Evolution of the groundwater system in the Chihuahua-Sacramento aquifer due to climatic and anthropogenic factors

Groundwater is the main source of water in arid cities, where precipitations are low and not evenly distributed. The continuous groundwater pumping causes a long-term water-level decline defined as groundwater depletion. The effects of groundwater depletion are complex and dependent on the combinations of aquifer characteristics, climate variability, human demand, land use, among others (Aeschbach-Hertig & Gleeson 2012). The water-level decline produces an increase in the cost of pumping or drying wells (Campana 2007; Fishman et al. 2011); reduced groundwater discharge, quantity, and quality, to streams and springs affecting ecosystems (Sophocleous 2000). Anthropogenic contamination can be generated by salt mobilization in irrigated regions or widespread of pesticides (Fogg & LaBolle 2006). There may also be problems produced by natural origin, related to aquifer geology most notably high levels of arsenic and fluoride (Fendorf et al. 2010; Giordano 2010).

In cities under arid conditions, groundwater depletion may cause water shortage during the dry season (Hoque et al. 2007; Táany et al. 2009; Garamhegyi et al. 2018). The sustainability of groundwater resources in many basins or plains in the world is threatened, as a result of continuous groundwater depletion through human activities and climatic stress (Wang et al. 2013). Likewise, the overpumping and population growth (linked to urbanization) are causing groundwater levels to decline around the world (Schwartz & Ibaraki 2011). Human activities, such as groundwater pumping near streams, irrigation, and the construction of reservoirs, etc., also impact groundwater dynamics and result in some side effects on the eco-environment (Hanson et al. 2004; Brekke et al. 2009; Famiglietti 2014; Yang et al. 2017). Likewise, authors have reported that in certain regions, the groundwater level decline is related to substantial aquifer dewatering (Garamhegyi et al. 2018), while the increase in annual mean temperature (due to climate change) has been a significant driving effect on the decrease in shallow groundwater levels (Chen et al. 2004). Overall, climate change and variability combined with increased anthropogenic demands on water resources are the two major stressors on reliable and sustainable water resources all over the world (Hanson & Dettinger 2005).

In a recent study focused on identifying the influencing factors on groundwater drought and depletion in north-western Bangladesh, it was shown that there is a vital need of including human-induced effects for drought analysis, and a latent urgency for new research and methods to identify the human influence on groundwater depletion (Mustafa et al. 2017). Several studies have improved the general understanding of how a groundwater system may respond to human influences, as well as on the potential impacts on groundwater drought (Parkinson et al. 2016; Jakóbczyk-Karpierz et al. 2017; Thomas et al. 2017; Han et al. 2019; Escriva-Bou et al. 2020; Kavianpour et al. 2020; Persaud et al. 2020).

Groundwater in Chihuahua, Mexico, is the most important source of drinking water and plays a vital role in the region, however, a few studies have been carried out about human influence on groundwater detriment in Chihuahua (Reyes et al. 2017; CONAGUA 2018; Sánchez-Navarro et al. 2019; Mendieta-Mendoza et al. 2020). Moreover, it is essential to identify which indicators of climate variability and human activities affect groundwater depletion and sustainability. This paper presents an adapted methodology that can be replicated in places with similar conditions and characteristics where fully integrated databases are non-existent. This manuscript allows a preliminary evaluation of the influential factors on the groundwater depletion in the Chihuahua-Sacramento (CHS) aquifer, which is a typical semi-arid area with a majority of urban use of groundwater. It is expected that the outcome of this study can provide a better understanding of the anthropogenic and climate impact on the groundwater system in the CHS aquifer, and thus contribute to groundwater management by the concerned authorities to ensure sustainability in the study area.

