Natural Gamma Logging

Natural gamma logging measures the total amount of gamma radiation emitted by materials surrounding a borehole (Keys, 1990). Two Mount Sopris Instruments (MSI) 2PGA–1000 polygamma tools were used to collect natural gamma logs. Gamma radiation is measured by the MSI 2PGA–1000 tool using a sodium iodide crystal that emits a pulse of light when struck with a gamma ray, which is then amplified and converted to a current pulse. Naturally occurring sources of gamma radiation commonly include potassium-40 and daughter products of the uranium and thorium-decay series (Johnson and others, 2011). Shales commonly contain higher fractions of radioactive minerals than other rock types, which produce higher gamma counts. Gamma logs can be used to differentiate geologic units and lithology, especially shale bedrock. The vertical resolution of the MSI 2PGA–1000 tool is about 0.3 to 0.6 meter (Johnson and others, 2011). A more complete background on gamma logging is provided by Keys (1990).

Gamma tools are calibrated by manufacturers to American Petroleum Institute units; however, tool calibration drifts over time and the data collected in American Petroleum Institute units may not be accurate (Johnson and others, 2011). Instead, tools can be operated so that natural gamma data are measured in units of counts per second. Natural gamma data collected during this study were measured in units of counts per second because the tools had not been calibrated by manufacturers since they were constructed. Additionally, two MSI 2PGA–1000 tools that likely were manufactured at different times were used in this study, and the drift rate of each tool was likely different; therefore, gamma emissions measured in counts per second from two or more tools were evaluated to ensure the tools were measuring similarly. Multiple wells were selected for repeat measurements where both tools were used to collect natural gamma data. The gamma data from repeat measurements indicated both tools measured gamma variations similarly, and therefore, no correction was needed.

Field procedures used for natural gamma log data collection in the “down” and “up” directions using the MSI 2PGA–1000 tool are described in this section. Before the gamma tool was inserted in a well, the depth to water and depth to bottom of the well were measured using an electric water-level tape using procedures described in Cunningham and Schalk (2011). Next, the tool was connected to a winch system that recorded the depth and speed at which the tool moved in the well. Additionally, a laptop was connected to the winch system to monitor data collection. The tool was then inserted in the well, and top of the probe was leveled to the top of casing (either above or below land surface). The probe was then turned on and lowered at a constant speed of about 10 feet per minute while collecting data until the tool reached near the bottom of the well. After recording and saving the log file, the tool was raised at a constant speed of about 10 feet per minute until the top of the tool reached the top of casing.

Natural gamma log data were collected at 45 wells in September 2020 and from July to August 2021. Natural gamma log data were processed using MSI WellCAD software (https://mountsopris.com/wellcad/). Before interpretation, natural gamma log data were depth corrected to land surface by subtracting the casing stickup, if applicable, from the depths recorded by the winch system so the depth to the top of the Pierre Shale (equivalent to surficial deposit thickness) was comparable among wells. After depth corrections were applied, a three-point moving-average filter was used to suppress data variations and produce a smooth log to document larger scale gamma count variations correlating to lithologic changes. Moving-average filters can introduce error by shifting the depths of gamma count measurements up or down. The depth error introduced from the three-point moving-average filter used in this study did not affect interpretation because the error was less than the vertical resolution of the tool. Depth corrected and filtered natural gamma data are available in the data release accompanying this report (Medler and others, 2022). Lithologic information from driller logs (South Dakota Department of Agriculture and Natural Resources, 2022), if available, was used to constrain interpretations of natural gamma logs. The Pierre Shale was expected to produce a higher gamma count than overlying surficial deposits, even those consisting mostly of clay and silt, because of its high content of radioactive minerals (Schultz and others, 1980).

