Alluvium

The highest yielding production wells in the study area are drilled into the alluvium. Previous investigators (Moyle, 1976; Woolfenden and Bright, 1988) distinguished the younger Holocene alluvium from the older Pleistocene alluvium and Pliocene to Pleistocene Bautista beds; these units were estimated from gravity data to be more than 1,000 ft thick in some places (Landon and others, 2015). Because the younger alluvium is mainly in the stream channels and above the groundwater table, the younger alluvium, older alluvium, and Bautista beds are considered one “alluvium” unit in this report. On a side note, the Interlocutory Judgment No. 33 (Santa Margarita River Watershed Watermaster, 2024a) describes the “shallow aquifer” to include the younger and older alluvial deposits to a maximum but variable depth of about 100 ft.

Wells completed in the alluvium are generally in the eastern Cahuilla Valley groundwater basin surrounding the town of Anza and in the central part of the Terwilliger Valley groundwater basin (fig. 5) where the alluvium is relatively thick. Because the younger alluvium is unsaturated in most places, most of the groundwater is from thick sequences of saturated older alluvium (Moyle, 1976). The range of specific capacity values for wells in the alluvium was 1.5–11 gal/min/ft at pumping rates ranging from 200 to 1,100 gal/min (Moyle, 1976; Woolfenden and Bright, 1988).

Decomposed and Competent Basement

The basement sourced materials that yield water to wells are moderately to highly decomposed or fractured. Wells open to the basement materials generally have a lower specific capacity and lower production rates than wells open to the alluvium unless the bedrock is highly fractured or decomposed. However, the DWR reported that a large irrigation well drilled in 1954 produced about 1,000 gal/min, and its high yield was notable because it was drilled in “crystalline rocks” (California Department of Water Resources, 1956). Landon and others (2015) reported that fractures of “considerable depth” potentially supply water to many wells completed in bedrock and that the permeable zone of the fractured bedrock extends through the weathered, or decomposed, part of the bedrock and into competent rock. Most of the wells completed in areas mapped as basement material are north of Durasno Valley and west of Anza, but Landon and others (2015) also reported that many wells in the groundwater basins have been drilled into the underlying bedrock where the alluvium is relatively thin. The range of specific capacity values from wells completed in the basement materials was 0.1–2.4 gal/min/ft and reported pumping rates ranged from 5 to 70 gal/min (Moyle, 1976; Woolfenden and Bright, 1988).

Groundwater-Level and Precipitation Data

Groundwater data have been collected for intermittent studies in the area since about the 1950s (California Department of Water Resources, 1956; Moyle, 1976), but more regular monitoring has occurred since fall 2013 (Woolfenden and Bright, 1988; Landon and others, 2015; U.S. Geological Survey, 2021). Since summer 2017, the USGS has collected groundwater data from selected wells throughout the study area as part of an ongoing monitoring network. Discrete and continuous groundwater-level data that were collected during this study from a total of 90 wells are shown in table 2; the number of wells monitored within the network changes each year depending on well accessibility and integrity, data quality, and wells that were added to replace wells no longer suitable for monitoring. In designing the current monitoring network, wells with historical records were given priority for inclusion to assess groundwater-level changes through time. Many wells with historical groundwater data targeted for inclusion could not be accessed, owing to difficulties in accessing the wells, well failure, or other obstructions.

To understand the relation between local recharge and discharge and the effects that these stresses have on groundwater levels, pressure transducers were installed in 18 monitoring wells in the Cahuilla Valley groundwater basin near the town of Anza and within the Durasno Valley (table 2). These transducers collected data at 1-hour intervals and were placed in areas where historical pumpage has been greatest or where rapid responses from recharge events were most likely to occur.

Groundwater-level measurements, reported as depth to water in ft bls, were collected by the USGS using calibrated steel or electric tapes in accordance with published USGS technical procedures (Cunningham and Schalk, 2011). All groundwater-level data were quality checked and entered into the USGS National Water Information System (NWIS; see the “Accessing Data” section; U.S. Geological Survey, 2021).

Two precipitation sites were installed by the USGS so that the data collected at a higher elevation could be compared to what is received at the lower elevation of the Anza Valley (fig. 5; table 2). Site 007S003E34E001S PRECIP is a climate response atmospheric site where groundwater-level and precipitation data are measured at 15-minute intervals. Another precipitation gage (THOMAS MTN PRECIP GAGE NR ANZA CA) was installed in 2017 to establish a record of local precipitation near Thomas Mountain where data are collected every 15 minutes. The site at the higher elevation on Thomas Mountain recorded a total cumulative precipitation of more than 48 in. from November 2017 to December 2021 (fig. 6A). Precipitation at the site near the town of Anza (007S003E34E001S PRECIP) recorded a cumulative total precipitation of about 43 in. from September 2017 to December 2021 (fig. 6B), indicating that the higher elevations received about 5 in. more precipitation than the valley over this 39-month period.

Electrical Resistivity Tomography

Previous USGS studies of the Cahuilla Valley and Terwilliger Valley groundwater basins defined the thicknesses and characteristics of the aquifer system and identified the location of wells completed in the groundwater-bearing units (Moyle, 1976; Woolfenden and Bright, 1988; Landon and others, 2015); however, the definition and connection between these units in most parts of the study area are not well understood.

Surface ERT (also known as direct-current resistivity) surveys provide a reliable method to distinguish between differing subsurface lithology types where sufficient contrast exists between the electrical properties of the deposits. These surveys can also provide valuable estimates for the cross-sectional area of the alluvium and estimates of depth-to-basement. Two ERT surveys were done about 2,150 ft apart to identify the thickness of the alluvium, its horizontal extent, and the depth-to-basement along two profiles perpendicular to Cahuilla Creek across a narrow section of Durasno Valley (fig. 5). The results from the ERT profiles were used to place four monitoring wells at two sites along these profiles in this valley and are discussed in the “Monitoring Wells” section; the lithology encountered by drilling the two monitoring sites were used to ground truth the ERT results.

