Christopher A. Scott

Research Scientist

International Water Management Institute

c/o CIMMYT, Lisboa 27, Col. Juárez, A.P. 6-641

C.P. 06600 México D.F. MEXICO

Tel. + 52 (5) 726-9091

Fax. + 52 (5) 726-7558

c.scott@cgiar.org

Carlos Garcés Restrepo

Program Leader, Mexico

International Water Management Institute

c/o CIMMYT, Lisboa 27, Col. Juárez, A.P. 6-641

C.P. 06600 México D.F. MEXICO

Tel. + 52 (5) 726-9091

Fax. + 52 (5) 726-7558

c.garces@cgiar.org

ABSTRACT

Rehabilitation and modernization programs aimed at saving water often neglect the influence of surface water management on groundwater levels. This paper assesses groundwater trends in the middle Río Lerma basin in Mexico through the development, verification and application of a simple, conjunctive surface-groundwater model that accounts for recharge resulting from surface irrigation. For the 1982-98 period, net aquifer extraction was predicted to result in an average static water level decline of 2.12 m/year, compared to a 1.81 m/year historical average in 398 wells in six aquifers in the basin. Eight alternative scenarios (producing average declines between 0.00 and 3.21 m/year) are presented, indicating that feasible changes in grain and vegetable cropping patterns and water management are unlikely to bring static groundwater depths back to historical levels. Contrary to conventional thinking, decreasing the relative water supply of surface irrigation (defined as the ratio of total surface water supply to crop demand) by 10%, as for instance through surface irrigation system rehabilitation, was simulated to result in an additional average decline of 0.91 m/year (combined average 3.03 m/year). Increasing the surface relative water supply by 10% (equivalent to increasing reservoir releases by 25%) was simulated to reduce average decline to 1.21 m/year. Increasing surface water supply by 23% (increasing reservoir releases by 57%, a level we consider to be unfeasible) was simulated to produce zero average decline. The results indicate that, in water-short basins, the sustainability of groundwater trends is inextricably linked to the management of surface water, and is highly sensitive to the area and type of crops irrigated, as well as surface water management practices.

ACKNOWLEDGEMENTS

The authors would like to thank the following agencies and individuals for data and conceptual inputs that made this paper possible: the offices of the Mexican National Water Commission (C.N.A., or Comisión Nacional del Agua in Spanish) for the Alto Río Lerma Irrigation District and for the State of Guanajuato; David Molden (IWMI); and Wim Kloezen (Wageningen Agricultural University, formerly IWMI). We are grateful to the organizers for the invitation to present the paper at the Workshop on Recursos Hidricos: Crisis, soluciones y vision futura; en Montevideo, Uruguay, Junio 15-18, 1999. It should be noted that the International Water Management Institute was previously the International Irrigation Management Institute (IIMI).

INTRODUCTION

In Mexico and around the world, a great deal of emphasis is currently placed on "water savings" through more efficient use of irrigation water. The underlying assumption is that water applied in excess of crop demand is "lost," presumably as drainage outflow from the system or evaporation of ponded water within the system. Following this line of reasoning, applying irrigation water more sparingly would result in some combination of the following three outcomes:

• increase in reservoir storage carryover for the next irrigation season
• increase in the area irrigated under present cropping patterns
• intensification of irrigation leading to higher water demand crops (higher value crops) planted on the present irrigated area

In order to achieve such outcomes, major rehabilitation and modernization programs are implemented in which canals are lined and efficient water application technologies (drip and sprinkler) supplant traditional flood or furrow irrigation. The question is: will these measures achieve real water savings?

The International Water Management Institute (Seckler, 1996; Molden, 1997) has questioned the classical conception of irrigation efficiency, particularly in the context of river basins where water is reused. The unforeseen outcome of conventional initiatives to rehabilitate or modernize irrigation infrastructure may simply be to reduce a third party’s water supply. These are referred to as "dry water savings" as distinct from "wet water savings" which do not produce such reductions. In river basins where surface and groundwater resources are coupled—this is the case in most basins—"low efficiency", or "leaky" surface irrigation systems play an important aquifer recharge function. This paper explores the implications of conjunctive surface and groundwater management for the middle Río Lerma basin in Mexico and shows that reducing surface supplies (i.e., "saving water") may have profound and negative effects on the basin’s aquifers.

