States of America, Kentucky, Barren

River Area Development District.

A. Croskrey (&)

Geologic Resources Division, National Park Service,

P.O. Box 25287, Denver, CO 80225-0287, USA

e-mail: [email protected]

A. Croskrey � C. Groves

Hoffman Environmental Research Institute,

Department of Geography and Geology,

Western Kentucky University,

1906 College Heights Blvd #31066,

Bowling Green, Kentucky 42101-1066, USA

123

Environ Geol (2008) 54:913–920

DOI 10.1007/s00254-007-0899-z

Ray and O’dell (1993a, b) and the US Environmental

Protection Agency (USEPA). Groundwater sensitivity is

considered to be the intrinsic ease with which groundwater

could be contaminated based on the basic hydrogeologic

characteristics of an area, independent of whether a con-

tamination source actually exists. Closely tied to this is the

fact that groundwater contamination is related to the land-

use in an aquifer’s recharge zone (e.g. Helsel and Ragone

1984; Grady 1994; Eckhardt and Stackelberg 1995). We

use the term groundwater vulnerability to consider sensi-

tivity as well as the presence or absence of contaminant

sources. Thus, for an area to be vulnerable in this sense

requires an area that is characterized by hydrogeologic

properties in which contaminants, if present, could be

easily transported to the groundwater supply, plus a source

of contamination. Identifying varying areas of groundwa-

ter sensitivity is clearly critical in developing strategies to

protect groundwater resources and helps in the rational

planning of urban growth, developing commercial and

industrial zones and in the remediation of environmental

accidents. Groundwater is a vital natural resource not only

within Kentucky, but also in the ten county Barren River

Area Development District (BRADD, Fig. 1) which has an

extensively developed karst landscape. This landscape

includes the globally significant Mammoth Cave system,

the longest known cave in the world, and the Pennyroyal

Sinkhole plain. Aquifers associated with these landscapes

are very susceptible to contamination and thus create a

high incentive for mapping groundwater sensitivity. The

dissolution of carbonate rock that is creating the high

permeability of the aquifer, places the groundwater in this

region at risk. The solutionally enlarged fractures and

larger conduits throughout the limestone allow rapid

transport of water. Significant conduits, for example, can

have hydraulic conductivities that are three or more

magnitudes greater than surrounding fractured rocks

(White 1988). This rapid transport also does not allow the

same filtration and lengthy transit time typical for

groundwater found in alluvium or sandstone aquifers. This

means that contaminants can travel quickly in karst lime-

stone and are dispersed in a manner that can make

remediation of spills difficult. Since the transport of water

in the Mississippian limestone in the BRADD area is

mainly through conduits and bedding planes, it is typically

not dispersed in a plume that has a location which can

easily be predicted. Sampling wells that miss contaminant-

carrying conduits in a karst aquifer by just a few meters

may never be able to detect the contaminants. The collapse

of dissolution conduits in the bedrock also places the

groundwater at an added risk by creating sinkholes and

swallets (locations where surface streams sink into the

subsurface) that transport surface water to the groundwater

typically with little filtration.

Previous investigations

Since Margat (1968) introduced concepts and terminology,

associated with the ‘‘vulnerability of groundwater to con-

tamination,’’ there have been numerous schemes

introduced to map the relative ease with which ground-

water might be impacted in the presence of one or more

contaminants. Examples of few of these schemes will

follow and will have some similar characteristics. One of

these characteristics includes an index or rating of sensi-

tivity or vulnerability. These indexed ratings are normally

created by assigning values to parameters of the hydroge-

ologic system. Examples of parameters commonly

included are rock type and water table elevations. Since

parameters like rock type are difficult to assign quantitative

values and measurements, qualitative interpretations are

applied and numbers assigned. Assigned values and num-

bers are then applied to an equation to create a combined

index score of sensitivity or vulnerability that is used to

assess and map the sensitivity or vulnerability of an area.

