Sinkholes and the Engineering and Environmental Impacts of Karst (2005)

This talk is an update of an address presented seventeen years ago at a similar meeting (Alexander, 1988) and builds on the common experience base of the participants at this Convention to address the following questions. Why do so many geoengineers and hydrogeologists in North America know so little about karst? Why do travesties such as the Walkerton Investigation (Worthington, 2005) happen in the 21^st century?

Geotechnical engineers and hydrogeologists often interact to resolve common problems, particularly on projects in areas underlain by karst. In these cases, the geotechnical engineer needs the hydrogeologist to help in the characterization of the karst features so that specific geotechnical analyses can be performed. The hydrogeologist, in turn, needs the geotechnical engineer to clearly explain the specific geologic and hydrogeologic characteristics that are needed for the analyses. Unfortunately, engineers and hydrogeologists are trained to approach and address problems differently and, therefore, the interaction can be a source of frustration for both. The purpose of this paper is to briefly identify some of the reasons for the frustrations and to propose recommendations for the project interactions that will hopefully aid both the geotechnical engineer and hydrogeologist in future interactions. The paper will consider several projects where the geotechnical engineer needs to perform specific geotechnical stability analyses and identifies specific data needs that can be provided by the hydrogeologist.

This paper presents the basic elements of a conceptual model for the development of epikarst in US mid-continent, horizontally-bedded carbonates in which flow is largely confined to secondary and tertiary porosity. The model considers the development of epikarst regimes in carbonate sequences beginning shortly after non-carbonate rocks are eroded away to expose the underlying carbonates and follows this through capture of the shallow flow by deeper dissolution conduits with reorientation of the epikarst to a more vertical form. The model does not require an underlying zone of vadose flow and in many cases considers development of such a zone to depend on the water supply provided by prior development of the epikarst. It is not claimed that all epikarsts form in the accordance with this model; rather this paper presents a viable additional model for epikarst formation under appropriate starting conditions. Factors influencing the development of epikarst are a combination of: 1) the pre-karst topography and modifications to this as the system evolves, 2) the original distribution and aperture of fractures as well as the distance and orientation of physically favorable fractures relative to potential discharge points, such as existing dissolutionally-enhanced channels with low head or nearby valleys, 3) character of soil cover as this affects percolation of water to the rock, erodability of the soil, sediment filling of conduits, and transport of sediment 4) variations in availability of dissolutionally aggressive water with time and location, and 5) low solubility layers, such as shale or chert, that promote lateral flow until a penetration point can be found. These interact to form an epikarst and deeper karst system that progressively increases its capacity both by internal improvement of its flow routes and extension into adjacent areas. The availability of water needed to promote dissolution also often has a positive feedback relationship to epikarst, in which locations of most active dissolution modify their vicinity to progressively increase capture of water, which promotes further dissolution. In early stages, lateral flow through the overlying soils and along top-of-rock must dominate the groundwater flow because the relatively intact carbonates have insufficient transmissivity to convey the available recharge through the body of the rock. Top-of-rock runnels developed by a combination of dissolution of their floors and piping erosion of their roofs would carry a significant portion of the flow. Horizontally-oriented epikarst develops with discharge to local drainage. Cutters and pinnacles, collapse-related macropores, and areas of concentrated recharge would begin to form at this stage. Initial downward propagation of this system would occur mostly due to lateral flow. Mixing corrosion could occur in sumps in these lateral flow routes when fresh, percolating rainwater mixes with older water with a higher dissolved load. Should conditions be suitable, leakage from this system promotes the migration of deeper karst conduits into the area by Ewers multi-tiered headward linking. Other sources of water may also bring in such deeper conduits. Once such deeper conduits are present, the epikarst can evolve into a more vertically oriented system, at least in the vicinity of master drains into this deeper system. Former shallow epikarst routes may then plug with sediment. In some areas, deeper systems may never develop due to unfavorable conditions. The epikarst may be the only significant system in these cases. This includes the case of poor karst formers such as interbedded shales and carbonates that may have very shallow horizontal epikarst flow paths that channel shallow subsurface flows.