In the following analyses, we operationalize the IWMI river basin water accounting framework (see Molden, 1997). Water inflows to the basin are categorized as gross inflows from precipitation and runoff plus (or minus) any change in storage. Outflows include committed (existing downstream water rights) and uncommitted flows (losses to the sea), whereas depletions are crop ET and evaporation from the soil and ponded water. Where no utilizable outflow occurs in the dry season, the basin is said to be "closed." This is the case for both the middle Lerma and the Lerma-Chapala basins, for which analyses are presented below.

CONCEPTUAL APPROACH

The concept of relative water supply (RWS) was originally introduced by Levine in the mid-70s and fully discussed later (Levine, 1982). It is used widely as an irrigation system diagnostic tool that not only describes the hydraulic performance of the system, but also sheds light on the causes of that performance. Thus, the variable constitutes a powerful analytical tool as it incorporates the "management" element and farmer reaction to perceived water availability. It is defined as:

where the denominator includes consumptive use and non-beneficial ET. In the original definition, precipitation is taken to be total (Levine, 1982).

RWS can be evaluated essentially at any location where the water supply can be measured or estimated, and the demand determined: on-farm, turnout, lateral, system or basin level. Given the nature and flexibility of this indicator, its use and definition are often adapted to the particular situation at hand. Most of the modifications deal with the definition of the numerator, that is, total water supply (TWS). A frequent change deals with the handling of rainfall—total or effective—and how surface water availability is treated. The reader should be aware that the groundwater contribution, as through capillary rise, is not included. However, there is no conceptual limitation to evaluating RWS for groundwater irrigation where TWS includes the volumes pumped.

In this paper, the relative water supply concept is applied in the context of a river basin (RWS_basin, or RWS_b); therefore, total surface water supply (TWS_s) is defined in the following manner to better reflect surface water availability:

where: P is precipitation (effective, using the U.S. Bureau of Reclamation method of 0.8*P_total)

STUDY AREA

The basin of the Lerma-Santiago river system drains much of the highlands of west-central Mexico and empties into the Pacific Ocean. The basin comprises a number of sub-basins, most importantly the Lerma River (Río Lerma) which flows into Lake Chapala (when lake levels are sufficiently high, the outflow is called the Río Santiago). The Lerma-Chapala Basin, which covers some 54,000 km², is undergoing rapid economic and social change. A dynamic agricultural sector and a rapidly growing industrial sector, which accounts for 35% of Mexico’s industrial GNP, characterize the basin. The first of Mexico’s river basin councils has been established in the basin, and will likely be a model for others to follow. The council has passed binding regulations on upstream surface water withdrawals in order to control the rapid decline in surface water availability and quality in the lower basin.

While the mean annual runoff in the Lerma-Chapala basin represents a little over one percent of Mexico's total runoff, the basin contains over 10 percent of the population and 14 percent of the irrigated area in the country. Hence, the basin is faced with growing water shortages, with current average annual water availability per capita at around 950 m³. Since at least the mid 1980s, water resources in the basin have been over-committed, resulting in significant pressure on surface water sources and alarming declines in the levels of groundwater and Lake Chapala (De Anda et al., 1998). Moreover, water quality is a rapidly increasing problem. Most significantly, water is being reallocated from the agricultural to the urban and industrial sectors, albeit in an uncoordinated manner. With over 700,000 ha of irrigation in the Lerma-Chapala basin, the study area is important from the perspective of agricultural production. Irrigation consumes approximately 78% of the water resources in the basin (Mestre, 1997).