Methods

In August 2018, an 8-channel SuperSting R8 resistivity/induced polarization meter (Advanced Geosciences, Inc., 2011) with a maximum of 56 electrodes was used to collect ERT data along two profiles in the Durasno Valley. In this report, the ERT profiles are numbered in downgradient order from the northeast (profile 1) to the southwest (profile 2), which is opposite of how they are numbered in the documentation by Ely and others (2020).

Along each profile, a linear array of electrodes spaced at regular intervals was placed in the ground, and several different combinations of current and potential electrode pairs were used to take apparent resistivity measurements. Stainless steel stakes, 18-in. long and approximately 0.5-in. diameter, were driven to a depth of approximately 12 in. Stakes were positioned every 23 ft along upgradient profile 1 (D–D′), which totaled about 1,590 ft in length, and every 16.4 ft along downgradient profile 2 (E–E′), which totaled about 1,120 ft in length (figs. 5, 7). The length of profile 2 was less than profile 1 because of the shorter distance to competent bedrock outcrops at each end of the downgradient profile 2.

Information about lateral variability in the subsurface is gained as different combinations of electrode transmitter/receiver pairs are translated across the array while the instrument reads from a command file. In this same manner, information about greater depths can be obtained by increasing the distance between transmitter and receiver electrodes (Minsley and others, 2010).

The transmitted currents are automatically controlled by the resistivity meter depending on the change in voltage between electrodes, which is controlled by the resistivity of the subsurface lithologic units. For profile 1, transmitted currents were between 6 milliamps (mA) and 1,080 mA, with an average current of 611 mA and a standard deviation of 213 mA. Transmitted currents for profile 2 were between 37 mA and 1,159 mA, with an average current of 783 mA and a standard deviation of 195 mA. An inverse Schlumberger array was used for its measurement efficiency and good contrast between lateral and vertical resolution (Minsley and others, 2010). Although the spacing of the electrodes differed slightly between the two profiles, these data were collected using the same array type and inverted using the same methods for interpretation, so the results are comparable. Details about the practical aspects of resistivity surveying techniques can be found in various references (Telford and others, 1990; Binley and Kemna, 2005; Day-Lewis and others, 2008; Milsom and Eriksen, 2011).

The resistivity data acquired were processed using the smooth, finite-element inversion method in EarthImager 2D software version 2.4.0, build 617 (Advanced Geosciences, Inc., 2011). The smooth-inversion method finds the smoothest possible model that fits the data to an a priori chi-squared statistic and assumes a Gaussian distribution of data errors. Images of the apparent resistivity data collected are shown in pseudosections, which are a conventional way to present the data but do not represent the true spatial distribution of resistivity values within the earth (Minsley and others, 2010). The pseudosections for the two profiles in Durasno Valley were used as starting models and are documented in Ely and others (2020). Surface topography was incorporated into the inversion to accurately account for electrode geometry and the effect of the irregular earth surface on the distribution of subsurface electric currents. Spatial data for all profiles were collected with the Trimble Geo7x differential global positioning system developed by Trimble, Inc., Sunnyvale, California. The data were corrected using a University NAVSTAR Consortium geodetic correction (UNAVCO Community, 2005). The inversion was allowed to run a maximum of eight iterations with a stop criterion of 5 percent or less root-mean-square error between the measured and modeled resistivity. If stop criteria were not met, the lowest quality data were considered outliers and removed using a percentage data misfit threshold, chosen to minimize data noise while retaining the maximum amount of data. The inversion was then repeated with the noisiest data removed. For a detailed discussion of the resistivity inverse problem, see Oldenburg and Li (1994) and Binley and Kemna (2005).

Monitoring Wells

As part of this study, the USGS installed four shallow monitoring wells along Cahuilla Creek in 2018 (figs. 5, 8) using the auger drilling methods described by Shuter and Teasdale (1989). Two wells were installed at each site to better understand the subsurface structure and the vertical and horizontal groundwater gradients in the eastern part of the Durasno Valley. The placement and depth of the monitoring wells were designed to coincide with the location of the ERT profiles. In this narrowest part of the Cahuilla Valley groundwater basin, where it is only about 1,500 ft wide, the volume of groundwater that flows from the Upper Cahuilla Valley through the Durasno Valley and into the lower part of Cahuilla Valley has not been evaluated previously.

As part of the drilling and construction of the four monitoring wells, downhole geophysical logs were collected to evaluate the subsurface lithology and to verify the ERT results (Ely and others, 2020; U.S. Geological Survey, 2023). Downhole geophysical logs, a generalized summary of the lithology encountered during the drilling, and the well construction information for the four wells are shown on figure 8. Wells 7S/3E-29M2 and -29M3, which are about 10 ft apart, are in the northeastern part of the Durasno Valley near the upgradient ERT profile 1 (figs. 5, 8B). Wells 7S/3E-30Q2 and -30Q3 are about 15 ft apart near ERT profile 2 (fig. 8A), about 2,150 ft downgradient to the southwest. The land-surface elevation at profile 2 is about 40 ft lower than at profile 1.

The lithology encountered during drilling at these sites was primarily sand and silts with interbedded clays. Determining the exact depths where the alluvium was present was difficult because the auger method can smear the materials when they are brought to the surface for examination. In these situations, the geophysical logs can help determine where predominantly finer-grained sediments—particularly clays—are present. At the upgradient site (wells 7S/3E-29M2 and -29M3; fig. 8B), the geophysical logs indicated that fine-grained sediments were present at about 39 ft bls. Heavily weathered bedrock was encountered at about 68 ft bls, and the auger was not able to penetrate any deeper. Basement (competent or decomposed) was not encountered in the downgradient site (wells 7S/3E-30Q2 and -30Q3; fig. 8A), but the ERT results indicate that the deeper well stopped at a transitional boundary that may have contained some basement rocks of unknown weathering or content, but this could not be definitively determined from the lithology.

