Published in Vadose Zone Journal 3:763-774 (2004)
© 2004 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
SPECIAL SECTION: RESEARCH ADVANCES IN VADOSE ZONE HYDROLOGY THROUGH SIMULATIONS WITH THE TOUGH CODES
The Role of the Unsaturated Zone in Artificial Recharge at San Gorgonio Pass, California
Alan L. Flinta,* and
Kevin M. Elletta,b
a Water Resources Division, United States Geological Survey, Placer Hall, 6000 J St., Sacramento, CA 95819
b Currently at Department of Civil and Environmental Engineering, University of Melbourne, Parkville, Victoria 3010, Australia
* Corresponding author (aflint{at}usgs.gov)
Received 23 October 2003.
 |
ABSTRACT
|
|---|
The hydrogeology of the unsaturated zone plays a critical role in determining the suitability of a site for artificial recharge. Optimally, a suitable site has highly permeable soils, a capacity for horizontal flow at the aquifer boundary, a lack of impeding layers, and a thick unsaturated zone. The suitability of a site is often determined by field and laboratory measurements of soil properties, field experiments, and numerical modeling. An artificial recharge site in the San Gorgonio Pass area in southern California, USA was studied to better understand the role of the unsaturated zone in artificial recharge by surface spreading. Field measurements and observations were used to characterize the site and to develop a conceptual model of the unsaturated zone. A numerical model was developed based on the conceptual model and calibrated using data from a 50-d artificial recharge experiment conducted in 1991 and borehole data collected between 1997 and 2002. Results indicate that an impeding layer exists 70 m below land surface that will cause lateral diversion of artificially recharged water, which would spread out and delay recharge to the water table 185 m below land surface.
Abbreviations: SGPWA, San Gorgonio Pass Water Agency
|
INTRODUCTION
|
|---|
ARTIFICIAL RECHARGE using water from the California State Water Project by surface spreading is being considered in the San Gorgonio Pass area of southern California, which is about 137 km east of Los Angeles (Fig. 1)
. Artificially recharged water must first move through a thick unsaturated zone (
185 m) before it reaches the underlying regional groundwater system. The suitability of an artificial recharge site is best determined by field and laboratory measurements of soil properties, field experiments, and numerical modeling.
View larger version (99K):
[in this window]
[in a new window]
|
Fig. 1. Landsat image of the San Gorgonio Pass Water Agency boundary area. Delineations of ground water storage units are defined by Bloyd (1971). The area proposed for artificial recharge (shown in inset) lies along the northern boundary of the Beaumont storage unit near Edgar Canyon.
|
|
The hydrologic properties of an unsaturated zone, such as porosity, permeability, and water retention characteristics, help determine the suitability of a particular site for artificial recharge. Optimally, areas used for artificial recharge should have highly permeable soils, the capacity for horizontal movement of water in the unsaturated zone and in the receiving aquifer, a lack of impeding layers, and a thick unsaturated zone. Under optimal conditions, water should reach the top of the saturated zone and spread laterally rather than building up a column of water toward the surface, which could greatly reduce recharge (Freeze and Cherry, 1979, p. 367–370). The available storage volume can also be reduced if recharged water is held tightly in the soil or if it drains slowly. For the most part, the unsaturated zone provides the underground storage space for recharge, although the amount of storage is dependent on the natural recharge occurring at the site. The greater the natural recharge at a site, the greater the pore space that is occupied by antecedent water moving through the unsaturated zone, which results in a smaller amount of available space for the artificially recharged water.
We present the methods and the field and laboratory data used to characterize the unsaturated zone beneath the Little San Gorgonio Creek spreading basins in San Gorgonio Pass. We will also present a conceptual and numerical model of the unsaturated zone that incorporates field and laboratory data collected at the site. The numerical model has been developed using TOUGH2, an integrated finite-difference numerical code (Pruess et al., 1999). The model will be used to help analyze the data collected at the site and to evaluate future artificial recharge at the site.
