Introduction
Groundwater is a natural resource often subjected to severe human impact. In Mediterranean regions, pres-sure on groundwater resources involves two factors: water has traditionally been scarce due to climatic con-ditions, i.e., low rainfall and high evaporation, and there has been an increase in the tourist population along the whole Mediterranean coast and particularly in the Costa del Sol area (Andalusia, southern Spain), where the population doubles during the summer months. In this respect, a new problem has arisen because many potentially contaminating human activities have been developed above the aquifers. Therefore, strategies such as vulnerability mapping are required to preserve
optimum groundwater quantity and quality. Therefore, management of this vital natural resource has become a worldwide priority.
Vulnerability maps have become an ever more essential tool for groundwater protection and envi-ronmental management. Between 1997 and 2003, the European Union supported COST Action 620 ‘‘Vul-nerability mapping for the protection of carbonate (karst) aquifers’’ to compare assessment methods for vulnerability mapping and to develop a general pro-cedure for application in European carbonate aquifers (Daly et al. 2002; Zwahlen 2004). Over large areas of Europe, groundwater from carbonate aquifers consti-tutes an important natural resource for drinking water supply and, for this reason, protection schemes have
J. M. Vı´as B. Andreo M. J. Perles F. Carrasco
A comparative study of four schemes
for groundwater vulnerability mapping
in a diffuse flow carbonate aquifer under
Mediterranean climatic conditions
Received: 18 May 2004 Accepted: 27 September 2004 Published online: 3 December 2004 ÓSpringer-Verlag 2004
Abstract This paper shows the re-sults of a comparative study involv-ing application of the vulnerability mapping methods known as AVI, GOD, DRASTIC and EPIK to a pilot carbonate massif in southern Spain, namely the Torremolinos aquifer. The main objectives of the study were to determine which methods are most suitable for diffuse flow carbonate aquifers such as in southern Spain, and to evaluate variations in the degree of vulnera-bility associated to the rainfall variations that normally occur in a Mediterranean climate. According to three of the above methods, the aquifer is moderately vulnerable, but the AVI method evaluated it as highly vulnerable—this, however, is improbable. The vulnerability maps reflect the great importance of
geol-ogy-related parameters (mainly those concerned with lithology) and, to a lesser degree, that of the depth of the groundwater table which is related to the rainfall. After this latter parameter, it is possible to distinguish between humid and dry climatic situations; thus, vulnerabil-ity increases in a humid year, espe-cially according to the GOD and AVI methods. In conclusion, the GOD method seems the most ade-quate of the methods applied in this work for vulnerability mapping of diffuse flow carbonate aquifers in the Mediterranean domains.
Keywords Vulnerability mappingÆ DRASTICÆGODÆ AVI and EPIK methodsÆDiffuse flow carbonate aquiferÆMediterranean climateÆ Southern Spain
J. M. Vı´asÆB. Andreo (&) M. J. PerlesÆF. Carrasco Department of Geography, University of Ma´laga, Ma´laga, 29071, Spain E-mail: [email protected] Tel.: +34-95-2132004 Fax: +34-95-2132000
been adopted. The project was given additional impe-tus by the European Water Framework Directive (2000), which is intended to provide a common framework for water resource policy and management. In the last 30 years, the international scientific com-munity has shown great interest in groundwater quality and, thus, many publications focused on environmental management for groundwater protection (Foster 1987; Adams and Foster 1992; Robins et al. 1994; Vrba and Zaporozec 1994; Ho¨tzl 1996; Daly and Drew 1999). Since the concept of vulnerability to the contamination was introduced by Albinet and Margat (1970), many methods have been proposed for vulnerability mapping of aquifers, including DRASTIC (Aller et al. 1987), GOD (Foster1987), AVI (Van Stempvoort et al. 1993), SINTACTS (Civita1994), EPIK (Doerfliger et al.1999); and PI (Goldscheider et al.2000). The above acronyms normally stand for the factors that are considered for vulnerability assessment. These are explained in the ‘‘Methodology’’ section.
