Tracer testing in the Edwards Aquifer

Four general areas of the Edwards (Balcones Fault zone) Aquifer have been evaluated since the mid-1980s using tracer tests. These are: (1) the Barton Springs area; (2) San Marcos and Comal Springs area, which are two of the largest springs in Texas; (3) the San Antonio and Bexar County area (Panther Springs Creek); and (4) the Las Moras Springs and Kinney County area, in the western reaches of the aquifer (Fig. 1).

Tracer testing can be categorized as either qualitative or quantitative. Qualitative traces are used to document the groundwater flow connection between tracer injection and recovery locations. Passive dye receptors (activated carbon dye receptors) provide continuous monitoring and high sensitivity to dye and can also provide an estimate of the time-of-travel depending upon frequency of sampling. Frequent sampling provides more realistic time estimates for the hydrologic conditions in the aquifer during the test period. Qualitative traces have been most commonly used in the Edwards Aquifer because of the logistics of long flow paths and related time-of-travel and difficulty in accessing large water wells. Quantitative tracing involves high-frequency water sample collection, typically with either automatic samplers or on-site equipment operated by experienced tracing personnel. Quantitative studies provide more accurate time-of-travel data and higher resolution of tracer breakthrough, which can assist with the determination of potential dilution/assimilative capacity of an aquifer. Quantitative techniques are employed when groundwater conditions are fairly well known and travel times are relatively short (Alexander and Quinlan, 1996).

During this testing, tracing agents, most commonly fluorescent organic dyes, were injected into sinking streams, surface streams, cave streams, and wells with direct connection to groundwater flowing in the aquifer. Springs, wells, and subsurface (cave) streams were sampled for tracing agents down gradient from the injection points. Velocities were typically calculated on the basis of straight-line (point-to-point) distances between injection and recovery points; this distance was then divided by the elapsed time between injection and the first appearance of the tracer at the monitoring site. In quantitative tests, the arrival time can also be estimated based on the time of the maximum concentration of the dye mass reaching the receptor. Actual groundwater flow paths are expected to be longer than straight-line distances. Thus, true groundwater velocities are greater than calculated (apparent) groundwater velocities from tracer tests.

Tracers used in the Edwards Aquifer are typically fluorescent organic dyes. The most common tracers used are: Uranine (sodium fluorescein), Rhodamine WT, Eosin, Sulforhodamine B, Phloxine B, and Pyranine. Other tracers include humic and fulvic acids and combustion products, which originated from infiltration of fire-fighting water from a mulch-pile fire in Helotes, Texas (Johnson, 2018). Another accidental tracer was from a diesel-fuel spill identified at the distal end of a 6.4 km flow path that resurged from Comal Springs. Another non-dye-tracing test resulted from an accidental spill of trichlorofluoromethane (C1₃CF), the refrigerant Freon 11, which was spilled in northwestern Bexar County (Thompson, 1976; Schultz, 1979; Buszka et al., 1995). The plume from this spill was detected at both Comal and San Marcos Springs.

Although not a formal tracing test, Zara Environmental (2010) described the biodiversity within the Edwards Aquifer (see Krejca and Reddell, this volume) based on the areal distributions of worms, gastropods, crustaceans, isopods, and fish that were collected from the aquifer. This inventory did not provide point-to-point connection nor time-of-travel data, but it did establish the connectivity of voids sufficient in size in the aquifer to allow organisms to move between various sampling locations.

Since 1996, the Barton Springs Edwards Aquifer Conservation District (BSEACD) has conducted numerous tracer tests with the City of Austin to infer flow paths between the recharge zone and Barton Springs (Hauwert et al., 2004), a major discharge from the aquifer and a popular recreational site (see Hunt et al., this volume). Dyes have been injected into 19 different natural recharge features and one well (Fig. 2) in all of the major contributing watersheds that supply water to Barton Springs (Hauwert et al., 2002; BSEACD, 2003).

Three groundwater basins (Hauwert et al., 2002; Smith et al., 2012) were delineated (Cold Springs, Sunset Valley, Manchaca) based on tracer tests. Each basin consisted of a network of flow routes coalescing with resurgence at the springs. Most of the flow within the Barton Springs segment occurs along preferential flow routes that are strongly influenced by faulting (Hauwert et al., 2002; BSEACD, 2003; Hauwert et al., 2004). Groundwater flow in the Manchaca basin follows two principal flow routes, the Manchaca and the Saline-Line flow routes. Groundwater flow within the Sunset Valley groundwater basin converges along the Sunset Valley flow route.