|
SITE ANALYSIS
|
|---|
The hydrogeology of the area has been described in previous studies by Bloyd (1971) and Boyle Engineering Corporation (1990)(1992, 1993a, 1993b). In 1991, the San Gorgonio Pass Water Agency (SGPWA) evaluated the feasibility of artificial recharge at the Little San Gorgonio Creek spreading basins (Fig. 1, inset) (Boyle Engineering Corporation, 1992; Shaikh et al., 1995). In 1997, the USGS, in cooperation with the SGPWA, began a study to evaluate the suitability of the unsaturated zone for artificial recharge at the spreading basins and to develop models of the unsaturated and the saturated zones of the San Gorgonio Pass area. Although well-defined guidelines are available for developing recharge spreading basins (Environmental and Water Resources Institute, 2001), spreading basins at this site were established in the 1960s before full analysis of subsurface hydrogeologic conditions and properties. Hydrogeologic data are essential in siting recharge spreading basins, particularly in alluvial basins where soils are highly stratified and contain continuous and discontinuous clay layers interbedded with sands and gravels (Flanigan et al., 1995).
The alluvial deposits that comprise the unsaturated zone underlying the spreading basins include younger surficial deposits (Qy), older surficial deposits (Qo), very old surficial deposits (Qvo), and the upper member of the San Timoteo beds (Qsu) (Fig. 2)
. In general, the surficial sedimentary materials (Qy, Qo, and Qvo) within the study area consist of interlayered sand and gravel deposits, with intermittent layers of clay, silt, and fine sand that become more compacted with depth. Unit Qsu consists of sand and gravel layers that are locally cemented into beds of sandstone and conglomerate.
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 2. Conceptual cross section of the layered stratigraphy, a fault, and the relative location of the cross-section (A–A', Fig. 1) and near-surface recharge ponds to features of the San Gorgonio Pass area, California.
|
|
As part of the USGS evaluation, several test wells were constructed in the unsaturated zone near the spreading basins and instrumented with deep tensiometers, heat-dissipation matric-potential sensors, temperature sensors, and suction-cup lysimeters (Ellett, 2002). Core samples and drill cuttings collected during the drilling of the test wells were analyzed in the USGS laboratory, in Sacramento, CA to determine particle-size distribution, water content, permeability, and lithology (Ellett, 2002). An interpretation of these data indicates that there are several alternating high- and low-permeability layers between the ground surface and the water table (185 m deep). A perched water table is present above a low-permeability layer present at the contact between geologic units Qo and Qvo, at about 70 m below land surface (Fig. 2). Data from other boreholes in the area indicate that this perched layer is areally extensive.
The combination of lithologic and geophysical logs from boreholes, surface-seismic reflection and refraction profiles, gravity measurements, and surface-resistivity measurements (Catchings et al., 1999; Christensen, 2000; Ellett, 2002) were used to develop a conceptual model of the layering and faulting in the area (Fig. 2). The Banning Fault forms the northern boundary of the study area, where it juxtaposes crystalline rocks against late Cenozoic sedimentary deposits. Water levels on the north side of the fault are more than 200 m higher than water levels on the south side of the fault, indicating that the fault is a barrier to groundwater flow (Fig. 2). Numerous faults were identified on the seismic profiles north of TW-1 (Catchings et al., 1999). These interpreted faults cumulatively offset the sedimentary deposits by as much as 50 m, with up-on-the-north displacement. For the purposes of this report, these faults are referred to as the Cherry Valley Fault zone. Water levels in Well TW-1 on the south side of the fault zone are about 10 m lower than water levels on the north side of the fault zone, indicating that the fault zone is a partial barrier to groundwater flow.