Depth to water table, Recharge, Aquifer media, Soil type,

Topography, Impact to vadose zone and hydraulic Con-

ductivity (DRASTIC), for example (Aller et al. 1985), was

developed for the USEPA and is perhaps the most widely

known groundwater sensitivity mapping scheme to stan-

dardize the evaluation of the groundwater pollution

potential within various hydrogeologic settings. This

approach of mapping groundwater sensitivity was based on

the DRASTIC. There have been a wide variety of efforts to

develop variations on the methodology of this strategy,

including modifications of DRASTIC in particular (e.g.

Whittemore et al. 1987; Crawford and Smith 1989;

Vermeulen et al. 1994; Rudek 1999). Other efforts have

Fig. 1 Map of study area, including the ten-county Barren River

Area Development District, Mammoth Cave National Park and the six

30 · 60 min geologic quadrangles

914 Environ Geol (2008) 54:913–920

123

developed conceptually similar indexing schemes but using

different sets of evaluation parameters and ways in which

they are weighted, including GOD (Foster 1987), SINT-

ACS (Civita et al. 1990), Land-surface Zoning (Adams and

Foster 1992), EPIK (Doerfliger and Zwahlen 1997; Do-

erfliger et al. 1999; Goldscheider et al. 2001a) and the PI

Method (Goldscheider et al. 2000). There have also been

studies to compare the effectiveness of different approa-

ches under various hydrogeologic conditions (Villumsen

et al. 1983; Schmidt 1987; Hoyer and Hallberg 1991; Ei-

mers et al. 2000; Amharref et al. 2001; Ceballos and Avila

2004; Vıas et al. 2005) and to evaluate the impact of data

uncertainty (Carbonell 1993). Rapidly evolving digital

technology has meant a steady refinement of the technical

processes and conceptual frameworks that these analyses

have been undertaken and the maps produced (e.g. Johnson

and Van Driel 1978; Vrba and Zaporozec 1994; Corwin

et al. 1997).

In the development of this work, it has been recognized

that well-developed karst provides challenges both to the

protection of groundwater and to the application of various

groundwater sensitivity and vulnerability methodologies.

In these areas, sinkholes, swallets and fissures can quickly

drain water directly from the surface to the water table,

bypassing the soil layers and vadose zones that in other

settings can impact the concentration of potential con-

taminants entering the groundwater supply as it is

recharged by surface runoff. Rapid downward vertical

movement of water, for example, can undermine the

importance of calculating the depth to the water table, an

element of DRASTIC. While in non-karst settings, slow

movement of recharge through tens of meters between the

surface and the water table can provide valuable prolonged

time, for example to allow for the degradation of bacteria.

In well-developed karst areas however, solutionally

enlarged fractures and in some cases even underground

waterfalls can allow water to travel downward essentially

unhindered under the influence of gravity. Therefore, the

weighting of these characteristics and the use of them in

karst areas, requires modification. In these settings and

even in lateral movement through karst aquifers, ground-

water can experience turbulent flow through conduits.

Hydraulic conductivities can exist that far exceed the

applicability of the standard rating scale set by DRASTIC.

Furthermore, DRASTIC does not account for the influ-

ences of overland runoff from non-karst regions into

valleys with karst-type drainage. For example, if steep

slopes of sandstone with low permeability, line the hillsides

of a karst valley, contaminants released on the surface of

that slope can travel by overland flow into the karst valley

(Goldscheider et al. 2001b). The contaminants can then

take a quick route to the water table via sinkholes, losing

streams and fissures. A question is then raised with regard

to the appropriate groundwater sensitivity rating for the

steeply sloped sandstone. If vertical transport of pollutants

is the only concern, then it should have a low sensitivity

rating. But if overall effectiveness of spills and leaks

contaminating the groundwater of the area is of concern,

then the area should be considered more sensitive to

groundwater contamination. Another significant compli-

cating issue is that karst groundwater recharge areas are

often independent of surface topography and require par-

ticular methods, such as groundwater tracer tests, to even

delineate contributing areas (e.g. Goldscheider et al. 2001a,

b; Meiman et al. 2001; Paylor and Currens 2001). There

has been considerable work in developing these methods in

karst areas in Europe, where people rely on karst aquifers

as sole-source water supplies (e.g. Hotzl et al. 1995; Triplet

et al. 1997; Cimino 2001; Goldscheider et al. 2001a, b;

Daly et al. 2002).

