Assessing Vulnerabilityof Groundwater

May 2001

This booklet is part of a series of educational brochures and slide sets that focuses on various aspects ofwater source protection. The series has been prepared jointly by the University of California AgriculturalExtension Service and the California Department of Health Services.

For further information about this and other documents in the series, contact the project team leader (seebelow) or visit the following website:www.dhs.ca.gov/ps/ddwem/dwsap/DWSAPindex.htm

Authors: Thomas Harter, Department of Land, Air, and Water Resources, University of California at Davis, andLeah G. Walker, California Department of Health Services, Santa Rosa, Calif.

Editor: Larry Rollins, Davis, Calif.

Layout crew: Larry Rollins and Pat Suyama

Photo credits:

• U.S. Natural Resources Conservation Service (cover)

Cover photo: Crop duster spraying pesticide. Vulnerability is a measure of the risk that such an activity mightaffect groundwater quality.

Project Team leader: Thomas Harter, Department of Land, Air, and Water Resources, University of California atDavis

Funding Agency: California Department of Health Services

This document is the result of tax-supported government projects and, therefore, is not copyrighted. Re-printed material, mainly figures, is used with permission, and sources are indicated. Statements, findings,conclusions, and recommendations are solely those of the author(s).

Reasonable efforts have been made to publish reliable information; however, the author and publishingagencies cannot assume responsibility for the validity of information contained herein, nor for the conse-quences of using such information.

1

This booklet introduces the topic of vulnerabilityassessment and mapping. Although it discussesgroundwater vulnerability exclusively, the approachesfor surface water and watershed vulnerabilityassessments are, in principle, very similar to the approachused for groundwater assessments. Hence, much of thepresentation here applies to watershed vulnerability aswell.

The booklet consists of three sections. The first sectionprovides a general overview, to answer the followingfrequently-asked questions:

• What is vulnerability, and how does it relate tosimilar terms, such as susceptibility, pollution risk,and contamination risk?

• Vulnerability of what groundwater?

• Vulnerability to what?

• Why assess vulnerability?

• Which factors determine the degree ofvulnerability?

• How do we characterize vulnerability?

• What is an appropriate scale of investigation fora vulnerability assessment?

• How are uncertainties taken into account whenmapping vulnerability?

• What are the limitations of vulnerabilityassessments?

• What is the role of a vulnerability assessment ingroundwater management?

The second section provides a summary of thevulnerability analysis method recommended byCalifornia’s Drinking Water Source Assessment andProtection (DWSAP) program, administered by theCalifornia Department of Health Services (DHS). Italso describes other commonly-used vulnerabilityassessment methods and provides examples.

The closing section contains a short guide describinghow to select a vulnerability assessment method andhow to verify and audit the results.

Overview

What is Groundwater Vulnerability?

Groundwater vulnerability is a measure of how easy orhow hard it is for pollution or contamination at theland surface to reach a production aquifer. Statedanother way, it is a measure of the “degree of insulation”that natural and manmade factors provide to keeppollution away from groundwater. Vulnerability is highif natural factors provide little protection to shieldgroundwater from contaminating activities at the land

surface. Vulnerability is low, on the other hand, if naturalfactors provide relatively good protection and if thereis little likelihood that contaminating activities will resultin groundwater degradation. The term first was usedin Europe, in the 1960s. The following offers a numberof definitions, from various sources, in chronologicalorder.

• “Aquifer vulnerability is the possibility ofpercolation and diffusion of contaminants fromthe ground surface into natural water-tablereservoirs, under natural conditions.” (Margat,1970, quoted in Vrba & Zoporozec)

• “Vulnerability is the degree of endangerment,determined by natural conditions andindependent of present source of pollution.”(Olmer and Rezac, 1974, quoted in Vrba &Zoporozec)

• “Vulnerability is the risk of chemical substances— used or disposed of on or near the groundsurface—to influence groundwater quality.”(Villumsen et al., 1983, quoted in Vrba &Zoporozec)

• “Groundwater vulnerability is the sensitivity ofgroundwater quality to anthropogenic activitieswhich may prove detrimental to the present and/or intended usage-value of the resource.”(Bachmat and Collin, 1987, quoted in Vrba &Zoporozec).

• “[Vulnerability of a hydrogeological system is]the ability of this system to cope with external,natural and anthopogenic impacts that affect itsstate and character in time and space.”(Sotornikova and Vrba, 1987, quoted in Vrbaand Zoporozec)

• “[Groundwater vulnerability is] a measure of therisk placed upon the groundwater by humanactivities and the presence of contaminants […].Without the presence of contaminants, even themost susceptible groundwater is not at risk, andthus, is not vulnerable.” (Palmquist, 1991,quoted in Vrba & Zoporozec)

• “[Groundwater vulnerability is] the tendency of,or likelihood for, contaminants to reach aspecified position in the groundwater system afterintroduction at some location above theuppermost aquifer.” (U.S. National ResearchCouncil, 1993, quoted in Vrba & Zoporozec)

• “Vulnerability is an intrinsic property of agroundwater system that depends on thesensitivity of that system to human and/ornatural impacts.” (International Association ofHydrogeologists, Vrba & Zoporozec, 1994)

• “Vulnerability is a combination of (a) the

2

inaccessibility of the saturated zone, in a hydraulicsense, to the penetration of pollutants; and (b)the attenuation capacity of the strata overlyingthe saturated zone as a result of physicochemicalretention or reaction of pollutants. […] It is […]a statement about the intrinsic characteristics ofthe strata (unsaturated zone or confining beds)separating the saturated aquifer from the landsurface, thus providing an indication of theimpact of land-use decisions at that point on theimmediately underlying groundwater.” (Foster,1998; in: Robins (ed.), 1998)

In describing and defining groundwater vulnerability,a number of issues must be examined and clarified, allof which refer to these two questions: vulnerability ofwhat?, and vulnerability to what?

