Introduction

Statement of the Problem

Increasing population and recent drought cycles in Cache Valley (figure 1), Utah and Idaho, have increased the need for additional water supplies. Several studies (Peterson, 1946; Beer, 1967; McGreevy and Bjorklund, 1970; Bjorklund and McGreevy, 1971; Kariya, et al., 1994) have considered ground-water resources in Cache Valley. Many of the studies of the hydrogeology of the valley include conceptual models of the valley’s aquifer system and estimates of the valley’s ground-water supply and the feasibility of using ground water to supply the growing need for water. Conceptual models are the foundation for computer simulations capable of calculating impacts of water use on a hydrogeologic system. Accurate, quantitative values of hydraulic conductivity, transmissivity, and storativity are necessary to properly conduct simulations that give reasonable estimates on the availability of Cache Valley’s water resources. Attempts have been made to perform a computer model of Cache Valley (Kariya et al., 1994; Myers, 2003), with limited success. The background of this work is described below. All the authors who conducted a model agree that better hydraulic parameters will enhance the accuracy of their models. Before an accurate computer model of Cache Valley’s hydrogeology can be constructed, I must determine the various hydrogeologic parameters of the hydrostratigraphic units in Cache Valley.

Previous Hydrogeologic Investigations

Bjorklund and McGreevy (1971) created an early hydrogeologic conceptual model (figure 2). Their conceptual model consists of an unconfined aquifer on the valley floor underlain by a continuous confining unit which overlies a confined aquifer. Their model shows the confined aquifer being recharged at the edge of the confining unit near the valley margins and through the fractured consolidated material that makes up the mountains. Kariya et al. (1994) created a conceptual model (figure 3) of Cache Valley’s hydrogeologic system. Their conceptual model differs from the original Bjorklund and McGreevy (1971) model in that it lacks a continuous confining layer overlying the principal aquifer.

Robinson (1999) conducted a thorough hydrostratigraphic and hydrologic examination on the valley. He created a revised conceptual model based on numerous cross sections made using well driller’s logs and water chemistry data. Similar to Bjorklund and McGreevy’s (1971) single continuous confining unit, Robinson’s (1999) conceptual model presented two continuous confining layers terminating within approximately one mile of Cache Valley’s eastern margin. Olsen (2007) later refined the appearance of Robinson’s conceptual model without significantly altering it (figure 4). With his model, Robinson described five major hydrostratigraphic units, which are summarized in detail by Table 1.

Kariya et al. (1994) used MODFLOW (McDonald and Harbaugh, 1988) model to simulate groundwater flow in the unconsolidated material in the valley, based off of their hydrologic budget and conceptual model (figure 3). Myers (2003) refined the Kariya et

al. (1994) model using Robinson’s conceptual model, and revised their boundary

conditions, water budget, and hydraulic properties. Using the same grid as Kariya et al. (1984) and the revised parameters, Myers (2003) also simulated Cache Valley groundwater conditions using MODFLOW (McDonald and Harbaugh, 1988). Myers (2003) concluded that the aquifers in Cache Valley are recharged along the margins of the valley and through subsurface flow from the surrounding mountain ranges. The model of Myers (2003) suggests that droughts may have a much greater influence on stream and spring discharges than increased pumping from wells. Myers (2003) recommended a comprehensive spring study to better simulate actual spring discharges and the effects of droughts and ground-water withdrawals with a MODFLOW (McDonald and Harbaugh, 1988) model, and well tests to obtain more accurate values of the hydraulic parameters used in the model, particularly the vertical hydraulic conductivity of the principal aquifer’s confining layers.

Following the similar recommendations of Myers (2003), Olsen (2007) also recommended that well tests be conducted on a select number of wells to determine accurate values of the hydraulic parameters of important hydrogeologic units in Cache Valley. Information from pumping tests would allow for the creation of credible simulation models that could be used to develop and manage the ground and surface water resources of Cache Valley most effectively.

Objectives

Based on the recommendations of previous workers (Myers, 2003; Olsen, 2007) I propose to determine transmissivity and its spatial distribution for the hydrogeologic

units in Cache Valley. Transmissivity is a measure of how easily water moves through a

unit thickness of aquifer. In order to determine transmissivity for each hydrogeologic unit, I must identify the units of which each well penetrates. I will also create groundwater contour maps of each decade to delineate major ground water level changes over time.

Methodology and Analysis

Data Treatment

Utah Division of Water Rights Data

Using Utah well databases available on the Utah Division of Water Rights (UDWR) web page I downloaded and plotted Utah’s wells in ArcMap. I then selected and isolated all the wells for Cache Valley using a cut or import tool. I then joined the tables of those wells to additional water level and well information databases provided by the UDWR.

Well driller’s records are stored as scanned originals (.tiff files) on the UDWR website. Although the UDWR databases do contain much of the information on many wells in Cache Valley, the database is not complete, and does not contain stratigraphic or specific capacity data. As a result, specific capacity and stratigraphic information must be added to the database manually.