In Warren County, Kentucky, Crawford and Smith

(1989) altered the normal DRASTIC scales and weights to

accommodate the well-developed karst of the Pennyroyal

Sinkhole Plateau. They used a modified DRASTIC model

created for use in Madison County, Alabama as their

starting point (Beard and Roland 1988). Crawford and

Smith (1989) added sinkholes to the flood prone areas

when calculating the depth to water table for the DRASTIC

model to properly accentuate the contamination potential

on the sinkhole plain. The result of their work was a paper

map of Warren County at a scale of 1:63,360 that divided

the county into areas of sensitivity, based on their modified

DRASTIC index. The modified DRASTIC index is the sum

of the weighted DRASTIC factors. This map is still used in

land use planning and in zoning laws in Warren County on

waste land spreading (Zoning Ordinance for Warren

County, Kentucky Article 5 Section 2.5E).

In 1994, a groundwater sensitivity map at a scale of

1:500,000, also produced in hard copy, was completed

for the entire Commonwealth of Kentucky (Ray et al.

1994). This assessment rated how susceptible the

groundwater would be to contamination in five physio-

graphic regions of Kentucky based on the general

recharge, flow rate and dispersion potential of the aquifer

media. This model was entitled ‘‘DIspersion/VElocity–

Rated SensitivITY or ‘‘DIVERSITY’’ (Ray and O’dell

1993a, b). By rating the hydrogeologic units using these

characteristics, they were able to portray the groundwater

sensitivity of karst aquifers more accurately in reference

with other aquifers than the original DRASTIC model.

Thus, this map based on hydrogeologic characteristics

was another step toward the goal of providing protection

and hazards information for groundwater resource man-

agement in Kentucky. Yet, the map is at a scale of

1:500,000 and was not used by individual communities

and counties in the BRADD area.

Environ Geol (2008) 54:913–920 915

123

The purpose of the current research is to develop and

implement a methodology for relatively rapid analysis of

groundwater sensitivity combining elevation models with

digital geologic data in order to develop more useful

products than have previously been available with regard to

detail, scale and coverage extent. These data recently only

have become available in Kentucky and these methods also

account for zones where there is potential for overland flow

of contaminants from relatively lower to higher sensitivity

areas. In addition, increasing availability of digital geologic

and topographic data allowed us to generate the maps in a

format that could be used in Geographic Information

Systems (GIS).

Materials and methods

In order to create a digital groundwater sensitivity map for

the BRADD area at a scale of 1:100,000 that accounted for

runoff, geologic and topographic data were obtained for the

Hopkinsville, Madisonville, Beaver Dam, Bowling Green,

Campbellsville and Tompkinsville 30 · 60 min quadran-

gles. Since bedrock between the surface and the groundwater

can inhibit the migration of contaminants and are the media

in which the water is stored, they were the main consider-

ation in creating this map. The majority of aquifers used in

the region are shallow so it was assumed that surface geology

is adequate to represent the media that contaminants must

travel through. The Kentucky geological survey’s digitally

vectorized geologic quadrangle maps (DVGQs) were

utilized (http://www.uky.edu/KGS/home.htm). These maps

were derived from the original 7.5 min, 1:24,000 scale paper

quadrangles. They display geologic information in the form

of points, vectors and polygons. This method of data storage

and display allowed straightforward categorization of the

geologic units in ArcMap (ArcGIS Version 8.3 by ESRI in

Redlands, CA, USA). The first step taken to create this

generalized map of groundwater sensitivity for the BRADD

area was to group the geologic units into categories based on

their gross lithology. Classification was similar to that by

Ray et al. (1994). The gross lithologies were then assigned to

a relative hydrogeologic sensitivity. In ArcMap, geologic

units were assigned to the appropriate sensitivity using

‘‘Select by Attribute’’ in ArcMap. Table 1 shows the hy-

drogeologic group and groundwater sensitivity to which

each of the geologic units in the study area were classified.