As a starting point, we must realize that anygroundwater is, in principle, vulnerable to humanactivity: no groundwater is completely isolated fromthe above-ground environment. And practically allgroundwater originates as recharge at the land surface,proving that there is a direct link between the surfaceand groundwater. (Some groundwater is indeed ofmagmatic origin and formed within the subsurface.)The degree of vulnerability will depend onenvironmental conditions, on how we definegroundwater and the part of groundwater we areinterested in (hence, the question, “vulnerability ofwhat?”), and on the time-scale of interest. It alsodepends on whether or not a vulnerability measure alsois intended to account for the presence and type ofpollutants (hence, “vulnerability to what?”).

Groundwater is a fairly ubiquitous resource. Whendefining vulnerability, we therefore need to ask whatpart of groundwater is to be assessed. For example, arewe to consider only groundwater at the water table?Or should we instead assess all groundwater within theproduction zone? Should we also consider groundwaterin deeper production aquifers? Should we use a simpler“black box” approach and consider groundwater onlyat the production or domestic well, where it is actuallyused?

Because the shallowest groundwater zone is typicallythe most vulnerable, vulnerability assessments aremostly concerned with the vulnerability of theuppermost aquifer (in a multi-aquifer system) or withthe water table (in an unconfined-aquifer system).

Some methods of assessing vulnerability also account—directly or indirectly—for the travel time between thewater table and a well, by considering aquiferpermeability or travel-time zones. Examples are U.S.EPA’s DRASTIC method by Aller et al., 1987,California DPR’s groundwater vulnerability assessmentmethod, and DHS’s vulnerability analysis method.

Time scales are important in defining vulnerability. Thepractical question is: should we consider groundwatervulnerable to today’s activities if the recharge time istens, hundreds, thousands, or tens of thousands ofyears? Many geothermal water resources—geysers, hotsprings, and geothermal wells, for example—delivergroundwater whose recharge period is measured inmillenia. Certainly, the vulnerability question has beenraised for such resources, since they are important asnatural history markers, as tourist attractions, as sourcesof drinking water with particular health benefits, or asan energy resource.

Consideration of the time element is an important partof defining a vulnerability assessment.Typically, if apollutant will take a very long time to reachgroundwater (because, for example, groundwater existsat great depth) or if the pollutant travels relatively slowlywithin the aquifer (because of low hydraulicconductivity, or because of low groundwater velocities),then groundwater vulnerability is thought of as beinglow. If, on the other hand, recharge reaches the watertable within a relatively short time because the aquiferis shallow or because geologic materials above theaquifer are highly permeable, then vulnerability isconsidered to be high. Deep groundwater is consideredless vulnerable than shallow groundwater, because ofthe longer travel times necessary for a pollutant to reacha well.

Time scales are also important when mappingvulnerability to specific pollutants that become lessharmful over time. For example, vulnerability topathogen contamination is only a concern where traveltimes to groundwater (or to groundwater wells) is lessthan a few months to a couple of years. Also, manyorganic chemicals degrade over long time periods,becoming less harmful substances. (For some pesticides,this time period may be on the order of weeks tomonths.)

Vulnerability may or may not include an assessment ofwhether pollutants are present or absent in the regionof interest. Vulnerability that is independent of whetheror not contaminants (pollutants) are present and whichfocuses primarily on a description of naturalenvironmental conditions is often (though not always)referred to as “susceptibility”, “natural vulnerability”,“aquifer sensitivity”, or “intrinsic vulnerability”.

Vulnerability is also a function of pollutant type.Different pollutants (or contaminants) behavedif ferently, depending on their chemical ormicrobiological make-up. Pollution-type-dependentvulnerability, or vulnerability to specific land uses, issometimes referred to as “specific vulnerability” or“integrated vulnerability”.

3

Why Assess Groundwater Vulnerability?

The purpose of vulnerability assessments is not to createscientific insight, but to provide a decision-making toolbased on the best available data and good scientificjudgement. In undertaking a vulnerability assessment,this perspective and the breadth of definitions shouldbe discussed and defined early in the process. In doingso, it is important to understand that vulnerability servesmore of an economic goal than a scientific analysis ofgroundwater resources.

Vulnerability assessments serve primarily to directgroundwater protection efforts such that the mostenvironmental and public health benefits are achievedat least cost. The Natural Research Council (1993)identified four general objectives typically achieved bygroundwater vulnerability assessments: (1) to facilitatepolicy analysis and development at the local andregional level; (2) to provide program management;(3) to inform land use decisions; and (4) to providegeneral education and awareness of a region’shydrogeologic resources. In California and other states,vulnerability assessments can be used to obtain waiversfor specific groundwater monitoring requirements.They are also used to define areas with specialregulations for agro-chemical applications.

The usefulness of vulnerability maps can be argued, asthere are a number of pros and cons to vulnerabilityassessments. Against the use of vulnerability maps forland use planning, one can argue that groundwaterflow conditions and the transport properties of thesubsurface are so complex that they cannot beappropriately captured by any vulnerability tool. Asecond group of arguments against vulnerability zoningarises from the fact that all groundwater is vulnerableto land pollution and that the only geographically-differing factor is the time-scale for the pollution toreach groundwater. On the other hand, land usedecisions will have to be made, land managementpractices need to be sensitive to the risk for groundwatercontamination, and not all anthropogenic activities canbe carried out in isolation of groundwater. Time scalesand distinctions based on travel time are important, atleast with respect to some pollutants. Time scales arealso important in placing groundwater monitoring wellsand in scheduling and providing for planning andaction-response time in case pollution does occur. Soa need exists to provide at least some general guidanceto land use planners, decision makers, and water usersthat allows them to make decisions that areeconomically sensible while at the same timegeologically reasonable.