The Cold Springs groundwater basin, north of the Barton Springs complex, discharges to Cold Springs and other minor springs along the Colorado River and is the northwesternmost spring basin in the Barton Springs area. In addition to the tracer tests, the four major Barton Springs (Main, Eliza, Old Mill, and Upper) have distinctive concentrations of chloride and sulfate. This is inferred to be the result of complex mixing of multiple subsurface sources of groundwater flow paths. Groundwater velocities to Barton Springs have been calculated from tracing to range from 915 m per day (m/d) to ~9150 m/d, depending upon aquifer stage. In another tracer test following a sinkhole collapse that occurred beneath a stormwater retention pond in the Sunset Valley groundwater basin in 2012, dye injected into this sinkhole arrived at Barton Springs in less than 4 d—a groundwater flow rate of at least 2100 m/d (Hunt et al., 2013).

Comal and San Marcos Springs, both first-magnitude springs, defined as having a flow of more than 2830 L per second (L/s), are located between Austin and San Antonio, Texas. Total average flow from the Comal Springs complex in New Braunfels, Texas, is ~9000 L/s, and total average flow from the San Marcos Springs complex is ~4300 L/s (Edwards Aquifer Authority, 2018). These, and other large springs in the area, are important from historical, archaeological, water supply, recreational, environmental, ecological, legal, and political standpoints (Edwards Aquifer Authority, 2012). The occurrence of rare and endangered species also drives the need to understand the hydrogeology of the Edwards Aquifer.

Rothermel and Ogden (1987) injected dye at three locations in Blieders Creek to identify recharge points for Comal and Hueco Springs. None of the dyes were detected at Comal Springs, but one of the dyes traveled ~3.2 km to Hueco Springs in 5 d, yielding a calculated time-of-travel of 640 m/d.

Edwards Aquifer Authority (EAA) completed 16 tracer tests (dye injections) between 2002 and 2017 near Comal Springs. Results indicated that water discharging from Comal Springs originates from the artesian, contributing, and recharge zones in Bexar, Medina, and Uvalde Counties to the west (see Sharp et al., this volume). Groundwater in the artesian fault block near Comal Springs had velocities generally less than 300 m/d, flowing to the northeast to discharge at Comal Springs. At some point up gradient from the Comal Springs complex, groundwater flow bifurcates into the artesian (downthrown) and Comal Springs (upthrown) fault blocks. Water from the downthrown block issues from Spring #7 and other springs beneath and along the northeastern stretch of Landa Lake, and water from the upthrown block issues from Springs #1, #2, and #3 and hundreds of other small springlets along the southeastern portion of the Comal Springs fault scarp (Fig. 3). Groundwaters in the two fault blocks also have slightly different geochemical and physical characteristics. For example, water discharging from springs from the artesian zone has slightly warmer temperatures (0.1–0.2 °C) than water issuing from the recharge zone fault block. Tracer tests revealed discrete flow paths in a three-dimensional groundwater flow system beneath Landa Lake (Schindel et al., 2002).

Initial tracer tests from Ezell’s Cave, Rattlesnake Cave, and Tarbutton’s Showerbath Cave to San Marcos Springs (Fig. 4) were conducted by Ogden et al. (1986), with additional tests by Hauwert et al. (2004). Ogden and Hauwert measured groundwater velocities between Ezell’s Cave and Deep and Catfish Hotel Springs in the San Marcos Springs complex of ~300 m/d. No dye was detected at the four other spring orifices that were monitored (Diversion, Weissmuller, Hotel, and Cabomba). Dye was also detected at the artesian well next to the Edwards Aquifer Research and Data Center (EARDC) at Texas State University (TSU) and at a City of San Marcos (COSM) municipal well next to Spring Lake (Ogden et al., 1986). Tracers injected by Ogden et al. (1986) into Rattlesnake Cave, ~1.2 km northeast of San Marcos Springs, traveled to all six of the monitored orifices in less than 40 d.

Ogden et al. (1986) attributed the slower velocity to drought conditions during this later tracing study that caused the water table to be relatively horizontal in the vicinity of the springs, thereby reducing San Marcos Springs discharge to ~2000 L/s. Dye was also detected in Sink Spring and Rattlesnake Well, ~150 m from Rattlesnake Cave. Ogden also injected dye at Tarbutton’s Showerbath Cave, located ~11 km northeast of San Marcos Springs on the Blanco River. This test indicated that dye injected in the cave appeared in all six orifices 359 d later and persisted for a month and a half. The report did not specify the frequency of dye receptor collection to constrain the travel-time estimate.

Hauwert et al. (2004) described a second tracer test involving Tarbutton’s Showerbath Cave in August 2000. Fluorescent dye injected into the cave was not detected in various individual springs monitored at San Marcos Springs before monitoring ended in August 2003.