The Bushkill Creek (Northampton County, Pennsylvania, USA) is a transitional fluvial system with characteristics attributable to both a straight channel stream and a meandering stream. Once within the Lehigh carbonate valley, the stream is an alluvial ponor (swallow hole). It is a losing stream with a significant percentage of the stream waters in a state of divergence through alluvial and glacial sediments that cover a well-developed karstic bedrock surface. The Bushkill is characterized by stretches of deep pools, riffles and periodic mid-stream bars composed of primarily pebbles and cobbles. These mid-stream bars are persistent features that can be observed in aerial photographs dating from the 1930's to the present. Aerial photographs that pre-date 1965 show a segment of the Bushkill between SR (state route) 33 and SR 2017 as having a split channel that circumvents a mid-stream bar. During the late 1960's, the two-channel stream was combined into one channel with the construction of SR 33. Throughout 2004/2005, both banks and the floodplain of the Bushkill Creek were being affected by an inordinate amount of sinkhole activity that was directly influencing the alignment of the main creek channel. It was hypothesized that the stream channel was making an effort to revert back to its pre-1960's configuration. Sinkholes that had opened along the north bank coalesced and eventually pirated water from the stream creating a small meander bend. Increased erosion due to subsequent rainfall and flooding eliminated the meander "neck" resulting in a recessed bank. This in turn encouraged a shift in the streams flow direction. The process repeated as existing sinkholes were connected with the now prograding stream creating a new stream channel. Although the sinkhole activity and subsequent stream response can be attributed to construction and mining activity as well as significant swings in precipitation amounts, the mechanism of stream migration is worth noting. This process of "sinkhole piracy" has been observed along other surface streams in Pennsylvania and what has been observed with the Bushkill Creek may serve as a time lapsed segment of stream evolution in a karstic terrain.

Several sinkholes occurred in 1988 in an urbanized development in Kuwait. The properties of the deep sub-soil and bedrock was investigated through a detailed research program consisted of topographic, geological/geotechnical, and geophysical investigations. The study revealed that the geological profile is made of 35–40 m thick sandy overburden soil uncomfortably overlaying a Karst limestone bedrock. The cause of these sinkholes was attributed to the dissolution of the limestone bedrock and the subsequent raveling of the overburden soil cover. New sinkholes occurred in the summer of 2004 in the same area, indicating the need to assess the extent of the Karst Terrain in the area and its surroundings. The investigation documented that the same general geological profile also exists along the eastern coastline of the Arabian Peninsula, which had experienced several cases of sinkholes due to Karst cavities. This paper presents a comprehensive description of the desert karst terrain that exists in the eastern coastline of the Arabian Peninsula in general and in Kuwait in particular. The paper also presents an analysis of the causes of sinkholes that occurred in Kuwait. Discussion of the consequences of the sinkholes incidents on the regulation of the local government in setting guidelines for preliminary investigations for futuristic urban development in similar geological terrains in order to minimize the deleterious effects resulting from ground failure is also presented.

Variations in the lithology of carbonate units can result in different morphologies in caves and conduits and in their hydrologic roles in karst aquifers. Examples are provided from the San Antonio area of central Texas. Solutional sinkholes are common in the Person Formation but less common in the Kainer Formation where small, solutionally enlarged fractures tend to open to large and deep caves. Members of the Person usually form broad rooms and passages along bedding planes and often contain considerable breakdown due to medium to thin limestone beds that frequently contain incompetent interbedded clay. Groundwater and conduits are occasionally perched for short distances on the Region Dense Member, with the longest reported horizontal extent estimated at 500 m before dropping into the Kainer Formation. The Kainer's Kirschberg Member produces few caves and karst features in San Antonio, but facies changes make it highly cavernous near Austin 150 km to the north. Conversely, the Kainer's Dolomitic Member near San Antonio forms some of the deepest pits in the state yet is essentially non-cavernous in north Austin. The Basal Nodular Member and the upper member of the Glen Rose Formation form some of the largest chambers and passages in the study area, yet are effectively non-cavernous elsewhere. However, even where the Glen Rose produces large caves, few significant karst features form in its outcrop. These and other characteristics can be derived from detailed analyses of cave and karst feature databases and can be used as a predictive tool to improve the efficacy of environmental impact assessments. Examples include the prediction of groundwater drainage area size and location, estimating the volume of flush water needed for successful tracer tests, aquifer conceptual model development, and improving the design and interpretation of geophysical studies.