The middle Lerma basin, for purposes of the analyses presented in this paper, is defined as the river reach below the Solís Dam to the Markazuza Diversion (see Figure 1). The 1,700 km² middle Lerma comprises 150,000 ha of irrigated area. The Alto Rio Lerma Irrigation District (ARLID, 112,000 ha command area) is supplied by surface water from four principal reservoirs (Solís, Tepúxtepec, Purísima, and Yuriria) in addition to groundwater. Because ARLID is the largest district within the Lerma-Chapala basin and located very much at its center (see Figure 1), these results can be considered representative of the basin as a whole.

ARLID takes approximately 44% or 880 MCM of all the water stored for use by the irrigation districts in the Lerma-Chapala basin. There are roughly 24,000 water users with 55% classified as communal farmers (‘ejidatarios’) and 45% as small private growers. The average land holding is 5 ha. Mean yearly evaporation is around 2,000 mm. On the other hand, the dry winter season with approximately 80 mm of rainfall, starts in November and ends at the end of April. Rainfall in the spring/summer and second seasons, from May until November, is approximately 670 mm. Wheat and barley are normally grown during the dry winter season while sorghum, maize and beans are the main crops grown in the wetter summer season. Farmers with access to groundwater tend to grow more vegetables (Kloezen and Garcés, 1998). The irrigation district is subdivided into 11 units, referred to as modules. An individual Water User Association (WUA) manages each module. The irrigation network comprises 475 km of main irrigation canals and 1,658 km of secondary and tertiary canals. Likewise, there is a network of approximately 1,031 km of drainage canals. In addition, there are 1,714 deep wells serving 35,075 ha. Roughly 80% of the wells in operation in the basin are for agricultural purposes (Chávez, 1998).

A number of smaller farmer-managed irrigation systems (‘unidades’) irrigate an additional 38,000 ha command area from a combination of sources, principally groundwater (for statistics on the national breakdown of districts and ‘unidades,’ see Garcés-Restrepo et al., 1996). Future IWMI research on the unidades will attempt to characterize these in sufficient detail to be able to distinguish water management and cropping practices from those prevalent in the districts.

Figure 1: Middle Río Lerma Basin (with Alto Río Lerma Irrigation District)

Six aquifers or hydro-geological units, as identified by the Mexican National Water Commission (C.N.A., Comisión Nacional del Agua), underlie the middle Lerma basin. The aquifers are not strictly independent since there are no strictly impermeable geological barriers that separate them. However, for management purposes it is more convenient to deal with these "fractions" or "units". In general terms, the first couple of hundred meters from the surface are composed of alluvial and lacustrine materials: gravel, sand, silt and clays, which form interconnected layers of irregular geometry and grading. The lower layers, several hundred meters in depth, are composed primarily of basaltic rocks and rhyolite tuff limited at the bottom by andesite and red conglomerate strata of low permeability (Chávez, 1998). The aquifers are naturally recharged through rainfall infiltration and surface run-off, the latter both in mountainous areas and along the main riverbeds. To this natural component of the recharge, a human-derived component stemming from agricultural development needs to be added: the water derived for irrigation from the Lerma and Laja rivers creates an incidental recharge generated by the conduction and distribution losses and the infiltration from excessive irrigation applications.

In previous research (from 1995 to 1996), IWMI calculated RWS_i at different levels in ARLID. Actual, planned and reported values of RWS_i were obtained at the system, module, and on-farm levels. Measurements were made to represent land tenure arrangements (ejido vs. private) and water sources (canal vs. well). In addition, the research layout followed the traditional head-middle-tail approach in the irrigation network. The results can be summarized in Table 1 below:

Table 1. Irrigation Relative Water Supplies (RWS_i) for the Alto Río Lerma Irrigation District

(Adapted: Tables 3 and 7a. Kloezen and Garcés, 1998).