What Factors Determine Vulnerability?

The thickness and hydraulic properties of the geologic

formations above the aquifer—the unsaturated zone andconfining layers above the aquifer—are the key factorsdetermining the vulnerability of an aquifer system. Theyare the principal natural controls determining therecharge rate and recharge time to the aquifer. Theunsaturated zone also provides key groundwaterprotection by:

• intercepting, sorbing, and eliminating pathogenicviruses and bacteria

• sorbing and degrading many synthetic organicchemicals

• attenuating heavy metals and other inorganicchemicals through sorption and complexationwith mineral surfaces within the unsaturated zoneand through uptake into plants and crops (e.g.,fertilizer)

Much of the degradation of synthetic chemicals (e.g.,pesticides, solvents) occurs in the soil layer, theuppermost part of the unsaturated zone. The soil layer,typically from 2 to 6 feet thick, is a very active zonemicrobiologically. The relatively high microbial activity,higher organic matter content, and the presence of rootsprovide more degradation and removal capacity in thesoil than in the underlying unsaturated zone. Thepotential of the soil and the unsaturated zone to sorb,degrade, or eliminate substances depends on the typeof pollutant and therefore is considered only invulnerability studies that are specific to certain land uses(also referred to as “possible contaminating activities”).

Unsaturated zone flow and transport processes aregenerally complex and difficult to measure, in partbecause of the large amount of natural variability foundin soils, sediments, and rocks. Hence, mappable soilproperties (e.g., infiltration capacity, permeability, soiltype) allow the analyst only to approximate actualtransport processes.

The amount of recharge occurring at the land surfaceis another important factor that determines vulnerability.If climatic conditions are such that little or no rechargeoccurs at the land surface, downward movement ofmoisture through the unsaturated zone will be verylimited, regardless of the hydraulic properties of theunsaturated zone. Many low-lying areas in central andsouthern California have effective natural recharge ratesthat are less than one quarter of one foot per year afteraccounting for the water uptake by the vegetationgrowing through the wet winter and spring months.Many of these areas are also farmed intensively, by useof irrigation. At irrigation efficiencies ranging from 50%to 85%, the recharge leaving the bottom of the rootzone (after crops have taken their share of water) andrecharging through the unsaturated zone to the watertable is typically much higher—as much as two feet peryear—than under natural climatic conditions.

4

On irrigated lands where significant precipitationoccurs, the amount of recharge to groundwater willdepend on a number of factors that are themselvesinterdependent. Such factors include: the hydraulicproperties of the unsaturated zone, the slope of theland surface, and the water uptake through plants andcrops. The steeper the slope, the more likely water is torun off into creeks and streams, especially if the soil hasa low permeability and resulting low infiltration capacity.On the other hand, water will have little opportunityto run off from a highly permeable soil (e.g., sand),regardless of the slope at the land surface. Once waterpenetrates into the soil, some or all of it will be takenup by roots for plant transpiration. Within a slope,water may also travel downslope in the shallowsubsurface, only to emerge as runoff near the bottomof the slope, recharging a stream. What water is leftover below the root zone, after passing through this“sorting” process at and near the land surface, is referredto as “net recharge”. It is available to percolate throughthe unsaturated zone to the water table. The higherthe net recharge, the higher the vulnerability of theaquifer.

How Do We Characterize Vulnerability?

A vulnerability assessment is a process by whichinformation relevant to characterizing groundwatervulnerability—however defined—is assembled toproduce a map that distinguishes areas of greatergroundwater vulnerability from areas of lessergroundwater vulnerability. Vulnerability assessment andvulnerability mapping are sometimes usedinterchangeably. Numerous schemes have beendeveloped for assessing and mapping vulnerability.These methods can be grouped into three majorcategories:

• index-and-overlay methods

• process-based computer simulations

• statistical analyses

Index-and-overlay methods are methods based onassembling information on the most relevant factorsaffecting aquifer vulnerability (soil type, geologicformation type, recharge, etc.), which then isinterpreted by scoring, integrating, or classifying theinformation to produce an index, rank, or class of“vulnerability”. The scoring, ranking, and integrationmethods are based on expert opinion rather thanprocesses and are inherently subjective to some degree.In the United States, the most prominent vulnerabilityassessment method in this category is “DRASTIC.”The vulnerability analysis specified by California’sDWSAP program also falls within this category.

The advantage of these methods is that they providerelatively simple algorithms or decision trees to integrate

a large amount of spatial information into maps ofsimple vulnerability classes or indeces. These types ofmethods are designed to rely on data that are readilyavailable from local, state, or federal governmentagencies, such as information on soils, water level depth,precipitation, geology, etc. These methods areparticularly suitable for use with computerizedgeographic information systems (GIS), which is a digitalform of map making, since they usually involve theoverlaying and aggregation of multiple maps showingsoil properties, depth to water table, recharge, etc.

Process-based computer simulations afford a greatamount of realistic complexity and detail to be builtinto the vulnerability assessment. Computer models canaccount for complex physical and chemical processesand at a very detailed scale. Unlike the two-dimensionalmaps and map layers utilized with other methods,computer modeling allows for a three-dimensionalresolution. Geologic and hydrogeologic variations withdepth can, therefore, be reproduced to evaluate theireffect on vulnerability. Process-based computer modelsfocus on recreating the flow and transport patternswithin the unsaturated zone or in an actual aquifer andcan be used to compute travel times or concentrationsof a contaminant in the unsaturated zone or the aquifer.Computer models do not directly computevulnerability. Rather, vulnerability is defined as afunction of what the computer models simulate. Forexample, high vulnerability may be defined as any regionin the aquifer for which the computer model shows atravel time of less than 5 years. Examples of unsaturatedzone models are PRZM, LEACH, and HYDRUS. Apopular groundwater computer model is MODFLOW.

