Stormwater Remediation in a Karst Watershed

Along with thousands of other small municipalities, Bowling Green, KY, became responsible for Phase II implementation of stormwater regulations under the National Pollutant Discharge Elimination System (NPDES) in 2003. As a part of its Phase II program, the city installed a structural water-quality device (SWQD) at the entrance of Bypass Cave. The cave entrance is located within a sinkhole and contributes recharge to the Lost River karst groundwater basin. The sinkhole is actively used for stormwater drainage and contains several drainpipes that empty into the cave entrance. The SWQD was installed to filter out surface debris and trash, and to improve the quality of the water entering the local karst groundwater system. City and county stormwater coordinators have been working with Western Kentucky University (WKU) to ensure that the NPDES Phase II stormwater program is rigorous and produces scientifically usable monitoring data.

For many years, the WKU, in coordination with the city of Bowling Green, has been studying Bypass Cave and its impact on local groundwater. Before installation of the SWQD, a set of water samples was collected to establish a water-quality baseline (Figure 1). An assessment of the cave passage was conducted to document the impact of stormwater flow and surface debris. A second set of samples was collected a year after the device was installed. Water-quality analyses subsequent to the SWQD installation indicated that the device had a negative effect on water quality and actually increased the amount of debris entering the cave system. A follow-up cave passage assessment documented a substantial increase in surface trash and debris in the cave, and also revealed that this material was being carried deeper into the cave system. This is likely a result of the overload of stormwater on the SWQD and an inadequate detention basin. Installing the unit caused a decrease in detention basin capacity, and nearly all infiltration within the basin was blocked so that the entire volume of stormwater must now pass through the cave system. The unit is unable to handle the increased volume of water that results from storm runoff that initially fills the detention basin behind the device. As the volume of water increases, it develops enough hydraulic head to flow over the device, and is injected into the far reaches of the cave passage. Implementation of a SWQD must include a corresponding increase in detention basin storage capacity so that the slower movement of water through the system and decreased infiltration can be accommodated. A sizable amount of infiltration generally occurs in sinkhole detention basins, and SWQD installation must actively minimize the negative effects on this process for remediation to be successful.

Karst Terrain
Karst is a type of landscape formed from the dissolution of soluble rocks including limestone, dolomite, and gypsum. Karst regions are characterized by sinkholes, sinking streams, caves, and springs. Nearly all surface karst features are formed by internal drainage, subsidence, and collapse caused by the development of underlying caves (Ford and Williams 2007). Karst terrain is present in 18% of the lower 48 states of the US and 25% of the world (Figure 2).

Rather than overland flow through streams, karst water flows belowground through systems of conduits and fractures until it emerges as a spring. Surface drainage through stream networks disappears, and sinkholes replace these features as the subsurface flow increases due to ever-enlarging conduits. Subsurface water in these systems moves very quickly to a spring, similar in speed to pipeflow (Brosig et al. 2008). A myriad of local planning problems are specific to karst landscapes, including sinkhole collapse, sinkhole flooding, and an easily pollutable groundwater supply (Crawford 2001). In non-karst areas, groundwater moves far more slowly, and this laminar flow and contact with soil and soil organisms allows for greater removal of contaminants from groundwater than in karst regions (White 1988).

Figure 3. Bypass Cave before installation of the structural BMP. The weir to measure water volumes blocks most of the entrance from view.

Bypass Cave is an insurgence cave in Bowling Green, KY, that has been dye-traced to the Lost River Cave system (a major tourist attraction). It most likely intersects the main Lost River Cave stream downstream from the terminal breakdown in Alexander Cave (Schafstall 1984) and eventually flows into the Barren River, which is the drinking water source for the city. Bypass cave has a surveyed length of 378 meters (1,247 feet) with a vertical extent of approximately 11 meters (35 feet). The cave entrance is situated within a small sinkhole located approximately 100 meters from US Route 31. A weir was installed by Crawford & Associates in the 1980s to monitor water quality and storm flow volumes (Figure 3). Before the SWQD installation, the entrance sink periodically contained small amounts of trash consisting of old clothes, aluminum cans, plastics, tire segments, and hubcaps. Many commercial businesses are located on US Route 31 near the entrance to the cave, including restaurants, clothing stores, a car repair shop, banks, and other small businesses. The Bypass Cave entrance sinkhole drains a large paved area of streets and parking lots. The cave also extends under residential areas located just off of the road.