The study area is located in the north-western part of Mexico as shown in Figure 1. The CHS aquifer is located in the Sacramento-Chihuahua Valley where the Chihuahua, Aldama, and Aquiles Serdan municipalities and agriculture occur between 28° 24′ 19.7″ and 28° 56′ 46″ to the North, 105° 57′ 41.8″ and 106° 32′ 48.5″ to the West and has an area of 1,889 km² with an average altitude of 1,415 m above sea level (CONAGUA 2018). The CHS aquifer is unconfined and is filled with quaternary alluvial, fluvial, and loess deposits with a thickness that goes from 350 to 700 m (CONAGUA 2009, 2018). In addition, this aquifer system is comprised of several stratigraphic units, such as rhyolite, basalts, sandstones, limestone rocks, conglomerates, and andesites. The climate in the study area is dry, similar to a desertic type, the mean annual temperature is 16.1 °C and the mean precipitation is 441.2 mm yr⁻¹ for the years range 1950–2010 (Sánchez-Navarro et al. 2019). Regarding the annual distribution of precipitation, almost 70% of rainfall occurs in July, August, and September.

This aquifer is located in Hydrological Region 24 called Bravo-Conchos basin (Figure 1; CONAGUA 2009). The Sacramento and Chuviscar rivers, which flow from the south to the north through the valley, are considered the main surface water sources in the study area. The inflow of surface water to the Sacramento River is from northwest to southeast while to Chuviscar River is from southwest to southeast. In the aquifer area, there are four dams, two of them on the Chuviscar River (Chihuahua and Chuviscar dams), one on the Sacramento River (San Marcos dam), and the last one on the San Pedro River (El Rejon dam). From those dams, only the water from the Chihuahua dam is used for human use and consumption while the others are for recreational uses and flood control. Figure 2 shows the geology and the distribution of the surface water network in the study area. The CHS aquifer also underlies most of the urban area of Chihuahua City. The water from this aquifer has several purposes, being the main use of the water supply for urban demand, and to a minor degree the use for agricultural, livestock, industrial, services, and mixed activities. The city of Chihuahua is the state capital with a population of 878,000 as of 2019 and is a growing urban area within a drought-prone region (David et al. 2020).

A semi-integrated layered approach with passive connections between separate estimates was employed for this study. The methodology by Mustafa et al. (2017) was used as a framework to adapt the methods, in order to achieve the objectives of the study. The methodological procedure examined the effects of (a) meteorological variables, (b) land-cover change, and (c) water abstraction on groundwater level (groundwater depletion) step by step over a recent historical period, 1986–2010.

The meteorological drought was assessed using the Multivariate Drought Monitor in Mexico (known in Spanish as MoSeMM) (Rangel 2017). The monthly groundwater recharge was calculated using the land-cover changes evaluated through the groundwater model previously developed for the study: feasibility of alternative sources and a preliminary draft of necessary hydraulic infrastructure, made by the State Water and Sanitation Board (known in Spanish as JCAS). The groundwater demand was assessed using data obtained by the Municipal Water and Sanitation Board (known in Spanish as JMAS). Figure 3 shows the conceptual model applied to identify the main factors that influence groundwater in the study zone. This approach provides an initial view of the potential drivers of supply and demand that can affect groundwater depletion and sustainability.

The land-cover change was assessed to determine the impact on groundwater recharge and actual evapotranspiration from natural, riparian, urban, and agricultural subregions. The main crops in the study area are (a) for fall-winter season forage: oats, grain oats, prairie, ryegrass, wheat, and triticale, (b) for the spring-summer season: forage oats, grain oats, green chili, beans, fodder corn, grain corn, potatoes, sorghum, and red tomato, and (c) the permanent crops are alfalfa, peach, apple, and walnut. The land-cover analyses were carried out for the years 1986, 1996, and 2011. Multispectral Landsat 5 Thematic Mapper (TM) and Landsat 7 ETM composite images of 1996–2011 were used to determine the location, area, and crop type. Landsat images were collected from USGS Global Visualization Viewer (GLOVIS). The Landsat 5 images are composed of seven different bands with a resolution of 30 m. The Landsat 7 ETM dataset is composed of eight bands with a 30 m resolution. The image analysis was conducted by ENVI 4.7, in which, through the combination of spectral bands, zones were discriminated. Also, Google Earth was used to obtain a resolution from 1 to 10 m in each image. Cultivated areas, water storage, bare soil, and lacustrine zones were the main zones obtained. For the identification of crops, the spectral signature was carried out by 6 bands from Landsat 5, based on the ratio of the amount of radiation reflected by a surface to the length of the electromagnetic wave. Knowing the spectral signature of the cover, allowed to classify the crop by pixel. Finally, a land-cover map with eight different classes was derived considering two years 1976 and 2000; one of those classes is the crops class having different values for the time series. Figure 4 shows the classes of land cover for analyzed years. Cultivated areas were found bordering the city of Chihuahua and the Chihuahua dam, having a greater presence in the extreme north of the aquifer, with an area of 143.95 km², equivalent to 7.62% of total coverage.