Electromagnetic Induction Logging

EMI logging measures the bulk electrical conductivity (in millisiemens per meter) of materials and fluids in materials surrounding a borehole (Williams and others, 1993). An MSI 2PIA–1000 tool was used to collect EMI logs in this study. The EMI tool generates an electromagnetic field to induce an electrical current in the materials surrounding the borehole. The induced current then generates a secondary electromagnetic field that is measured as a voltage in receiver coils in the tool. Electrical conductivity is determined by the tool because it is proportional to the strength of the secondary electromagnetic field measured by the receiver coils when induction numbers are low (less than 100 millisiemens per meter; McNeill, 1986). Electrical conductivity changes are caused by borehole construction and the composition of materials and fluids surrounding the borehole, such as borehole diameter, casing material (steel, polyvinyl chloride, and so on), porosity of materials, specific conductance of groundwater, and the abundance of metallic minerals in materials. The effect of water in the borehole on electrical conductivity measurements is minimized because the tool was designed to be most sensitive to materials and groundwater about 1 foot (ft) away from the borehole (Johnson and others, 2011). Although the EMI tool measures electrical conductivity variations, the data file from the tool also reports electrical resistivity, the reciprocal of electrical conductivity, and is commonly plotted with electrical conductivity on composite geophysical logs. The vertical resolution of the tool is about 2 ft. A more complete background on EMI logging is provided by McNeill (1986).

The MSI 2PIA–1000 tool was calibrated before each measurement following procedures provided by the manufacturer. First, the tool was powered on and lowered into the borehole until it was either fully or partially submerged in the borehole fluids for 15–20 minutes to stabilize and equilibrate the internal temperature of the tool. Second, the tool was removed from the borehole and calibrated using two-point calibration with a free-air value of zero and (or) a calibration ring with the dial set at 465 or 2,060 millisiemens per meter based on the expected range of electrical conductivity values at each site. The two-point calibration involved holding the tool vertically in the air either with or without the calibration ring and was completed as quickly as possible to avoid major thermal changes. At some wells, the two-point calibration could not be completed because of tool malfunctions and was instead operated at “full scale” using manufacturer settings. EMI logs collected when operating at “full scale” were evaluated at three wells using two EMI tools—one calibrated and the other at “full scale.” The EMI logs from the calibrated and “full scale” tools were nearly identical and indicated minor variations. The tool was submerged for 15–20 minutes in the borehole fluid when the “full scale” settings were used to minimize thermal effects on data collection.

EMI log data were collected at 43 wells from July to August 2021. EMI log data were processed using MSI WellCAD software (https://mountsopris.com/wellcad/). Before interpretation, EMI log data were depth corrected to land surface from the depths recorded by the winch system so the depth to the top of the Pierre Shale (equivalent to surficial deposit thickness) was comparable among wells. After depth corrections were applied, a two-point moving-average filter was used to suppress variations and produce a smooth log to document larger scale variations correlating to lithologic and (or) groundwater chemistry. Moving-average filters can introduce error by shifting the depths of electrical conductivity measurements up or down. The depth error introduced from the two-point moving-average filter used in this study did not affect interpretation because the error was less than the vertical resolution of the tool. Depth corrected and filtered EMI data are available in the data release accompanying this report (Medler and others, 2022). Lithologic information from driller logs (South Dakota Department of Agriculture and Natural Resources, 2022) and natural gamma logs, if available, were used to constrain interpretations of EMI logs. The Pierre Shale was expected to produce higher electrical conductivity values than overlying surficial deposits because of its relatively high clay content and saline groundwater (Schultz and others, 1980; Williamson and Carter, 2001).

Borehole Nuclear Magnetic Resonance Logging

The bNMR logging estimates the water content, pore-size distribution, and hydraulic conductivity of materials in the saturated zone surrounding a borehole. Either a Javelin JP–175 or JP–238 logging tool, both manufactured by Vista Clara, Inc. (Walsh and others, 2013), was used to collect bNMR logs depending on the diameter of the well casing. The JP–175 and JP–238 have a diameter of 1.75 inches (in.) and 2.38 in., respectively, and were used to collect data in 2-in. and 6-in. diameter wells, respectively. The JP–175 tool was used for bNMR data collection for all wells, except wells SVE0101 and SVE0106, where the JP–238 tool was used because the wells had a 6-in. diameter casing. Both bNMR tools use static magnets and radio frequency fields to estimate hydraulic properties of a thin shell (0.08 in. thick) of material surrounding a borehole at a specific distance (7.5 in. for the JP–175 and 11.5 in. for the JP–238) from the borehole. Static magnets in the tool cause water molecules in the geologic material surrounding the tool to align to the magnetic field generated by the tool. The tool emits radio frequency fields to disrupt the magnetic field and cause the protons in the water molecule to be tipped into the transverse plane normal to the background field. The bNMR tool then measures the strength of the magnetic field produced as protons in the water molecules undergo precession to the background magnetic field imposed by the static magnets before disruption from radio frequencies. The signal strength decreases exponentially with time and is called the transverse relaxation time (T₂). In these measurements, the initial magnetic field strength in the T₂ is directly proportional to the total water content in the measurement zone. In saturated sediments, the total water content is equivalent to the formation porosity. The rate of the T₂ indicates the pore-size distribution; faster T₂ indicates smaller pore sizes and slower T₂ indicates larger pore sizes. The echo spacing, the time between each pulse during the measurement, for the JP–175 and JP–238 are 1.0 and 1.2 milliseconds, respectively. Additional background on bNMR tools used for this study is provided by Walsh and others (2013).