The most advanced computer models also allow theanalyst to compute the uncertainty that is inevitablyassociated with the computer model predictions dueto shortcomings in the database fed into the computerand due to our limited knowledge of what the“underground world” actually looks like.

Computer models are not commonly used forvulnerability assessment due to their considerable datarequirements and the expertise required to implementthem. In other words, computer simulation models arerarely an economic alternative for vulnerability mapping.However, computer modeling is an excellent andeconomic tool of vulnerability mapping if:

• a more localized analysis of specific vulnerabilityto particular land uses (particular contaminants)is required and sufficient data are available orcan be collected to prepare the computer model

• a number of “what-if” scenarios involvingcomplex processes need to be evaluated formaking important land use planning decisions

Statistical methods are used to quantify the risk of

5

groundwater pollution by determining the statisticaldependence or relationship between observedcontamination, observed environmental conditions thatmay or may not characterize vulnerability (e.g.,unsaturated zone properties, recharge), and observedland uses that are potential sources of contamination(e.g., fertilizer applications, septic tank occurrence).Once a model of this dependence or relationship hasbeen developed with the statistical analysis, it can beused to predict—in a similar area elsewhere—the chanceor risk of contamination. Such an application requires,of course, knowledge of significant environmentalconditions for that area. When statistical methods areused, the risk of contamination is essentially aquantitative measure of “vulnerability”. The higherthe contamination risk, the higher the vulnerability.

In principle, statistical methods are not much differentfrom index-and-overlay methods: both establish arelationship between inherent natural conditions andgroundwater vulnerability (which the statisticians referto as groundwater contamination risk). In the overlaymethods, the relationship is established by a team ofexperts. In the statistical methods, the relationship isestablished by statistical analysis. The advantage of thestatistical method is that the statistical significance canbe explicitly calculated. That provides a measure ofuncertainty or certainty of the model. The disadvantageis that statistical methods are difficult to develop and,once established, can only be applied to regions thathave similar environmental conditions to the regionfor which the statistical model was developed.

Few statistical methods exist, and all have been appliedto specific regions. In California, the Department ofPesticide Regulations uses a statistical vulnerabilityassessment that was developed from a statistical analysisof pesticide occurrence in San Joaquin Valleygroundwater. The vulnerability assessment is used todelineate and discern areas that are particularlyvulnerable to pesticide contamination (called“groundwater protection zones”). Within those areas,certain management practices must be followed whenpesticides are applied, to reduce the risk for pesticideleaching.

What Is an Appropriate Scale of Investigationfor a Vulnerability Assessment?

Vulnerability maps are typically done at a sub-basin,basin, or regional scale. They are not normally usedfor site-specific assessments involving areas smaller thana few tens of square miles. The broad generalizedcategories of parameters used to determine vulnerabilityprovide only broad distinctions of vulnerability.Vulnerability maps are therefore best used todemonstrate large scale, regional differences ingroundwater vulnerability.

Vulnerability mapping of an area or region must bedistinguished from determining the vulnerability of aspecific location, e.g., an individual well or well field,which is but one point on a larger map. Vulnerabilityassessment of a well location does not involve anymapping, only the computation of a vulnerability indexor vulnerability ranking for that particular location (e.g.,DWSAP vulnerability analysis).

How Are Uncertainties Taken Into AccountWhen Mapping Vulnerability?

Only statistical methods allow us to quantify (albeitroughly) the degree of uncertainty associated with avulnerability ranking. Index-and-overlay methods,which rely on qualitative interpretation of broadly-described natural conditions, do not include anymeasures of uncertainty. The scoring or ranking usedin these methods is generally designed to err on theconservative side, that is, it tends to overestimate ratherthan underestimate groundwater vulnerability.

What Are the Limitations of VulnerabilityAssessments?

Vulnerability assessments are a general planning anddecision making tool. They should not be mistaken fora scientifically precise prediction of futurecontamination. Rather, they are a general assessmentof the risk that contamination may occur ingroundwater. As with any risk analysis, there is noguarantee that the contamination does or doesn’t occur.

Because of the implied imprecise nature of vulnerabilityassessments and the inevitable subjectiveness of theunderlying interpretive scheme, the National ResearchCouncil (NRC) issued “three rules of groundwatervulnerability”. These rules, or limitations, should bespelled out explicitly with every vulnerability assessment:

• All groundwater is to some degree vulnerable

• Uncertainty is inherent in all vulnerabilityassessments

• There is a risk that the obvious may be obscuredand the subtle may become indistinguishable

The latter refers to the danger, especially when usingcomplex vulnerability assessment tools, that in light ofthe final vulnerability index or ranking one may losesight of the data used for the analysis and of theassumptions underlying vulnerability assessmentschemes.

Vulnerability Assessment Methodsand Examples

In the previous section, we distinguished three majorgroups of vulnerability assessment and mapping

6

methods: index-and-overlay methods, process modelingmethods, and statistical methods. In this section weprovide a brief overview of a few specific methods. Wealso provide several examples.