Once all the geology for the region was classified into a

sensitivity index of high, moderate, low, very low or open

water, areas of ‘‘high risk runoff’’ were classified. Overland

flow can carry contaminants from impermeable rocks (such

as shale) that protect groundwater onto very permeable

rocks (such as karst limestone) that can quickly transport

runoff to the groundwater. This situation can be observed

in and around the Mammoth Cave Plateau as well as many

other karst regions. As a result of the study area containing

such a landscape, drainage basins and topography were a

very important consideration in the creation of these maps.

Topographic data for all six 30 · 60 min quadrangles, in

the form of United States Geological Survey (USGS) for-

mat Digital Elevation Models (DEMs), were obtained from

the Kentucky Division of Geographic Information website

(http://www.dgi.ky.gov/gisdata.htm). The DEMs were

based on 1:24,000 7.5 min quadrangles and therefore share

the United States National Map Accuracy Standards issued

by the United States Bureau of the Budget on June 17 1947,

90% of the points are within the 40 feet of their true

position. These were then transformed to ESRI grid rasters

using Arc toolbox since this format is applicable to the

watershed calculation in spatial analyst tools. Thus the

watershed of the highest sensitivity areas could be

extrapolated. In order to calculate the watershed, there are

a series of steps that need to be taken to prepare the raster

for analysis. First the raster was used to calculate a flow

direction raster. This process determines the greatest

Table 1 Geologic formations categorized into hydrogeologic groups and groundwater sensitivity rating

Hydrogeologic groups Formation codes from DVGQs Groundwater

sensitivity rating

Karst limestone ls, Mba, Mbp, Mbr, Mg, Mgd, Mgdh, Mgg, Mgh, Mmg, Mmk, Mpbr, Mpr,

Mr, Mre, Mrsg, Mrss, Msg, Msgl, Msgr, Msl, Msll, Msls, Mslu, Msrf, Mssu

High

Interbedded sandstone,

limestone, and shale

Cs, cu, Db, Dsj, Kt, Mb, Mbe, Mbk, Mbm, Mbss, Mbw, Mc, Mcb, Mcl, Mcp,

Md, Mdc, Mdm, Mdp, Mfp, Mgb, Mgbc, Mgc, Mgcc, Mgcy, Mgo, Mgs, Mh,

Mha, Mhb, Mhc, Mhg, Mhgc, Mk, Mkc, Mkd, Mkp, Ml, Mme, Mp, Mpb,

Mpc, Mpt, Mpv, Mrs, Ms, Msa, Msh, Msw, Mswsh, Mtg, Mth, Mts, Mv,

Mw, Mwl, Mwv, Oa, Oag, Oat, Oc, Oca, Occ, Od, Odb, Odr, Ods, Ogl, Ol, Pac,

Pb, Pbm, Pc, Pca, Plr, Pm, Pp, Ps, Psh, Pt, Ptc, rl, Sb, Slo, Slwl, Sob, Solc, ss

Moderate

Unconsolidated sediments Af, g, Kc, Qal, Qala, Ql, Qs, Qt, QTb, QTg, QTs, QTt Low

Shale Dc, Dna, sh Very Low

Open water Lake, river Open Water

916 Environ Geol (2008) 54:913–920

123

difference in elevation between a cell and its neighbors and

then assigns a value to the cell to identify the direction

runoff would flow. With the flow direction determined,

sinks or cells that have an elevation lower than all its

neighbors were identified. The sinks were then filled and a

new flow direction raster was created. To delineate the

watershed using the spatial analysis tool, pour point data

were needed to identify the area for which the watershed

was to be delineated. A pour point is the outlet of a

watershed, which is also the lowest point in a watershed.

Thus, when using digital elevation data to calculate a

watershed, all cells with elevations increasing in value

away from the chosen pour point (uphill from the pour

point) are included in the watershed. Since the runoff onto

the high sensitivity regions was desired, the shapefile of the

high sensitivity areas was converted to a raster and used as

the pour point data. Then the pour point raster was applied

to the new flow direction raster and the watershed tool

located all the cells that flowed onto the high sensitivity

regions. The high risk runoff for the Beaver Dam and

Campbellsville quadrangles were also digitized using

editing sessions in ArcMap, tagged vectorized contours and

the sensitivity shapefiles. These were used for comparison

with the high risk runoff identified using spatial analyst

tools.