During the crisis of a flooding limestone or dolomite mine, a thorough site investigation of the karst flow conduit must be performed before any remedial work is attempted. A plan for remediation which is based on an accurate understanding of the flow conditions, rather than a concept, obviously has a much greater chance of success. The considerable risk of attempting to seal a flow conduit is reduced with an accurate site model. Three different models of flooding flow conduits have emerged. Site 1 is a Pennsylvania dolomite quarry which intersected a flooding flow conduit at 130 feet ((40 m) depth, allowing 10,000 gpm (38,000 L/min) of river water to flood the quarry. A conduit of 1,500 feet (460 m) connects the river and quarry. The local geology is Ontelaunee dolomite. River water enters a bedding-controlled conduit aligned with a master joint structure. The floodwater flows from here into a solutioned high-calcium limestone bed which follows the limb of an anticline structure to the quarry pit. Site 2 is a Pennsylvania limestone quarry with inflows of river water at 60 feet (18 m) and 120 feet (37 m) depth. The local geology is Myerstown-Annville-Ontelaunee-Epler. Water flows into epikarstic caverns beneath the river and to a conduit near the Annville-Ontelaunee contact. A zone of densely-spaced high angle faults pass east to west through the floor of the quarry, are solutioned, and channel the flow to several artesian outflows. Site 3 is a West Virginia Dolomite quarry with in excess of 10,000 gpm (38,000 L/min) of river inflow at 120 feet (37 m). This is the site of North America's largest bitumen grout curtain in karst limestone. A new conduit developed adjacent to the remediated feature. The local geology is Tomstown dolomite-Antietam sandstone. Water enters sinkholes in the riverbed over a zone of karstified limestone that is intercalated within dolomite. The conduit follows the strike of limestone bedding which has been intensively deformed within a plunging syncline structure. Recent mining has deepened the quarry into the projected pathway of the flow conduit. More models will be offered as guidance as projects are completed. These case studies are offered as templates to apply to other sites for expediting the conceptual site modeling, and characterization work, thus improving on the chances of a successful remediation.

In a mantled karst terrain engineering practices can vary often requiring innovative approaches that are significantly different from traditional ones. Three situations are considered as an example. The problems were as follows: (1) Pre-construction assessment of sinkhole development probability for a residential area located in the vicinity of Moscow with insufficient karstological data. (2) Pre-construction karst risk assessment for an industrial site in Dzerzhinsk, Nizhny Novgorod region. (3) Quantitative classification of karstified terrains for the purposes of selection of waste disposal area for industrial and household wastes. The researchers dealing with problem number 1 had limited data on previously recorded sinkholes but sufficient information on previous overburden studies in the region. Assessment of collapse risks was performed with due attention to the specificities of the karstified terrain relief, karst location depths and the properties of the overburden rock. Among the methods used there was assessment by experts (rating scales), entropy and a round cylinder model of karst sinkholes. In the second case related to designing of an industrial structure there were no difficulties with basic data on registered sinkholes and hence a traditional method could be used. But the customer's demand was to assess the karst risk for both the entire industrial site and some separate constructions in order to plan adequate karst monitoring and protection measures of a long- and short-term period. The assessment method proposed was to obtain the risk level as a ratio between probable construction collapse risk and the allowable risk. Risk levels were assessed at different stages: after engineering protection of the building had been performed and after regular maintenance procedures. Problem number 3 considered a method of classifying karstic terrain according to the level of geological environment contamination for the purposes of selection of a site for waste disposal (household/industrial wastes). The classification is based on a probable level of the environmental contamination resulting from sinkhole development.

The meaning of terms "karst hazard" and "karst risk" are discussed. Quantitative measures of considered risk, at any level of forecast assessment include the following: probable damage from karst deformations, frequency (probability) of its realization, and also the product of the mentioned parameters characterizing annual physical, economic, social or ecological losses within a certain territory. The general procedure of regional assessment of karst collapse intensity (cases of failures / years ⋅ km², m²/ years ⋅ km²), vulnerability of objects of national economy and population for these deformations, as well as karst economic ($/year, $/km² ⋅ year), social (per. /year) and individual (per. /per. ⋅ year) risks of losses are discussed.