The bNMR data were collected at 34 wells from July to August 2021 using Javelin JP–175 or JP–238 tools. The diameter of the drilled borehole and the screened interval were reviewed in driller logs for each well, if available, before data collection to ensure the measurement zone of the tool was outside of the drilled hole and in surficial deposits overlying the Pierre Shale. Measurements in packing material within the drilled hole and the Pierre Shale were not the focus of this study and were removed where possible when collecting bNMR data; however, in some cases, bNMR data were collected within the disturbed zone of the borehole and (or) in wells screened entirely in the Pierre Shale. Additionally, bNMR data were not collected in the unsaturated zone of each well because hydraulic conductivity and pore-size distribution can be estimated only in the saturated zone.

At each site, the water level and total depth of each well were measured using an electric water-level tape. Next, the tool was connected to a winch system that recorded the depth and speed at which the tool moves in the well. Additionally, a laptop was connected to the winch system to monitor data collection. The tool then was inserted in the well and referenced to the top of casing (either above or below land surface). The bNMR data were collected in the up direction at 1.6-ft intervals in the saturated zone of the well starting from the bottom of the well. Dual-frequency measurements were made at center frequencies of about 300 and 250 kilohertz. Additionally, two relaxation times were used for both tools: a full relaxation time of 4 seconds and a shorter “burst-mode” relaxation time of 1 second. Measurements were stacked 30 times for the full relaxation time and 180 times for the burst-mode relaxation time.

The bNMR data were processed using the manufacturer-supplied software, VC_Javelin_Processor, version 4.1.1. In postprocessing, the full and burst-mode measurements for each frequency were combined for each 1.6-ft depth increment. An impulse noise filter was used to remove noisy T₂ data. The bNMR data also were adjusted by removing the ambient noise collected with an external reference coil measured concurrently with the subsurface measurements. The resultant T₂ data from the combined frequencies was fit with a multiexponential decay curve, which was inverted to produce a pore-size distribution model. For each depth, the pore-size distribution was determined as a fraction of the total water content. For each depth interval, the total, mobile, capillary, and clay fractions were determined using empirically derived cutoff T₂ values (Straley and others, 1997). The total water content includes the mobile and immobile fractions of water, which consist of the capillary and clay fractions. The mobile fraction, which is the fraction that decays after the 33-millisecond cutoff, represents the effective porosity. The immobile fraction, which includes the clay and capillary fractions that decay before the 33-millisecond cutoff, represents bound water. The clay-fraction cutoff was set at a T₂ value of 3 milliseconds.

Hydraulic conductivity estimates from bNMR data were calculated using default parameters in VC_Javelin_Processor software (version 4.1.1) with two equations: (1) the Schlumberger-Doll research (SDR) equation (Kenyon and others, 1988) and (2) the sum of echoes (SOE) equation (Allen and others, 2000). The SDR equation uses the measured values of total porosity and the mean log T₂, as follows:

Data Collection for Slug Tests

Data collected for slug test analysis included well-construction information, general aquifer characteristics, and water-level changes recorded during each test. Driller logs, when available (table 2; Medler and others, 2022), provided various well-construction information and general aquifer characteristics including lithologic descriptions, well-screen length, well-construction material, inside radius of the well casing, and the well borehole diameter. Well-construction information such as total well depth and measuring-point height for each well was verified by electric water-level tape and a surveyor tape, respectively, using methods described by Cunningham and Schalk (2011). Wells without driller logs (table 2) were examined in the field for well-construction material (normally polyvinyl chloride) and inside radius of the well (normally 1 in.). Wells without borehole diameter information were assumed completed with a 4.25-in. radius auger based on wells constructed similarly during the same year. Additionally, geophysical logs were used to determine the well-screen length and screened interval for wells without driller logs by identifying the depth of the bentonite plug or seal in the annular space of the well using high gamma counts in natural gamma logs. Specific well-construction data and general aquifer characteristics used for slug test analysis are available in the accompanying data release (Medler and others, 2022).