A general understanding of stormwater quality during wet weather can be ascertained by monitoring water within the cave to evaluate how effectively specific best management practices (BMPs) improve water quality. Water sampling within the cave was conducted for baseline water-quality and post-SWQD water-quality comparisons. The SWQD was installed with the intention of filtering the flood-water runoff, which would ultimately improve water quality entering the Barren River system where Bowling Green’s drinking water intake is located (Figure 4). The SWQD has six baffles—three high, three low—and a collection grid in the back (Figure 5). The three top baffles filter floating debris, and the lower three baffles collect all of the contaminants that are heavier than water.

Figure 4. The sinkhole entrance after structural BMP installation. A student is standing at the cave entrance.

Figure 5. The structural BMP installation and internal construction

Figure 6. Water quality in the cave directly below an auto repair facility

Figure 7. Trash and debris deep within the cave in an area that had been clean for decades prior to the installation of the structural BMP

Cave Passages
Most of the passages in Bypass Cave are narrow stream canyons with dimensions of 3 meters high by 1 meter wide. Several small infeeders bring water into the stream canyon during wet weather. The floor of the stream canyon contains a series of discrete, shallow pools of water, which form a stream during very wet weather. The canyon is decorated with flowstone and small stalactites, which indicate input from epikarstic water.

The foundation of a building intersects the cave at approximately 60 meters into the cave. The foundation extends one-half meter below the natural roof of the cave, but does not block water flow through it. At this point in the cave, the water quality changes from a stormwater gray color with oily residue (which occupied all interior pools) to include an orange organic growth, possibly associated with the deposition of metals and other automobile wastes (Figure 6). This change begins approximately below an automobile repair facility located on US Route 31.

A significant amount of surface debris (both natural and anthropogenic) was observed to litter the floor, walls, and ceiling of the passage during the initial assessment. Natural surface debris consists of leaves, sticks, and branches. The manmade debris includes glass bottles, aluminum cans, plastic bags, hubcaps, tire pieces, old pipe segments, bricks, and pieces of rebar—likely from the building foundation. The debris reaches almost to the ceiling in places, indicating that the first 200 meters of passage can flood to the ceiling.

At this point, the cave passage changes morphology from a simple canyon to a T-shaped canyon. The upper segment of the canyon is much wider in cross-section than the lower segment. A series of two flowstone ramps have formed short segments above the canyon floor. In between the ramps, the stream canyon can be observed below. After the second ramp, the passage remains above the stream channel and is very dry. After 40 meters, this dry (i.e., never flooded) section becomes nearly blocked with breakdown from past roof collapses. Beyond this point, the passage opens into the eastern edge of a large terminal chamber, 40 meters long by 20 meters wide. One has to climb down the breakdown to enter the room. A second passage also enters the chamber from the east. This passage contains an actively flowing stream, and the passage is a low crawl space that has been mapped for 90 meters, at which point it becomes too low to continue.

Secondary deposits of flowstone, rimstone dams, stalactites, stalagmites, and small soda straws are present throughout the cave, but mostly in the first 200 meters of the cave. The speleothem deposits appear to be active, which means that surface water is entering the cave through the cave’s bedding planes. Water is also entering through the various small joints and fissures that occur throughout the cave. Additionally, there are two minor and one major side passage water infeeders in the cave that do not flow past the SWQD location. The constant inflow of water from a variety of locations, even during dry periods, would limit the overall effectiveness of a cave-mouth SWQD for downstream water quality.

Because the biology of stream passages within a cave is an important indicator of water quality, a preliminary inventory of cave life was done at Bypass Cave. Organisms observed included tubifex worms, pouch snails, small flies, cave crickets, spiders, washed-in earthworms, beetles, and on the upper walls and ceiling small patches of actinomycetes (a filamentous bacteria). Tubifex worms are indicative of low oxygen concentrations, which characterize sewage-type pollution. Pouch snails generally indicate nutrient-enriched conditions and poor water quality. The fauna of the entire Lost River Cave system consists of troglobitic amphipods, isopods, crayfish, and blind fish. Yet none of these was observed in the stream canyon of Bypass Cave, which indicates that the water quality of the cave is seriously compromised.

Stormwater Impact Assessment
A preliminary investigation of Bypass Cave was conducted on July 23, 2004, and it was determined that water-quality sampling prior to SWQD installation would be appropriate. A second visit was made on September 30, 2004 to continue stormwater assessment of the cave and conduct baseline water sampling at five locations within the cave (Figure 1) to evaluate the changes in water quality as it moves through the cave system. A cave cleanup by Western Kentucky University students was organized in conjunction with the city stormwater program on October 23, 2004 to ensure that all anthropogenic debris was removed from the cave before SWQD implementation. The SWQD, which was designed to handle 25 cubic feet per second, was installed on February 21, 2006. A second set of samples was collected on November 29, 2006.