Gogu and Dassargues (2000) have done a complete overview of several existing methods on groundwater vulnerability assessment, especially on the way these methods have been developed and applied and the future challenges on vulnerability mapping. The present man-uscript completes the former work because it is not a theoretical discussion on the different methods of vul-nerability mapping, but a practical application of several methods in a pilot site with climatic, geological and hydrogeological characteristics representative of Medi-terranean carbonate aquifers poorly karstified. Thus, the results obtained in this manuscript could be potentially useful for vulnerability mapping in this type of aquifer. This work contains new contributions to the paper written by Gogu and Dassargues (2000) in connection with two main aims: (1) to analyze the results obtained by different methods of vulnerability mapping to deter-mine which is the most suitable for diffuse flow car-bonate aquifers in southern Spain and (2) to evaluate how vulnerability depends on the quantity of precipita-tion. The work was carried out in the Torremolinos carbonate aquifer, in the NE part of Sierra de Mijas (Andalusia, S Spain), to the west of the town of Torre-molinos, in the Costa del Sol area (Fig. 1). This aquifer was selected as a Spanish experimental area as part of the COST 620 Action (Zwahlen 2004), because the two conditions mentioned at the beginning of this chapter are present (Andreo et al. 2000,2002): water resources are scarce and there is a potential risk of contamination due to human impact.
Characteristics of the pilot site
The Torremolinos aquifer extends over 56 km2and its topography, like that of most carbonate aquifers in
southern Spain, is very abrupt except at the northern and eastern edges. The average annual temperature is 18.3°C and the rainfall is 630 mm per year, although it is highly irregular in time. Two rainwater patterns can be distinguished: humid years (precipitation >700 mm) and dry years (<500 mm). The average net recharge of the Torremolinos aquifer is approximately 20 hm3 per year (Andreo1997).
Two major units of relief can be distinguished in the study area (Fig.1): the Alpine massif of Sierra Mijas and, to the North and East, the Neogene-Quaternary basin of Ma´laga. From a geological point of view, Sierra de Mijas comprises two lithological formations (Andreo1997): the lower one is made up of Paleozoic metapelites with a thickness of 400 m and the upper formation is formed by Triassic marbles with metapelite intercalations, 600 m thick, which constitute a carbonate aquifer. The Neogene-Quaternary rocks are mainly Pliocene marls with a thickness of hundreds of meters and a tabular disposition. From a hydrogeological standpoint, the groundwater flow is from W to E, where the main points of natural discharge, the Torremolinos springs, are located at an elevation of 55 m a.s.l. (Fig. 1). Analysis of the hydro-geological (hydrodynamic and hydrochemical) data from this system shows that it is a fissured aquifer with a diffuse flow behavior (Andreo1997; Andreo et al.1997; Andreo and Carrasco 1999): the response (hydrodi-namic, hydrochemical and isotopic) to the rainfall is very slow; it can occur several months after precipita-tion. The average hydraulic conductivity is 10 m per day in the marbles and 10)2m per day in the Pliocene-Quaternary sediments (Andreo1997). Piezometric vari-ations occur very slowly due to the great inertia of the system, as a result of the extended travel time of pre-cipitation. Nevertheless, the Torremolinos aquifer is intensively exploited (nearly 30 hm3per year) by means of pumping for water to supply the tourist area of Costa del Sol; and, consequently, variations of up to 70 m in the water table have been measured depending on the climatic conditions (Andreo et al. 2000): normally the spring is dry and the piezometric level decrease because of the exploitation but after wet years the water table can rise and emerge water in the spring.