Based on the data provided by the JMAS and the information obtained by the (JCAS report 2013), it was approximated the total water withdrawal from the CHS aquifer for the years 1986–2010. This aquifer has 564 units (deep and shallow wells) of water abstraction with volumes of 72 hm³ yr⁻¹ (1986), 71 hm³ yr⁻¹ (1996), and 78 hm³ yr⁻¹ (2011) (JCAS 2013). Likewise, groundwater abstractions for agricultural use were estimated using the well-crop allocation process by satellite images, where the active crop areas are related to the corresponding well according to its location, hydraulic infrastructure, as well as the type and crop area. Thus, the water abstraction volumes for agriculture reported by JMAS are 12, 12, and 13 hm³ yr⁻¹ in 1986, 1996, and 2011, respectively. In the study area, 91 observation wells are the ones that have complete information over time to perform data analysis. Figure 5 shows the location of wells under study.

The model evaluated different recharge zones based on: the historical data of precipitation from six climatic stations placed in the aquifer area; the induced recharge from agricultural irrigation; the leaks of the water distribution system; land use and vegetation distribution; these recharge zones were determined by spatial analysis in the ArcGIS 10.3.1, considering the vegetation of 1976 and the vegetation of the year 2000 (Figure 4). The agricultural irrigation return was estimated from the crops identified in the area establishing the return volume of the range 20–40%, in the case of this model, the lower limit was taken based on climatology and geology, reaching values from 2 to 2.41 hm³ yr⁻¹. The induced recharge from leaks in the distribution network was determined from the percentage of area that the city occupies in the CHS aquifer and from the leakage percentages estimated according to the diagnostic study, modeling, and planning of sectors in the drinking water distribution network (JMAS-IMTA 2008), reaching recharge values from 1986 to 1996 of 4.53 hm³ yr⁻¹; from 1996 to 2009 of 5.20 hm³ yr⁻¹; and from 2009 to 2011 of 5.35 hm³ yr⁻¹.

It was found that 7 of the 15 zones within the model belong to agricultural zones. The spatially distributed recharge was calculated a priori based on water and land-cover use, specifying 15 zones for the groundwater model (Figure 6). The groundwater recharge was a boundary condition applied to the top layer of the groundwater model. The groundwater recharge zones were analyzed by the simulation results from the groundwater model using the postprocessor Zonebudget and the zonation file that groups model cells into the five zones was used as an input for a spatially distributed map of the groundwater budget for the years 1997, 2009, and 2011.

The JCAS (2013) developed a groundwater flow model of the CHS aquifer for the historical period 1986–2011 using VISUAL MODFLOW 2011.1 v.4.6.0.160. The model uses three layers to represent the current understanding of the geological environment. The hydraulic properties of the materials within these hydrostratigraphic units used to represent groundwater flow and selected boundary conditions were assigned to all three model layers that included groundwater pumpage, natural and artificial recharge, river gains and losses, and evapotranspiration. The initial conditions of groundwater flow and hydraulic loads were modeled in transient time.

The CHS model shows an area of 3,567 km² by the spatial discretization of the aquifer, with a finite-difference grid with 3 layers, 124 rows, and 119 lines, with a grid size of 0.5 km×0.5 km. The stratigraphic model was determined based on geological sections reported in the geological mining cross-sections derived from electrical resistivity modeling and well logs, lithological cuts provided by the National Water Commission (known as CONAGUA in Spanish), and the geophysical surveys provided by JMAS and JCAS. However, the lithological units bordering the aquifer were deactivated, leaving 7,556 active cells per layer, equivalent to an equal areal extent in all three model layers of 1,889 km² (Figure 7).