In May, 2000, the Tennessee Department of Transportation (TDOT) initiated a road-widening construction project on State Route 75 in Washington County in a region underlain by folded and faulted carbonate bedrock. The project involved deepening an existing shallow cut through a small hill (∼20–30 m high) to remediate a blind intersection. During the course of excavation, the contractor removed several meters of yellowish-brown, chert-rich regolith across the crest of the hill and encountered an underlying soft dark-brown to black, fine-grained, graded to laminated, organic- and clay-rich deposit that presented a stability problem for road construction. Later drilling delimited thicknesses of at least 30 meters of these sediments surrounded and overlain by the yellowish-brown cherty regolith. Geological results suggest a steep-walled structure of karstic origin at one time existed within the original bedrock. Subgrade stability failures were anticipated and a remedial plan was recommended. In June, TDOT and Tennessee Division of Geology personnel made a major paleontological discovery on the road project. Unusually well preserved vertebrate fossils including tapir, elephant, rhinoceros, alligator, and a number of smaller vertebrates including fish, frogs, and turtles were identified by scientists from The University of Tennessee and East Tennessee State University. These are the first fossils of probable Late Miocene age discovered in Tennessee and constitute the largest and best-preserved terrestrial Late Miocene to Early Pliocene (ca. 7–3 Ma) localities known in the United States east of the Mississippi River. The site is a highly significant scientific discovery, containing new species, including an animal related to the red panda, and the largest assemblage of tapir ever discovered. Abundant plant remains include leaves and leaf impressions, tree trunks, limbs, and twigs and seeds. In September, 2000, the Governor of Tennessee announced that TDOT would relocate a portion of the SR 75 road project to bypass the fossil-bearing sediments and preserve the site, which had been designated by the State Archeologist as the Gray Fossil Site. This action saves the site for scientific research and educational purposes, under the auspices of East Tennessee State University, but has mandated relocation efforts that are both expensive and time-consuming. Mounting evidence—especially in finds of fossil plants, but also lignite and other regolith—suggests that similar regolith deposits may exist in karst belts elsewhere within the Appalachians, where they may be encountered by future construction activities. Increased knowledge of localization factors for such features will aid in prospecting for infilled deposits so that they can be avoided during planning for future projects.

Many sinkholes collapsed since 1998 in Wu Han, China. Sinkhole collapses have put local people's properties and safety in great danger. This paper introduces a large-scale experimental study to implement physical models of sinkhole collapse in this area. Two conceptual models of sinkhole collapsing have been established based on the processes of sinkhole formation, basic geologic conditions, hydrodynamic features, and human activities in this area. Rock, soil, and water samples from sinkhole collapsing areas have been used for a large-scale experiment to verify the conceptual models. High resolution fluid pressure transducers and soil pressure transducers were used in the experimental study to monitor pressure changes in fluids and soil. Measurements of pressure changes were automatically collected and inspected during the experiment. Quantitative studies of Quaternary groundwater use, karst water use, Quaternary soil texture and structure, and the size of the pipelines of karst openings have been conducted to investigate their relationships with the sinkhole collapsing processes. The threshold values of dominant physical properties to trigger a sinkhole collapse have also been studied in this area. Results from this research project provided important information and guidelines on how to prevent future sinkhole collapse in this area.

In this study, the numerical model CAVE (Carbonate Aquifer Void Evolution) was used to simulate the development of solution conduits beneath a weir that is constructed on river sediments overlying gypsum bedrock. The weir site is located at Hessigheim on the Neckar River in the State of Baden-Württemberg, Germany. After flowing through the State of Baden-Württemberg over a length of 367 km, the Neckar flows into the Rhine River near Mannheim and Heidelberg. At this site, a non-soluble layer overlying soluble rock limits the maximum leakage from the reservoir. The model was used to identify the key parameters controlling void evolution under flow conditions controlled by soluble and non-soluble rocks. In all, three scenarios were simulated. Sensitivity analyses indicate that the breakthrough time was influenced by permeability of the stratigraphic units, initial conduit diameter, conduit length and the number of nodes coupled to the gypsum layer. Leakage rates of the order of 10^–7 m³ s^–1 (about 9 L day ^–1) were observed after breakthrough for the standard scenario. Thus, under limited flow conditions, if the hydraulic conductivity of the intervening layer is sufficiently low, leakage rates may not pose a serious threat to the dam's ability to store water. However, the development of conduits having diameters of the order of tens of centimeters or more may pose a threat to the long-term physical stability of the dam.

