Groundwater Vulnerability Assessment for Organic Compounds: Fuzzy Multicriteria Approach for Mexico City

DOI: 10.1007/s00267-005-0059-8

ENVIRONMENTAL ASSESSMENT Groundwater Vulnerability Assessment for Organic Compounds: Fuzzy Multicriteria Approach for Mexico City MARISA MAZARI-HIRIART* Departamento de Ecologa de la Biodiversidad Instituto de Ecologa Universidad Nacional Autnoma de Mxico Tercer Circuito Exterior Ciudad Universitaria Coyoac n, 04510 Mxico, D.F., Mxico GUSTAVO CRUZ-BELLO Instituto Nacional de Investigaciones Forestales, Agrcolas y Pecuarias Carmen Coyoac n, 04110 Mxico, D.F., Mxico ´ RQUEZ-TAPIA LUIS A. BOJO ´ REZ-MARUSICH LOURDES JUA Programa Universitario de Medio Ambiente Universidad Nacional Autnoma de Mxico Circuito Exterior Ciudad Universitaria Coyoac n, 04510 Mxico, D.F., Mxico ´ PEZ GEORGINA ALCANTAR-LO Direccin General de Poltica Ambiental e Integracin Regional y Sectorial Secretara de Medio Ambiente y Recursos Naturales Tialpan, 14210 Mxico, D.F., Mxico LUIS E. MARI´N Departamento de Recursos Naturales Instituto de Geofsica Universidad Nacional Autnoma de Mxico Circuito Exterior Ciudad Universitaria Coyoac n, 04510 Mxico, D.F., Mxico

ABSTRACT / This study was based on a groundwater vulnerability assessment approach implemented for the Mexico City Metropolitan Area (MCMA). The approach is based on a fuzzy multicriteria procedure integrated in a geographic information system. The approach combined the potential contaminant sources with the permeability of geological materials. Initially, contaminant sources were ranked by experts through the Analytic Hierarchy Process. An aggregated contaminant sources map layer was obtained through the simple additive weighting method, using a scalar multiplication of criteria weights and binary maps showing the location of each source. A permeability map layer was obtained through the reclassification of a geology map using the respective hydraulic conductivity values, followed by a linear normalization of these values against a compatible scale. A fuzzy logic procedure was then applied to transform and combine the two map layers, resulting in a groundwater vulnerability map layer of five classes: very low, low, moderate, high, and very high. Results provided a more coherent assessment of the policy-making priorities considered when discussing the vulnerability of groundwater to organic compounds. The very high and high vulnerability areas covered a relatively small area (71 km2 or 1.5% of the total study area), allowing the identification of the more critical locations. The advantage of a fuzzy logic procedure is that it enables the best possible use to be made of the information available regarding groundwater vulnerability in the MCMA.

ERNESTO SOTO-GALERA Instituto Mexicano del Petrleo San Bartolo Atepehuacan, 07730 Mxico, D.F., Mxico

KEY WORDS: Analytic hierarchy process; Aquifer; Contamination; Geographic Information System; Organic compounds; Megacity Published online January 24, 2006. *Author to whom correspondence should be addressed; email: [email protected]

Environmental Management Vol. 37, No. 3, pp. 410–421

Urbanization poses major challenges for water planning and management, especially for groundwater resources in developing countries. The United Nations (2002) reported that there are 389 cities around the world with a population of 1 million or more, and the majority of these lie in regions experiencing mild to severe water stress (UNESCO–WWAP 2003). Moreover, in 2000, 11 megacities (i.e., urban conglomerates with ª 2006 Springer Science+Business Media, Inc.

Groundwater Multicriteria Approach for Mexico City

more than 8 million inhabitants; Fuchs 1999) that depended on groundwater were located in developing countries (Howard and Gelo 2003). By the year 2015, it is estimated that there will be 27 megacities in developing countries (UN 2002). Because groundwater is the only source of water for about half of the worldÕs population (Llamas and Custodio 2003), it is reasonable to surmise that rapid urban growth will only accelerate groundwater exploitation in such megacities. Therefore, as more and more of the worldÕs population will be living in urban zones within the not too distant future, aquifer protection is increasingly becoming a priority in public policy and should be considered critical in developing countries. Groundwater vulnerability mapping is an essential component for the design of aquifer protection and management strategies (Daly and others 2002). Groundwater vulnerability refers to the relative ease with which imposed contaminant loads can migrate to an aquifer under a given set of criteria such as land use, contaminant characteristics, and hydrogeological conditions (Tait and others 2004; US EPA 1993a). The most widely adopted approach for mapping groundwater vulnerability consists of implementing a multicriteria model using a geographic information system (GIS), the so-called overlay/index method, to produce a relative ranking of groundwater vulnerability in different zones within a study region (Gogu and Dassergues 2000; US EPA 1993b). One advantage of this approach is the increased objectivity it produces for identifying the necessary priorities for aquifer protection. One downside of conventional groundwater vulnerability mapping techniques, such as DRASTIC, EPIK, and SEEPAGE (Gogu and Dassergues 2000), is that they overlook the location and characteristics of contaminant sources. However, contaminants such as organic compounds have been regarded as a serious health threat (Thomas 1990; Thornton 2000), to such an extent that they have been monitored in the developed world to prevent their downward migration (Pye and others 1983; Schwille 1988; Westrick 1990). Similar programs are yet to be implemented in developing countries, even though it is already known that aquifer contamination by organic chemicals might be practically irreversible (Mackay and Smith 1993; Pankow and Cherry 1996). Therefore, in developing countries, it is of the utmost importance to make use of mapping approaches to support decision-making with respect to the vulnerability of groundwater pollution (Lobo-Ferreira and Oliveira 1997), especially by organic compounds. Responding to the dire need of a groundwater vulnerability assessment for the Mexico City Metropolitan