|
NUMERICAL MODELING
|
|---|
Model Development
The conceptual model of the unsaturated zone at San Gorgonio Pass was used to develop a numerical model to further analyze existing data, to help confirm the conceptual model, and to evaluate future artificial recharge at the site. TOUGH2, an integrated finite-difference numerical code (Pruess et al., 1999), was used to develop the three-dimensional numerical model using the equation of state module EWASG (Battistelli et al., 1997). This code simulates the flow of heat, air, water, and dissolved component (defined here to be NO3 associated with septic tank leach fields in the area) in three dimensions under saturated and unsaturated conditions. The geometry of the site requires a three-dimensional model because of downdip migration of recharged water through the alluvial deposits (north to south), as well as lateral flow of natural recharge (generally east to west) from the nearby stream. The modeling domain is approximately 2.5 km (east to west) by 1.3 km by 185 m deep and contains more than 50000 grid elements. Vertically the model was divided into seven layers (Table 1). Layer 1 represents Qy, Layers 2 through 4 represent Qo, Layer 5 represents the perching layer at the contact of Qo and Qvo, Layers 5 and 6 represent Qvo, and Layer 7 represents the bottom of Qvo and the top of Qsu (Table 1). The lateral model boundaries are the Banning Fault on the north, the southern extent of the Cherry Valley Fault zone on the south, and the edges of the alluvial basin where they encounter the mountain block on the east and west. The bottom boundary is the water table and the upper boundary is represented as a specified flux. The specified flux is temporally and spatially variable depending on the artificial recharge scenario and on the location and amount of recharge from precipitation, streamflow, and septic tank return flow.
Model Calibration
The model initially was developed using the hydrologic properties measured or estimated from the laboratory data (Ellett, 2002) and was then simplified by assuming isotropic permeability and homogeneous layers. Hydraulic conductivity was measured in the laboratory using cores collected in situ in a few intervals and cores that were repacked from the cuttings collected during drilling in other intervals. The temperature profile collected from Borehole TW-3 near the spreading basins was then used to estimate the vertical hydraulic conductivity of the perching layer (Layer 5).
Borehole temperature data collected from Boreholes TW-2 and TW-3 (Fig. 1) indicate that the coldest water temperature occurs at the perched water body in Borehole TW-2 and TW-3 (Fig. 3)
. The lower temperature in the perched water body was used with the temperature profile for water beneath the perched water body to estimate the vertical hydraulic conductivity of the perching layer using inverse modeling of convective heat transport. The calibration process involved changing the hydraulic conductivity of the perching layer (Layer 5) until the simulated temperature profile matched the measured profile below the perching layer. We assumed that the hydraulic conductivity values for the other layers remained the same as estimated from laboratory data. The thermal conductivity (Kt) of all layers was assumed to be 1.64 W m–1 °C–1, the water table in the perching layer was held constant at 72 m, and water temperature of the perching zone was held constant at 15.4°C. The hydraulic conductivity of the perching layer was estimated to be approximately 9.57 x 10–4 or 0.35 m yr–1 under these assumptions (Fig. 4a)
. Because there is a unit hydraulic gradient in the perching layer, the hydraulic conductivity of the perching layer is equal to the volumetric flux of water moving through the perching layer.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4. Results from inverse modeling of convective heat transport, which show (a) the sensitivity to water flux of the measured temperatures at TW-3, which were best approximated by a flux value of 0.35 m yr–1, and (b) the sensitivity of modeling results to thermal conductivity using a flux value of 0.35 m yr–1. The undefined thermal conductivity (Kt = Undefined) would be the profile for any conductivity under a no-flow boundary condition.
|
|
The estimate of vertical hydraulic conductivity is relatively insensitive to the Kt of the unsaturated zone. A sensitivity analysis was done by varying the Kt of the unsaturated zone from the values based on laboratory measurements of similar alluvial samples. The best-fit vertical hydraulic conductivity value (0.35 m yr–1) was used, and Kt was varied between a high and low estimate of 2.14 and 1.64 W m–1 °C–1, respectively. The undefined thermal conductivity, Kt = Undefined (Fig. 4b), is a simulated temperature profile for any conductivity under a no-flow boundary condition (no recharge). The coupled effects of hydraulic conductivity and thermal conductivity on the temperature profile provide a nonunique solution and introduce uncertainty in both values.
Once the vertical hydraulic conductivity of the perching layer was determined, the other model layer parameters could be calibrated (Table 1). Textural data were used to estimate porosity and the water retention function using pedotransfer functions and the van Genuchten equations (Schaap et al., 1998; van Genuchten, 1980). The model was calibrated by adjusting the vertical hydraulic conductivity value for the different layers until simulated results matched measured borehole temperature data, matric potential data, and the occurrence of perched water. The model was calibrated under steady-state conditions assuming natural recharge from precipitation of about 37 mm yr–1 over the modeling domain (120250 m3 yr–1) and from streamflow of about 2000 mm yr–1 over the width of the stream channels (32070 m3 yr–1). The estimates of recharge are based on preliminary results of a groundwater flow model being developed for the study area (D. Rewis, USGS, personal communication, 2003). The recharge temperatures of precipitation and streamflow were assumed to be 18 and 5°C; respectively. The streamflow temperature was estimated from data collected during streamflow events along Little San Gorgonio Creek (Fig. 5)
. The underlying groundwater temperature was held constant at 16.4°C.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 5. Stream bed temperature time series suggests the possible source for cold water in the perched zone is from low temperature stream-flow events that infiltrated into the stream channel.