Groundwater-level data from April 2019 from the wells at both sites showed that the hydraulic heads in the upgradient wells near profile 1 were about 6 ft bls, but the hydraulic heads were very different at the downgradient wells near profile 2 where artesian (flowing) conditions were initially observed in both wells. At the downgradient site, the initial hydraulic head in the deeper well 7S/3E-30Q2 was about 12 ft above land surface and had a higher hydraulic head than the shallower well 7S/3E-30Q3, where groundwater was only slightly above land surface (fig. 8A). Continued groundwater-level monitoring of these sites indicates that artesian conditions have been present in the deep well 7(S/3E-30Q2) since it was installed and that groundwater flows from the well at about 1 gal/min when the well cap is removed. The movement and character of groundwater flow in this valley is discussed further in the “Groundwater Flow, Levels, and Movement” section.

Well Logs

Lithology information from drillers’ log descriptions from about 1,185 well logs (fig. 4; Landon and others, 2015; Fenton and others, 2020) provided information on the subsurface geometry of the alluvium, decomposed basement, and competent basement, and the variability of major lithologic textures within the alluvium. Most well logs were obtained from the DWR Well Completion Report database (California Department of Water Resources, 2022). Some well logs that were not available in the DWR Well Completion Report database were available from hardcopy or digital files of the USGS San Diego Projects office files; these logs were primarily from private water wells drilled more than 50 years ago or on Native American land. Well-test information was also compiled from the well logs where available (Fenton and others, 2020). Summary descriptions of well construction, location (including source and accuracy), regularized lithology information, and well-test data were published by Fenton and others (2020) and Shepherd and others (2022).

Drillers’ log descriptions were of variable quality and had a range of detail in the lithologic information (Landon and others, 2015; Fenton and others, 2020; Shepherd and others, 2022). The logs varied in terms of the quantity and thickness of described subsurface lithologic intervals, the detail of the descriptions, and the terminology used to describe sediment types and textures. Quality assurance analysis of the well logs and lithologic descriptions verified the physical location of each well and evaluated the suitability of the lithologic descriptions for understanding the distribution of different lithology types across the study area. Some well logs were removed from consideration because their reported well locations could not be verified or were inconsistent with nearby wells, and some well logs were removed because they had ambiguous or imprecise lithologic descriptions. A final set of 579 well logs and lithologic descriptions was compiled by Shepherd and others (2022) from existing well-log data (Fenton and others, 2020) for use in developing the GFM and lithologic characterization.

The drillers’ log descriptions were classified into 54 regularized categories using an approach similar to Faunt (2009) and Landon and others (2015). These regularized lithology categories were used to help interpret the depth and thickness of geologic units in the GFM (Shepherd and others, 2022). The regularized lithology classes were further simplified into three generalized texture categories that loosely correspond to grain size: (1) coarse gravel, cobbles, boulders, and crystalline rock types; (2) medium sand, sandstone, and decomposed material; and (3) fine clay, silt, mud, and soil. These three textural groups were used to map the distribution of lithology in the alluvium and to show the vertical and spatial distribution of textures in the alluvium along selected sections (see the “Framework Model Construction” section).

Framework Model Construction

The GFM was constructed from geologic maps (Rogers, 1965; Dibblee and Minch, 2008), well-log lithologic descriptions (Shepherd and others, 2022), estimated productivity of wells (reported as specific capacity data; Shepherd and others, 2022), and gravity-derived depth-to-basement estimates (Landon and others, 2015). Three geologic units were modeled in the GFM: (1) alluvium, (2) decomposed basement, and (3) competent basement. The alluvium unit is primarily within the Cahuilla Valley and Terwilliger Valley groundwater basins (fig. 4). In the GFM, the alluvium unit was assumed to be present everywhere that it was mapped at land surface and was assumed to fill the subsurface volume between the top of the decomposed basement unit and land surface. The decomposed basement unit is comprised of plutonic or metasedimentary rocks (fig. 4) that have been physically or chemically weathered in situ. Although there may be some areas where basement occurs without the presence of overlying decomposed basement, most well-log lithology descriptions (Shepherd and others, 2022) and field observations indicated that decomposed basement was present in most areas across the study area. Therefore, in the GFM, a relatively thin zone of decomposed basement is assumed to overlie basement everywhere that the competent basement unit is present. The competent basement unit is comprised of plutonic or metasedimentary rocks that have not been physically or chemically weathered. Competent basement rocks are mapped in outcrop and are assumed to underlie the alluvium and decomposed basement. In the GFM, the competent basement unit is assumed to be present throughout the study area.

Surface contacts of the geologic units were determined using geologic map information (fig. 4). Subsurface contacts of the three geologic units were determined using well-log lithology descriptions (Shepherd and others, 2022), gravity-derived depth-to-basement estimates (Landon and others, 2015), and estimated well productivity data (Shepherd and others, 2022). Varying quality and detail of well-log lithology descriptions sometimes made subsurface geologic unit interpretations from lithology descriptions difficult to ascertain. Therefore, the gravity-derived depth-to-basement estimates and estimated well productivity data were used as supporting information for interpreting geologic unit contacts for each well log.

Gravity-derived depth-to-basement estimates corresponded to well-log lithology descriptions of either the decomposed or competent basement in different parts of the study area. Estimates of depth-to-basement from the gravity data corresponded to decomposed basement in most well logs in the Cahuilla Valley groundwater basin, but corresponded to the competent basement in the Terwilliger Valley groundwater basin. Estimated productivity of wells, represented by specific capacity values, was used to help distinguish between the three geologic units. Specific capacity is defined as pumping rate divided by drawdown (unit volume per unit time per unit length; Freeze and Cherry, 1979), and is expressed as gal/min/ft. Following Woolfenden and Bright (1988), specific capacity values greater than 2.4 gal/min/ft were assumed to indicate wells perforated in alluvium, and wells with specific capacity values less than or equal to 2.4 gal/min/ft were generally assumed to be perforated in decomposed or competent basement.