The last significant rainfall occurred on August 12, 2004, prior to collection of the pre-SWQD installation samples, and, thus, they represented “low-flow” conditions. This was the longest dry period for several months prior to sampling, and only intermittent pools of water covered passage floor. Normally, these pools are much deeper. Because the sampling was done at low flow, water quality is indicative of the contaminants that remain in the cave after flood flows and is a better representation of baseline water quality. The second set of samples was collected in 2006, shortly after a storm event with several prior weeks of dryness so that conditions within the cave were similar to those on the initial sample date.

Methods. Two sets of samples were collected from the cave stream in 40-ml and 1,000-ml bottles. A visual assessment of the cave passage was conducted and photo-documented. Analysis was conducted at Western Kentucky University’s WATERS Laboratory both before (2004) and after (2006) the installation of the SWQD. Analysis was performed for cations, anions, metals, atrazine, fecal coliform, and oil and grease (Table 1). In addition to water sampling, other properties were also measured (pH, conductivity, turbidity, temperature, and dissolved oxygen) using the Horiba U-10 Multiparameter Water Quality Meter. This sampling regime was developed by the Bowling Green and Warren County stormwater coordinators and Western Kentucky University scientists in a cooperative endeavor to ensure that Phase II stormwater monitoring leads to scientifically usable data. This sampling framework is used at a variety of surface karst features through Bowling Green to evaluate water quality and stormwater program effectiveness quarterly. This project extended the research belowground to evaluate water quality under more difficult circumstances.

Sampling Locations. Sample 1 was collected 3 meters after an in-feeding stream shortly past the cave entrance at the end of an overflow passage. Sample 2 was collected shortly just beyond the 90-degree turn in the main passage before the first climb. Sample 3 was collected between the second and third up-climbs.

Sample 4 was collected from an in-feeding stream shortly before its confluence with the main stream in the terminal big room. The water quality at this site should have minimal improvement from the SQWD because it is primarily fed by a second cave stream.

Sample 5 was collected in a stream after the confluence of the main stream and infeeding stream, located several meters upstream of a monitoring well. Because the water in this area is downstream of the confluence of the main stream and the last infeeding passage and it is impossible to ascertain contaminant sources, no sample was collected in 2006.

See Figure 1 for a map of sampling locations.

Results. The results of the analysis for 2004 and 2006 (Table 1) indicate that there is no significant improvement in water quality and that there is a decline in water quality in some areas. The post-installation cave passage assessment showed a major increase in the amount of surface debris and garbage entering the cave. The debris traveled far deeper into the cave than in previous years (Figure 7). Sediment coatings continued to cover all flooded surfaces, and trash was distributed much farther downstream in the cave passage than had previously been observed. There was an increase in lead (Figure 8) from 2004 to 2006. Chromium and other metals were pushed farther into the cave, from sample site 2 to sample site 3 (Figure 8). Copper and other metals increased close to the surface of the cave and were being carried in from parking lots and other paved areas. Calcium and chloride both increased dramatically as well. Fecal coliform increased at all sample sites except site 4 (Figure 8).

The most notable result for both sampling dates was the fecal coliform levels from site 1 (10,000 c/mL) and the lesser, but still high, levels at the other stations. Site 1 is located adjacent to the first infeeder encountered in the cave directly below a Chinese food restaurant. During a cave cleanup by Western Kentucky University students (October 23), a very septic odor filled the cave starting at the infeeder (where it was strongest) and continuing to the main room. Fecal coliform levels were also high at sites 4 and 5. The high fecal counts at site 1, 4, and 5 are strongly indicative of multiple septic system failures. This situation will not be resolved by using a structural water-quality device at the entrance of the cave.

Levels of several metals and of oil and gasoline at sites 2 and 3 were noticeably higher than at the rest of the sites for both sampling dates. These levels are indicative of possible wastewater leakage or chemical dumping from industrial or commercial sources. A transmission repair shop is located directly over the cave near this location and is almost certainly the source of these contaminants.

Dissolved oxygen levels at sites 1 through 3 are too low to support aquatic life (and lack of aquatic organisms was observed in the cave). Dissolved oxygen levels increased toward the back of the cave and reached nearly 6 mg/L, which is enough to support aquatic life, although no organisms were observed there.

The observed sediment deposits within the cave far exceed natural levels observed in other area caves. Sedimentation can negatively affect aquatic communities. Nutrients and toxic chemicals can attach to sediment particles, ride the particles in the water, and either settle with the sediment or become soluble in the water column. SWQD installation should have most directly improved this situation but appears to have made the situation worse.

Though nitrate levels at each site are below the EPA’s drinking water standards (10 mg/L), sites 4 and 5 had levels that were higher than natural background levels (>2 mg/L). Because this part of the cave does trend toward a residential area, it is possible that the above-background nitrate levels result from lawn fertilizer application.