|
|
A comparison of the simulated and measured matric potential is presented in Fig. 6
. As shown on the figure, the calibrated model closely matches the measured data. The simulated temperatures also are in good agreement with measured temperatures at TW-2 and TW-3. A two-dimensional cross section of the simulated temperature profile was taken for visualization from the three-dimensional model (Fig. 7)
. The simulated temperature profiles are in good agreement with the measured temperature profiles at TW-3 (Fig. 8)
. The simulated values in Boreholes TW-2 and TW-3 show a decrease in temperature from the ground surface to the perched water body, then a gradual increase toward the water table (Fig. 7 and 8). The simulated temperature at TW-3 is warmer than TW-2 because TW-3 is farther from the stream than TW-2, which is consistent with the measured temperature profiles (Fig. 7 and 8).
View larger version (45K):
[in this window]
[in a new window]
|
Fig. 7. Simulated temperature profiles under two parallel streams show the decrease in temperature in Boreholes TW-2 and TW-3, reaching a minimum at the perching layer with a gradual increase toward the water table that matches the response in TW-3 in Fig. 8.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 8. Subsurface temperatures measured in Borehole TW-3 fall between the simulated temperature centered on the model nodes on either side of TW-3.
|
|
Model Applications
The calibrated model was used to evaluate future artificial recharge scenarios. The model was run under transient conditions to simulate the period 1960 to 2005. These model simulations assumed that natural recharge conditions were the same as used for the steady-state model. Septic tanks are the only source of sewage disposal in the modeling domain and are generally considered as point sources of recharge. However, because of the large modeling domain and relatively large grid cells recharge from septic-tank return flows were distributed uniformly over the modeling domain at about 54 mm yr–1 (about 175500 m3 yr–1). Artificial recharge was assumed to occur at the spreading basins from 2001 through 2005. A total of 1.23 million cubic meters of artificial recharge were applied during 50 d at the beginning of each year.
The simulated water content after the first 50 d of water application is shown in Fig. 9
. In the simulation, the application of water is discontinued for the remainder of the year except for that representing natural recharge and septic-tank return flow. Figure 10
shows the results 5 d into the second year of artificial recharge. By this time, the simulation indicates that the initial application had reached the perched water body and moved downdip, backing up against the Cherry Valley Fault (a no-flow boundary). By the end of the fifth year of simulation, which included three more 50-d applications to the spreading basins, a considerable amount of water had accumulated against the fault (Fig. 11)
. Matric potential, temperature, and pressure after the fifth year of application are shown in Fig. 12, 13, and 14
, respectively. The thickening of the perched water body, indicated in Fig. 12 as nearly 0 MPa, indicates an increased head at the no-flow fault boundary condition representing the Cherry Valley Fault of approximately 0.55 MPa (or 45 m of water height assuming an atmospheric pressure of 0.10 MPa) (Fig. 14). The increase of water levels (pressure) in the perched water body (Fig. 14) results in increased percolation through the perching layers and increased saturation below the perching layer (Fig. 11).
View larger version (123K):
[in this window]
[in a new window]
|
Fig. 9. Simulated water content after 50 d of application of water at spreading basins during the first year of simulation.
|
|
View larger version (120K):
[in this window]
[in a new window]
|
Fig. 10. Simulated water content after 5 d of application of water at spreading basins during the second year of simulation.
|
|
One of the concerns near the spreading basins is the potential for artificial recharge to entrain septic-tank effluent as it moves through the unsaturated zone, and subsequently contaminate the regional aquifer. In another artificial recharge program in a nearby desert basin, rising groundwater levels resulting from the artificial recharge entrained high-nitrate septage stored in the unsaturated zone, resulting in NO3–N concentrations in excess of the drinking water standard (10 mg L–1 as NO3–N) (Nishikawa et al., 2003). Because all the homes in the area use septic systems, many of which have been in use for more than 40 yr, the possibility for contamination from the entrainment of septic-tank effluent was addressed by this study. Septic tank return flow was assumed to average 54 mm yr–1 for the entire model domain with an average NO3–N concentration (NO3 reported as N) of 80 mg L–1 (P. Martin, USGS, personal communication, 2003).