In Sierra de Mijas, karst features are poorly devel-oped, mainly karren field. Two large soil zones can be distinguished in the study area: one is composed of soils with a thickness of less than 70 cm, namely anthrosol, regosol and calcisol on the northern and eastern borders of the study area, where marls outcrop. The rest of the area is mainly uncovered or composed of leptosol, less than 30 cm thick, over the marbles. The land use is mainly as woodland (brushwood and reforested pines), while the rest (15%) is dedicated to non-irrigated farming and to urbanization on the northern and southern borders of the aquifer, especially in recent years. Human impact is greater at the borders of the
aquifer due to activities which are potentially contami-nant (urban development, landfill filling old quarries, petrol stations, roads and cemeteries), occuring over the marbles without any important protection measures (Andreo et al. 2002; Vı´as2003).
Methodology
In the framework of COST 620 Action, intrinsic vul-nerability is the term used to define the susceptibility of groundwater to contaminants generated by human activities. It takes into account the inherent geological, hydrological and hydrogeological characteristics of an
area, but is independent of the nature of the contami-nants (Daly et al.2002; Zwahlen2004).
Overlay analytical functions and non-coincident spatial variables were incorporated into vulnerability mapping, and thus application of the Geographic Information System (GIS) was an essential element. Arc/info and Arc/view GIS were used in this study.
Five methodological steps were followed: (1) the data necessary to evaluate the different parameters or factors were compiled using several approximation scales, such as the 1/10,000 scale of the topographic map, the 1/25,000 original scale of the geological map, and the 1/100,000 scale of the soil map created by the Lucdeme Project (ICONA 1991); (2) a map for each factor was Fig. 1 Geographical location of
the Torremolinos carbonate aquifer in the eastern part of the Sierra de Mijas. Hydrogeologi-cal map and cross-sections of Torremolinos aquifer (after Andreo1997). Legend:1
Plio-Quaternary sediments,2
Marbles,3Metapelites,4
Unconformity,5Fault,6
Strike-slip fault7Normal fault,
8Reversed anticline,9Springs and its reference,10Well and its reference,11Water table,12
watercourse,13Possible trans-fer of groundwater
digitized by georeferencing on a digital tablet or by automatic interpolation; (3) the topologic relations of connectivity, superposition without connection, influ-ence, proximity, coincidence and inclusion were established and the spatial analysis of the coverage was determined, with special emphasis on the overlay; (4) the vulnerability index for each surface unit was calculated applying different methods; and (5) vulnerability rank-ings or classes were established for the catchment area, taking into account the values of the vulnerability indi-ces.
In the Torremolinos aquifer, four methods were ap-plied to evaluate vulnerability (Table1) and two hydrological years were considered for vulnerability mapping (Vı´as 2003): 1996–1997, a humid year (1,213 mm rainfall), which occurred after other humid years (1995–1996, 1,139 mm) and thus the aquifer was in a high water table condition, and 1994–1995, a dry hydrological year (230 mm) with low water table con-ditions. The average difference in the piezometric height between the two years was 40 m.
DRASTIC (Aller et al.1987) is a weighting and rat-ing method that assesses vulnerability by means of seven parameters: depth of groundwater (D), net recharge (R), aquifer media (A), soil media (S), topography (T), im-pact of the vadose zone media (I) and the hydraulic conductivity of the aquifer (C). In this paper, parameters A, I and C were evaluated taking into account the lithology (Table 1). Each parameter involves several ratings, some (D, T) with a large range and with values
between 1 and 10, whereas the weights were between 1 and 5. In every case, however, the maximum values correspond to the highest vulnerability. The vulnerabil-ity index is obtained by multiplying the value of the parameters by a weighting coefficient, according to the following equation:
I ¼5Dþ4Rþ3Aþ2Sþ1Tþ5Iþ3C ð1Þ
The DRASTIC method establishes a standard clas-sification of the vulnerability index by eight materials, although a ranking has been made, from ‘‘very low’’ to ‘‘very high’’ with fire degrees of vulnerability.
GOD (Foster 1987) is a rating system method that assesses vulnerability by means of three variables: groundwater occurrence (G), overall lithology of aquifer (O) and depth to groundwater table (D). This method uses fewer parameters than DRASTIC, although two of them (Gand D) also depend on the lithology (Table 2), and the range of values for each rating is short, varying from 0 (minimum vulnerability) to 1 (maximum vul-nerability). The final index is obtained from the formula:
I ¼GOD ð2Þ
The value of the index may vary from 0 to 1 and five vulnerability classes are differentiated by the method.