Installation of the SWQD does not appear to have reduced the overall amount of trash and large debris entering the cave. Observations on the physical impact of floodwater in the cave passage have been ongoing for many years. Prior to the SWQD installation, all reports noted thick sediment coatings on flooded surfaces. However, the trash was typically reported to reach the area before the first up-climb in the cave—just before sampling site 3. The terminal chamber contained some plastic bag segments and a few bottles and cans, which is the typical flood debris that one would expect in a cave of this type. However, unlike the stream passage, the main walls and ceiling of the main room were not coated with flood debris. Past research teams have reported that this area remained dry and safe even during heavy rain and flooding of the entrance passage. The low stream passage that enters the main room from the east contained a gasoline odor, so it could not be safely assessed.

Since the SWQD installation, trash is now found in all sections of the cave, including as far as the terminal breakdown room at sampling station 5. Our assessment also shows that the trash has not simply been washed in, but has entered under hydraulic head. Trash has been observed crammed in small cracks and crevices in the walls and ceiling, and between and under breakdown blocks. Much of this material is broken, torn, and tattered into much smaller pieces during transport into the cave passage. The location and state of the trash indicates that the SWQD has a limit on how much stormwater it can filter. Surface runoff from a typical storm pulse begins to accumulate and then back up at the entrance of the cave, and infiltration into the sinkhole epikarst that had previously occurred has been blocked by the SWQD and its concrete apron. As more water accumulates, the flow develops enough hydraulic head to flow over the device and into the entrance of the cave under pressure. Trash that has washed off the surrounding streets and parking lots is literally injected into the cave and transported more than 300 meters to the terminal room of the cave.

Conclusions
Based on two sets of water samples collected before and after installation of the SWQD, the device does not appear to have improved the quality of the water within the cave, nor did it effectively filter any trash to keep it from entering the cave. The baseline water-quality data are from only one set of samples and provide only a snapshot of the pre-SWQD water quality. A series of samples, ideally for an entire year, would have given a more accurate assessment of existing water quality. The post-SWQD sampling set is also only a snapshot of the water quality at Bypass Cave. Our water-quality monitoring program will provide more accurate information on the state of the water in the cave in the future.

Visual inspection of the trash within the cave passage does indicate a significant increase in trash within the cave and also shows that the trash is being transported farther into the cave passage than it was previous to the installation of the SWQD. Treating the water before it enters the cave would be very effective in improving the overall groundwater quality if the entrance was the only source of water. However, two infeeders that enter the cave beyond the entrance have significant water-quality issues that will not be addressed with the entrance SWQD. In addition, water also enters the cave by infiltration from the surface and into fractures, fissures, and bedding planes that occur at the soil-rock interface. In order to improve the overall groundwater quality in Bypass Cave, the pollution for two infeeders and from nonpoint sources at the surface need to be identified and remedied. Commercial and residential land use above the cave most certainly impacts the groundwater not only during storm events, but also during normal flow conditions.

These initial findings are crucial for evaluating stormwater mitigation activities in Bowling Green. It appears that much of the pollution (i.e., fecal coliform) entering the cave cannot be directly tied to stormwater flows. If subsequent research demonstrates this to be true, then total maximum daily load (TMDL) levels in karst surface water bodies cannot be used to measure the success or failure of the stormwater program. Instead, this type of targeted in-cave monitoring will best demonstrate the contribution to water quality made by Bowling Green stormwater mitigation activities.

Overall, structural water-quality devices in karst areas are not going to provide a magic bullet to address stormwater quality. Land use controls and planning, combined with required sewering of homes and business, are critical. It is also critical to determine the locations of caves, which are the major conduit of subsurface water flow, so that polluting activities are not located directly above cave streams, or so they are effectively regulated if they are currently located in the area. Most importantly, while structural water-quality devices have a role to play as part of an integrated stormwater management program, it is critical that their installation provide for an increased volume in the attached detention basin and that infiltration into the epikarst is maximized so that the devices are not regularly overwhelmed and, thus, create a worse water-quality situation.

Author's Bio: Pat Kambesis has more than 30 years of experience mapping and conducting inventories in caves and karst areas both nationally and internationally. She is a karst hydrogeologist and the assistant director of Hoffman Environmental Research Institute.

Author's Bio: John All, Ph.D., is an associate professor of environmental geography at Western Kentucky University. He is also a lawyer who specializes in environmental law. He works with the Hoffman Environmental Research Institute and is the head of the Human-Environment Linkages Program.

Author's Bio: Jeremy Goldsmith is a student at Western Kentucky University in the Environmental Management Program.

Author's Bio: Ronson Elrod is a student at Western Kentucky University in the Environmental Management Program.