The model simulated 40 yr of septic tank return flows before the artificial recharge scenarios were started. The artificially recharged water entrained some of the septic tank return flows and moved it below the perched water body after 5 yr (Fig. 15)
; however, NO3–N concentrations remained below the drinking-water standard as the artificially recharged water migrated to the regional water table. The artificial recharge water, which was assumed to have no NO3, diluted the NO3–containing soil moisture in the unsaturated zone beneath the spreading basins.
View larger version (99K):
[in this window]
[in a new window]
|
Fig. 15. Simulated NO3 as N after 40 yr of accumulation under septic leach fields, followed by 5 yr of artificial recharge.
|
|
Before the application of artificial recharge, the simulated travel time from the ground surface to the water table was approximately 50 yr for locations directly beneath the stream and more than 250 yr for locations away from the stream. The simulated artificial recharge from 2001 to 2005 decreased the travel time in the unsaturated zone to <10 yr directly beneath the spreading basins. The velocity of the recharge water beneath the perching layer in the vicinity of the spreading basins was <2 m yr–1 at the end of 5 yr of artificial recharge. During the simulation period most of the artificially recharged water mounded above the perching layer at 70 m below land surface. The simulated mound extends from the spreading basins to the Cherry Valley Fault (no-flow boundary) located about 1200 m south of the spreading basins.
The model results are sensitive to the location and permeability of the Cherry Valley Fault. If the fault is closer to the spreading basins the mounding would be greater and if the fault is at a greater distance the mounding would be less. Note that the fault was assumed to be a no-flow boundary. If the fault is not a complete barrier to flow, water would migrate laterally across the fault and the mounding would be reduced and recharge in the spreading basins would be reduced. Microgravity station transects will be used in conjunction with water-level measurements from the perched and regional water tables in future artificial recharge experiments to track the lateral migration of water. If the fault is a barrier, then water will collect against the fault as indicated by model simulations (Fig. 11). The location and degree of mounding could be used in the model to estimate the location and permeability of the fault. There are several management options, depending on the degree with which the fault acts as a permeability barrier. Production wells can be installed directly into the perched water body at some optimal location, or multiple wells can be drilled through the perching layer to perforate it and increase its effective permeability, which would allow gradual infiltration from the perched water to the unsaturated zone below and eventually to the water table. Another option is drilling wells through the perching layer for direct injection into the unsaturated zone. These options can be included in modeling scenarios using the existing model to determine the number of dry wells required for increasing the permeability of the perching layer or the optimal location of production wells. As more data become available, the model can be refined and recalibrated, providing a flexible tool for enhancing research and management decisions.
|
SUMMARY AND CONCLUSIONS
|
|---|
Generally, artificial recharge projects apply water in surface and near-surface spreading basins, pits, and trenches, using the unsaturated zone to transport and store water. The hydrogeology of the unsaturated zone plays a critical role in transporting and storing artificially recharged water. Evaluating this zone will determine if the area is suitable for artificial recharge and will help to identify the most effective methods of surface or subsurface application of water. Field and laboratory data and field experiments were used to develop a conceptual and a numerical model of the unsaturated zone at San Gorgonio Pass in southern California. Calibration exercises indicate good matches to matric potential and temperature measurements. The results of the model simulations were used to refine the conceptual model and to test scenarios for artificial recharge. Results of the numerical model simulations of this site indicate that little recharge will reach the regional aquifer beneath the spreading basins during the 5-yr simulation period. The simulations indicate that most of the water will remain above a perching layer at 70 m below land surface, mounding along the assumed no-flow fault boundary located about 1200 m south of the spreading basins. The simulations indicate that the perching layer will delay recharge to the water table 185 m below land surface. Although the recharged water intercepts NO3–rich round water from septic tank leach fields as it spreads laterally and vertically through the unsaturated zone, the simulated NO3–N concentration of water in the perched water layer is <10 mg L–1, the maximum level set as a drinking-water standard. Further work on the characteristics of the fault and extension of the modeling domain farther downgradient of the fault are required to provide more conclusive results for the characterization of the site for the application of artificial recharge.