This model was designed with three layers: the first layer represents a combination of alluvial deposits representing the alluvial plain and part of the mountain range composed of the fractured rocks; the second layer is also a combination of the older alluvial deposits and fractured volcanic rocks; and the third layer corresponds to the same material as the second, but with more volcanic fractured rocks.

From the temporal point of view, the CHS model simulated the historical period from 1986 to 2011, assigning as the initial flow conditions, those of the year 1986, because it is the oldest date with extraction volumes reported within the aquifer. The temporal discretization was mostly annual stress periods with additional temporal refinement for 1996 and 2010 with 5–10 amplified time steps per stress period.

The CHS aquifer system is heterogeneous and anisotropic on a regional scale; it is mostly composed of alluvial filling with water flow direction from north to south. The relevant parameters of this aquifer system, hydraulic conductivity, specific yield, and storage coefficient, were assigned using zones with uniform properties. The initial parameter values of each zone were specified based on parameters estimated from the stratigraphic lithology, geomorphic unit, and sedimentary type of the aquifer. Sixty pumping tests were reinterpreted using the programs Visual Two-Zone Model and Aquifer Test, as well as the traditional methods of Logan and Eden-Hazel to determine the hydraulic conductivity, giving values from 0.08 to 16.53 m d⁻¹. Both, the hydraulic conductivity and storage coefficient were adjusted in the calibration of the transient flow model.

Analysis of trend is necessary to determine if conditions in a waterway, aquifer, or watershed are improving or deteriorating (Hirsch et al. 1982). Distribution-free trend analysis is ideal due to the unknown nature of the data, making non-parametric methods better suited for these data. The widely used non-parametric Mann-Kendall is considered one of the strongest correlation trend tests (Berryman et al. 1988). It is appropriate for data that are not normally distributed, tolerates missing values, and is relatively unaffected by extreme values or skewed data. The non-parametric Sen's method (Gilbert 2006) was performed to determine the slope of the trend for all variables (SPI, MSDIc, groundwater recharge, groundwater abstraction, and groundwater level) (Mustafa et al. 2017).

Related to the Mann-Kendall test, the Seasonal-Trend decomposition procedure based on Loess (STL) (Cleveland et al. 1990) was performed to determine whether or not significant changes have occurred over time while considering the variation due to seasonal effects (Hirsch et al. 1982). Average monthly time series of recharge and meteorological drought were decomposed by the STL method using the statistical software R. Groundwater abstraction and depth were not decomposed, as monthly data were not available.

Multiple linear regression was performed for each variable (abstraction, recharge, SPI, and MSDIc) to evaluate the relative contribution in increasing groundwater depth. Annual time series of 1986–2010 were used in the regression analysis to ensure the same length for all data. The partial coefficient of determination (R²) was estimated for each of the influencing factors.

To determine the most influencing factors, stepwise multiple linear regression analysis was conducted (Mustafa et al. 2017) according to Draper & Smith (1998).

The monthly time series of MSDIc and SPI (Figure 8) both show a negative trend. Exceptional meteorological drought occurs in 1994–1995 and 1999–2000, but in some years severe to extreme drought occurs mainly from 1995 to 1996 and 2000 to 2003. The patterns of the SPI cycle generally agree with those of MSDIc, displaying that SPI derived from rain gauges can detect severe drought episodes (De Jesús et al. 2016; Kavianpour et al. 2020).

Figure 8 represents the spatial distribution of MSDIc over the study area. MSDIc drought categories are varying from exceptional to abnormal or no drought within the study area. There are no clear spatial trends observed in MSDIc in the study area when analyzed on an annual basis, suggesting that a seasonal analysis may yield a more meaningful result.

The relation between the average groundwater depth with the annual MSDIc can be evaluated qualitatively, which shows that the groundwater depth is continuously increasing with little correlation to MSDIc or SPI. It is observed that although there is a negative trend in MSDIc and SPI values, both indexes do not have a continuous decrease in value, unlike groundwater depth which is continuously increasing. In 1986, the average groundwater depth was about 32 m from the surface. After 25 years, in 2010, the average groundwater depth had increased to about 92 m from the surface. On average, the difference increased monotonically with time, which can indicate a slight constant decrease in storage due to abstraction as stated in the study made by Van Loon & Van Lanen (2013).