n.a. = not available

Table 1 shows that the irrigation district operated during the winter 1995-96 and the summer 1996 seasons under conditions of relatively abundant water availability. It was possible for the system to meet crop water requirements with a margin of safety, as indicated by the RWS_i value at different district levels. Two other important results can be derived from the table: RWS_i values for private wells are generally lower than those for canal water in the winter season, but high for the summer season. However, given that the RWS_i values for canals are calculated at their offtake points, while those for the private wells represent on-farm level water supply, it is concluded that the farmers who use wells use more water. Two reasons explain this. First, private well owners generally do not wait for the rains to come but start irrigating as soon as they can. As a result of the late rainfall onset during the summer of 1996, private well owners had already completed one irrigation, which explains the slightly higher actual RWS_i summer values for wells as compared to surface water. Second, owing to subsidized energy tariffs the cost of pumping water had not yet exceeded the cost of surface water and as such, this has never been an incentive for well owners to economize on water (Kloezen and Garcés, 1998). Finally, the difference in RWS values for surface irrigation between the two modules suggest quite different management approaches. The study revealed that Salvatierra’s management set-up needs to be upgraded and that the existing "relaxed" management has led to significant deterioration of the infrastructure. These data, particularly for the ARLID District, compare well with the RWS_b and RWS_i values calculated in the present analyses, as shown in Table 2.

The IWMI Mexico Program is presently undertaking field research in the middle Lerma basin to determine the fate of excess water by quantifying the magnitude of drainage flow and ponding "losses." Two lateral canal command areas, each 250 ha in area, which are conjunctively irrigated from surface and groundwater sources have been identified where daily field measurements of all inflows and outflows are made. It is not our intent here to describe the details of the field measurement program; however, it is important to indicate that the results reinforce our contention that the "real water losses" are minimal. During the May – November 1998 rainy season, drainage outflow, which occurred primarily during storm events comprised only 5% of the total outflows from the field area, crop evapotranspiration (ET) was 66% and soil moisture change accounted for 5%. The evaporative loss resulting from ponding was negligible, and 24% of the total outflow was calculated to recharge the aquifer. On the other hand, pumping rates were sufficiently high so as to result in net aquifer extraction as described in this paper.

Table 2. Basin and Irrigation Relative Water Supplies for the Middle Lerma Basin

PROBLEM STATEMENT

In water-short river basins, such as the middle Lerma, exploitation of water resources currently exceeds the renewable annual supply. Intra-annual variations in runoff may alleviate or exacerbate water availability in the short-term; however, the medium and long-term effects are clearly significant only for the aquifers. In other words, a condition of low surface water supply can be overcome in just one year of high rainfall and runoff, whereas the accumulated deficit of years of aquifer overexploitation will similarly take years to reverse. Based on the reasoning and data we show here, modernization programs to make water use more efficient are likely to have adverse impacts on aquifer levels in basins with similar conjunctive surface – groundwater characteristics as described here.

Nevertheless, reduced groundwater recharge resulting from more efficient surface water management implies a number of important tradeoffs with more than purely physical consequences. We refer specifically to social equity and economic productivity considerations, i.e., who irrigates with water from each source, and what costs and benefits are accrued. In the middle Lerma basin, the ‘ejido’ sector (with smaller landholdings producing primarily cereals) is more dependent on surface water than the private sector (commercial farms with a greater degree of diversification in vegetable production for export). Conversely, groundwater tends to be the principal water source for private growers, although some ‘ejido’ production units have access to groundwater. In this context, the present arrangement of a "leaky" surface water supply system coupled with the large irrigation depths observed in the field (Kloezen et al, 1997) amounts to an indirect water transfer from the ‘ejido’ to the private sector. While the constraints to producing higher value crops with surface water are many (including poor access to credit), lack of flexibility in the scheduling of surface deliveries represents a primary water management limitation. At the same time, it appears as though pumping costs still do not represent a major cost constraint in the production of higher value vegetables (primarily broccoli, cauliflower, peppers, carrots, strawberries, etc.) and fodder (alfalfa) crops.

The trend towards increased water shortages has resulted in higher inter-sectoral competition for the resource. In this regard, the urban sector has some advantage over others given existing policy priorities (not always clearly justified), higher economic capacity and sheer demographic numbers. This situation has contributed to migration from rural areas toward the cities and the corresponding deterioration of the rural quality of life.