Aquifer vulnerability index (AVI) is an analogical relation or numerical method that uses two parameters (Van Stempoort et al.1993): the thickness of each sed-imentary layer above the uppermost saturated aquifer
Table 1 Rating values of the vulnerability parameters for DRASTIC method
D(depth to groundwater)
Range (feet) 0–5 5–15 15–30 30–50 50–75 75–100 +100
Rating 10 9 7 5 3 2 1
R(recharge)
Range (inches) +10 7–10 4–7 2–4 0–2
Rating 9 8 6 3 1
A(aquifer media)
Range Dolomitic marbles Calcareous marbles
Rating 7 7
S(soil media)
Range Leptosols Regosols Anthrosols Calcisols
Rating 10 9 6 3
T(topography)
Range (%) 0–2 2–6 6–12 12–18 +18
Rating 10 9 5 3 1
I(impact to zone vadose)
Range Dolomitic marbles Calcareous marbles Metapelites Metapelites interbedded Marls
Rating 7 7 5 5 1
C(conductivity hydraulic)
Range Dolomitic marbles Calcareous marbles
(d) and the estimated hydraulic conductivity (k) of each of these sedimentary layers. This method does not con-sider ratings and/or weights. The index is determined from the relation between the two parameters, taking into account variations of an order of magnitude (Ta-ble 3), according to the following equation:
AVI¼Xd
k ð3Þ
The AVI method also establishes five classes of vulner-ability, which reflect the variations of the index equiva-lent to an order of magnitude.
EPIK (Doerfliger et al. 1999) is a parameter weighting and rating method especially developed for karst aquifers to protect water supply sources (springs and wells). This method does not consider parameters depending on time (i.e. rainfall, recharge) but only the intrinsic parameters of the aquifer (Table4): presence of epikarst (E), the characteristics of the protective cover (P), the infiltration conditions (I) and the karst network development (K). A protection factor (Fp) is
calculated by summing the values of parameters E,P,I and K (each between 1 and 4) and applying a weight varying from 1 to 3. In this case, unlike other methods, lower values of Fp correspond to higher vulnerability,
because the vulnerability index is converse to the pro-tection factor:
Fp¼3Eþ1Pþ3Iþ2K ð4Þ
The protection factor is divided into four vulnerability classes, from ‘‘low’’ to ‘‘very high’’.
Results
After using each method to evaluate the vulnerability indexes, these are expressed within a range of five intervals, from ‘‘very high’’ to ‘‘very low’’, to stan-dardize the legends of the figures and to obtain com-parable vulnerability maps for different rainfall conditions.
The map obtained by the DRASTIC method (Fig.2) shows a ‘‘low’’ degree of vulnerability for the marl rocks and a ‘‘moderate’’ vulnerability for the marbles. The areas of ‘‘very low’’ vulnerability are due to the existence of clay soils (mainly calcisol according FAO classifica-tion) with low hydraulic conductivity. Only small dif-ferences in vulnerability were found between a humid year (Fig.2a) and a dry one (Fig.2b) but, in any case, these are due to variations in the piezometric level and, especially, to the variations in the recharge. Thus, the vulnerability is lesser, although not greatly so, in a dry year than in a humid year. Moreover, the recharge is highly dependent on the hydraulic conductivity of the unsaturated zone and, therefore, on the lithology.
Using the GOD method, the Torremolinos aquifer has a ‘‘moderate’’ degree of vulnerability in the car-bonate rocks and a ‘‘low’’ degree in the Pliocene-Qua-ternary materials (Fig. 3a, b). The variations in vulnerability between the maps corresponding to humid and dry years are also due to differences in the depth of the water table. Thus, in several zones near the borders of the aquifer, where human activities are more evident (Andreo et al. 2002, Vı´as 2003), the vulnerability in-creases from a ‘‘moderate’’ to a ‘‘high’’ degree if a dry year is compared with a humid year, because the water table rises closer to the surface.