|
ACKNOWLEDGMENTS
|
|---|
The San Gorgonio Pass Water Agency and Stephen Stockton, General Manager and Chief Engineer, supported this work.
|
REFERENCES
|
|---|
- Battistelli, A., C. Calore, and K. Pruess. 1997. The simulator TOUGH2/EWASG for modelling geothermal reservoirs with brines and non-condensible gas. Geothermics 26:437–464.[ISI]
- Bloyd, R.M. 1971. Underground storage of imported water in the San Gorgonio Pass area, southern California. USGS Water-Supply Paper 1999-D. USGS, Washington, DC.
- Boyle Engineering Corporation. 1990. Well completion report, Test Well No. 1. San Gorgonio Pass Water Agency, Beaumont, CA.
- Boyle Engineering Corporation. 1992. Water importation program: Beaumont-Calimesa groundwater storage project, artificial recharge feasibility investigation. San Gorgonio Pass Water Agency, Beaumont, CA.
- Boyle Engineering Corporation. 1993a. Alternative analysis: Groundwater recharge impacts. San Gorgonio Pass Water Agency, Beaumont, CA.
- Boyle Engineering Corporation. 1993b. Groundwater investigation Beaumont-Banning area. San Gorgonio Pass Water Agency, Beaumont, CA.
- Catchings, R.D., G. Gandhok, M.R. Goldman, E. Horta, M.J. Rymer, P. Martin, and A. Christensen. 1999. Subsurface, high-resolution, seismic images from cherry valley, San Bernardino County, California: Implications for water resources and earthquake hazards. USGS Open-File Rep. 99-26. USGS, Menlo Park, CA.
- Christensen, A.H. 2000. A gravity survey of the San Gorgonio Pass Area, California. M.S. thesis. California State University, San Diego.
- Ellett, K.M. 2002. Hydrologic characterization of the deep vadose zone of the San Gorgonio Pass area for artificial and natural recharge analysis. M.S. thesis. University of California, Davis.
- Environmental and Water Resources Institute. 2001. Standard guidelines for artificial recharge of ground water. ASCE, Reston, VA.
- Flanigan, J.B., P.A. Sorensen, and M.A. Tucker. 1995. Use of hydrogeologic data in recharge pond design, Vol. II. p. 139–148. In A.I. Johnson and R.D. Pyne (ed.) Artificial recharge of ground water. ASCE, New York.
- Freeze, R.A., and J.A., Cherry. 1979. Groundwater. Prentice Hall, Englewood Cliffs, NJ.
- Nishikawa, T., J.N. Densmore, P. Martin, and J. Matti. 2003. Evaluation of the source and transport of high nitrate concentrations in ground water, Warren Subbasin, California. USGS Water Resources Investigation Rep. 03-4009. Available at http://water.usgs.gov/pubs/wri/wrir034009/text.html (verified 18 June 2004). USGS, San Diego, CA.
- Pruess, K., C. Oldenburg, and G. Moridis. 1999. TOUGH2 user's guide. Version 2.0. Rep. LBNL-43134. Lawrence Berkeley Natl. Lab., Berkeley, CA.
- Schaap, M.G., F.J. Leij, and M.Th. van Genuchten. 1998. Neural network analysis for hierarchical prediction of soil hydraulic properties. Soil Sci. Soc. Am. J. 62:847–855.[Abstract/Free Full Text]
- Shaikh, A., R.B. Bell, M.E. Ford, and S.P. Stockton. 1995. Feasibility of recharge by surface spreading. Vol. II. p. 159–167. In A.I. Johnson and R.D. Pyne (ed.) Artificial recharge of ground water. ASCE, New York.
- van Genuchten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44:892–898.[Abstract/Free Full Text]
TOP
ABSTRACT
INTRODUCTION