Using the well driller’s records, geologic maps, and hydrogeologic cross-sections (Robinson, 1999; Smith, 1997) I determined which hydrostratigraphic unit each well penetrates. I assigned hydrostratigraphic units based on Robinson’s descriptions (Table 1) and included that information in the database. I also added a “comments” field onto water level measurements that span the last 30 years. Each water level is recorded with a sample site number, water level, elevation, elevation accuracy, and latitude and longitude. The measurements listed in the USGS database are recorded by scientists or experienced water level measurers who are about the quality of the data. The locations are accurately placed by latitude and longitude measurements. I used zonal statistics and the aforementioned 5 meter DEMs to assign elevation measurements to wells with an elevation accuracy of 20 feet or less. I then created a water level elevation field and filled it in using the calculator function. I sorted this column to find any zero values and remove them in an editing session. I also extracted the wells that are only important to Cache Valley, because the USGS measures water levels all over the state of Utah.

Interpolation

After standardizing water levels with elevation data and finding transmissivity from specific capacity, I was ready to interpolate. I tried many different methods of interpolation and found that kriging, although processor intensive, was the most robust method. I used kriging both for the creation of water level surfaces and to display the distribution of hydrogeologic properties.

Both the UDWR and the USGS databases have water level information with a date specified. I selected points into decades based on water level date from the two databases using “select by attributes” and exported the data into several different point files. I attempted to merge the water level data from the UDWR and the USGS databases, but found that the databases were structured too differently to allow for an easy join. Using the exported points from each database for each decade, I interpolated the database to record any important comments on the hydrostratigraphic unit identification.

On some of the driller’s records, specific capacity (figure 5) data are recorded. I systematically went through over one hundred individual well logs to find the specific capacity data and add them to my database. The specific capacity data were then applied to the Theis, Brown, and Meyer (1963) method to estimate the transmissivity from each well. Each well analyzed will be assigned to its corresponding area. Then the transmissivity data for each area was compiled and statistically analyzed to obtain a value for transmissivity for each area.

Using new 5 meter resolution DEMs for the valley from the Utah Automated Geographic Reference Center (AGRC), I assigned elevation values to each of the wells in Cache Valley (figure 6) using a zonal statistics tool in ArcInfo. Because the UDWR has depth to water, depth of screened interval, and depth of well in their database, I used the well elevations to compare the three dimensional distribution well depths, water levels and screened intervals of Cache Valley’s wells.

Using DEMs to assign well elevation can introduce significant error in elevation. The wells in the UDWR database are placed at their location using the Utah cadastral system. Although many wells are plotted within 5 feet of the actual location, some wells are off of location by as much as 50 feet. In a area of great relief, 50 feet of horizontal error can introduce significant vertical error, even though the DEM is 5 meter resolution.

United States Geological Survey Data

United States Geological Survey (USGS) has water level data recorded at various

intervals for over 100 wells in the valley. At least seven of these wells have over 200

water level elevations with kriging and created raster files. The “geospatial analysis: ordinary kriging” tool allowed me to average multiple measurements of water level taken at the same station during the same decade. I then used the “extract by mask” tool to make the sizes of the raster files equal to allow for easy comparison. The “raster to tin” tool was used to project the USGS water level data into three dimensions (figure 7) to allow for easy visual comparison. I used raster math to compare 1970’s water levels to current water levels.

The interpolation tools in ArcMap are especially useful in characterizing spatial distributions of hydrogeologic characteristics. I used smooth ordinary kriging to interpolate all of the transmissivity values to show a distribution of transmissivity throughout the valley. I also kriged the transmissivity of the principal aquifer (units A1 and A2). There was an insufficient amount of points for the other aquifer units for a raster to be helpful, so I left them as points.

Results and Discussion

Water Levels

Comparison of decade water level rasters indicate water levels are changing over time, and central tendencies from raster statistics indicate a drop in water level. The resultant raster created from the “raster math: difference” tool shows the change in water levels from 1970 to 2000 (figure 8). There appear to be significant decreases in water level near the Smithfield area, while the water levels in the northwest section of the

valley (in Utah) have increased.

The UDWR water level contour map (figure 9) shows the general shape of the piezometric surface (water table) of Cache Valley, Utah. As expected, the surface is similar to the ground elevation, higher near the edges, and lower in the center, with a possible outlet near the outlet of the Bear River. Ground water flow would move from a high potentiometric contour to a low one, meaning the groundwater will move toward the center of the valley. Although water level elevations are lowest near the center of the valley, they are higher than the ground surface in the center of the valley. This is observed by the numerous flowing wells near the center of the valley.

Using the geostatistical analysis toolbar in ArcMap, I found that the error from interpolation of the water levels can be as high as 200 feet and averages around 20 feet. This was verified by cross checking each raster using randomly selected water level readings not used to make the raster. I attribute the error to temporal variations in water level and an insufficient sample size. Although the resultant water level rasters can allow for a general understanding of the shape of the piezometric surface, they are unreliable quantitative measures.