As in many other areas of the country, the extraction of groundwater in the basin began, on a large scale, in the early 50s driven by urban growth, industrial development and emerging irrigated agriculture. During this decade and the next one, hundreds of wells were sunk to extend the area under irrigated agriculture or to complement irrigation requirements in areas already under irrigation. Through time, the fall in static water levels led to the need to sink deeper wells: indigenous water buckets (‘norias’) were soon displaced by wells up to 100 m deep. Nowadays, depths between 200 and 400 m are not uncommon, and in the Salamanca area depths of 500 to 1,000 m have been reported (Chávez, 1998). The resulting drawdown of water levels has also affected small surface water impoundments, which lose their storage rapidly through percolation.

The principal effects of over-exploitation, or "groundwater mining," in the basin include:

• increased costs of groundwater exploitation related to deeper wells, large motors or pumps and higher energy consumption
• lower discharges and efficiencies of older devices
• subsidence, settling and terrain fissures
• deterioration of water quality due to rapid transport of industrial contaminants and increased use of fertilizer and pesticides

As static groundwater depths approach 60 – 80 m in many of the aquifers (with dynamic depths greater than 100 m), the need for well rehabilitation will soon exert cost constraints, even for the production of higher value crops. A number of the aquifers are underlain by deeper constrained and semi-constrained strata with piezometric levels reportedly in the range of current static levels, i.e., at 100 – 150 m in the aquifers presently being exploited (J.V., local well equipment supplier, personal communication, 1998). This may lead to a sense of false optimism that operational costs will be stable (once the investment has been made to drill into the deeper aquifer). It is our view that continued groundwater mining will deplete the present aquifers to a point where all but the highest value crops cease to be profitable. Given that the installed electricity generation capacity in Mexico will soon be surpassed by demand, unit energy costs will invariably increase. The only safe conclusion that can be drawn is that water scarcity from both surface and groundwater sources in the middle Lerma is inevitable in the short- to medium-term (10 –20 years), and that this future scenario will only be hastened by programs to make surface water irrigation more efficient.

MODEL DESCRIPTION

A simple conjunctive surface water – groundwater model was developed to simulate the behavior and response of the middle Lerma basin to climatic processes and water allocations. The primary objective was to characterize the influence of irrigation water management practices on groundwater levels, in order to understand the implications of alternative surface water management and cropping scenarios for groundwater. Based on a vertical water balance, the model accounts for changes in groundwater storage resulting from the difference between inflows—both natural (precipitation and river flow) as well as reservoir releases for irrigation within the simulated system—and outflows including river flows, irrigation releases for areas outside the system and the evapotranspirative demand of crops and natural vegetation. The residual term in the mass balance is net change in aquifer storage, expressed as:

P is precipitation (effective)

ET_c is crop evapotranspiration

ET_v is vegetation (non-crop) evapotranspiration.

The change in static water level is calculated as:

A is the aquifer area

k is the aquifer storage coefficient

The system boundaries have been defined as the spatial limits of the irrigated area for which monthly inflow and outflow data are available. We recognize that urban pumping places additional demand on the aquifer; because of high return rates (in the range of 70%), this is not considered in the model. On the other hand, with the spatial concentration of wells in urban areas, there is a pronounced cone of depression in those areas.

An annual mass balance timestep was used to simulate basin response to agricultural water demand. The historical flow and cropping data were checked and found to be consistent and reliable. The following parameters were calculated or estimated, based on our knowledge of the study area:

• 10-day reference ET₀ and evapotranspiration coefficients for crops, trees, fallow (uncultivated) land, and bare soil (between crops) are based on CROPWAT (FAO, 1996; Doorenbos and Pruitt, 1975)
• area under trees = 10% of the total area
• aquifer storage coefficient, k = 0.1 Values are reported in the range of 0.03 – 0.2 (CEASG, unpublished data, 1998).