411

Area (MCMA), a GIS-based fuzzy multicriteria approach was developed using readily available information. This alternative approach was necessary because the nature of available data precluded the implementation of a conventional model. More precisely, information on contaminant loads, aquifer attributes, and hydrogeological conditions for the region consisted of rather anecdotal, qualitative, and schematic descriptions. Due to the lack of accurate data, local experts were consulted in order to identify proxy variables for pollutant sources and aquifer sensitivity, which were then integrated into the fuzzy multicriteria procedure (Malczewski 1999). Results identified zones that were classified according to their priority for aquifer protection. The zones were determined to be very high or high vulnerability areas, which encompassed 71 km2, or moderate, low, and very low vulnerability areas, which in total covered of 4545 km2. This research shows how fairly simple approaches can be used to develop sound and useful information for policy-making related to aquifer protection in developing countries, even when only limited data are available.

Study Area The MCMA is a megacity of 18 million inhabitants that extends over 16 administrative units of the Federal District and 27 neighboring municipalities in the State of Mexico. It encompasses 4616 km2, of which 24% is urban (INEGI 1993) (Figure 1). From the geomorphologic viewpoint, the MCMA is located on a high, elevated lacustrine plain (2240 m above sea level), which for the most part has been artificially desiccated. The clays of the lacustrine plain form an aquitard of depths from 30 m, in most of the urban area, to 300 m in the south. The depth of the aquifer varies from 80 to 300 m (Lesser and Corte´s 1998). The specific characteristics of the aquifer of the MCMA are a matter of debate because hydrogeologists are yet to reach consensus on its composition and specific hydraulic conductivities. Ortega and Farvolden (1989) described it as an aquifer system of both volcanic and sedimentary origin, whereas Rudolph and Frind (1991) and Warren and Rudolph (1997) described it as an ‘‘enormous granular semiconfined aquifer.’’ In contrast, Lesser and others (1990) considered that the regional aquifer should be divided into three subaquifers of different origin (lacustrine deposits, volcanic rocks and pyroclastics, and alluvial deposits). Birkle and others (1995) and Marı´n and others (2002) described the MCMA aquifer to be regional that can be either confined or semiconfined.

412

M. Mazari-Hiriart and others

Figure 1. Mexico City Metropolitan Area including 16 administrative units of the Federal District and 27 neighboring municipalities in the State of Mexico (outer boundary shown by the dark line) and the urban area (shown in black).

Despite this continuing debate, it is clear that the MCMA aquifer is under enormous stress due to urban sprawl, groundwater extraction, and poor wastewater management. Urban sprawl is a major factor associated with an increased risk of groundwater contamination in the MCMA. Widespread land-cover transformation is causing an increase in the number of potential pollutant sources in groundwater recharge areas of the MCMA. Until 1930, the MCMA was restricted to the original downtown area. From the 1930s to 1950s, peripheral expansion took place as households were built mainly in the southern and western sections of the MCMA. From the 1960s to 1980s, urban sprawl accelerated as low-income housing converged on dry lacustrine land in the east and northeast sections, middle-income housing and industry converged on the northern sections of the MCMA, and residential subdivisions converged on mountain slopes in the south and southwest. Groundwater extraction has increased along the sprawl of the MCMA. Currently, it supplies 72% of the total demand (72.5 m3/s), and continued pumping since the mid-1940s has caused a differential in land subsidence from 6 cm/year in the downtown area to 30 cm/year in the eastern and southern sections of the

MCMA (Jime´ nez-Cisneros and others 2004). Records show that since the 1900s, some downtown locations have subsided by 9 m (Marsal and Mazari 1990). By 2010, it is expected that the population will reach 21 million. If the current trends of water consumption continue, groundwater abstraction will have to increase to meet the expected demand of 80 m3/s (Merino 2000). Wastewater treatment is rather limited in the MCMA (about 15% of the total discharge of 45 m3/s), so it is discharged out of the basin in Mexico through an open canal (known as the ‘‘Gran Canal’’) that has been in operation since the early 1900s and a ‘‘Deep Sewer System’’ (‘‘Drenaje Profundo’’), which was built in the 1970s. As groundwater extraction has caused land subsidence in the MCMA, the open canal has lost its original downward gradient and, currently, auxiliary pumping is needed to discharge storm runoff, industrial wastewater, and domestic sewage out of the basin. The risk of downward migration of surface contaminants to the aquifer is high in areas where the aquitard has been breached by drilling, excavations, pile driving, and large underground infrastructures, such as the Metro and the Deep Sewer System (Mazari and Mackay 1993; Mazari and others 1996). Last but

Groundwater Multicriteria Approach for Mexico City

413

Figure 2. Permeability as reported by Mooser and others (1996 and personal communication).

not least, groundwater extraction has reversed the natural groundwater flow that created springs; thus the vulnerability of the aquifer is now higher because pollutants might migrate downward (Lesser and others 1986; Mooser and Molina 1993; Ortega and Farvolden 1989; Durazo and Farvolden 1989).