The location and type of data used to interpret geologic unit contacts are shown on figure 9. The top of the alluvium unit was interpreted solely from land surface elevation; however, there were wells that encountered the alluvium (fig. 9A), which were used to map the distribution of textures throughout the alluvium with a method described below. Subsurface data for the top of the decomposed basement unit were from well-log lithology data, gravity-derived depth-to-basement estimates, and synthetic control points. Synthetic control points were used to enforce unit geometry and thickness in the absence of appropriate well-log or depth-to-basement data. Subsurface data for the top of the competent basement unit were from well-log lithology data and gravity-derived depth-to-basement estimates.

Interpreted geologic unit contacts were used as input data to construct the GFM. Input data for each geologic unit were initially gridded for EarthVision (version 11.0), a three-dimensional geologic-modeling software package that uses a biharmonic cubic-spline interpolation algorithm to produce minimum-tension (minimum-curvature) grids from x, y, z point data (Dynamic Graphics, Inc., 2021). The input data were interpolated to create a grid for the top surface of each unit. The horizontal discretization of the grids was 160 ft in the x (east-west) and y (north-south) directions.

The initial geologic unit grids were exported from EarthVision to a geographic information system (GIS) software (Esri ArcGIS version 10.7.1) and were manually adjusted to ensure that the units’ top grids were consistent with basic geologic principles and the geologic understanding of the Cahuilla Valley and Terwilliger Valley groundwater basins. The units’ top grids were clipped to land surface using a discretized digital elevation model (DEM) grid based on the National Elevation Dataset 10-meter DEM (U.S. Geological Survey, 2019). A minimum thickness of 10 ft was assigned to the alluvium wherever it was interpreted to be present. This value is typically the smallest interval of reported lithologic data from well logs. A minimum thickness of 20 ft was assigned to the decomposed basement within the Cahuilla Valley and Terwilliger Valley groundwater basins wherever the alluvium was present at land surface; this value was chosen because it was generally the smallest thickness of decomposed basement identified in well logs. In areas where plutonic and metasedimentary rocks were mapped at land surface, and in all areas outside of the Cahuilla Valley and Terwilliger Valley groundwater basins (fig. 4), the decomposed basement thickness was set to 75 ft. This value was the mean thickness of decomposed basement from the well-log data (for reference, the median thickness of decomposed basement was 68.5 ft with a standard deviation of 49.9 ft).

Lithology data (Shepherd and others, 2022) from well logs shown on figure 9A were interpolated within the alluvium to evaluate the spatial and vertical distribution of lithologic texture classes within that unit. Lithologic data from well logs were characterized into three generalized textural groups (coarse, medium, and fine; see the “Well Logs” section) and interpolated across the alluvium using Rockworks three-dimensional geologic-modeling software (version 20; RockWare, 2021). The interpolation used a lateral blending algorithm that extends lithologic data from a given well and randomizes lithologic correlations within the middle one-third space between adjacent wells (RockWare, 2021). The resulting lithologic interpolation is a solid voxel model of the three generalized texture classes within the alluvium. The lithologic interpolation is continuous between wells in areas with many well logs but is absent in areas with few or no well logs (fig. 10).

Framework Model Results

The GFM provides an improved high-resolution understanding of the hydrogeology in the Cahuilla Valley and Terwilliger Valley groundwater basins. The resulting thickness and extent of the alluvium and decomposed basement, the variability of lithology and grain size within the alluvium, and the structural geometry of the competent basement are discussed in this section. Maps of the modeled thicknesses of the alluvium and decomposed basement and the modeled elevation of the top of the competent basement are shown on figure 9. Sections through the GFM show the geometric relation of the modeled units and vertical faults, along with the depth of selected wells (fig. 10). Additionally, the sections show the interpolated lithologic texture distribution within the alluvium.

The alluvium is present throughout much of the Cahuilla Valley and Terwilliger Valley groundwater basins and overlies the decomposed basement. The average thickness of the alluvium is about 140 ft (fig. 10A); this unit is thickest in the Cahuilla Valley and the central part of the Terwilliger Valley groundwater basins, where it is estimated to exceed 450 ft thick, and in the northeastern part of the Cahuilla Valley groundwater basin along the San Jacinto fault zone, where it is estimated to exceed 4,000 ft thick. Estimates of thick alluvium are in areas with limited borehole data and in areas where the tops of the decomposed basement and competent basement are estimated to be relatively deep.

The decomposed basement unit is present throughout the study area and underlies the alluvium where alluvium is present; it overlies the competent basement unit everywhere. Decomposed basement is modeled to outcrop everywhere at land surface except where it is overlain by alluvium. Within the groundwater basins, the minimum thickness of the decomposed basement is 20 ft, the maximum thickness is about 230 ft, and the average thickness is about 71 ft (fig. 10B).

The competent basement forms the hydraulic base of the study area, underlying the decomposed basement and alluvium (fig. 10C). The elevation of the top of the competent basement follows the general topographic trend of land surface in the study area, whereby the unit is at the highest elevations at the northeastern and southern margins of the study area (elevations greater than 6,700 ft), and at the lowest elevations at the western margin of the study area (elevations less than 3,000 ft). The competent basement forms topographic highs at the boundaries of the Terwilliger Valley and Cahuilla Valley groundwater basins (figs. 9C, 10A). Within the groundwater basins, competent basement forms topographic lows, providing depocenters in which alluvium has accumulated. The most substantial topographic low is in the northern part of the Cahuilla Valley groundwater basin along the southwestern side of the San Jacinto fault zone; there, the top of the competent basement varies more than 6,500 ft between its lowest point in the groundwater basin and its highest point near Thomas Mountain (fig. 10C).

Lithology texture data form the geologic basis for estimating the hydraulic properties that are used in numerical groundwater-flow models (Faunt and others, 2010; Alzraiee and others, 2022). The alluvium is comprised mainly of medium-grained lithologic textures (fig. 10) in most places; the percentage of fine-grained textures is greater in parts of the Terwilliger Valley groundwater basin (figs. 10A, 10C). Coarse-grained textures are present in the alluvium throughout the study area, although typically in small pockets.