Results

This methodology resulted in classification of all 136

geologic units included in the six 30 · 60 min quadrangles

(Table 1; Fig. 2). This and the runoff data were accurate to

a scale of 1:24,000. Selection of attributes and export of

selected units created five shapefiles: one of each for high,

moderate, low and very low sensitivity and one for major

lakes and rivers in the study area called open water.

Another shapefile for designating the area of high risk

runoff was created using DEMs and the watershed tool.

The high sensitivity, moderate sensitivity, low sensitivity,

very low sensitivity, and open water shapefiles do not

overlap since they were derived from the DVGQs and the

high risk runoff shapefile overlaps some areas of the

moderate, low, very low and open water shapefiles.

Discussion

Since DVGQs are a product of assembling the 7.5 min

quadrangles together, there were continuity problems

where different geologists had split or combined forma-

tions. Thus, boundaries between 7.5 min quadrangles were

visible in the merged geology if a continuous formation

was split into members on one quadrangle and not on the

neighboring quadrangle. However, the breaks in mapped

geologic unit boundaries were eliminated by creating gross

lithologies since the breaks between quadrangles were

between similar rock types, but with different names.

There were five hydrogeologic groups into which geo-

logic units from the DVGQs were grouped. These are

similar to those created by Ray and O’dell (1993a, b).

Geologic units known to develop karst geomorphology

were assigned to the hydrogeologic group called karst

limestone. Bedrock in this group transports water at rates

that can reach kilometers per day. Furthermore, solution

features on the surface allow direct contamination of

groundwater from surface runoff and subsurface conduits

complicate the process of drilling to locate contaminant

plumes. Therefore, areas with karst limestone were

assigned as areas with the most sensitive groundwater and

are rated as having a high sensitivity.

A second hydrogeologic group included interbedded

sandstone, limestone and shale. Geologic formations and

Fig. 2 Map showing analysis

of the groundwater sensitivity

for the Beaver Dam

30 · 60 min quadrangle

Environ Geol (2008) 54:913–920 917

123

members added to this group had heterogeneous charac-

teristics of aquifer lithology that changed on a scale not

recorded on the 7.5 min geologic quadrangles used in this

analysis. The areas with this hydrogeology are not at an

extremely high risk of contamination but have local

anomalies and variances that require investigation at a site-

by-site scale. These areas were assigned moderate sensi-

tivity. Low sensitivity was assigned to a hydrogeologic

group for unconsolidated sediments in stream valleys.

These small aquifers tend to be groundwater discharge

areas in humid climates, such as Kentucky, and have

laminar flow. Contaminants entering the groundwater sys-

tem in these regions would be a greater threat to surface

water they feed than the groundwater near the recharge

area. The exception to this are losing streams in the karst

regions that drain onto karst that were included in the high

risk runoff zones.

The two final hydrogeologic groups are shale and open

water. Though shale has high porosity, its permeability is

so low that it typically acts as an aquitard and decelerates

the transport of fluids (Fetter 2001). Consequently, shale

has groundwater that is typically the least sensitive to

contamination in the study area and was rated as very low

sensitivity. Finally, open water areas are shown to make the

coverage complete since groundwater sensitivity was not

evaluated for these areas.