EAA completed 31 tracer tests between 2002 and 2010 at various locations in the vicinity of San Marcos Springs and Barton Springs in collaboration with BSEACD and the City of Austin (Fig. 4; Johnson et al., 2012). The tracer tests revealed discrete groundwater flow paths and rapid groundwater velocities connecting the recharge zone to San Marcos Springs and Barton Springs, consistent with the karstic nature of the Edwards Aquifer. Straight-line velocities ranged from <1 to 3600 m/d. Positive tracer tests were acquired from each injection point in the vicinity of San Marcos Springs and Barton Springs (Smith et al., 2012). This confirmed that both spring complexes are important discharge points for the greater Edwards Aquifer system. Groundwater flowed freely both parallel to and perpendicular to the Balcones fault zone, from injection points to both spring complexes and other monitoring sites, revealing the aquifer’s three-dimensional groundwater flow system. Infiltration from Onion Creek controls the boundary between the San Marcos Springs and Barton Springs springsheds. During wet conditions, the potentiometric mound beneath the creek blocks groundwater flowing from San Marcos Springs toward Barton Springs, a connection that normally occurs during dry weather (Fig. 4; Johnson et al., 2012).

The EAA injected nontoxic fluorescent dyes (Field et al., 1995) into six caves within the San Antonio segment of the Edwards Aquifer to trace groundwater flow paths and estimate groundwater-flow velocities in the Panther Springs Creek groundwater basin (Fig. 5) in northern Bexar County (Johnson et al., 2010). An additional objective of the tracer tests was to investigate the connectivity of the Edwards Aquifer with the underlying Trinity Aquifer, both of which are present in the basin.

The monitoring array consisted of 32 public and private wells, including irrigation wells at the Club at Sonterra (a golf club), Bexar Metropolitan Water District public water supply wells in the Hollywood Park area, and EAA monitor wells. The wells were completed in either the Edwards Aquifer or the Trinity Aquifer.

Tracer tests revealed discrete groundwater flow paths near Panther Springs Creek (Fig. 5). Dyes were detected primarily in well 68-28-608, with lower concentrations and groundwater velocities in other wells. Groundwater velocities to well 68-28-608 ranged from 1100 to 5300 m/d. Velocities to the other wells where dye was detected ranged from 13 to 2330 m/d. This demonstrated the rapid groundwater velocities in conduits as a result of steep hydraulic gradients in the recharge zone. It also indicated that groundwater flows between injection points in the upper member of the Glen Rose Formation (Trinity Aquifer) and monitoring sites in the Edwards Aquifer. Dye was injected into Boneyard Pit and Poor Boy Baculum Cave, a section that includes ~40 m of unsaturated Edwards Limestone, with enterable cave terminating in the upper member of the Glen Rose Formation. The local depth of groundwater indicates that Blanco Road Cave most likely extends through the Edwards Group limestone to the Glen Rose Formation limestone. Seven of the wells where dye was detected are completed in the Edwards Aquifer, and one was completed in the Trinity Aquifer. Flow paths between caves and wells crossed several northeast-southwest–trending faults across which members of the Edwards and Glen Rose Formations are juxtaposed. Faults with up to 104 m of vertical displacement did not impede groundwater flow. These traces demonstrated excellent communication between groundwater in the Upper Trinity Aquifer and the Edwards Aquifer in the study area. This is significant because the two aquifers are regulated differently by the City of San Antonio and State of Texas agencies. One trace was also initiated through a 1 m² site thinly covered by soil in an interstream upland area with no observable surface karst features. Dye was injected in this site followed by 180,000 L of water (at an average rate of 250 L/h) over a 1 mo period. Dye was subsequently detected in three wells with an apparent groundwater velocity of 16 m/d over a distance of more than 8 km. This trace demonstrated that vulnerability to contamination is not limited to recognizable karst landforms such as caves and sinkholes.

Groundwater flows southward and eastward from the western reaches of the Edwards (Balcones Fault zone) Aquifer to the San Antonio area, which is divided into three pools for management purposes. Tracer tests were undertaken to investigate the connections between the aquifer in Kinney County and Uvalde County and the San Antonio pool (see Sharp et al., this volume, and Green et al., this volume).

Between 2007 and 2013, EAA completed 13 tracer tests at various locations in Kinney County to investigate groundwater flow paths to Las Moras Springs and to the adjacent Uvalde Pool (Johnson and Schindel, 2015). More than 100 monitoring sites, including Las Moras and Pinto Springs and public and private wells completed in the Edwards Aquifer formed the data collection network for these studies (Fig. 6).