Natural Recharge

The primary sources of natural recharge to the study area are from the infiltration of runoff from the surrounding higher elevations of the San Jacinto Mountains to the northeast and from southeast of the Terwilliger Valley. Recharge also occurs, though in lesser quantities, as surface water from precipitation that is not captured and used by plants (evapotranspiration [ET]). When present, this precipitation can infiltrate the ground surface along the many small ephemeral creeks and washes into the alluvium in the valleys or into the fractures and weathered zones of the basement rocks in upland areas. Recharge and runoff do not occur in the same amounts every year; when very wet years occur, most of the water becomes runoff, and a lesser component becomes recharge. Both recharge and runoff are controlled by the variable cycles of wet and dry weather and are temporally variable across the region (Stern and others, 2021). In addition to being temporally variable, recharge and runoff are spatially variable across the region; recharge frequently occurs outside of the groundwater basins and rarely over the groundwater basins’ footprint (Flint and Flint, 2007). Because natural recharge is related to the variable cycles of precipitation, estimates are difficult to quantify without hydrologic models; however, the previous estimate of recharge by Moyle (1976) and those done as part of this study (Stern and others, 2021) are presented in table 3.

Moyle (1976) estimated the amount of groundwater recharge based on annual precipitation records for the 50-year period from 1897 to 1947 and assumed that 95 percent of precipitation was used by native vegetation or lost through flood flow in stream channels. The precipitation from that period ranged between 5.21 in. (in 1956) and 22.38 in. (in 1943), and the average annual precipitation ranged from 16 to 30 in. Based on these averages, Moyle estimated that the average annual groundwater recharge was about 3,800 acre-feet/year (acre-ft/year).

In arid basins where surface-water data are lacking, the potential in-place recharge and potential runoff into the groundwater system can be estimated using the regional-scale Basin Characterization Model for California (CA-BCM; Flint and others, 2013). The CA-BCM applies a monthly regional water-balance model to simulate hydrologic responses to climate and estimate basin recharge and runoff. The CA-BCM has been used in other desert basins where surface-water data are sparse or nonexistent (Flint and Martin, 2012; Faunt and others, 2015). For the Anza area, stream-flow data from nearby and similar basins, snowpack and the timing of snowmelt, temperature, types of vegetation, and precipitation in the San Jacinto Mountains were used to estimate the average annual potential runoff and recharge. The area that was locally calibrated and validated by Stern and others (2021) was slightly larger than the footprint from Moyle (1976) and encompassed streams and creeks to characterize the water-budget components, including recharge, runoff, stream flow, actual ET, and climatic water deficit. Although some portion of the estimated runoff becomes recharge, the CA-BCM currently does not route water to the groundwater system, and runoff in the CA-BCM is routed only through stream channels; therefore, the amount of runoff that potentially becomes recharge could not be estimated.

Recharge and runoff have extreme interannual variability in arid regions, including the study area; the occurrence of recharge and runoff can be sporadic, so the long-term average value for recharge is not a reliable estimate for a particular year in this region (Stern and others, 2021). For years when precipitation is less than about 8–12 in., the CA-BCM estimated that recharge was negligible, so it is important to note that the estimates in table 3 represent long-term averages, including years when recharge values were zero. The relation of recharge to precipitation is exponential in arid and semi-arid regions and often requires the exceedance of a precipitation threshold to produce substantial recharge or runoff (Stern and others, 2021). This relation is noted because in drier years, the ratio of recharge to runoff is higher, and recharge per unit area can be equal to, or higher than, runoff in some areas. For this area, the CA-BCM estimates for recharge and runoff were highly variable and highly dependent upon which years were used.

The variability of recharge and runoff is best illustrated when comparing estimates from different periods. For example, during water years 1896–2018, average monthly recharge ranged from 0 to 11,000 acre-ft (a water year is the period from October 1 to September 30 and is designated by the year in which it ends; for example, water year 2021 was from October 1, 2020, to September 30, 2021). During that period, the average recharge was lower than the average runoff and was estimated to be about 4,900 acre-ft/year (Stern and others, 2021; table 3). During 1981–2010, when dry years occurred more frequently than wet years, the average annual runoff remained about the same, but recharge was estimated to be about 4,400 acre-ft/year. For 1971–2000, which included the highest peak flow on record in 1980, the long-term average recharge was estimated to be about 5,900 acre-ft/year. These examples demonstrate that applying one average annual value of recharge and runoff to this region is not a reliable estimate for any specific year because of the extreme interannual climatic and spatial variability.

Anthropogenic Recharge

The town of Anza (fig. 5) and the surrounding community are unincorporated rural areas and therefore rely on private septic disposal systems to dispose of wastewater. The infiltration of water from this source is difficult to quantify, and the volume that reaches the groundwater table likely is small; however, a rough estimate of the potential recharge from private domestic septic systems can be made using an assumed volume of effluent discharged to septic systems and the reported population. Umari and others (1995) used an average septic tank effluent of 70 gallons per day (gal/d) per person to estimate the quantity of septic wastewater that mixed with the underlying groundwater in Victorville, a desert community about 100 mi to the northwest that relies solely on groundwater. Using the 2020 population estimates for the study area of about 6,480 residents (Esri Data Development, 2023), the amount of potential recharge from septic effluent may have been as much as 500 acre-ft.

Irrigation return is groundwater that has been pumped locally to irrigate agricultural fields but is not consumed by the crops and subsequently infiltrates the groundwater system. The amount of water that is lost through plant use and evaporation and does not return to the groundwater table is considered consumptive use. Estimates of recharge based on crop type, crop efficiencies, and irrigation requirements can be made with accurate estimates of pumpage, percentage of irrigated land, reference ET, and crop coefficients. Estimates of irrigation-return volumes are complicated by many factors, including evaporation and potential significant lag times in arid environments, and are best estimated by hydrologic models that account for variable irrigation efficiencies, soil moisture, the thickness of the unsaturated zone, and complicated crop-irrigation practices and rotation. Stamos and others (2001) estimated irrigation-return flow in the upper part of the Mojave Desert that ranged between 29 and 46 percent of pumpage; however, modern farming methods have increased the efficiency of irrigation, resulting in decreased irrigation-return rates over time.