Using the AVI method, ‘‘high’’ and ‘‘very high’’ de-grees of vulnerability are found in the marbles and ‘‘low’’ and ‘‘very low’’ degrees in the Pliocene-Quater-nary marls (Fig.4a, b). The hydraulic conductivity of the unsaturated zone (especially the soil) is the param-eter that most dparam-etermines vulnerability, but in the Tor-remolinos aquifer, soil is practically absent and this fact greatly increases the vulnerability. Within the marbles, the vulnerability varies, depending especially on the thickness of the unsaturated zone and, thus, on the Table 2 Rating values of the vulnerability parameters for GOD method
G(groundwater occurrence) Range Pliocene Marls (semi-confined aquifer)
Triassic marbles (unconfined aquifer)
Rating 0.3 1
O(overall lithology of aquifer) Range Triassic marbles
Rating 0.9
D(depth to groundwater) Range (meters) 0–2 2–5 5–10 10–20 20–50 50–100 >100
Rating 1 0.9 0.8 0.7 0.6 0.5 0.4
Table 3 Rating values of the vulnerability parameters for AVI method
Parameters Unsaturated layers
Pliocene marls Triassic marbles Soils
d(thickness)
Order magnitude (m) 102 102 1
k(hydraulic conductivity)
depth of the water table. The vulnerability increases by one degree where the water table is less than 100 m deep (Vı´as 2003). The vulnerability distribution in the map for a dry year is similar to that obtained by the GOD method for a humid year, although the degrees of vul-nerability are higher in the map deduced by the AVI method.
Finally, the EPIK method permits us to obtain only one vulnerability map (Fig.5) because it only considers the intrinsic parameters of the aquifer; thus the vulner-ability map is the same for humid and dry climatic conditions. In general, this method gives a similar vul-nerability distribution to that obtained with the DRASTIC and GOD methods, that is, a ‘‘low’’ degree Table 4 Rating values of the vulnerability parameters for EPIK method
E(epikarst) Range Highly fractured in quarries and roads Rest of catchment area
Rating E1=1 E3=4
P(protection cover) Range Leptosols and soils on quarries
Regosols, anthrosols, calcisols
Soils on layers that have very low hydraulic
conductivity and thickness >400 m
Rating P1=1 P2=2 P4=4
I(infiltration) Range (out of catchment area)
Areas collecting runoff water (buffer 50 and 100 m) and slopes
feeding those areas (slope higher
than 10% for cultivated sectors and 25% for meadows and pastures)
Rest of area
Rating I3=3 I4=4
K(karst network) Range Triassic marbles
Rating K3=3
Fig. 2 Vulnerability maps for 1996–1997 humid hydrological year (a) and 1994–1995 dry hydrological year (b) obtained by the DRASTIC method
of vulnerability in the Pliocene-Quaternary materials and a ‘‘moderate’’ degree of vulnerability in the Triassic marbles (Table 5). However, there are exceptions in
zones where an important degree of fracturing occurs, such as in quarries or in the slopes of the roads, which raises the vulnerability from ‘‘moderate’’ to ‘‘high’’; Fig. 3 Vulnerability maps for
1996–1997 humid hydrological year (a) and 1994–1995 dry hydrological year (b) obtained by the GOD method
Fig. 4 Vulnerability maps for 1996–1997 humid hydrological year (a) and 1994–1995 dry hydrological year (b) obtained by the AVI method
where infiltration conditions are supposed favorables, vulnerability becomes ‘‘very high’’.
Discussion
Each method of vulnerability evaluation results in a different map, although they all show the same dis-tribution of spatial variability, i.e., a smaller degree of vulnerability of the Pliocene-Quaternary marls with respect to the Triassic marbles. To compare and dis-cuss the results of the four methods, the authors cal-culated the percentage of the surface of the study area assigned to each degree of the vulnerability mapping method (Table 2).