Quantitative groundwater tracing is being used in the Woodville Karst Plain (WKP) of North Florida to characterize the hydraulic complexities of conduit flow for numerical modeling purposes. Accordingly, seven tracer tests were performed along both mapped and unmapped pathways ranging from 2.5 to 19.5 km in length utilizing between 0.7 and 4.5 kg of fluorescent tracer. Tracer recovery curves were measured at one or more points along a flow path for each tracer test. Computed mean velocities through the conduits ranged from 500 to 4560 m/day with peak concentrations between 0.12 and 125 ppb. The data obtained from the tracer tests have been incorporated into a finite element numerical model of groundwater flow in the WKP. Through-conduit velocities and mapped flow paths identified by the first five tracer tests were used in conjunction with cave map data to build discrete conduit pathways into the model framework and set flow parameters within the numerical conduits. Particle tracks generated from the resulting model were then used to highlight flow paths associated with a sinking stream that was the focus of the last two tracer tests. Subsequent tracer test results closely confirmed the model predictions. Current model results indicate that this approach will provide an effective method of realistically simulating spring flows and delineating individual springsheds within the karst basin. Further details about the tracing projects are available at www.hazlett-kincaid.com/FGS.

Numerical ground-water flow models for the Edwards aquifer in the San Antonio region of Texas generally have been based on a diffuse-flow conceptualization. That is, although conduits likely are present, the assumption is that flow in the aquifer predominantly is through a network of small fractures and openings sufficiently numerous that the aquifer can be considered a porous-media continuum at the regional scale. Whether flow through large fractures and conduits or diffuse flow predominates in the Edwards aquifer at the regional scale is an open question. A new numerical ground-water-flow model (Edwards aquifer model) that incorporates important components of the latest information and an alternate conceptualization of the Edwards aquifer was developed. The conceptualization upon which the Edwards aquifer model is based emphasizes conduit development and conduit flow, and the model can be considered a test of one of two reasonable conceptualizations. The model incorporates conduits simulated as generally continuously connected, one-cell-wide (1,320 feet) zones with very large hydraulic-conductivity values (as much as 300,000 feet per day). The locations of the conduits are based on a number of factors, including major potentiometric-surface troughs in the aquifer, the presence of sinking streams, geochemical information, and geologic structures (for example, faults and grabens). The model includes both the San Antonio and Barton Springs segments of the Edwards aquifer in the San Antonio region, Texas, and was calibrated for steady-state (1939–46) and transient (1947–2000) conditions. Transient simulations were conducted using monthly recharge and pumpage (withdrawals) data. The predominantly conduit-flow conceptualization incorporated in the Edwards aquifer model yielded a reasonably good match between measured and simulated hydraulic heads in the confined part of the aquifer and between measured and simulated springflows. The simulated directions of flow in the Edwards aquifer model are most strongly influenced by the presence of simulated conduits and barrier faults. The simulated flow in the Edwards aquifer is appreciably influenced by the locations of the simulated conduits, which tend to facilitate flow. The simulated subregional flow directions generally are toward the nearest conduit and subsequently along the conduits from the recharge zone into the confined zone and toward the major springs. Structures simulated in the Edwards aquifer model that tend to restrict ground-water flow are barrier faults. The influence of simulated barrier faults on flow directions is most evident in northern Medina County.