Index-And-Overlay Methods

DWSAP Vulnerability Analysis

The DWSAP Vulnerability Analysis has been developedspecifically for determining the vulnerability of drinkingwater sources in California. That includes bothgroundwater and surface water sources. An example ofan index method, the DWSAP method is based onchecklists and a point system derived from generalizedhydrologic and hydrogeologic principle. It also relies onexpert knowledge, including knowledge ofcontamination sources and knowledge of travel time. Itis a “specific vulnerability” assessment method, asopposed to an “intrinsic vulnerability” or “susceptibility”assessment (see previous section). That’s because itexplicitly accounts for the presence and type of possiblecontaminating activities (PCAs) at the land surface andfor the known presence of specific contaminants,regardless of a known contamination source.

The DWSAP method defines vulnerability as “adetermination of the most significant threats to thequality of the water supply that takes into account thephysical barrier effectiveness of the drinking watersource.” It further states that “the vulnerabilitydetermination also considers the type and proximity tothe water supply of activities that could releasecontaminants.”

The DWSAP method prescribes a three-part procedure:(1) determination of the PCAs within the source area ofa well or surface water intake (a complete list of possiblePCAs is provided in the DWSAP documentation); (2)determination of the relative likelihood thatcontaminants can travel from a potential PCA locationto the well or surface water intake (DWSAP calls thisthe “Physical Barrier Effectiveness” or PBE); and (3)determination of the travel time from various PCAs tothe well or surface water intake.

The focus on individual water sources (wells, well fields,water intakes) distinguishes this method from overlaymethods such as the popular DRASTIC method or theBritish GOD method. The viewpoint of the DWSAPmethod (and of other methods focusing on wellheadprotection zones, e.g., that used by the Texas NaturalResource Conservation Commission, Blodgett, 1993)is from the drinking water source (well, intake) backwardto the source of the water coming to that well or intake.This viewpoint assesses vulnerability not only as afunction of the unsaturated zone overlying the aquiferat the location of the well, but as a function of the soiland unsaturated zone in the entire source area, the aquifer

type, and the travel time to the well from individualareas of potential contamination. The product of aDWSAP analysis is a ranked list of all PCAs within theten-year travel time limits of the source area of adrinking water well. Unlike other vulnerabilitymapping methods, the DWSAP analysis does notassign a single, specific vulnerability index to particularwell or well-field of concern or to a location on a map.

To create a ranked list of all PCAs within the sourcearea (wellhead protection area), each PCA obtains ascore for:

• the contamination risk associated with the PCA(1 for low, 3 for medium, 5 for high, 7 for veryhigh),

• the aquifer horizontal travel time from the watertable underneath the PCA to the well (5 if lessthan 2 years, 3 if less than 5 years, 1 if less than10 years),

• and the physical barrier effectiveness (PBE) withwhich soils and aquifer can preventcontamination from reaching the well (5 forlow, 3 for medium, or 1 for high).

The contamination risk ranking of a PCA is predefinedor can be modified through the use of a definedchecklist. The travel time is determined by the locationof the PCA on a map showing the source area(wellhead) protection zones delineating zones A, B5,and B10 (which correspond to areas of less than 2, 5,and 10 year travel time, respectively). Note, that theaquifer travel time does not include the vertical traveltime through the unsaturated zone. The latter isindirectly accounted for in the Physical BarrierEffectiveness. The PBE is determined through the useof a checklist, where points are given depending onaquifer type (confined, unconfined, fractured), aquifermaterials, depth to water, screened well depth, andwell construction. Points for each item are added. Thetotal score determines whether the PBE is low,moderate, or high. The intent of the PBE analysis isto highlight sources that have low or high PBE. Byintent, most sources will have moderate PBE.

To obtain the final score, the three scores(contamination risk, travel time, PBE) are addedseparately for each PCA within each of the threeprotection zones (A, B5, B10). The PCA scores arelisted—in order, from highest to lowest—for each PCAthat exists within each travel time zone. The rankedlist of PCA scores is the final result of the vulnerabilityanalysis. A well is considered vulnerable to those PCAsthat score 8 points or more.

The DWSAP vulnerability analysis itself does notproduce a vulnerability map. However, by assemblingthe vulnerability information developed for many wells

7

within a region, a number of maps could be produced:maps representing intrinsic vulnerability (from the PBEscores), maps showing whether a well is vulnerable atall (PCAs with scores of 8 or higher within the 10-yeartravel time zone), and specific vulnerability maps ofindividual PCA types. Such maps can then be used bylocal land use planners and decision makers for zoningdecisions, etc. Note that PCAs outside the 10-year travelzone are not considered in the determination of wellwater vulnerability.

As with other index-and-overlay methods, the checklist items (e.g., for determining the PBE) and final PCAvulnerability scoring are based on simple, broadcategories. For example, only three types of aquifer aredistinguished: confined, unconfined, and fractured rockaquifer. For unconfined aquifers, three types of aquifermaterials are recognized: unconsolidated sedimentswith a significant clay layer above the water table,unconsolidated sediments without a clay layer, andfractured rock. Depth to the water table is categorizedinto four groups: less than 20 ft., 20–50 feet, 50–100ft, and greater than 100 ft. The reason for the simplicityof the classifications is two-fold: First, the informationrequired to answer the check list questions must berelatively easy to obtain. And second, the assignmentof scores to individual items such as aquifer type anddepth to water is not a precise science.