The model was verified using measured static water levels from 398 wells in the six aquifers underlying the middle Lerma basin study area (INEGI, 1998). For the 1982-98 simulation period, the average simulated decline was 2.12 m/year (yearly results are presented below with the water management scenarios). The simulated decline compares favorably to the average measured decline of 1.81 m/year (for the first part of the simulation period), as shown in Table 3a, which presents data on declines for those aquifers in the study area for the dates which cover the simulation period. Table 3b shows that additional aquifers in the region are in similar processes of static groundwater decline (again for the simulation period only), while Table 3c shows similar historical trends (longer than the simulation period) in regional aquifers. Only wells with 3 or more static level measurements were used in the regression analyses. Figures 2 – 4 present time series plots of three representative aquifers (for purposes of visual comparison, all groundwater plots in this paper use the same axes scales). It should be noted that the number of wells plotted in these figures may differ from the number of wells used for the regression analyses in Tables 3a – 3b. The average declines of 1.81 m/year in both Tables 3a and 3b are purely coincidental.

Table 3a: Groundwater Trends: Aquifers Studied, Simulation Period

Table 3b: Groundwater Trends: Regional Aquifers, Simulation Period

Table 3c: Groundwater Trends: Regional Aquifers, Historical Record

Figure 2: Measured Static Groundwater Levels, Presa Solís Aquifer

Figure 3: Measured Static Groundwater Levels, Irapuato - Valle de Santiago Aquifer

Figure 4: Measured Static Groundwater Levels, Silao - Romita Aquifer

Based on these results, we are confident that the model accounts for a representative set of physical hydrologic processes. Unfortunately, the Mexican government program to monitor static water levels in pilot wells throughout the middle Lerma basin was suspended in the late 1980s, with a few exceptions. Figure 4 shows data for 16 wells in the Silao-Romita aquifer from Nov-94, which appear to indicate that static groundwater levels may have stabilized, or even increased slightly. While this would be a logical outcome of increasing energy costs (both as a result of declining water levels and increasing unit energy costs), the lack of data for the May-86 to Nov-94 period makes it difficult to draw a final conclusion. Time series data for additional aquifers would need to be analyzed to test such a hypothesis.

ALTERNATIVE CROPPING AND WATER MANAGEMENT SCENARIOS

• • area under grains in the fall/winter season
  • area under vegetables in fall/winter
  • area under grains in spring/summer and second season
  • area under vegetables in spring/summer and second season

Table 4 summarizes the five cropping scenarios (S1 – S5) and three water management scenarios (S6 - S8) simulated. Actual historical hydrologic conditions (rainfall, river flow, and reservoir releases) were used to assess the groundwater response to each of the cropping scenarios. S1 reflects a reduction in agricultural production that could result from increased inter-sectoral competition over water. S2 is the average case, and should result in static groundwater declines similar to those simulated for actual cropping patterns and to measured groundwater trends. S3 is an unlikely scenario of increases in both grain and vegetable production (this will be shown to have the most extreme effect on groundwater levels). S4 and S5, different degrees of conversion of grain production to vegetables, represent a trend that is likely to result from the signing of the North American Free Trade Agreement (NAFTA). Under this scenario, large-scale grain production units in the United States and Canada (with their built-in subsidy and price support structures) out-compete Mexican grain production, but where climatic and labor conditions favor the production of vegetables in Mexico. This trend is already evident in the middle Lerma basin, where commercial production units are increasingly entering into direct export contracts with foreign buyers (local informants, personal communication, 1998-99).

The basin-level surface relative water supply RWS_b is an indicator of water available relative to crop ET demand. Using the actual historical cropping patterns, three additional water management scenarios were simulated. S6 and S7 represent +10% and –10% changes in RWS_b, respectively, while S8 represents a +23% change in RWS_b, as will be explained in the next section.

Table 4. Scenarios Simulated

RESULTS AND DISCUSSION

Scenarios S1 – S7 represent changes in cropping and water management that we consider to be feasible. It should be pointed out at the outset of this section that none of these scenarios resulted in the stabilization of static groundwater levels; instead, all resulted in continuing declines. It is extremely interesting, however, to consider which scenarios produced the highest rates of decline and which even reversed groundwater decline in years of high surface water supply. Further, these processes are clearly reflected in the calculated relative water supplies. On the other hand, S8 was designed to produce a zero average groundwater decline for the 1982-98 period.