Another important element of this project was to iden-

tify zones of high risk runoff flowing from lower to higher

sensitivity areas. A point of serious consideration when

using spatial analyst to identify these zones was that one of

the steps to calculate watersheds is to fill sinks. In general,

the fill sinks step is used to identify cells that were assigned

erroneous values when the DEMs were created, such as

cells that were void of data. Filling sinks also creates a

smoother surface that helps to simplify calculations and

models. Yet, sinks occur in nature so the fill sinks step is

usually limited to sinks less than 10 m deep. This is done to

remove sinks that are a result of error and keep sinks that

are result of nature. In this study, we wanted areas that

drained onto karst limestone and there are sinks that are

greater than 10 m and still drain onto these high sensitivity

regions. Therefore, all sinks were filled for this analysis. To

check the accuracy of using the spatial analyst tool to

identify high risk runoff zones, the Beaver Dam and

Campbellsville quadrangles were digitized by hand. There

were a few differences found between the zones. The high

risk runoff zones identified by the tools did not go as far

into the ridges as the zones digitized by hand and had

angular boundaries. This is probably a result of the flatness

of the ridge tops. When digitized by hand, a flat ridge top is

divided symmetrically in half between the highest con-

tours, as is typical for catchment boundary identification in

non-karst areas. It is difficult for the flow direction tool to

calculate the flow direction in very flat areas since the tool

is unable to make this step that hand digitizing with con-

tours can. Therefore, the spatial analyst tool missed the

tops of some flat hills and ridges that should have been

included in the high risk runoff. The other difference

noticed was that the spatial analyst tool was able to identify

zones draining into limestone sinks that broke through the

sandstone cap rock that were missed in hand digitizing.

Therefore, the runoff identified using the spatial analyst

tool will have to be reviewed and corrected.

Since both the data sources used in creating the different

zones for the map were at a scale of 1:24,000, they can be

accurately projected at any scale between 1:100,000 and

1:24,000, a significant advantage of digital products. With

the added high risk runoff areas and the availability of

these maps in a digital format that can be used in ArcMap,

these maps have expanded on what has been done in the

past. A portion from the 1:500,000 scale map created by

Ray et al. (1994) is compared (Fig. 3) with the same area

from a map created in this project. An increase in the

definition between sensitivity zones, the identification of

unconsolidated sediments in the streams and the addition of

high risk runoff zones can be seen.

Continuing work includes the creation of a template

groundwater vulnerability mapping within the study area

Fig. 3 Two groundwater

sensitivity maps of Edmonson

county at a scale of 1:500,000.

Map on the left was created by

Ray et al. in 1994. Map on the

right is from shapefiles created

by this project

918 Environ Geol (2008) 54:913–920

123

(using Barren County and parts of Mammoth Cave

National Park) that will combine the groundwater sensi-

tivity data with land use data to estimate the risks to

groundwater associated with human activity. The meth-

odology developed in this project will also now readily

allow groundwater sensitivity maps to be created at a scale

of 1:100,000 for the rest of the state. We are also investi-

gating ways to add other layers that could impact

groundwater sensitivity (soils, for example), as more data

continue to become available in digital formats at suffi-

ciently detailed scales.

Acknowledgments Funding was provided by the US Environmen-

tal Protection Agency (we appreciate the assistance of Kentucky

Senator Mitch McConnell), the Barren River Area Development

District, the National Park Service—Mammoth Cave National Park,

the Natural Resource Conservation Service and the Western Kentucky

University Applied Research and Technology Program and Action

Agenda Program. We also appreciate assistance in this research from

Heather Veerkamp, Weldon Hawkins, Ben Tobin, Pat Kambesis,

Kevin Cary, Shwu Jing and Amy Edwards.

References

Adams B, Foster SD (1992) Land-surface zoning for groundwater

protection. J Inst water Environ Manage 6:312–320

Aller L, Bennett T, Lehr JH, and Petty RJ (1985) DRASTIC—a

standardized system for evaluating ground water pollution

potential using hydrogeologic settings. US Environmental Pro-

tection Agency Report EPA/600/2-85/018

Amharref M, Mania J, Haddouchi B (2001) Adaptation of a

vulnerability evaluation method to groundwater pollution.In:

Proceedings of the first international conference on saltwater

intrusion and coastal aquifers—monitoring, modeling, and

management. Essaouira, Morocco, 23–25 April 2001

Beard LM, Roland B (1988) Madison County, Alabama—DRASTIC

application, Report No. WR28-1520-146, 22

Carbonell A (1993) Ground water vulnerability assessment: predict-

ing relative contamination potential under conditions of

uncertainty. In: Committee on techniques for assessing ground

water vulnerability. National Academy Press, Washington, p 204

Ceballos RP, Avila JP (2004) Vulnerabilidad del agua subterranea a la

contaminacion de nitratos en el estado de Yucatan. Ingenierıa

8:33–42

Cimino A (2001) Hydrogeological and geophysical contributions to

karst aquifer protection: the Carburangeli Wildlife Reserve,

North-west Sicily. In: 7th conference on limestone hydrology

and fissured media. Besancon, France 20–22 September 2001, pp

83–86

Civita M, Chiappone A, Falco M, Jarre P (1990) Preparazione della

cartado vulnerabilita per la rilocalizzazione di un impianto pozzi

dell’ Aquedotto do Torino. In: Proceedings of the 1st National

Convention ‘‘Protzione e Gestione delle Acque Sotterranee:

Metologie, Tecnologie e Objettivi 2:461–461

Corwin DL, Vaughn PJ, Loague K (1997) Modeling nonpoint source

pollutants in the vadose zone with GIS. Environ Sci Technol

31:2157–2174

Crawford NC, Smith JH (1989) ‘‘DRASTIC’’: an assessment of

groundwater pollution vulnerability in Warren County, Ken-

tucky, Warren County. Comprehensive Plan Technical Report,

Bowling Green, Kentucky

Daly D, Dassargues A, Drew D, Dunne S, Goldscheider N, Neale S,

Popescu IC, Zwahlen F (2002) Main concepts of the European

approach for (karst) groundwater vulnerability assessment and

mapping. Hydrogeol J 10(2):340–345

Doerfliger N, Zwahlen F (1997) EPIK: a new method for outlining of

protection areas in karstic environment. In: Gunay and Jonshon

(eds). International symposium on Karst waters and environ-

mental impacts, Antalya, Turkey. Balkema, Rotterdam, pp 117–

123

Doerfliger N, Jeannin PY, Zwahlen F (1999) Water vulnerability

assessment in karst environments: a new method of defining

protection areas using a multi-attribute approach and GIS tools

(EPIK method), Environ Geol 39(2):165–176

Eckhardt DA, Stackelberg PE (1995) Relation of ground-water

quality to land use on Long Island, New York. Ground Water

33:1019–1033

Eimers JL, Weaver JC, Terziotti S, Midgette RW (2000) Methods of

rating unsaturated zone and watershed characteristics of public

water supplies in North Carolina. US Geological Survey Water-

Resources Investigations Report 99-4283, p 31

Fetter CW (2001) Applied hydrogeology, 4th edn. Prentice-Hall,

New Jersey

Foster S (1987) Fundamental concepts in aquifer vulnerability,

pollution risk and protection strategy. In: Proceedings, vulner-

ability of soil and groundwater to pollutants, International

conference (Noordwijk aan Zee, The Netherlands), proceedings

and information No. 38, pp 69–86

Goldscheider N, Klute M, Sturm S, Hotzl H (2000) The PI method—a

GIS-based approach to mapping groundwater vulnerability with

special consideration to karst aquifers Zeitschrift Angew.

Geologie 46:157–166

Goldscheider N, Hotzl H, Fries W, Jordan P (2001a) Validation of a

vulnerability map (EPIK) with tracer tests. In: Proceedings of the

7th conference on limestone hydrology and fissured media,

Besancon, 20–22 September 2001, pp 167–170

Goldscheider NC, Hotzl H, Neukum C, and Werz H (2001b) Tracer

tests and vulnerability mapping in the alpine karst system

Winterstaude as the scientific basis of a drinking water

protection strategy for the community of Bezau (Austria). In: J

Mudry and F Zwahlen (eds) 7th conference on limestone

hydrology and fissured media, Besancon, pp 171–174

Grady SJ (1994) Effects of land use on quality of water in stratified-

drift aquifers in Connecticut: US Geological Survey Water-

Supply Paper 2381, p 56

Helsel DR, Ragone SE (1984) Evaluation of regional ground-water

quality in relation to land use. US Geological Survey toxic

waste-ground-water contamination program. US Geological

Survey Water-Resources Investigations Report 84-4217, p 33

Hotzl H, Adams B, Aldwell R, Daly D, Herlicska H, Humberg G,

Ketelaere D, Silva ML, Sindler M, and Tripet JP (1995) Karst

groundwater protection. Regulations—European Commission

Report EUR 16547EN, COST-ACTION 65, pp 403–434

Hoyer BE, Hallberg GR (1991) Groundwater vulnerability regions of

Iowa. Iowa Geological Survey Bureau, Iowa City (1:500,000

map sheet)

Johnson RH, Van Driel JN (1978) Susceptibility of coastal plain

aquifers to contamination, Fairfax County, Virginia, A computer

composite map: US Geological Survey Open-File Map 78-265,

1 sheet

Kentucky Division of Water (2004) Overview of Kentucky’s Waters.

http://www.water.ky.gov/homepage_repository/overview.htm.