Methods A spatial multicriteria analysis (Malczewski 1999) was implemented to depict the areas prone to groundwater contamination by organic compounds in the MCMA. This approach involved three steps: (1) identification of evaluation criteria and their weighted importance, (2) development of evaluation criteria map layers, and (3) implementation of decision rules or aggregation functions. Step 1 involved both the identification and ranking of contaminant sources and the spatial distribution of the hydraulic conductivity of geological materials in the MCMA according to Mooser and others (1996), to be used as a proxy of the hydrogeological attributes of the geologic materials (Figure 2). The weight of the contaminant sources was obtained following the Analytic Hierarchy Process (AHP) (Saaty 1980). In short, the AHP is a systematic procedure used to derive weighted priorities for decision criteria using expert judgment [because the AHP is a

widely used approach, its formulation will not be repeated here; however, for further details, refer to Banai-Kashani (1989) and Ramanathan (2001)]. Accordingly, the potential contaminant sources were identified from the available literature (DGCOH 1993; IMP 1997; PEMEX 1997) (i.e., point sources, including landfills, fuel deposits, gas stations, and extraction wells, and nonpoint sources, including the primary sewer network, Deep Sewer System, industrial zones, and the urban zone). Extraction wells were considered as contaminant sources because of the risk of downward migration of contaminants when not pumped. Likewise, the latter source was included to depict areas lacking sewage or pavement, as well as the effect of contamination by runoff from roads to either clear or vegetated land within the city. The next step was to distribute questionnaires listing the potential contaminant sources to a group of 20 well-known specialists in the areas of environmental sciences, engineering, geology, soils, and groundwater. Specifically, the specialists were asked to perform the required pairwise comparisons among the contaminant sources in the 1 to 9 SaatyÕs (1980) importance scale. Responses were received from eight specialists, and the respective pairwise comparisons were aggregated through the geometric mean method. These aggregated values were then used to derive the weight of each potential contaminant source using the left eigenvector method.

414

M. Mazari-Hiriart and others

Table 1. Geological formations, permeability, and hydraulic conductivitiy for the MCMA Geologic formation or rock type Lacustrine clays Unweathered tuffs Unfractured rhyolites, andesites, and dacites Quaternary domes Undifferentiated volcanic rocks Alluvium Fractured lavas and tuffs

Permeability (%)

K (m/s) )9

5 · 10 ) 5 · 10)12 5 · 10)6 ) 5 ·10)7 7 · 10)6 1 · 10)5 2.4 · 10)5 4.8 · 10)5 ) 9 · 10)5 6 · 10)5

0 2 3 5 10 20 25

Sources: Mooser and others 1996; Marsal and Mazari 1969; Va´ zquez 1995; Ortega and Farvolden 1989; Rudolph and others 1991; Warren and Rudolph 1997.

Step 2 involved the development of a set of raster map layers (pixel size of 1 ha) in a GIS. The layers included in the GIS were the following: geology (Mooser and others 1996) and urban cover, generated from the interpretation of two LANDSAT TM satellite images (26/46 and 26/47) from March 1997 using the ERDAS IMAGE 8.4 (ERDAS 1999), and a binary layer for each of the potential sources of groundwater contamination identified in the previous step, in which 1 indicated the presence of a contaminant source in a pixel and 0 indicated otherwise; the geographic coordinates of both the locations of point contaminant sources and the non–point sources were obtained from secondary data (i.e., DGCOH 1993, 1996; INEGI 1994; IMP 1997; PEMEX 1997; Riosvelasco 1994). The use of geology as a proxy variable for hydraulic conductivity was necessary given the rather qualitative and schematic descriptions of the regional aquifer. Given the lack of geographic specificity of available hydrogeological data, the geology map produced by Mooser and others (1996) was used to obtain permeability values for each geologic formation (Figure 2). These values were generated by Mooser (personal communication) by judging the percentage permeability of each geologic formation and relating the resulting values to hydraulic conductivities as reported by Marsal and Mazari (1969), Ortega and Farvolden (1989), Rudolph and others (1991), Va´ zquez (1995), and Warren and Rudolph (1997), as summarized in Table 1. Step 3 entailed a series of three GIS operations to produce an overall score for each pixel in a groundwater vulnerability map layer. All of the spatial operations in this step were performed using GRASS v.5.8 (Geographic Resource Analysis Support System; USA– CERL 2003). The first operation involved applying the simple additive weighting method (SAW) to aggregate the sources of contamination in a single map layer. The SAW was operationalized in the GIS through a scalar multiplication of the criteria weights obtained in Step 1