The Edwards Aquifer of central Texas is a karst aquifer developed in faulted and fractured Cretaceous-age limestones and dolomites. Numerous groundwater models have been developed for the three segments of the Edwards Aquifer since completion of the first major flow model in 1979. Groundwater models have helped refine our understanding of the relationships among flowpaths, recharge, groundwater pumping, and springflow. Tracer test studies have been performed on the Edwards Aquifer since 1982. These studies have also brought about a better understanding of aquifer flowpaths, yet the results of groundwater modeling and dye trace studies do not have a high level of agreement. A comparison of the two types of studies has indicated the strengths and weaknesses of each method. Groundwater models of the Edwards Aquifer have been effective in simulating spring discharge and determining water budgets. The models have simulated water levels reasonably well, but there are significant discrepancies between measured and simulated water levels in areas that are more strongly influenced by conduit flow. Tracer testing is the best method for measuring rates of flow from a recharge feature, or a well, to springs and other wells. However, tracer tests provide little useful information about water levels or water budgets. Traditional groundwater models are poor tools for simulating contaminant transport and delineating areas for source-water protection of mature karst aquifers with well-developed conduit networks.

Feicheng Basin is an important energy base in Shandong Province, North China. The basin is underlain by the Ordovician Limestone, in which the groundwater has been used for agricultural and industrial purposes for many years. Each individual water user controls the current exploitation of the karst water. Agricultural irrigation uses significantly more water in dry seasons, which often results in water shortage for industries. The dewatering activities in the coalmines within the basin discharge approximately 35 million m³ of water annually into local rivers. As a result of the current practice, the regional groundwater level has dropped more than 30 m over the recent 20 years. The groundwater is overdrawn in approximately 300 km², which has restrained the development of local economy. A numerical model with distributed parameters is developed for the karst aquifer to help properly manage the water resource and sustain its utilization. The responding matrix method is used to couple the prediction model and the management model. The objective of the modeling exercises is to obtain the maximum amount of water while keeping the water level drawdown within an acceptable level. The optimum management scheme is selected among two proposed designs. The recommended measures from this study include (1) comprehensive utilization of the water pumped from the mines, (2) improvement of water use efficiency in industries, and (3) improvement of the irrigation practice. When the domestic water supply and corporation water demand take the top priorities, the numerical modeling provides a useful tool to help with the rational distribution of the karst water resources among different water users.

Broad investigations into the extent of landfill groundwater impacts are often discouraged by adjacent property owner concerns. As a result, inferred groundwater potentiometric surfaces, flow directions, and monitoring well networks are often confined to the landfill property. In many circumstances, particularly in karst settings, this confined, myopic view can result in gross misinterpretations of groundwater flow directions and contaminant transport. Monitoring techniques utilizing existing offsite spring and well networks enable a better understanding of the distribution and migration of landfill groundwater impacts. Characterizing the extent of offsite impact in karst often reveals that the flow paths surrounding landfill sites situated in karst settings are radial rather than linear. Several case studies at closed landfill sites situated in karst areas of the Valley and Ridge geologic province of Tennessee are presented to illustrate the potential for complex conduit radial flow and offsite impacts. At each case study site, initial interpretations of groundwater flow were linear, and monitoring well networks were installed, sampled, and analyzed accordingly. Just one problem, the upgradient well(s) were impacted! Why? Radial Flow. By stepping back, and understanding groundwater flow from a more regional perspective, it was apparent that the supposed upgradient wells were actually downgradient due to radial flow patterns at these sites. Although newer Subtitle D landfills are considered dry tombs, many older landfills are inadequately lined and capped, leading to significant groundwater and surface water intrusion. This often creates mounding of groundwater within the disposal area. Radial flow patterns can evolve when groundwater elevations overcome former groundwater divides and create new flow patterns.

The Barton Springs segment of the Edwards Aquifer is the sole source of water supply for about 45,000 people in and immediately south of Austin, Texas. For water management purposes, it is important to be able to predict the availability of groundwater in response to future development and potential droughts. The Edwards Aquifer comprises heterogeneous carbonate rock strata that have developed a well-connected network of karst conduits. This karstic geological structure makes aquifer characterization and groundwater modeling challenging. Previously, Scanlon et al. (2001) developed a two-dimensional groundwater model for the Barton Springs segment using MODFLOW. The karst conduits were not explicitly represented in the model. Instead, the study area was divided into 9 zones for which the transmissivities were obtained through calibration. We revisit Scanlon et al.'s work in this study by using MODFLOW-DCM, a MODFLOW module developed to represent flow through karstic aquifers. MODFLOW-DCM adopts a dual-conductivity approach in which the aquifer is conceptualized as being composed of two interacting flow systems, i.e., the background matrix and the karst conduits. This approach allows karst aquifers to be modeled as coupled systems, and thus allows aquifer dynamics related to karst conduit flows to be accurately simulated. Our preliminary results show improved matching of both water level measurements and spring discharges records.