Cited 20 Mar 2006

Margat J (1968) Vulnerabilite des nappes d’eau souterraines a la

pollution–Bases de la cartographie. BRGM # 68. Bureau de

recherches geologiques et minieres, Departement eau) #68, SLG

198 HYD. Orleans, France

Environ Geol (2008) 54:913–920 919

123

Meiman J, Groves C, Herstein S (2001) In-cave dye tracing and

drainage basin divides in the Mammoth Cave karst aquifer,

Kentucky,US Geological Survey Karst Interest Group Proceed-

ings, Water-Resources Investigations Report, 01-4011, pp 179–

185

Paylor RL, Currens JC (2001) Mapping karst ground-water basins in

the inner bluegrass as a nonpoint-source pollution management

tool. In: Beck BF, Pettit AJ, Herring JG (eds) Proceedings of the

2001 sinkhole conference, Louisville, pp 229–234

Ray JA, O’dell PW (1993a) DIVERSITY: a new method for

evaluating sensitivity of groundwater to contamination. Environ

Geol 22:345–352

Ray JA, O’dell PW (1993b). Dispersion/velocity-rated groundwater

sensitivity. In: Beck BF (ed) Applied karst geology, multidis-

ciplinary conference on sinkholes and the engineering and

environmental impacts of karst (4th, Panama City, Florida,

1993), proceedings. Balkema, Rotterdam, pp 189–198

Ray JA, Webb JS, O’dell PW (1994) Groundwater sensitivity regions

of Kentucky. Kentucky Department for Environmental Protec-

tion—Division of Water—Groundwater Branch. 1:500,000

Rudek J (1999) Improvements to the DRASTIC ground-water

vulnerability mapping method. US Geological Survey Fact

Sheet FS-066-99, p 6

Schmidt R (1987) Groundwater contamination susceptibility in

Wisconsin. Wisconsin Department of Natural Resources

Groundwater Management Plan Report # 5. WR 177-87,

Madison, pp 1–27

Triplet JP, Doerfliger N, Zwahlen F (1997) Vulnerability mapping in

karst areas and its uses in Switzerland. Hydrogeologie 3:15–57

Vermeulen H, Lobo-Ferreira JP, Oliveira MM (1994) A method for

estimating aquifer recharge in DRASTIC vulnerability mapping.

In: Proceedings of the second European conference on advances

in water resources technology and management. Lisbon, June

1994. AA Balkema and EWRA, Rotterdam

Vıas JM, Andeo B, Perles MJ, Carrasco F (2005) A comparative study

of four schemes for groundwater vulnerability mapping in a

diffuse flow carbonate aquifer under Mediterranean climatic

conditions. Environ Geol 47:586–595

Villumsen A, Jacobson O, Sonderskov C (1983) Mapping the

vulnerability of groundwater reservoirs with regard to surface

pollution. Geological Survey of Denmark Yearbook 1982.

Copenhagen, pp 17–38

Vrba J, Zaporozec A eds. (1994) Guidebook on mapping groundwater

vulnerability: Verlag Heinz Heise, International Association of

Hydrogeologists, vol 16, International contributions to hydrog-

eology, p 131

White WB (1988) Geomorphology and hydrology of karst terrains.

Oxford University Press, New York, p 464

Whittemore DO, Merchant JW, Whistler J, McElwee CE, Woods JJ

(1987) Ground water protection planning using the ERDAS

geographic information system, Automation of DRASTIC and

time-related capture zones. In: Proceedings of the NWWA

FOCUS conference on Midwestern Ground Water Issues.

National Water Well Association. Dublin, pp 359–374

920 Environ Geol (2008) 54:913–920

123