It seems that in some years there is a relation between MSDIc/SPI and groundwater depth in the study area (Figure 9; Kavianpour et al. 2020). The exceptional meteorological drought in 1994 corresponds to an increase in groundwater depth in the consequent year of 1995. Between the years 2000 and 2003, an extreme long-term meteorological drought occurred, which can be related to a high increase in the depth of groundwater in 2003. In 2004, the MSDIc level was positive, which can be related to a moderate stabilization of the depth of the groundwater. The year 2007 had an abnormally wet MSDIc probably contributed to a recovery of the groundwater depth in the year 2008. The same behavior is observed in the whole period of study. The meteorological drought and groundwater depth reflect a similar behavior throughout the whole period of study (Breña-Naranjo et al. 2015). This means that the meteorological drought seems to have some effect, leading to groundwater level fluctuations, as Mustafa et al. (2017) reported in their study using SPI. Only a couple of studies present a conclusive relationship between drought and the water-level behavior (e.g. Edossa et al. 2016 or Garamhegyi et al. 2018) which establishes the level fluctuations with periodic behavior of drought, just as can be observed in this study.

Monthly groundwater recharge (Figure 10) shows that there has been a significant long-term trend in groundwater recharge over the last 30 years. Groundwater recharge is not uniformly distributed over the year, however, it shows that groundwater recharge is very low during the dry season (November to April) and very high in July, August, and September, in accordance with the wet season. Yearly groundwater recharge in the study area has varied between 72 and 253 mm yr⁻¹ over the last 30 years with a yearly average groundwater recharge value of 150 mm. Some parts of the study area were characterized by null or very low recharge, due to impermeable soil, mainly in the urban area. The area of greatest recharge occurs in the outcrop areas determined in Figure 4, as well as in the forested area located in the Sierra el Mogote and Azul.

For the year 1986, a surface of 1,104.73 ha was planted, with the extraction of groundwater of 12.04 hm³ yr⁻¹, for the year 1996 an area of 914.98 ha was planted, with the extraction of groundwater of 11.9 hm³ yr⁻¹ and finally for 2011, an area of 705.08 ha was planted with the extraction of 13.23 hm³ yr⁻¹ for the year 2009, there is only the volume of groundwater extracted, which is equivalent to 10.0 hm³ yr⁻¹. Concerning the previous data, there was a recharge for return of irrigation for the year 1986 of 2.41 hm³ yr⁻¹, for 1996 of 2.38 hm³ yr⁻¹, for the year 2009 of 2.0 hm³ yr⁻¹, and the year 2011 of 2.65 hm³ yr⁻¹.

Figure 11 represents the comparison of the deviation of groundwater recharge from the long-term average with the annual MSDIc. Here, the long-term average was calculated for each year. Negative values of the deviation mean that the recharge is less than the long-term average value (negative anomalies). Positive values of the deviation mean that the recharge is higher than the long-term average value (positive anomalies). Figure 11 shows that MSDIc has a direct relation with the groundwater recharge deviation (recharge anomalies), inferring that the variables that compose the MSDIc (rainfall, runoff, and soil moisture) are a very sensitive input for the recharge estimation (Van Loon & Van Lanen 2013). Both MSDIc and groundwater recharge deviation (recharge anomalies) show almost the same temporal pattern and dynamics. Based on the results, if the periodic behavior of water levels in the wells changes, it is expected that recharge from precipitation will decrease as well (Garamhegyi et al. 2018). The irrigation areas in the study area seem to be the most vulnerable to the detriment of groundwater recharge.

The public water use of the city and the exploitation of the CHS aquifer (Figure 12) present opposite trends, public use tends to increase over time due to various factors such as population growth rate, economic, and industrial growth. The public use is satisfied through: CHS aquifer, Sauz-Encinillas (SE) aquifer, Tabalaopa-Aldama (TA) aquifer, and superficial water. The main supplier of water for the city is the CHS aquifer, in the year 1985 it supplied 75% of the public use and the remain proceeding from TA aquifer, for the year 1997 the necessary infrastructure was used to boost the water supply through aquifers TA and SE increasing to 36% of the public use, at the end of the study period the aquifers TA and SE supplied a 49% of the water used in the city.