Evapotranspiration and Evaporation

Evapotranspiration is the process that removes soil water or groundwater from the subsurface through plant transpiration or direct evaporation. To calibrate the CA-BCM model for estimating recharge (see the “Sources of Recharge” section), Stern and others (2021) used estimates of actual ET from remote sensing and water-balance data for January 2000–December 2013 at 1-kilometer resolution. Estimates of actual ET for four native vegetation types from that study varied greatly and ranged from less than 0.5 in., about 10 millimeters, to about 4 in., or 110 millimeters.

Estimates of potential evapotranspiration (PET), which is the amount of water that would be evapotranspired from an unlimited water supply, were used for this study to determine the maximum amount of water lost through plants. Often, PET is derived using the value for crop coefficient (Kc), a unitless number that represents a crop’s maximum potential water use that varies by plant type, phenology, soil moisture, and crop distribution, multiplied by reference ET (Eto). Defined as the amount of water evapotranspired from well-watered grass of a uniform height, Eto varies based on regional climate patterns (Allen and others, 1998). The California Irrigation Management Information System (CIMIS; California Department of Water Resources, 2012) divides California into zones based on long-term monthly average Eto using data from CIMIS weather stations. Based on the CIMIS weather station data, an Eto of 62.5 inches per year (in/year) was used to determine a course estimate of ET in this area.

Estimates of PET rates for this study area for selected years were made by applying Kc values (Allen and others, 1998; Pereira and Alves, 2005; Kjelgren and others, 2016; table 4) to publicly available land-use data for the years 1934 (figs. 3, 11; Swift and Sabourin, 2000), 1945, 1972, 1973 (Moyle, 1976), 1986 (Woolfenden and Bright, 1988), 1990, 1993, 2001, 2005 (Southern California Association of Governments, 2005), 2012, and 2016 (Southern California Association of Governments, 2019). The fourteen different land-use classes and associated Kc values ranged from 0.20 for bare soil to 1.15 for irrigated potatoes and grain. The Kc values for crops represent mid-season single crop classes (Allen and others, 1998; Pereira and Alves, 2013; Kjelgren and others, 2016). Based on the Kc values and the crops present during the selected years shown on figure 3, the maximum PET rates in those years, assuming crops were able to obtain all water required for the entire year, ranged from 24.7 in/year in 1972 to 22.3–21.3 in/year in 2016 (fig. 11). Lower rates in the later years reflected the conversion of native lands to urban developments and agricultural lands.

Direct evaporation occurs from the surface of open water and in areas where groundwater is near land surface. Lake Riverside, a man-made reservoir built in 1962 and fed by pumped groundwater, has an area of about 75 acres and is west of the Cahuilla Reservation (fig. 3K). A few much smaller irrigation ponds near the town of Anza, shown as “Water” on figure 3, occasionally are used to temporarily store groundwater when crop demand is greater than the yield of agricultural pumps. Moyle (1976) estimated that the evaporation from these lakes and the reservoir was about 750 acre-ft/year; Woolfenden and Bright (1988) estimated that the evaporation rate was about 740 acre-ft/year in 1973 and about 580 acre-ft/year in 1986. Based on the area of Lake Riverside in 2016, the Kc for “Deep water” (table 4), and the annual evaporation rate of 62.51 in. from CIMIS (California Department of Water Resources, 2012), estimates of the volume of water evaporated from the reservoir was about 390 acre-ft/year.

When considering the potential amount of water evaporated from the irrigation ponds, it is important to note that these ponds do not contain water year-round like a reservoir. Assuming that these ponds, which covered an area of about 58 acres in 2016 (fig. 3K), contain water about one-third of the year and have a Kc of 1.05 (see “Shallow water” in table 4), then an estimate of the amount of water evaporated from them was only about 13 acre-ft/year. The combined potential amount of water evaporated from the open water bodies is estimated to be about 400 acre-ft/year.

Determining the potential rate of groundwater lost through evaporation where the water table is shallow includes factors that are variable and estimates for this differ widely. Moyle (1976) estimated that the rate of evaporation from groundwater at land surface over 315 acres was about 1,670 acre-ft/yr in 1973, but Woolfenden and Bright (1988) estimated a value that was much larger for 1986, about 3,870 acre-ft/yr from 730 acres. These estimates probably assumed that groundwater was near the land surface over a much larger area than was observed over the course of this study; also, it is likely that the groundwater table was shallower in many places in those years. During this study, evidence of shallow groundwater as saturated terrain and marshy conditions was observed mainly in the area of about 485 acres along Cahuilla Creek, east of Durasno Valley. Assuming a Kc value for pasture, the estimated amount of groundwater lost through evaporation was lower and likely about 2,400 acre-ft/yr at most for wet years and much less for drier years.

Groundwater Pumpage

Groundwater has been used as the sole source of water for Native American Tribes, agriculture, and for municipal and domestic supply. Most of the groundwater has been used for agriculture, which likely started near the turn of the 20th century, sometime before 11 irrigation wells were first observed in 1916 by Waring (1919). Pumpage has not been metered, and previous investigators have attempted to estimate the volume of groundwater extracted by pumpage based on land use. In 1953, the DWR inventoried 37 wells (California Department of Water Resources, 1956), and in 1968–69, the DWR compiled information for about 415 unused and active wells, noting that about 535 acres of land were being irrigated (California Department of Water Resources, 1974). Moyle (1976) estimated that, in 1973, 457 acres were being used for irrigated crops and that groundwater was being used at a rate of about 4,200 acre-ft/yr for irrigation, natural pasture, replacement of evaporation from lakes and reservoirs, domestic supply, and livestock. Woolfenden and Bright (1988) reported that, by 1986, there was an increase of about 6,000 acre-ft/yr in consumptive groundwater use since 1973, and the authors estimated that agricultural use increased by 1,820 acre-ft/yr. In comparison, evaporation from natural pasture was estimated to have increased by 2,200 acre-ft/yr and domestic use by 420 acre-ft/yr.