In fact, it is a bit like trying to compare apples withoranges. For example, as part of the score for PBE, thesame score is assigned for aquifer material consisting ofunconsolidated sediments without a protective clay layeras is assigned for water table depth being more than100 ft. Both receive 10 points for their effectiveness asa protective barrier. Ten points are also given for asanitary seal that exceeds 50 feet depth. Ten extra pointsare given if the unconsolidated sediment has a naturalprotective clay layer above the water table (20 pointstotal). The same “effectiveness” score of 20 is assignedto the following combinations of unconfined aquifertype, water table depth, and well construction:

• Unconsolidated sediments with clay layer (20pts.), water level depth 0–20 ft (0 pts), no sanitaryseal (0 pts)

• Unconsolidated sediment without clay layer (10pts), water level depth greater than 100 ft (10pts), no sanitary seal (0 pts)

• Unconsolidated sediment without clay layer (10pts), water level depth 0–20 ft (0 pts), sanitaryseal 50 ft or thicker (10 pts)

• Fractured rock aquifer (0 pts), water level depthgreater than 100 ft (10 pts), sanitary seal 50 ftor thicker (10 pts)

But are any of these combinations really equivalent in

their barrier effectiveness? How would we know? Onemeasure would be to model and compare the transportand fate of contaminants in either one of those scenarios.The outcome most likely would be different for eachscenario. More importantly, even if we investigated ormodeled the contaminant transport at ten different sitesthat all fell within the same classification above, resultswould greatly differ between sites.

The aforementioned example illustrates the reason forkeeping the check list and scoring simple: because ofthe uncertainty and imprecision of the scoring system,it is of little additional value to distinguish betweenmore than a couple of different classes for each checklist or scoring item. Even the final DWSAP vulnerabilitylist of PCAs should be interpreted carefully and not toomuch weight should be given to differences betweenindividual scores. But the analysis is useful to provide alist of those PCAs that should be of high concern (PCAsscoring above the cutoff) and those that are not a majorconcern (PCAs scoring below the cutoff). The list ishelpful in prioritizing planning and protection efforts,and in providing guidance for additional detailedvulnerability analysis, e.g., with groundwater models,where ambiguity exists.

DRASTIC Vulnerability Mapping (Aller et al.,1987)

DRASTIC is perhaps the most popular vulnerabilitymapping tool in the United States. Put together by agroup of experts and the EPA in the mid-1980s, it hasbeen applied to a number of groundwater basins,regions, even states, including some regions inCalifornia.

The name stands for Depth to groundwater, Rechargerate, Aquifer media, Soil media, Topography, Impactof vadose zone media, and hydraulic Conductivity ofthe aquifer. Similar to Physical Barrier Effectivenesscomputation in the DWSAP vulnerability analysis, eachof these seven parameters are assigned a score. InDRASTIC, scores of 0 to 10 are assigned to eachparameter, with 0 meaning low risk for groundwatercontamination, 10 meaning high risk for contamination.(By contrast, high PBE scores in DWSAP’s methodmean low risk for contamination and high barriereffectiveness.) Each parameter is weighted. That is, thescores are multiplied with a parameter-specific weight.The weighted scores of all seven parameters are thenadded for the final DRASTIC score. Two sets of weightscan be used: one for general (intrinsic) vulnerabilityanalysis, one for specific vulnerability to pesticides.

In DRASTIC, the scores are computed for everylocation within a larger region and mapped. (The samealgorithm could be applied to individual wells, whichare points on the map.) For the mapping, sevenindividual maps are prepared, one for each of the seven

8

parameters. The seven maps are then overlayed, and ateach point of the map, the final DRASTIC vulnerabilityscore is obtained by computing the score from the sevenparameters valid at that location of the map. A GISsystem makes this task extremely simple; it would bevery cumbersome to implement manually, without theGIS system.

Unlike in the DWSAP analysis, DRASTIC does notaccount directly for contaminating activities orgroundwater contamination already present in the areaof interest. It also does not account for the travel timewithin the aquifer.

DRASTIC scores, like the DWSAP’s PBE scores, aremeant as a rough measure of the likelihood or risk thatgroundwater contamination can (or cannot) occur. Theadvantages and pitfalls are similar to those of theDWSAP PBE scoring. DRASTIC has been critizedspecifically for underestimating the vulnerability offractured rocks, when compared to unconsolidatedaquifers.

GOD (Foster, in Robins, 1998)

GOD is a vulnerability assessment method developedin Great Britain, where most groundwater resourcesare in hardrock aquifers, primarily sandstone andlimestone aquifers. Unconsolidated overburden or soillayers cover the fractured hardrock in many places. LikeDRASTIC, GOD is an index-and-overlay methoddesigned to map groundwater vulnerability over largeregions based on a few important parameters; GODstands for Groundwater occurrence, Overall lithologyof the unsaturated zone or overlying aquitard, andDepth to groundwater table. Scores are assigned to eachof the three categories and then multiplied to yield afinal score. In developing GOD, the method’s authorshave given particular consideration to the likelihood offractures or fracture systems to develop in the soils,overburden, or overlying geologic units of the aquifer.

Process-Based Computer Modeling

Process-based computer models imitate the actualphysics and chemistry of the contaminant transport inthe subsurface, to make predictions about the amountof degradation, sorption, or remobilization that acontaminant may experience as it travels through thesubsurface. Computer models can be used to computethe travel time and the extent of contaminant plumesin the subsurface. Such computations are based not ongeneralized expert or statistical knowledge, but on thefundamental scientific principles controlling themovement of water and contaminants in the subsurface.