In contrast, the CHS aquifer presented a variability in the abstraction with a negative trend. From 1988 to 1993, there was a constant decrease in the abstraction of the CHS aquifer (Figure 12), this period presented mostly wet years that were combined with the implementation of intermittent water supply (IWS) (David et al. 2020) to satisfy the growing demand of the population.

From 1993 to 1997, a constant increase in the abstraction of the CHS aquifer can be appreciated along with the boost of water supply from the SE and TA aquifers. From 1998 to 2000, there is a decrease in the abstraction of the CHS aquifer despite being a dry season, in consequence, JMAS reduced the hours of water supply. The year 2004 is the one with the highest MSDIc value and coincides with the highest precipitation value with 577.7 mm, showing a decrease in the extraction of water in that year.

From 2005 to 2010, there is a decline in water extraction reaching the lowest values observed for the year 2008. As is seen in many other agricultural areas, extraction as groundwater pumping varies inversely with precipitation and directly with acres of production. The exploitation of the CHS aquifer, in general, is variable throughout the study period due to diverse factors, such as population growth rate, economic and industrial growth, agricultural use, precipitation, and the exploitation of other sources of groundwater (JCAS 2013).

The influence of groundwater abstraction on average groundwater depth can be evaluated qualitatively. Between 1988 and 1993, there is a stability of the groundwater depth that can be related to the decrease in abstraction that exists in the same range of years. From 1995 to 1997, there is a minor recovery of the static level that adjusts with the considerable decrease in groundwater exploitation in the same range of years. The trend of the average groundwater depth displays a monotonically increase in the deepening of the groundwater.

Table 2 gives the annual and monthly trend values from the Mann-Kendall trend test and Sen's method.

A positive (negative) value of Z_MK indicates that the data tend to increase (decrease) with time. It is observed that groundwater depth had a significant increasing trend, in contrast, groundwater abstraction, SPI, and MSDIc had a significant decreasing trend meaning that meteorological drought worsens (Hao et al. 2014). The Sen's slope value shows the average groundwater depth at a rate of 2.97 m yr⁻¹ even though the groundwater abstraction has been reduced at a rate of 0.41 hm³ yr⁻¹. Groundwater recharge demonstrates a significant decreasing trend, resulting from a complex approximation from induced recharge and natural recharge, the amount of natural recharge from precipitation is reduced as proven by the variables of SPI and MSDIc on the contrary the induced recharge increases due to various factors such as the need to use groundwater for irrigation (irrigation return), possible leaks in the drinking water network, and the increase of losses due to the setting of IWS.

The STL test was used to determine whether or not significant changes have occurred over time while taking into account variation due to seasonal effects. Figure 13 provides the details about decomposed time series of monthly average rainfall, groundwater recharge, SPI, and MSDIc. The patterns of meteorological index present a similar seasonal component emulating the results obtained by Rangel (2017). These findings are concordant with Teng et al. (2018), where precipitation decreased over the past five decades, varying from year to year.

MSDIc decreased with a slope of .0017 from 1986 to 2010, this value unambiguously represents a drought trend. As Rangel (2017) stated, the use of MSDIc has managed to capture the persistence of drought through its propagation process based on specific information. These meteorological trends accord with the global climate change warming trend (Meehl & Stocker 2007; Lin et al. 2018).

Table 3 gives the T-value and P-value of the different variables in predicting groundwater depth by multiple linear regression. In this study, only the groundwater abstraction contributes significantly to predict groundwater depth. These findings are in accordance with Bui et al. (2012) and Lu et al. (2014), who mentions that groundwater abstraction is the main influence factor in the changes of the groundwater level. The model obtained in this study differs from the findings of other researchers, where a larger percentage of the total variation in the groundwater level can be explained by the different models proposed by Li et al. (2017) and Mustafa et al. (2017). The distribution of the data showed obvious peaks, fat tails, or heteroscedasticity; making the robustness of the multiple regression model to be poor. The low model R² value (39.62%) indicates that the multiple linear regression assumptions could not be achieved, causing errors in the case of looking to produce a precise prediction.