More recent estimates for “substantial water users” were compiled for 1991–2021 from annual reports by the Santa Margarita River Watershed Watermaster (Santa Margarita River Watershed Watermaster, 2024b; referred to hereafter as the Watermaster) and are shown on figure 12. As defined by the Watermaster, substantial water users are water purveyors, Indian reservations, mobile parks, Lake Riverside (fig. 3K), and private landowners who irrigate 8 or more acres or use the equivalent quantity of water. The Watermaster estimated that the total annual pumpage by substantial water users in the Santa Margarita River watershed for 1991–2021 ranged from about 510 acre-ft in 2019 to about 3,080 acre-ft in 1994. Pumpage from about 1,200 “de minimis” users (those who irrigate less than 8 acres or the equivalent amount of water) were not included in the pumpage estimates by the Watermaster (M. Preszler, Santa Margarita River Watershed Watermaster, written commun., 2023).

Pumpage for agriculture was highest between 1991 and 2007—during that time, the average pumpage rate was about 2,030 acre-ft/yr. Pumpage decreased after 2007, and the average agricultural pumpage during 2008–21 was about 1,190 acre-ft/yr (fig. 12). In 2016, 2019, 2020, and 2021, agricultural pumpage was about the same or less than other uses, particularly for Lake Riverside. Although Lake Riverside is considered a substantial water user by the Watermaster, it does not deliver water to customers, and the groundwater pumped for Lake Riverside is used to maintain the water in the lake that is lost through evaporation.

Groundwater used on the Cahuilla Reservation varied between about 40 and 420 acre-ft/yr and was used for grazing of cattle on non-irrigated acreage; it was also used for domestic and commercial uses, which included watering of turf grass, dust control, and for supplying water to a casino (Santa Margarita River Watershed Watermaster, 2024b). The amount of reported pumpage for municipal use was much less than other uses and did not exceed about 50 acre-ft/yr.

Domestic pumpage other than on the Cahuilla Reservation was not reported by the Watermaster, so domestic use was estimated from population estimates and per capita water use within the study area boundary. Population data in the study area from Manson and others (2019) for 1990, 2000, and 2010 was about 3,470, 4,550, and 6,110, respectively, and was about 6,480 in 2020 (Esri Data Development, 2023). The population was estimated during each decade by linearly interpreting between years when population data were reported; population estimates for 2021 were assumed to be the same as 2020. The population estimates, excluding the Cahuilla Reservation, were then multiplied by the 2010 per capita water usage of 87 gal/d per person (Dieter and others, 2018) to estimate domestic water use. Based on these data, domestic pumpage was estimated to range between about 340 and 610 acre-ft/yr between 1991 and 2021.

The estimated total pumpage for 1991–2021 ranged from about 1,140 acre-ft in 2019 to about 3,450 acre-ft in 1994. When summed, the cumulative amount of estimated pumpage between 1991 and 2021 was about 81,400 acre-ft. Although domestic pumpage has been estimated for this report, the amount of pumpage by de minimus users is unknown, and their use may be significant in some years (M. Preszler, Santa Margarita River Watershed Watermaster, written commun., 2023); therefore, the actual annual pumpage volumes were higher than those shown on figure 12.

Short-Term Trends in Groundwater Levels

An examination of short-term trends in groundwater levels can show how some parts of the basin respond quickly to recharge and discharge, at least hydrologically, because depths to groundwater are shallow in many places and are affected by the variable cycles of wet or dry climatic periods. The effects of the interannual variability on recharge in this arid region were discussed in the “Natural Recharge” section and are evident when comparing groundwater-level responses after short-term precipitation events.

To show any potential effects of short-term local recharge events and stresses from pumpage in different areas, hydrographs were constructed using groundwater-level elevations from August 2017 to December 2021 from 10 wells and combined with data collected from two precipitation sites installed as part of this study (figs. 14, 15). The short-term responses to precipitation events provide information about the thickness of the alluvium and depth of the decomposed and competent basement, and the effects of local recharge, if present. Groundwater levels in wells that are outside the Cahuilla Valley groundwater basin, or in areas where the alluvium is thin, showed more pronounced responses from precipitation events than wells where the alluvium is thickest—mostly within the groundwater basin boundary (fig. 9A). For example, the hydrograph for well 7S/2E-13R1 (hydrograph 1 on figs. 14 and 15) shows that groundwater levels increased rapidly after a large storm event on February 14, 2019. As a result of that storm and others in the following winter and spring months, groundwater levels in well 7S/2E-13R1 rose almost 50 ft by June 2019, demonstrating a response to local recharge over about a 4-month period. This well is not near any local creeks, and according to the driller’s log (California Department of Water Resources, 2022), it is completed mostly in decomposed basement and competent basement which extend to at least 70 ft bls in this well. Heavy local precipitation may have occurred near the well that was not recorded by the precipitation gages, but the response in this well more likely represents lateral movement of groundwater through the decomposed basement. The rapid response to storm events in this and other wells completed in the decomposed basement and competent basement shows that local recharge infiltrates and then dissipates quickly and that these materials do not have much capacity for the long-term storage of groundwater compared to the alluvium.

Another well on the edge of the Cahuilla Valley groundwater basin boundary also showed a rapid, but more muted and prolonged response, to the storms in early 2019. Well 7S/3E-20A2 (figs. 14, 15, hydrograph 2) is about 100 ft from Cahuilla Creek, which had an estimated peak flow of 59 ft³/s from the February 2019 storm event (table 1). Groundwater-level elevations in this well rapidly increased by almost 5 ft, reached a peak in early March, and plateaued through May (figs. 14, 15, hydrograph 2). In this well, the rise in groundwater elevations began about 3 weeks after the largest storm event in February 2019. The driller’s log from this well indicates that it is perforated in granite (California Department of Water Resources, 2022), or basement rocks, as described in this report.