and the binary layers generated in Step 2 and then by adding together the resulting map layers. In this way, an aggregated score in the interval [0, 0.5] for each pixel in the layer of potential sources of groundwater contamination was obtained. The second operation consisted of reclassifying the geology layer to produce a vulnerability layer, using the corresponding hydraulic conductivity values. To be consistent with the previous operation, the permeability values (Table 1) of the geology map were normalized using a linear scale transformation (Malczewski 1999), which matched the range of values [0, 0.5] produced by the SAW. The third operation consisted of a fuzzy logic approach to combine the layers of vulnerability and potential sources of groundwater contamination to generate a final layer of groundwater vulnerability. It should be noted that this operation implied that an equal weight was assigned to both map layers, so that the final scale was consistent with that of the [0, 1] interval in SAW. The fuzzy logic operation was carried out in three steps: fuzzyfication, combination, and defuzzyfication (Cox 1994; Bojadziev and Bojadziev 1995; Bojo´rquezTapia and others 2002). For both the vulnerability (v) and the potential sources (s) of groundwater contamination layers, the fuzzification was based on the following five linguistic variables: vk, sk = {VL (very low), L (low), M (moderate), H (high), VH (very high)}, where vk is the value of the kth pixel in the vulnerability layer and sk is the value of the kth pixel in the potential source groundwater contamination layer. The linguistic variables were converted to fuzzy sets (Figure 3) by the following membership functions:  lVL ðxÞ ¼  lL ðxÞ ¼

1 10x þ 1:5

for 0  x  0:05 for 0:05 < x  0:15

10x  0:5 10x þ 2:5

for 0:05 < x  0:15 for 0:15 < x  0:25

Groundwater Multicriteria Approach for Mexico City

Figure 3. Fuzzy sets corresponding to the five linguistic variables used for determining the groundwater contamination hazard of Mexico City (VL = very low; L = low; M = moderate; H = high; VH = very high). The abscissa shows the index score for either the aggregate value of potential sources of contamination or the normalized value of hydraulic conductivity, and the ordinate shows the membership value, l(x), for the corresponding fuzzy sets.

 lM ðxÞ ¼  lH ðxÞ ¼

10x  1:5 10x þ 3:5

for 0:15 < x  0:25 for 0:25 < x  0:35

10x  2:5 10x þ 4:5

for 0:25 < x  0:35 for 0:35 < x  0:45

 lVH ðxÞ ¼

10x  3:5 1

for 0:35 < x  0:45 for 0:45 < x  0:50

where x is the value for vk or sk, as appropriate. The two fuzzy numbers or a-cuts for both vk and sk were obtained by substituting the appropriate values in x in the membership functions. These a-cuts were then used to scale down the original fuzzy sets, which, in turn, were combined by means of a fuzzy additive system to generate a fuzzy solution space P (Kosko 1992). Next, this fuzzy space was defuzzified to a crisp number by means of the composite moment method (Cox 1994). A nominal layer for groundwater vulnerability was then generated by transforming the crisp numbers in the kth pixel, pk, back to the proper linguistic variable, using the following value ranges: VL = {pk Œ0 £ pk £ 0.1}, L = {pk Œ0.1 < pk £ 0.2}, M = {pk Œ0.2 < pk £ 0.3}, H = {pk Œ0.3 < pk £ 0.4}, and VH = {pk Œ0.4 < pk £ 0.5}.

Results The AHP identified industrial zones and landfills as the most important contaminant sources (w = 0.113 and w = 0.088, respectively), followed by primary sewers (w = 0.063), fuel deposits (w = 0.059), the urban zone (w = 0.053), gas stations (w = 0.050), Deep Sewer System (w = 0.042), and extraction wells (w = 0.032).

415

With the exception of fuel deposits, the major proportion of the total area of each contaminant source was found on lacustrine clays, followed by alluvium and undifferentiated volcanic rocks. Remarkably, the major proportion of the area occupied by fuel deposits was found on alluvium (Table 2). The distribution pattern differed among the contaminant sources (Figures 4A–4H). The primary sewer network, the Deep Sewer System, and gas stations converged in the center of the urban zone, which is the oldest part of the MCMA. About half of the extraction wells were located outside of the urban zone, in the eastern part of the MCMA. The majority of the industrial zones were located in the western area, whereas the landfill/dump sites and fuel deposits were found on the periphery of the urban zone. The concentration of groundwater contaminant sources was located in the central zone of Mexico City, where four contaminant sources were identified per square kilometer. In areas incorporated recently into the megacity, there was a lower concentration of contaminant sources. Most of the MCMA (98.4%) presented scores that corresponded to very low, low, and moderate groundwater vulnerability. The three classes tended to be located on the periphery of the MCMA (Figure 5), which accounted for the absence of contaminant sources in a major proportion of their areas (>70%). Very low vulnerability areas were located in 35% of the MCMA. The major proportions of these areas (72%) were located on impermeable lacustrine clays of very low permeability, and the remaining were located on quaternary domes of low permeability (Figures 2 and 5). Urban was the single contaminant source in 28% of the area. Low vulnerability areas were found in 24% of the MCMA. The major proportion of the area (75%) was located on undifferentiated volcanic rocks of moderate permeability and on lacustrine clays in the remaining area (Figures 2 and 5). In the latter case, the main contaminant sources were urban, primary sewer network, Deep Sewer System, and landfills/dump sites. Moderate vulnerability areas covered 39% of the MCMA. The major proportion of the area (82%) was located on alluvium and fractured lavas and tuffs of high permeability (Figures 2 and 5). Urban was the single contaminant source in 16% of the area. In contrast, a relatively small proportion of the MCMA was classified as either very high vulnerability (0.1%) or high vulnerability (1.5%). Both vulnerability classes were located in the central, northwestern, and southeastern areas of the MCMA. Both of these vulnerability classes were located on alluvium and

416

8.3 37.4 10.5

43.2 0.5 0.1

Extraction wells

M. Mazari-Hiriart and others

fractured lavas and tuffs of high permeability. The main contaminant sources were urban, industrial, and primary sewer system (Figures 2 and 5).