The detriment of the groundwater level despite the decrease in the abstraction of the aquifer is related to the overcoming of the balance. The balance of the aquifer is estimated by quantifying the inputs, outputs, and the change of storage in a period, according to the JCAS (2013) study, the CHS aquifer presents for the years 1996 and 2009 inputs of 50.23 and 50.93 hm³ yr⁻¹ and outputs of 72.06 and 65.67 hm³ yr⁻¹, so there is a storage change of −21.83 and −11.73 hm³ yr⁻¹ respectively, exemplifying that despite the decrease in the exploitation of the aquifer, there is a constant negative evolution of the groundwater level because the recharge capacity (natural and induced) in the aquifer is constantly exceeded.

It is well established that groundwater pumping affects surface water availability by intercepting water that would otherwise discharge to streams, conversely, surface water management affects groundwater availability by altering the timing, location, and quantity of groundwater recharge and pumping. A lack of modeling tools capable of simulating interactions between surface water and groundwater affect the analyses of climate change impacts on water resources (Hanson et al. 2004) it is essential to create a model that sheds light on this interaction and is capable of baring the importance of interdecadal-climate cycles as controls on rates and mechanisms of climate-varying recharge and support the conclusion that understanding natural climate variability is a necessary step toward predicting groundwater response under climate change. Such understanding may help managers to better plan for the long-term sustainability of groundwater resources (Hanson et al. 2012).

This study develops a methodology for the analysis of influential factors in groundwater in a study area in which there is no complete and comprehensive database. This proposed methodology allows a primary evaluation of the state of groundwater use. This study comprehensively analyzes the influential factors on groundwater depletion in CHS aquifer in Chihuahua México. It was found that there are serious impacts of human activities on the groundwater system. This study area experienced a trend towards an increase in the intensity of the climatic drought from 1986 to 2010. The univariate standardized indexes (SPI) and the multivariate standardized index (MSDIc) allowed weighing the intensity of the main influential meteorological variables in climatic drought. The combined impacts of the change in land cover and the trend of increase in climatic drought have caused impacts in the reduction of groundwater recharge. According to the model obtained, it can be determined that the decrease that exists at the static level is significantly correlated with the increase in the amount of groundwater abstraction. The average groundwater level was 32 m for the year 1986, increasing to 92 m in 2010.

From the multiple linear regression model, it can be observed that the only significant variable that produces an effect on the groundwater level is the groundwater abstraction. The detriment of the groundwater level, despite the decrease in the abstraction of the aquifer, is related to the overcoming of the balance. Meteorological variables and groundwater recharge, despite maintaining a diminishing trend as the static level does, do not prove to be significant to the detriment of groundwater level. Due to various factors, such as low model robustness, the need for a greater number of observations (years analyzed), and the inaccuracy of the calculation of water withdrawal volumes, causes the model to have a low R². This low percentage of the response variable variation explained by the model does not allow to establish accurate predictions regarding the evolution of groundwater levels. An invariable monthly measurement of the static level in the different piezometers that the JCAS has, is indispensable to be able to evaluate the variability of the evolution of the static level according to the seasonality of the year.

The study area will face unprecedented challenges concerning the management of the groundwater resources to meet the increasing water demand of a growing population. The findings of this study could be beneficial to decision-makers and could help ensure adequate preparations of effective climate variability and change adaptions plans at national and local levels. Strengthening meteorological, hydrological, and groundwater monitoring is vital to ensure the data that can allow researchers to continue to elaborate knowledge that will shed light to endure water usage sustainability. Seasonal monitoring data would allow improving this model, making it possible to predict the behavior of the aquifer under different conditions. Additionally, the changes generated to the detriment of the groundwater level can cause changes in native water quality. Thus, a study to provide comprehensive data on groundwater quality to evaluate the sustainability to protect the groundwater source should be further explored.

Data cannot be made publicly available; readers should contact the corresponding author for details.