Specific Capacity and Transmissivity

The transmissivity of the hydrogeologic units varied significantly over space, but seemed to show significant spatial patterns. When considered for all of the hydrogeologic units, transmissivity is highest near the Logan area and decreases radially (figure 10). The transmissivity for the principal aquifer shows a similar trend, with a mean transmissivity of 20,000ft²/day (figure 11). The transmissivity values of alluvium (Qal) and Salt Lake Formation (Tsl) are generally low, with averages of 1,900 ft²/day and 3,400ft²/day, respectively.

The results of the transmissivity were as expected. Clay becomes more prevalent in the principal aquifer to the west and to the north. The Salt Lake Formation has been previously mentioned to have low transmissivity, while the principal aquifer is known to have a high transmissivity (Bjorklund and McGreevy, 1971; Kariya, et al., 1994).

Remaining Work

There are still several hundred well records with available data to improve accuracy of average transmissivity and interpolated hydrogeologic characteristics. I hope to add enough to the database to create a statistically satisfactory dataset. I want to add pump test data, which can provide more reliable indicators of transmissivity. I also need to compile more water level data to make more realistic interpretations of temporal water level fluctuations.

Conclusions

There is an immense amount of hydrogeologic data available through various sources for Cache Valley. However, the quality of data can vary significantly, therefore making cautious data handling very important. Water levels have fluctuated over time, but without a more reliable surface maps, it is hard to determine definite total increases or decreases in water level. Specific capacity is highest near Logan decreases radially from that area. The principal aquifer has the highest mean specific capacity while it is much lower in the Salt Lake Formation, and the alluvium.

There is still much work necessary to characterize Cache Valley’s complex

hydrogeologic setting. Addition of specific capacity, water level, and pump test data will make the results more significant. Updated and accurate transmissivity rasters can be used in GIS based ground-water models for resource optimization.

DATA SOURCES United States Geological Survey (USGS) water level data taken from http://nwis.waterdata.usgs.gov/usa/nwis/gwlevels Division of Water (UDWR) databases and data taken from http://www.waterrights.utah.gov Base layer data and DEMs taken from Utah Automated Geographic Reference Center (AGRC), http://gis.utah.gov

REFERENCES CITED

Beer, L.P., 1967, Ground-water hydrology of southern Cache Valley, Utah: unpublished Ph.D. dissertation, University of Utah, Salt Lake City, Utah, 104 p.

Bjorklund, L.J., and McGreevy, L.J., 1971, Groundwater resources of Cache Valley, Utah and Idaho: Utah Department of Natural Resources Technical Publication No. 36, 72 p.

Kariya, K.A., Roark, D.M., and Hanson, K.M., 1994, Hydrogeology of Cache Valley, Cache County, Utah, and adjacent part of Idaho, with emphasis on simulation of ground water flow: Utah Department of Natural Resources Technical Publication No. 108, 120 p.

McDonald, M.G., and Harbaugh, A.W., 1988, A modular three-dimensional finite-difference ground-water flow model: U.S. Geological Survey Techniques of Water-Resources Investigations, Book 6, Chapter A1, 586 p.

McGreevy, L.J., and Bjorklund, L.J., 1970, Selected hydrogeologic data, Cache Valley, Utah and Idaho: Utah Department of Natural Resources Utah Basic-Data Release No. 21, 51 p.

Myers, B., 2003, Simulation of ground water flow in Cache Valley, Utah and Idaho: unpublished M.S. thesis, Utah State University, Logan, Utah, 89 p.

Olsen, A.A., 2007, Discharge monitoring, chemical characterization, and source identification of springs along the east side of southern Cache Valley, Utah: unpublished M.S. thesis, Utah State University, Logan, Utah, 185 p.

Robinson, J.M., 1999, Chemical and hydrostratigraphic characterization of ground water and surface water interaction in Cache Valley, Utah: Unpublished M.S. thesis, Utah State University, Logan, Utah, 184 p.

Theis, C.V., Brown, R.H., and Meyer, R.R., 1963, Estimating the transmissibility of aquifers from the specific capacity of wells: U.S. Geological Survey Water-Supply Paper 1536-I, p. 331-341.

Figure 1. Location of study area.

Figure 2. Bjorklund and McGreevy’s (1971) conceptual model. Note the continuous confining layers.

Figure 3. Kariya et al.’s (1994) conceptual model. Note the discontinuous confining layers.

Figure 4. Conceptual diagram of hydrogeologic setting modified from Olsen (2007) using Robinson’s (1999) notation.

Figure 5. Conceptual diagram of specific capacity. Long dashed line indicates water level in response to pumping, s_c is specific capacity, Q is discharge, and s is drawdown.

Figure 6. Screen capture of database showing elevation, screened units, and comments columns.

Figure 7. Ground water level TINs made from USGS data.

Figure 8. Difference in water level from the 1970s to the 2000s. Red and orange indicate an increase in water level while green indicates a decrease.

Figure 9. Water level contour map using all (every date) of UDWR data. Values are in feet above mean sea level.

Table 1. Robinson’s (1999) hydrogeologic unit nomenclature with explanations.