Computer modeling codes for unsaturated zone andgroundwater modeling abound, often serving veryspecific applications. Four computer model codes are

particularly popular for computing the fate andtransport of contaminants as those contaminants traveldownward through the unsaturated zone to the watertable: VLEACH (EPA), PRZM (EPA), LEACH(Cornell University), and HYDRUS (InternationalGroundwater Modeling Center). All four codes (listedhere roughly in order of complexity and user expertiserequired, with VLEACH being least complex andHYDRUS being the most complex) compute thedownward movement of water (flow) and contaminants(transport). Each is strictly one-dimensional: onlydownward movement is considered. Horizontal orinclined movement through the unsaturated zone isnot considered. (HYDRUS can be adapted for 2-dimensional evaluations, however). These codes aredesigned specifically to determine the amount ofsorption and degradation that occurs in soils and inthe unsaturated zone, and are typically used to computethe residual, non-degraded mass of a contaminant thatarrives, over time, at the water table. These models areexcellent tools to predict water flow and contaminanttransport under specific hydrogeologic conditions inthe unsaturated zone, in particular those that are highlylayered (heterogeneous), and for chemicals thatundergo multiple chemical processes or chemicalreactions. For vulnerability assessments, these modelsare sometimes applied to evaluate the attenuationcapacity within and travel time through the unsaturatedzone above the water table. Subregions or sections withsimilar unsaturated zone properties are selected andgrouped. One simulation is performed for each groupor section and the results mapped in various ways, e.g.,a map of the travel time to the water table (perhapsspecific to a contaminant), a map of the percent removalof a contaminant within the vadose zone, etc.

The most common computer code used forgroundwater modeling (in all three dimensions) isMODFLOW (USGS). This and other 3D computermodels can be used for highly specific site studies ofgroundwater contaminations that are far beyond thepurpose of a typical vulnerability assessment. The datarequirements for running these models are tremendousand require careful data preparation and dataprocessing, particularly in a fully three-dimensionalsimulation. Sometimes, groundwater models are usedto explicitly compute the complex source area of thewater reaching a well. In that case, the same computermodel can also be used to compute specific vulnerabilityfor particular contaminants of concern or for specificPCAs within the source area of the well. If a well-calibrated and well-documented groundwater modelexists, it may be preferable to utilize the groundwatermodel to address specific vulnerability issues rather thanto use a basic index or a scoring-based vulnerabilityassessment.

9

Statistical Methods

Few statistical methods have been developed forvulnerability assessment, primarily because they rely onlarge data sets from regions already contaminated. Thekey to the development of a statistical method is a large,high quality data set of a specific contaminant or severalspecific contaminants commonly found in a region.That data set is then correlated in one fashion or anotherto specific properties of the subsurface (depth to water,soil permeability, slope, hydraulic conductivity, etc.) tocreate a statistical predictive model that can be used tomake a quantitative statement about the contaminationrisk, such as: “if this kind of unsaturated zone, aquifer,and land use properties exist in a location, then the riskfor groundwater contamination is X percent, with aconfidence interval of plus or minus Y percent”.

CALVUL (Troiano et al., 1999)

In California, a statistical method has been developedby the Department of Pesticide Regulations (DPR) todetermine the specific vulnerability of groundwater topesticide residues. The method is nicknamed CALVUL(California Vulnerability approach, Troiano et al.,1999). DPR uses the vulnerability analysis to determine,for each section of agricultural land in California,whether groundwater in that section of land isvulnerable or not. If it is vulnerable, the land section(square mile) becomes part of a groundwater protectionarea (GWPA), a term that has been formally defined byDPR: “’Groundwater protection area’ means an areaof land that contains soils that have been determinedby the director, in consultation with the countyagricultural commissioner, to be sensitive to themovement of pesticides to ground water.”

The DPR method distinguishes two types ofvulnerability: (1) vulnerability to pesticide leaching incoarse soils with shallow water table, and (2)vulnerability to pesticide runoff over hardpan soils (andsubsequent leaching in dry wells). Within thegroundwater protection zones, certain pesticidemanagement practices must be followed, depending onwhether pesticide leaching or pesticide runoff is ofconcern.

How is the vulnerability determined with CALVUL?Originally, the department’s regulations declaredvulnerable any section of land that had a qualifieddetection of pesticide residue in groundwater. Thatmethod was found to be unsatisfactory, because it didnot encourage application of mitigation measures untilthe contamination had already occurred. The pesticidedatabase for the San Joaquin Valley was used todetermine the statistical relationship between pesticideresidue occurrence in groundwater and a large numberof soil (unsaturated zone) properties, including waterholding capacity, texture, organic matter content,

permeability, shrink-swell potential, slope, infiltration,etc. By determining what soil and geographic propertiesare good predictors of pesticide residue occurrence ingroundwater, the statistical model can be used toidentify sections of land that have similarly “vulnerable”soil and geographic conditions but no past or currentpesticide residue detections (Troiano et al., 1999).Presumably, the most vulnerable areas are identifiedand can be targeted for implementation of mitigationmanagement practices.

In principle, the CALVUL approach is identical to thevulnerability assessment in DRASTIC or in the DWSAPanalysis: environmental factors are aggregated(combined) to make a determination of which areasare vulnerable and which areas are not vulnerable. Here,however, factors are aggregated by utilizing a statisticalmodel. In DWSAP or DRASTIC, factors are aggregatedby utilizing an expertly-derived scoring system. Becauseit is based on actually measured contamination, thestatistical analysis lends the approach quantifiablecredibility and validity that is not available with theindex-and-overlay methods.

Texas Case Study (Evans and Maidment,1995)

In Texas, a statistical analysis similar to DPR’s SanJoaquin study, albeit based on a different set of statisticaltools, was developed by Evans and Maidment (1995).Instead of pesticide residues in groundwater, thismethod uses an analysis of nitrate in groundwater asthe basis for delineating vulnerable groundwater areas.As in CALVUL, the assumption made by the developersof the method was that where high contaminationexists, the aquifer is more vulnerable than elsewhere.Nitrate was used as the target variable because a largeamount of data was available throughout Texas andbecause nitrate is neither sorbed nor signifantlydegraded in groundwater (except under anaerobicconditions). The analysis was done by using data fromthe entire state. The analysis units in this model areindividual 7.5 minute maps. (Compare: the analysisunits in CALVUL are land sections and the analysisunit in the DWSAP method is the protection area of awell.)