11.5 11.4 10.0 25.7 2.4 0.1 14.3 21.1 9.0 8.6 82.2 21.9 6.1 10.9 5.8 36.0 1.9

0.1 11.5 15.6 9.1

74.6 1.4 1.1 61.4 0.6 49.7 4.5 1.3 8.6 0.6 63.2 0.2 0.4 60.4 0.7 55.9 0.1 0.3

Lacustrine clays Unweathered tuffs Unfractured rhyolites, andesites, and dacites Quaternary domes Undifferentiated volcanic rocks Alluvium Fractured lavas and tuffs

Gas stations Urban zone Fuel deposits Primary sewer Landfill/ dump site Industrial zone Geologic formation

Contaminant source

Table 2. Percentage of the MCMA area occupied by contaminant sources versus geologic formation permeability

Deep sewer

Discussion and Conclusions It is true that protection of the water supply in urban areas is one of the most important concerns for sustainable development. In developing countries, however, urban expansion has not been matched by policies to prevent groundwater contamination. In the MCMA, the situation is critical because most of the water for human consumption is currently supplied by groundwater (Ezcurra and others 1999; Jime´ nezCisneros and others 2004). Thus, groundwater vulnerability assessment for the MCMA and other megacities in the developing world should take into account the location of contaminant sources, in addition to using the best available information. One particularly challenging issue in the MCMA is groundwater contamination by organic compounds, which should be closely monitored to prevent serious health problems. In fact, these contaminants represent a major risk to public health because they have deleterious and irreversible effects in humans at low concentrations. Therefore, this long-overdue research is a first approximation of the problem of groundwater contamination by organic compounds. However, the general approach could be used for any contaminant source. Despite the fact that groundwater vulnerability in the MCMA has been a concern for environmental agencies and academia for many years, its assessment has been a matter of controversy. In fact, setting priorities for groundwater protection in the MCMA is challenging because of both the complexity of the aquifer system and the lack of adequate hydrogeological information. Therefore, current policies dealing with groundwater vulnerability are based on little scientific evidence and high levels of subjectivity. Evidently, more research is needed to settle such controversies, but the fact is that aquifer protection in the MCMA should not wait until exhaustive scientific research is carried out to fill critical gaps in objective information. As asserted by Villa and McLeod (2002), what is needed under such conditions is an approximate quantification of groundwater vulnerability based on easily measurable attributes while maintaining conceptual accessibility for decision-makers. Consequently, this study used an expert consultation approach to supply key data for assessment and a fuzzy logic model to take into account the uncertainties in both cartographic data and expert determinations.

Groundwater Multicriteria Approach for Mexico City

417

Figure 4. Location of contaminant sources in Mexico City: (A) urban zone, (B) industry, (C) primary sewer, (D) deep sewer, (E) gas stations, (F) fuel deposits, (G) landfills, and (H) extraction wells. Subdivisions correspond to the 16 administrative units of the Federal District and 27 neighboring municipalities in the State of Mexico.

418

M. Mazari-Hiriart and others

Figure 5. Groundwater vulnerability map of Mexico City.

The use of any groundwater vulnerability mapping approach is problematic. For example, the numerical rankings and weightings scheme of DRASTIC, perhaps the best known groundwater vulnerability assessment index, has been criticized and validation attempts have had mixed results (Tait and others 2004). In our case, the fuzzy multicriteria approach allowed us to subdivide the MCMA into groundwater vulnerability categories. However, because of the lack of validation and the uncertainties involved in the prioritization of areas, the groundwater vulnerability map (Figure 5) should be used as a screening tool to identify priority areas for further investigations. Given that the vulnerability categories relate to the type of contamination source, our results allow the implementation of specific strategies to protect the aquifer and to monitor potential contaminant sources. It is evident that this approach presupposes that the organic compounds produced by contamination sources at the surface will eventually reach the aquifer at the same location, and it disregards important factors such as the depth of the water table and the amount of precipitation. Nonetheless, this assumption is reasonable because the approach does take into account the permeability of the geologic layers beneath the contaminant sources. Hence, the assessment results should be interpreted with respect to both the potential discharge of contaminants and the hydrogeological data. More specifically, the lacustrine clay zone presents the lowest vulnerability because it has been described as a natural barrier that protects the