The vulnerability classed as ‘‘moderate’’ by the DRASTIC, GOD and EPIK methods (Figs. 2, 3 and
5) basically concurs with the presence of marbles (Fig.1), approximately 85% of the study area. The AVI method, however, evaluates near 75% of the aquifer marbles as presenting a ‘‘High’’ degree of vulnerability and assigns a ‘‘very high’’ degree to a further 10% (Fig.4 and Table2). Therefore, the AVI method reports a higher degree of vulnerability of the aquifer than do the other methods; this seems improbable because no evidence of contamination in the groundwater has been detected in several decades, despite the existence of various human activities
potentially contaminant developed over the marbles (landfill, waste pipe lines in urbanisations, petrol sta-tion). Thus, a ‘‘Moderate’’ vulnerability for the mar-bles is coherent with the characteristics of the aquifer (Andreo 1997; Andreo et al. 1997, 2000; Andreo and Carrasco 1999): high thickness of the unsaturated zone, relatively low hydraulic conductivity, strong inertia and, therefore, diffuse flow behavior.
The highest degrees of vulnerability, ‘‘high’’ and ‘‘very high’’, obtained with the GOD and AVI methods respectively, coincide with zones where anthropic pres-sure is highest because of the presence of built-up areas, landfill, roads and crops, all of which increase the risk of contamination. This agrees with the results obtained previously by the SINTACS method (Longo et al.2001). In any case, these aspects underline the importance of contamination vulnerability mapping in carbonate massifs in southern Spain. Vulnerability mapping con-stitutes an important tool for environmental manage-ment to preserve the quality of groundwater. In fact, after this pilot experience in the Torremolinos aquifer, landfills and cemeteries have been closed.
The highest vulnerability classes in the map obtained by the EPIK method correspond to the quarries, which are considered as ‘‘artificial dolines’’, and the slopes of the roads where the absence of protective cover and the high fracturation provoke an increase in infiltration (after Doerfliger et al.1999). However, in these ‘‘artifi-Fig. 5 Vulnerability map
ob-tained by the EPIK method
Table 5 Percentage of surface area according to the degree of vulnerability calculated by the different methods for humid and dry climatic conditions
Vulnerability DRASTIC GOD AVI EPIK
Humid Dry Humid Dry Humid Dry
Very high 0 0 0.1 0 10 2 1
High 0.1 0 2 0.1 76 84 2
Moderate 86 83 84 86 0 0 83
Low 13 16 14 14 14 14 14
cial dolines’’ swallow holes, which permit rapid infil-tration into the aquifer, do not exist. The Torremolinos system has a diffuse flow behavior (Andreo et al.1997,
2000), and infiltration normally occurs slowly because the marbles are highly fractured but poorly karstified.
The influence of rainfall variations and therefore of the water table is clear in the maps obtained by the AVI and GOD methods, in contrast to the DRASTIC vul-nerability maps, which are less influenced by the groundwater table variations arising from differences between years of higher or lower rainfall. This could be because the rating range of parameters Dand R in the DRASTIC method are not well suited to the Torremo-linos aquifer. In any case, it is clear that the interannual variations of rainfall that normally occur in the Medi-terranean climate affect the aquifer’s vulnerability to contamination. Thus, in a humid year the vulnerability is higher than in a dry year because the groundwater table rises closer to the surface.
The similarity in the distribution of the vulnerability maps obtained by the different methods is due to the importance of the lithology, and to the influence of the final classification of the intervals used in each method. These aspects are particularly relevant in the DRASTIC method, which evaluates several parameters by means of the lithology; moreover, it does not establish a universal classification in intervals and consequently is exposed to subjective interpretations (Vı´as2003).