Wells that are further within the Cahuilla Valley groundwater basin, such as wells 7S/3E-21D4 and 7S/3E-23D3 (figs. 14, 15, hydrographs 3, 4), showed little response to the storms in the winter and spring of 2019. These wells are in the area where the alluvium is much thicker and have small but steady groundwater-level declines. For example, the groundwater levels in well 7S/3E-21D4 declined by slightly less than 1 ft between 2018 and 2021. Other wells, such as 7S/3E-34E1 (figs. 14, 15, hydrograph 6), show small, short-term fluctuations in groundwater levels, which are mostly due to varying pumping cycles in nearby wells.

Well 7S/3E-13C2 (figs. 14, 15, hydrograph 5) had subtle changes in groundwater levels, but these changes were not concurrent with any recorded precipitation event(s) nor did they show any effect from nearby pumping. The groundwater levels in this well increased steadily over 4 ft from 2019 to 2020 and reached a maximum in summer 2020, then steadily decreased through December 2021. This well is within the San Jacinto fault zone and showed almost no changes in groundwater levels related to precipitation events. The data for all the wells discussed herein are available on the USGS NWIS website (U.S. Geological Survey, 2021), as described in the “Accessing Data” section.

In general, the differences in groundwater-level responses in wells were due to several factors, including but not limited to, the following: (1) the hydrologic properties of the aquifer to transmit water both vertically and horizontally; (2) proximity to recharge sources; (3) variation in the spatial and temporal distribution of precipitation; (4) hydraulic properties of the aquifer to store water; (5) the thickness of the alluvium or basement materials in which wells are screened; or (6) a combination of these factors. Short-term groundwater-level fluctuations occurring over weeks or months are indicative of the limited capacity of the aquifer system for groundwater storage in some areas and the sensitivity of this basin to the amount and availability of recharge. In areas where sustained historical pumpage has occurred, primarily from the alluvium for agriculture near and northeast of the town of Anza, groundwater levels are deeper and generally do not show the short-term effects of local recharge from storm events or surface runoff.

Long-Term Trends in Groundwater Levels

Compiling long-term groundwater-level data from wells gives insight as to how stresses such as recharge and discharge (mainly pumpage) have affected the movement and direction of groundwater flow over decades. In most areas, a historical record of continuous groundwater-level data from a single well is rare, so data from wells that are near each other are often combined. Using groundwater-level data that were available from 1916 (Waring, 1919), 18 long-term hydrographs were constructed within the Cahuilla Valley groundwater basin, showing trends from 1916 to 2021. Most wells near the boundaries and outside of the Cahuilla Valley and Terwilliger Valley groundwater basins showed more rapid responses to stresses, and some wells outside of the groundwater basin had groundwater levels that remained mostly unchanged or had very little decline. In contrast, wells within the groundwater basins generally showed the effects of long-term pumping.

The data collected between 1916 and 2021 show that changes in groundwater levels within the boundary of the Cahuilla Valley groundwater basin were variable and dependent on location. As mentioned earlier, wells outside of both groundwater basins showed little or no long-term groundwater-level declines in contrast to wells within the groundwater basins. Wells within the groundwater basins had variable trends in groundwater levels depending on their location relative to recharge that originates mostly in the higher elevations and areas of long-term groundwater withdrawal for agricultural pumpage.

Groundwater levels in wells that are farthest from where most recharge occurs in the northeast had some of the largest long-term groundwater-level declines. Groundwater levels in areas farther from recharge sources do not respond rapidly to recharge events because recharge takes longer to arrive at these areas. For example, well 7S/3E-34E1 (figs. 16, 17, hydrograph 11), which is in the southeastern part of the Cahuilla Valley groundwater basin, declined by about 30 ft between 1951 and 2021, less than about 0.5 ft/year; groundwater levels in well 8S/3E-2D1, which is in the northwestern part of the Terwilliger Valley groundwater basin, declined by about 40 ft between 1960 and 2021, or about 0.7 ft/year (figs. 16, 17, hydrograph 10). In addition to the distance these wells are from the recharge sources in the higher elevations, the depth to groundwater ranged between 60 and 80 ft bls, so any potential locally-derived recharge would take longer to migrate downward to the water table than in places where the groundwater table is shallower (see the “Short-Term Trends in Groundwater Levels” section).

Areas of long-term agricultural pumpage had the largest groundwater-level declines because pumping has controlled the movement and direction of groundwater flow. Combining the records from one of the wells Waring (1919) reported in 1916 with the more recent measurements from well 7S/3E-22J2 shows that groundwater-levels declined about 60 ft between 1916 and 2021 (figs. 16, 17, hydrograph 8). Groundwater levels in well 7S/3E-14P3 have declined almost 40 ft over the shorter period between 1971 and 2021—a rate of about 0.8 ft/year (figs. 16, 17, hydrograph 6). Although this well is in the northeastern part of the Cahuilla Valley groundwater basin and is near recharge sources, the depth to groundwater in this well dropped from about 75 ft bls to more than 112 ft bls over those 50 years.

The long-term data show that several factors affect groundwater flow and levels. In addition to a well’s distance from recharge sources, some quantity of recharge is captured upgradient by pumping wells in the northeastern and southern parts of the groundwater basins, which diminishes the amount of recharge that reaches downgradient areas. The long-term declines in groundwater levels also suggest that the amount of water removed by pumpage has exceeded that which has been recharged and that groundwater has been removed from the aquifer system; that is, the amount of groundwater from aquifer storage has decreased. Moyle (1976) estimated that the total volume of groundwater removed from aquifer storage, or the amount of groundwater depletion, between 1950 and 1973 in two areas in the Cahuilla Valley groundwater basin was about 14,000 acre-ft, or about 600 acre-ft/yr. Long-term groundwater-level data are invaluable for understanding changes in groundwater flow, trends through time, and the effects of recharge and discharge. These long-term data indicate where imbalances between the volume of water that is recharged and amount of groundwater that is discharged, primarily by pumpage, exist. The continued evaluation of these long-term data is important to successfully managing the basins’ groundwater resources and exploring potential management strategies in the future.