aquifer, whereas the alluvium and fractured lavas and tuffs present the highest vulnerability because of their permeability. Thus, the fact that landfills, the second highest ranking contaminant source according to expert opinion, correspond to low vulnerability areas can be explained by the fact that landfills are built either in low permeability areas or when they occur in higher permeability areas, they are the single contaminant source on pixels. On the other hand, urban, industrial, and primary sewer systems correspond to areas of very high vulnerability because they are relevant contaminant sources that occur together in high permeability areas. With respect to expert consultation, results are based on the judgments of eight experts and, thus, the final results might not reflect the true importance of the potential contaminants. It is a self-evident truth that consulting more experts would have avoided bias in the weight determination. Nonetheless, the weights assigned to the contaminant sources are consistent with the ranking of contaminant sources reported by Knox and Canter (1996) and Rail (2000), especially in the case of the most important ones (i.e., industrial zones and landfills). Overall, the approach presented here contributed to a more coherent assessment of the priorities involved in policy-making and groundwater vulnerability. The analysis identified the most important sources and the location where preventive or corrective actions should take place (see Figure 5 and Table 2). Importantly, the analysis allowed a better understanding of

Groundwater Multicriteria Approach for Mexico City

the implications of the relationship among contaminant sources and the aquifer characteristics. In that respect, it could be claimed that the approach was transparent, given that it provided a systematic examination of the relevant factors involved. One important issue in environmental policy-making is public participation. In the case of the MCMA, the problem of groundwater vulnerability is yet to be a concern for the general public. Hence, the final map (Figure 5) should be used not only for setting priorities for action but also as a communication tool of those priorities to the different stakeholders and the general public. The approach presented in this study should be considered at least as easy to understand, transparent, and simple to use as any other groundwater vulnerability assessment technique. Importantly, it incorporates location data and the relevance of potential contaminant sources, which is typically neglected. Therefore, under a more general framework, this approach could be implemented in other megacities, especially in the developing world, with an appropriate readjustment of weights to reflect particular conditions. It is believed that this fuzzy multicriteria mapping approach makes the best use of expert knowledge and available, but imprecise, information regarding groundwater vulnerability.

Acknowledgments We thank F. Mooser, J. M. Lesser, M. Mazari, C. Siebe, M. Maass, J. L. Palacio, M. A. Ortiz, and A. Zarco as well as the three anonymous reviewers for their invaluable contributions to the development of this article. This research was supported by DGAPA– UNAM, through project IN-225399. We also thank the Institutional Program of the Environment and Security at the Instituto Mexicano del Petro´leo.

Literature Cited Banai-Kashani, R. 1989. A new method for site suitability analysis: The Analytic Hierarchy Process. Environmental Management 13:685–693. Birkle, P., V. Torres-Rodrı´guez, and E. Gonza´ lez-Partida. 1995. Effects of evapotranspiration on the water balance of the Valley of Mexico. Geofı´sica Internacional 35:63–72. Bojadziev, G., and M. Bojadziev. 1995. Fuzzy sets, fuzzy logic, applications. World Scientific, River Edge, NJ. Bojo´rquez-Tapia, L. A., L. Jua´ rez, and G. Cruz-Bello. 2002. Integrating fuzzy logic, optimization, and GIS for ecological impact assessments. Environmental Management 30:418– 433.

419

Cox, E. 1994. The fuzzy system handbook: A practitionerÕs guide to building, using, and maintaining fuzzy systems. AP Professional, Boston, MA. Daly, D., A. Dassargues, D. Drew, S. Dunne, N. Goldscheider, S. Neale, I. C. Popescu, and F. Zwahlen. 2002. Main concepts of the ‘‘European approach’’ to karst-groundwatervulnerability assessment and mapping. Hydrogeology Journal 10:340–345. DGCOH (Direccio´n General de Construccio´n y Operacio´n Hidra´ ulica). 1993. Fuentes de contaminacio´n al agua subterra´ nea y alternativas de saneamiento. Departamento del Distrito Federal Contract 3-33-1-0172, Mexico D.F., Mexico. DGCOH (Direccio´n General de Construccio´n y Operacio´n Hidra´ ulica). 1996. Plan maestro de drenaje de la Zona Metropolitana de la Ciudad de Me´ xico 1994–2010. Departamento del Distrito Federal, Mexico D.F., Mexico. Durazo, J., and R. N. Farvolden. 1989. The groundwater regime of the Valley of Mexico from historic evidence and field observation. Journal of Hydrology 112:171–190. ERDAS. 1999. ERDAS IMAGINE 8.4 Tour Guide. ERDAS, Inc., Atlanta, GA. Ezcurra, E., M. Mazari-Hiriart, I. Pisanty, and A. G. Aguilar. 1999. The Basin of Mexico. Critical environmental issues and sustainability. United Nations University Press, Tokyo. Fuchs, R. J. 1999. Introduction. Pages 1–13 in R. J. Fuchs, E. Brennan, J. Chamie, F. Lo, and J. I. Uitto (eds.), Mega-city growth and the future. United Nations University Press, Tokyo. Gogu, R. C., and A. Dassergues. 2000. Current trends and future challenges in groundwater vulnerability assessment using overlay index methods. Environmental Geology 39:549– 559. Howard, K. W. F., and K. Gelo. 2003. Intensive groundwater use in urban zones: The case of megacities. Pages 35–58 in M. R. Llamas and E. Custodio (eds.), Intensive use of groundwater. Challenges and opportunities. Swets & Zeitlinger B.V., Lisse, The Netherlands. IMP (Instituto Mexicano del Petro´leo). 1997. Sistemas de informacio´n geogra´ fica aplicados a la informacio´n cartogra´ fica del proyecto de investigacio´n sobre materia particulada y deterioro atmosfe´ rico (IMADA) GCA-097085, Instituto Mexicano del Petro´leo, Mexico, D.F., Mexico. INEGI (Instituto Nacional de Estadı´stica, Geografı´a e Informa´ tica). 1993. Ciudad de Me´ xico (a´ rea metropolitana). Perfil Sociodemogra´ fico, XI Censo General de Poblacio´n y Vivienda. Me´ xico, D.F., Me´ xico, pp. 2. INEGI (Instituto Nacional de Estadı´stica, Geografı´a e Informa´ tica). 1994. Censos Econo´micos. Me´ xico, D.F., Me´ xico. Jime´ nez-Cisneros, B., M. Mazari-Hiriart, R. Domı´nguez-Mora, and E. Cifuentes-Garcı´a. 2004. El agua en el Valle de Me´ xico. Pages 5–32 in B. Jime´ nez and L. Marı´n (eds.), El Agua en Me´ xico vista desde la Academia. Academia Mexicana de Ciencias. Literal D. de R. L, Me´ xico, D.F., Me´ xico. Knox, R. C., and W. Canter. 1996. Prioritization of ground water contaminants and sources. Water, Air and Soil Pollution 88:205–226.