The assignation of ratings in the methods involves a certain degree of subjectivity that is difficult to eliminate. In this sense, the DRASTIC method is the most sub-jective, because of the wide range of the rating of some parameters. The EPIK and GOD methods involve a more selective choice of variables and a reduction of the ratings; as a result, the risk of subjectivity, with respect to the ratings, is smaller. The AVI method does not have any subjective element, as it does not rate the parameters. Finally, concerning the scale, the EPIK method needs a more detailed scale to obtain theEand Iparameters, and so it is advisable to use a larger scale (1/25,000) to delimitate the karst landforms (mainly karren field in Sierra de Mijas) with influence in the vulnerability map. The DRASTIC, GOD and AVI methods do not need such a detailed scale to evaluate vulnerability, and a moderate scale (1/50,000) could be very effective, be-cause the differences in vulnerability depend on the geology and the groundwater table depth, both of which can be determined at a scale of 1/50,000.
Conclusions
For the DRASTIC, GOD and AVI methods of vulner-ability mapping, the parameters related to geology, and
especially to the lithology, are most relevant, while the depth of the groundwater table has less influence. The latter does, however, determine variations in vulnera-bility between humid and dry years, especially when applying the AVI and GOD methods. In the EPIK method, the vulnerability is defined by the characteris-tics of the protection cover and, to a lesser degree, by the presence of highly fractured zones.
The higher or lower number of parameters used for each method does not establish significant differences in the final vulnerability map. Thus, with the DRASTIC method, which uses seven parameters, it is possible to obtain a vulnerability mapping similar or even with less class than the one obtained by the GOD method, which only uses three variables.
The kind of assessment model used (parametric or analogical relations) influences the different degrees of vulnerability. Thus, the AVI method reports a higher vulnerability than that found with the parametric methods (GOD, EPIK and DRASTIC) but this is in disagreement with the hydrogeological knowledge available.
The high vulnerability zones deduced by EPIK for very fractured areas (quarries and the slopes of roads) are in contradiction with the very low karstification in such areas and, consequently, the slow infiltration into the marbles because of the diffuse flow behavior of the aquifer.
Vulnerability mapping with the DRASTIC, GOD and AVI methods seems very useful for land use man-agement, using moderate and small scales that provide an overall view. However, there is a need for methods with a larger scale to establish protection zones for the aquifers. For carbonate aquifers, the EPIK method of-fers better performance for the establishment of pro-tection zones.
From the results of this work, the authors conclude that the GOD method could be adequate for vulnera-bility mapping in poorly karstified carbonate aquifers in southern Spain, at least at small–moderate scales. If a large scale is used, or if the study is performed in areas where karstification is well developed, it could be useful to compare the results of the GOD method with those obtained by the EPIK method, or other methods created specifically for karst aquifer in the framework of COST Action 620, before adopting a vulnerability map for groundwater protection.
Acknowledgements This paper is a contribution to European COST Action 620, to the projects IGCP 448 of UNESCO, REN2002-01797/HID and REN2003-01580/HID of the DGES and to the Research Groups RNM 308 and HUM 776 of the Junta de Andalucı´a. The comments of Prof. Alain Dassargues (University of Lie`ge, Belgium), Prof. Jacques Mudry (University of Franche-Comte´) and an anonymous reviewer are much appreciated.