The changes in static groundwater levels resulting from the five cropping scenarios with the historical hydrologic data are shown in Figure 5; changes in static groundwater levels resulting from the three water management scenarios are shown in Figure 6. In Figure 5, it is apparent that increased area under grains produces groundwater declines. This is a result of the high crop demand (over a longer period) as compared to vegetables. It should be noted that multiple vegetable crops within a single season were not considered, although this may become more common for economic reasons. Based on the definitions of very low, low, average, etc. defined for Table 4, the magnitude of changes in grain area are significantly greater than the magnitude of changes in vegetable area, i.e., the absolute values of the average and standard deviation for grain are considerably higher than for vegetables. Nevertheless, it is clear that producing grain in a water-short basin with high vegetable production potential is neither profitable nor sustainable from a water resource perspective. The reasons for this continued practice have been discussed, and are primarily due to lack of flexibility in scheduling surface water deliveries for vegetables.

While it appears that a major shift to vegetable production would be beneficial from a groundwater perspective, it should be noted that this would imply significant changes in the current management arrangements for irrigated agriculture. Specifically, we refer to marketing, credit, transport, and support systems in general.

Figure 5: Simulated Static Groundwater Response to Alternative Cropping Scenarios

Figure 6: Simulated Static Groundwater Response to Alternative Water Management Scenarios

Table 5. Relative Water Supply Resulting from Alternative Cropping Scenarios

Under Historical Hydrologic Conditions

Figure 7: 16-year Average Static Groundwater Response to Increased Surface Water Supply

The "hump" in Figures 5 and 6 for 1985-86 is the result of a particularly wet year, with high precipitation and net surface inflows. From Table 5, it is evident that RWS values (using both basin and irrigation definitions) for such conditions are very high. On the other hand, dry years with low total surface water supplies produce precipitous groundwater declines, even under low crop demand scenarios (see 1987-88 particularly for S5 or S1). In 1990-91, the year of average TWS_s, the relative water supply values for irrigation are extremely high for all five scenarios, because the RWS_i index does not consider outflows that leave the basin, making the numerator extremely high. This clarifies our earlier statement that RWS_b better reflects the true supply and demand conditions in river basins. The results of the scenarios simulated clearly indicate that groundwater levels are extremely sensitive to changes in surface water management.

These findings run contrary to conventional thinking, and must be incorporated into the planning and implementation of rehabilitation and modernization programs. The cornerstone of the current Mexican irrigation modernization program is to tackle the problem of "water losses" through a variety of strategies to improve water use efficiency at all levels. Our analysis indicates, however, that such an approach would negatively impact the already declining groundwater resources.

CONCLUSION

Surface water and groundwater resources are linked, although the mutual influences are only apparent when viewed from a river basin perspective. This paper analyzes historical cropping and water management data, combined with a simple water balance model, to assess the outcomes of several crop and water management alternatives. The results are highly illustrative of the tradeoffs involved in conjunctive surface water – groundwater systems. For the middle Lerma basin, increases in the irrigated area of vegetables coupled with decreases in the irrigated area of grains would have a beneficial impact on groundwater levels, but presupposes major changes in current water management practices and institutional arrangements.

There is a need to continue or reestablish groundwater monitoring programs in the water-short Lerma-Chapala basin. While previous studies indicated the high relative water supplies in the middle Lerma basin, it is apparent to us that this resulted from significant groundwater depletion. Definitions of water scarcity must account for groundwater trends.

The approach followed in this paper could be applied to other areas in Mexico in order to protect both surface and groundwater resources. In particular, the true extent of "dry" and "wet" savings must be evaluated. Furthermore, the specific results strongly indicate the need to assess more carefully the groundwater impacts of irrigation modernization programs that are underway or being planned in Mexico.

REFERENCES