420

M. Mazari-Hiriart and others

Kosko, B. 1992. Neural networks and fuzzy systems: A dynamical systems approach to machine intelligence. Prentice-Hall, Englewood Cliffs, NJ. Lesser, J. M., and M. A. Corte´ s. 1998. El hundimiento del terreno en la ciudad de Me´ xico y sus implicaciones en el sistema de drenaje. Ingenierı´a Hidra´ulica en Me´xico 13:13– 18. Lesser, J. M., F. Sa´ nchez, and D. Gonza´ lez. 1986. Hidrogeoquı´mica del acuı´fero de la Ciudad de Me´ xico. Ingenierı´a Hidra´ulica en Me´xico Septiembre–Diciembre: 64–77. Lesser, J. M., F. Sa´ nchez-Dı´az, and D. Gonza´ lez-Posadas. 1990. Aspectos geohidrolo´gicos de la Ciudad de Me´ xico. Ingenierı´a Hidra´ulica en Me´xico Enero–Abril: 52–60. Llamas, M. R., and E. Custodio. 2003. Intensive use of groundwater: A new situation which demands proactive action. Pages 13–31 in M. R. Llamas and E. Custodio (eds.), Intensive use of groundwater. Challenges and opportunities. Swets & Zeitlinger B.V., Lisse, The Netherlands. Lobo-Ferreira, J. P., and M. M. Oliveira. 1997. Drastic vulnerability mapping of Portugal. Pages 132–137 in Proceedings of the 27th Congress of the International Association for Hydraulic Research, Groundwater: An Endangered Resource, 10–15 August 1997, San Francisco. Mackay, D. M., and L. A. Smith. 1993. Organic contaminants. Pages 323–343 in W. M. Alley (ed.)., Regional ground-water quality. Van Nostrand Reinhold, New York. Malczewski, J. 1999. GIS and multicriteria decision analysis. Wiley, New York. Marı´n, L. E., O. Escolero-Fuentes, and A. Trinidad-Santos. 2002. physical geography, hydrogeology, and forest soils of the basin of Mexico. Pages 44–67 in M. E. Fenn, L. I. de Bauer, and T. Herna´ ndez-Tejeda (eds.), Urban air pollution and forests. Resources at risk in the Mexico City air basin. Springer-Velag, New York. Marsal, R. J., and M. Mazari. 1969. El subsuelo de la Ciudad de Me´ xico. Universidad Nacional Auto´noma de Me´ xico, Me´ xico, D.F., Me´ xico, pp. 148–168. Marsal, R. J., and M. Mazari. 1990. Desarrollo de la meca´ nica de suelos en la Ciudad de Me´ xico. En: Anon. El Subsuelo de la Cuenca del Valle de Me´ xico y su Relacio´n con la Ingenierı´a de Cimentaciones a Cinco An ˜ os del Sismo. Sociedad Mexicana de Meca´ nica de Suelos, A.C. Me´ xico, pp. 3–25. Mazari, M., and D. M. Mackay. 1993. Potential for groundwater contamination in Mexico City. Environmental Science and Technology 27:794–802. Mazari, M., J. Alberro, M. Mazari-Hiriart, and S. Gonza´ lez. 1996. Agrietamiento de arcillas lacustres y su relacio´n con el uso y reu´ so del agua. Pages 113–156 in M. Mazari (ed.), Hacia el Tercer Milenio. El Colegio Nacional, Mexico, D.F., Mexico.