Water levels were measured before slug testing with a calibrated electric water-level tape (Cunningham and Schalk, 2011). Water levels were used to determine saturated thickness and the depth to suspend a submersible transducer below the water surface to record water levels before and during the slug test. The transducer was an unvented Solinst Levelogger LT F30/M10 (https://www.solinst.com/) with a radius of 0.036 ft. The transducer was suspended between 2 and 6 ft below the water surface and was the only downhole equipment in the well. The transducer recorded at 1-second intervals, and water levels were corrected for barometric pressure using a Solinst Barologger LT F15/M5 (https://www.solinst.com/) placed within 50 ft of the well during the slug test. The anisotropy ratio (vertical to horizontal hydraulic conductivity ratio, K_z/K_r, where K_z is vertical hydraulic conductivity and K_r is horizontal hydraulic conductivity) was assumed to be 0.1, which is a common value for unconsolidated sediments (Freeze and Cherry, 1979). Water-level data collected during slug tests are available in the accompanying data release (Medler and others, 2022).

Water slugs were used rather than solid slugs to prevent cross contamination between wells. The water slug also ensured a rapid head displacement in the well during the slug test. One to three slug test trials were completed at each well with differing volumes of water for each trial; the first trial used 0.5 liter (L) of water, the second trial used 1.0 L, and the third trial used 1.5 L. Water-level response during slug testing was variable among the wells; some tests recovered in less than 1 minute but other tests did not fully recover within 60 minutes. If water-level recovery times exceeded 20 minutes, only one trial was completed. Water-level displacement was calculated by subtracting the barometrically corrected water level from a reference (static) water level. The reference (static) water level was equal to the median of 30 consecutive seconds of stable barometrically corrected water-level data recorded by the transducer before or after the slug test.

Sampling intervals for some slug test trials were edited to ensure an appropriate and manageable number of data points for analysis because some slug test trials recorded as many as 60 minutes (more than 3,000 data points) of water-level data. Editing the sampling interval also reduced “noisy” data caused by spurious water-level data points. Python programming language (van Rossum and Drake, 2011) was used to edit the sampling interval of water-level data for trials exceeding 30 seconds. The initial 30 seconds of data during each trial were retained, and the sampling interval was changed for the remaining data based on the total water-level recovery time of the trial: one data point every 30 seconds was retained for slug test trials exceeding 1,000 seconds, one point every 20 seconds was retained for trials between 500 and 1,000 seconds, one data point every 10 seconds was retained for trials between 200 and 500 seconds, and one point every 5 seconds was retained for trials between 30 and 200 seconds. Editing the sampling rate did not affect analysis because the largest water-level displacement changes were within the first 30 seconds of the trials, and edited sampling rates were applied when water-level changes were relatively small.

Analytical Methods for Slug Tests

Water-level changes for each slug test trial were analyzed with AQTESOLV Pro, version 4.50.002 (Hydrosolve, Inc., 2007), using the Kansas Geological Survey (KGS; Hyder and others, 1994) method for unconfined aquifers without well skin. The KGS method was selected because the method best represents well conditions in the study area. The method incorporates the storage properties of the well and general aquifer characteristics, such as saturated thickness (Butler, 2020). Additionally, the KGS model considers the effects of partial well penetration into the aquifer and aquifer anisotropy (Hyder and others, 1994).

AQTESOLV uses curve fitting to provide an estimate of hydraulic conductivity and specific storage. The curve-fitting algorithm creates a best-fit curve by varying the estimated hydraulic conductivity and other hydrogeologic parameters until a theoretical curve best fits the measured time and water-level observations (Hydrosolve, Inc., 2007). The automatic curve-fitting feature in AQTESOLV using the KGS method without well skin was used for all slug test trials. AQTESOLV-readable files used for analysis are available in the accompanying data release (Medler and others, 2022). Although calculated by AQTESOLV, estimates of specific storage using the KGS method were not reported because of the questionable accuracy of storage and specific storage estimates from slug tests (Butler, 2020). Storage is the ability of an aquifer to accept or release water over time, and changes in storage can take considerable time; therefore, the short water-level recovery time and the small volume of the aquifer tested by the slug tests resulted in uncertain storage estimates for this study.