From a 30-year statewide database, the probability (risk)that a well would have nitrate levels in excess of somethreshold value (1, 2, 5, or 10 mg/l NO3-N) wascomputed separately for each analysis unit. Theexceedance probability is used as a quantifiable measureof vulnerability: the higher the chances that nitrate isfound in wells within an analysis unit (area), the higherthe vulnerability must be. For example, if 4 out of 12wells within an analysis unit exceeded the thresholdlevel, then the risk was computed as 4/12 = 0.333.This exceedance probability was then correlated to theunsaturated zone thickness, to the organic matter

10

content, the amount of precipitation, and the amountof fertilizer sales (as a surrogate for contaminant sourcepresence). Results were compiled statewide as well asfor individual aquifer basins. The regression modeldeveloped from the correlation analysis showed nosignificant correlation between vulnerability (nitrateexceedance probability) and any of the indicatorvariables except depth to water table. However, therewas significant correlation between vulnerability andaquifer basin, indicating that the different regionalaquifers investigated had significantly differentvulnerability. From least to most vulnerability, the fouraquifer types were ranked as follows: deep bedrock(14%), shallow bedrock (37%), deep unconsolidated(46%), and shallow unconsolidated (49%), wherenumbers in parentheses indicate the exceedanceprobability for 1 mg/l NO3-N.

This study illustrates that overall results of the statisticalanalysis follow the general patterns laid out by the index-and-overlay methods, at considerable higher expense.

How to Decide Which Method WorksBest

The decision of which method to use will dependprimarily on the objectives of the vulnerability analysis.It also depends on the available data and the availablefunding. In defining the objectives, it is important toconsider what needs to be achieved with thevulnerability analysis, who will use the results of theanalysis, what decisions it will influence, and what thecost will be if a wrong decision is made because ofinadequate or poor information. That cost should beweighed against the cost of an appropriatelyimplemented vulnerability assessment.

Index-and-overlay methods probably will continue tobe the main staple for vulnerability mapping. They aremost suited for application by planning departmentsand non-hydrologically-trained users, since the decisiontree comes in a “black box”. However, theinterpretation of the results requires some professionaljudgment, which can be sought as part of thevulnerability assessment.

Process modeling with computer methods will workbest where the hydrogeology and unsaturated zoneconditions are well known, where data exist for buildinga groundwater model, or where a well-calibrated andwell-documented groundwater model is alreadyavailable. If an existing groundwater model is used, oneshould carefully and professionally review what themodel was created for, what assumptions went into themodel, and how good the data are that went into themodel. Groundwater models are like cars: not everymodel is good for every purpose (even if they all havethe same basic components, namely an engine, wheels,

etc.).

Statistical methods are useful where widespreadcontamination of pesticides or nitrates exist to build awell-founded statistical prediction model that can beused for adjacent areas. Statistical methods are usuallyregion specific and not suitable for transfer to other,geographically different regions. Unlike index methods,the statistical method implies a certain degree ofvalidation and quantifiable measure of vulnerability.

After the Assessment: Verificationand Post-Audit

Verification refers to some independent procedure thatcan verify the results of the vulnerability analysis. Post-audit is essentially the same as verification, but typicallyoccurs years later, when additional data have beensampled that can be compared to the predictions madeby the vulnerability analysis.

Verification and post-audits of vulnerability assessmentscan be done in many different ways. The most commonapproach, particularly for verification of assessmentsdone with index-and-overlay methods, is to comparethe vulnerability map with the actual occurrence of somecommon pollutant in groundwater. Typical pollutantsused are nitrate and pesticides. However, suchverification works well only where the appropriatepollution source is actually present and has been presentfor some time.

11

• Blodgett, R. 1993. Drinking Water Vulnerability Assessments in Texas, Proceedings, 1993 Annual Con-ference, American Water Works Association.

• Evans, Thomas A., and David R. Maidment. 1995. A Spatial and Statistical Assessment of the Vulner-ability of Texas Groundwater to Nitrate Contamination. Technical Report CRWR 260, Center for Re-search in Water Resources, Bureau of Engineering Research, The University of Texas at Austin, Austin,TX. 257p.

• Holtschlag, D. J., and C. L. Luukkonen. 1998. Assessment of ground-water vulnerability to Atrazineleaching in Kent County, Michigan: Review, comparison of results of other studies, and verification,Water-Resources Investigations Report 98-4006, U.S. Geological Survey, Lansing, MI. 32p.

• Robins, N. S. (ed.). 1998. Groundwater Pollution, Aquifer Recharge and Vulnerability. Geological Soci-ety Special Publication No. 130, Geological Society, London, GB. 224 p.

• Snyder, Daniel T., J. M. Wilkinson, and L. L. Orzol. 1998. Use of a ground-water flow model with particletracking to evaluate ground-water vulnerability, Clark County, Washington, Water-Supply Paper 2488,U.S. Geological Survey, Denver, CO, 61 p.

• Troiano, J., F. Spurlock, and J. Marade. 1999. Update of the California Vulnerability Soil Analysis forMovement of Pesticides to Ground Water. October 14, 1999. Department of Pesticide Regulation, Sac-ramento, CA 95814-3510, Document EH 00-05. Document available on the internet: http://www.cdpr.ca.gov/docs/empm/pubs/ehapreps/eh0005.pdf

• Vrba Jaroslav and Alexander Zoporozec (eds.). 1994. Guidebook on Mapping Groundwater Vulnerabil-ity. International Association of Hydrogeologists: International Contributions to Hydrogeology Volume16, Verlag Heinz Heise, Hannover, Germany, 129p.

For Further Reading