The tests revealed discrete groundwater flow paths with a range of slow to rapid groundwater velocities connecting the injection points to down-gradient wells and springs. Velocities ranged from 1 m/d to more than 1350 m/d. Injection points were Alamo Village Cave, HF&F Cave, Grass Valley PW-1, Whitney Cave, and Pratt Sink in the northern part of the study area, the Dooley irrigation well in Pinto Valley, and the Boerschig well located approximately 3 km northwest of Brackettville, Texas. Dyes from Grass Valley PW-1, Whitney Cave, and Pratt Sink traveled radially to the south, east, and west, influenced by a structural embayment in the aquifer that maintained relatively low groundwater gradients. Deep flow paths probably influenced by geologic structures resembling anticlines likely diverted dyes southward to Las Moras Springs and nearby Brackettville municipal wells. Groundwater compositions varied at those locations, which suggests discrete flow paths to each well. Dye from Alamo Village Cave traveled to Pinto Springs. Dye from HF&F Cave was detected in wells north of Brackettville. Dyes from the Boerschig well were detected at Las Moras Springs and a municipal well, although the dye detections seemed to change with aquifer stage and Las Moras Springs discharge. Igneous intrusions in the vicinity of Las Moras Mountain created a barrier to groundwater flow and diverted dyes westward toward Pinto Valley.

The tracer tests confirmed that recharge from the West Nueces River in the north-central part of the county infiltrates into a structural embayment. Groundwater flows south toward Las Moras Springs, east toward Uvalde County, and west toward Pinto Valley. Tracer tests revealed the highly heterogeneous, three-dimensional groundwater flow system in the Edwards Aquifer.

Tracer tests are powerful tools with which to investigate groundwater flow paths in the artesian, contributing, and recharge zones of the Edwards Aquifer, where groundwater velocities have been measured in thousands of meters per day. The tests revealed that the karstic nature of the Edwards Aquifer is characterized by high-velocity, discrete flow paths to a few large regional springs with three-dimensional hydraulic systems.

• (1) Tracer tests in the vicinity of San Marcos Springs and Barton Springs indicated that both spring complexes are important discharge points for the Edwards Aquifer system. Results indicated that groundwater in the Edwards Aquifer, virtually anywhere in Hays County, may discharge at San Marcos Springs or Barton Springs, depending upon aquifer stage. Consequently, both springs are vulnerable to activities in their springsheds that may degrade water quality. Endangered species and other aquatic wildlife are potentially vulnerable to water-quality impacts in the recharge zones of these springs.

• (2) The Barton Springs segment of the aquifer is demonstrated to have three groundwater basins with distinct flow paths. The Cold Springs basin does not discharge to Barton Springs, but rather to Cold Springs and directly to the Colorado River.

• (3) San Marcos Springs is recharged by both regional and local sources of groundwater. Groundwater from the artesian zone, flowing northeastward along the strike of the faults (see Ferrill et al., this volume), supplies the largest volume of spring discharge, based on the fast apparent groundwater velocities measured in tracer tests (Johnson and Schindel, 2008). Apparent velocities from the southwest flow paths were significantly faster than from other directions. Tracer tests also indicated that the Blanco River is a regional source, especially during dry periods, when it helps to sustain both San Marcos and Barton Springs.

• (4) The groundwater divide between the San Marcos Springs and Barton Springs segment is located near Onion Creek. Groundwater moves east and west toward San Marcos Springs and Barton Springs during wet conditions when infiltration from Onion Creek forms a potentiometric mound.

• (5) Tracer tests showed that Blanco River recharges both the San Marcos and Barton Springs complexes, especially under low-flow conditions. To reach San Marcos Springs from the Blanco River injection points, groundwater had to flow through multiple members of the Edwards Aquifer under both unconfined and confined conditions. These findings revealed the three-dimensional groundwater flow system in the Edwards Aquifer.

• (6) Groundwater flows to Comal Springs through the artesian and recharge zone fault blocks in discrete flow paths, as shown by both tracer tests and water-quality contrasts.

• (7) Tracer tests showed that preferential groundwater flow paths are associated with the Panther Springs Creek basin in northern Bexar County. Injection points consisted of several caves that all demonstrated a connection to groundwater, with little, if any, attenuation. Although some caves transmitted dye to groundwater more quickly than others, all had a direct and unfiltered connection.

• (8) Groundwater carried dyes both parallel to and perpendicular to the Balcones fault zone from injection points in northern Bexar County to the San Marcos and Barton Springs complexes. This indicates that faults may not act as flow barriers. However, the faults and juxtaposition of the Edwards Group members do affect the permeability field and, consequently, hydraulic gradients and apparent velocities.

• (9) Solution features associated with high-transmissivity flow paths crossed several northeast-southwest–trending faults in which members of the Edwards Group and the Glen Rose Formation are juxtaposed by as much as 104 m of vertical displacement.

• (10) In Kinney County, groundwater flows from up-gradient areas to Las Moras and Pinto Springs at velocities ranging from 1 m/d to more than 1350 m/d. Tracer tests also showed a connection to groundwater in the Uvalde Pool.