Mooser, F., A. Montiel, and A. Zu´ n ˜ iga. 1996. Nuevo mapa geolo´gico de las cuencas de Me´ xico, Toluca y Puebla. Estratigrafı´a, tecto´nica regional y aspectos geote´ rmicos. Comisio´n Federal de Electricidad. Gra´ fica, Creatividad y Disen ˜ o, S.A. de C.V., Me´ xico, D.F., Me´ xico. Ortega, A., and R. N. Farvolden. 1989. Computer analysis of regional groundwater flow and boundary conditions in the Basin of Mexico. Journal of Hydrology 110:271–294. Pankow, J. F., and J. A. Cherry. 1996. Dense chlorinated solvents and other DNAPLs in groundwater: History, behavior, and remediation. Waterloo Press, Portland, OR. PEMEX (Petro´leos Mexicanos). 1997. Datos ba´ sicos de estaciones de servicio en el Valle de Me´ xico. Internal report. Gerencia Comercial Zona Valle de Me´ xico. Pye, V I., R. Patrick, and J. Quarles. 1983. Groundwater contamination in the United States. University of Pennsylvania Press, Philadelphia, PA. Rail, C. D. 2000. Contamination, sources and hydrology. Technomic Publishing, Lancaster, PA. Ramanathan, R. 2001. A note on the use of the analytic hierarchy process for environmental impact assessment. Journal of Environmental Management 63:27–35. Riosvelasco, P. 1994. La disposicio´n final de desechos so´lidos en la Ciudad de Me´ xico. Direccio´n General de Servicios Urbanos, Departamento del Distrito Federal, Me´ xico, D.F., Me´ xico. Rudolph, D. L., and E. O. Frind. 1991. Hydraulic response of highly compressible aquitards during consolidation. Water Resources Research 27:17–30. Rudolph, D. L., J. A. Cherry, and R. N. Farvolden. 1991. Groundwater flow and solute transport in fractured lacustrine clay near Mexico City. Water Resources Research 27:2187–2201. Saaty, T. 1980. The Analytic Hierarchy Process. McGraw-Hill, New York. Schwille, F. 1988. Dense chlorinated solvents in porous and fractured media: Model experiments. Lewis, Ann Arbor, MI. Tait, N. G., D. N. Lerner, J. W. N. Smith, and S. A. Leharne. 2004. Prioritisation of abstraction boreholes at risk from chlorinated solvent contamination on the UK Permo.Triassic Sandstone aquifer using a GIS. Science of the Total Environment 319:77–98. Thomas, R. D. 1990. Evaluation of toxicity of volatile organic chemicals: General considerations. Pages 451–463 in N. F. Ram, R. F. Christman, and K. P. Cantor (eds.), Significance and treatment of volatile organic compounds in water supplies. Lewis, Chelsea, MI. Thornton, J. 2000. PandoraÕs poison. Chlorine, health, and a new environmental strategy. MIT Press, Cambridge, MA.

Merino, H. 2000. Sistema Hidra´ ulico. Pages 344–351 in G. Garza (ed.), La Ciudad de Me´ xico en el fin del segundo milenio. El Colegio de Me´ xico-Gobierno del Distrito Federal, Mexico, D.F., Mexico.

UN (United Nations). 2002. World urbanization prospectus. The 2001 revision. Data tables and highlights. Department of Economic and Social Affairs. ST/ESA/SER.A/216. UN Secretariat, New York.

Mooser, F., and C. Molina. 1993. Nuevo modelo hidrogeolo´gico para la Cuenca de Me´ xico. Centro de Investigacio´n Sı´smica de la Fundacio´n Barros Sierra, Mexico, D.F., Mexico, pp 68–84.

UNESCO–WWAP (United Nations Educational, Scientific and Cultural Organization–World Water Development Report). 2003. Water for people water for life. World Water Assessment Program. Barcelona, Spain, pp. 157–175.

Groundwater Multicriteria Approach for Mexico City

USA–CERL. 2003. GRASS 5.8 (Geographic Resource Analysis Support System). United States Army Corps of Engineers Construction Engineers Research Laboratory. Champaign, IL. US EPA (US Environmental Protection Agency). 1993a. Ground water resources assessment. EPA 813-R-93-003. Office of Groundwater and Drinking Water, US Environmental Protection Agency, Washington, DC. US EPA (US Environmental Protection Agency). 1993b. A Review of Methods for Assessing Aquifer Sensitivity and Ground Water Vulnerability to Pesticide Contamination, EPA 813-R-93-002. U.S. Environmental Protection Agency, Washington, D.C. Va´ zquez, E. 1995. Modelo conceptual hidrogeolo´gico y caracterı´sticas hidra´ ulicas del acuı´fero en explotacio´n en la parte

421

meridional de la Cuenca de Me´ xico. M.Sc. thesis. Universidad Nacional Auto´noma de Me´ xico. Posgrado en Geofı´sica. Villa, F., and H. McLeod. 2002. Environmental vulnerability indicators for environmental planning and decision-making: Guidelines and applications. Environmental Management 29:335–346. Warren, C. J., and D. L. Rudolph. 1997. Clay minerals in basin of Mexico lacustrine sediments and their influence on ion mobility in groundwater. Journal of Contaminant Hydrology 27:177–198. Westrick, J. 1990. National survey of organic compounds in ground and surface waters. Pages 103–125 in N. F. Ram, R. F. Christman, and K. P. Cantor (eds.), Significance and treatment of volatile organic compounds in water supplies. Lewis, Chelsea, MI.