References
Adams B, Foster S (1992) Land surface zoning for groundwater protection. J Inst Water Environ Manage 6:312–320 Albinet M, Margat J (1970) Cartographie
de la vulnerabilite´ a la pollution des nappes d’eau souterraine. Bull BRGM 2e se´r. 3(4):13–22
Aller L, Bennett T, Lehr JH, Petty RH, Hackett G (1987) DRASTIC: A stan-dardised system for evaluating ground-water pollution potential using hydrogeologic settings, US EPA Report 600/2-87/035, Robert S. Kerr Environ-mental Research Laboratory, Ada, Oklahoma, p 622
Andreo B (1997) Hidrogeologı´a de acuı´f-eros carbonatados en las Sierras Blanca y Mijas, Cordillera Be´tica, Sur de Espan˜a. Service of publications of the University of Ma´laga, p 490
Andreo B, Carrasco F (1999) Application of geochemistry and radioactivity in the hydrogeological investigation of car-bonate aquifers (Sierras Blanca and Mijas, southern Spain). Appl Geochem 14:283–299
Andreo B, Carrasco F, Sanz de Galdeano C (1997) Types of carbonate aquifers according to the fracturation and the karstification in a southern Spanish area. Environ Geol 30(3/4):163–173 Andreo B, Carrasco F, Dura´n JJ,
Ferna´n-dez del Rı´o G, Linares L, Lo´pez-Geta JA, Mayorga R, Vadillo I (2000) Hydrogeological investigations for groundwater exploitation in the Sierras Blanca and Mijas (Ma´laga, southern Spain). Hydroge´ologie 3:19–33 Andreo B, Vı´as JM, Perles MJ, Carrasco F,
Vadillo I, Jime´nez (2002) Ensayo met-odolo´gico para la proteccio´n de aguas subterra´neas en acuı´feros carbonatados. Aplicacio´n al sistema de Torremolinos. Jornadas sobre Presente y futuro del agua subterra´nea en Espan˜a y la Directiva Marco Europea. Zaragoza (Spain). IAH-Spanish Chapter, pp 147–153
Civita M (1994) La carta della vulnerabilita` degli acquiferi all’inquinamiento. Pit-agora, Bologna
Daly D, Drew D (1999) Irish methodolo-gies for karst aquifer protection. In: Beck B (ed) Hydrogeology and engi-neering geology of sinkholes and karst. A.A. Balkema, Rotterdam, pp 267–272 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 10:340–345 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
European Water Framework Directive (2000) Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. European Commission, Brussels
Foster S (1987) Fundamental concepts in aquifer vulnerability, pollution risk and protection strategy. In: Van Duijven-booden W, Van Waegeningh HG (eds), Vulnerability of soil and groundwater to pollutants. Proc Inf TNO Comm Hydrol Res, The Hague 38:69–86 Gogu RC, Dassargues A (2000) Current
and future trends in groundwater vul-nerability assessment. Environ Geol 39(6):549–559
Goldscheider N, Klute M, Sturm S, Ho¨tzl H (2000) The PI method—a GIS-based approach to mapping groundwater vulnerability with special consideration on karst aquifers. Z Angew Geol 46(3):157–166
Ho¨tzl H (1996) Scientific basis for karst groundwater protection: guidelines and regulations. In: Antigu¨edad I (ed) Pro-ceedings of the conference on ground-water resources on karst regions, Vitoria (Spain), pp 147–157
ICONA (1991) LUCDEME project. Soil map of Andalusia. Sheet 1066 (Coı´n) and 1067 (Ma´laga)
Longo CA, Andreo B, Carrasco F, Cucchi F, Vı´as JM, Jime´nez P (2001) Compar-ison of two contamination vulnerability maps obtained by the SINTACS meth-od in two carbonate aquifers (S Spain). In: Mudry J, Zwahlen F (eds) Proceed-ings of the 7th conference on Limestone Hydrology, Besanc¸on, pp 233–236 Robins N, Adams B, Foster S, Palmer R
(1994) Groundwater vulnerability mapping: the British perspective. Hydroge´ologie 3:35–42
Van Stempoort D, Ewert L, Wassenaar L (1993) Aquifer Vulnerability Index (AVI): A GIS compatible method for groundwater vulnerability mapping. Can Water Res J 18:25–37
Vı´as JM (2003) Vulnerabilidad y peligro de contaminacio´n en el acuı´fero carbona-tado de Torremolinos (Ma´laga). Servi-cio PublicaServi-ciones de la DiputaServi-cio´n de Ma´laga, p 180
Vrba J, Zaporozec A (1994) Guidebook on mapping groundwater vulnerability. International Association of Hydroge-ologists, International Contributions to Hydrogeology, 16, Verlag Heinz Heise, Hannover
Zwahlen F (ed) (2004) Vulnerability and risk mapping for the protection of car-bonate (karst) aquifers, final report (COST Action 620). European Com-mission, Directorate-General XII Sci-ence, Research and Development. Brussels, p 297