We used feeler gauges and image analysis to quickly characterize and interpret hundreds of fracture apertures and the cross-sectional area, aspect ratio, and perimeter of thousands of dissolution features in eight exposures of the Edwards aquifer in south-central Texas. Using frequency-size distribution plots, we showed that fracture aperture and the cross-sectional area and hydraulic diameter of dissolution features follow fractal scaling, with fractal dimensions ranging from 0.85 to 0.93 for aperture and from 0.62 to 1.15 for cross-sectional area of dissolution features. We used fractal scaling with the Riemann zeta function to estimate secondary porosity of the exposures. Estimates of secondary porosity using fractal scaling are two to five times higher for fractures and on average twice as high for dissolution features than those determined directly from the measured values.

Internal drainage microbasins were found constitute 10% of the 98 square km (38 square mile) portion of the Barton Springs segment of the Balcones Fault Zone Edwards Aquifer contributing to Barton Springs. Internal drainage basins have subtle geomorphological expressions that have historically been overlooked as a major recharge source within the Barton Springs segment. Internal drainage basins can be recognized and mapped based on large solution sinkhole bowls, internal contours on surface topographic maps, aerial photos, inspection of sinkholes during and after rain events, field delineation of catchment areas with GPS, and accounts from individuals such as property owners and cave explorers. We measured water balance components within a large internal drainage basin, the 19-hectare (46-acre) Headquarter Flat Sink to more directly quantify the amount of recharge contributed from internal drainage microbasins to the Edwards Aquifer. The monitored water balance components of HQ Flat that were measured for a year period (348 days), included the discrete recharge to the cave drain, evapotranspiration flux using an eddy covariance system mounted on a 15-meter tower, and soil moisture using insitu sensors and gravimetric analysis. Over the year test period, where rainfall totaled 106 cm, 58% left as evapotranspiration, 8% recharged directly through the cave drain, and 34% recharged diffusely across the internal microbasin of the cave site. The total recharge was measured to be 42% of the rainfall within the test basin. Only the discrete recharge portion was directly measured in a second internal drainage microbasin, the 22-hectare (54-acre) Hint Ridge Cave, since June 2003. The discrete recharge from Flint Ridge Cave was measured to be 9% over the same test year. During the 348-day test period, it can be estimated that known internal drainage basins contribution comprises about 5% of the Barton Springs flow reported by the USGS of 62,297,153 m³ (2,200,003,200 ft³). The water balance data from this study shows that internal drainage basins appear to be relatively efficient in recharging runoff for a given amount of source area because: 1) they do not lose runoff to areas downstream of the Recharge Zone; 2) naturally developed channels and bowls divert runoff rapidly into the subsurface; and 3) the recharge is typically limited to intervals during and shortly after rainfall events when the humidity is near saturation and net losses to evaporation are reduced. Recharge quantity can be significantly enhanced through preservation and restoration of flow to internal drainage basin sinkholes.

Recharge in the karst terrain may exhibit distinct characteristics. For the Edwards aquifer in the south central Texas, recharge mainly occurs in the form of stream transmission loss in the outcrop area—a clear manifestation of karst influence on recharge processes. This part of recharge has been quantified fairly accurately using stream gauging stations. However, recharge estimation from the upland areas is much less certain and is generally restricted to the modeling method. In order to understand the strong control of karst features on the recharge potential, and to quantify spatial and temporal variability of recharge in this region, a detailed analysis on the relationship between precipitation and recharge is conducted in this study. The water budget analysis of an upland watershed using synthesized data indicates that recharge in this area may have been underestimated by more than 10% of annual precipitation. We speculate that such high recharge potential and variability are closely related to the karst features and vegetation structure in the region. Process-based hydrologic study has provided us critical conceptual understanding of the recharge processes and regional hydrology. Further investigation to validate the information is warranted.

An experimental rainfall simulation plot (7 x 14 m) in the Edwards Aquifer region of Texas was established in dense juniper land cover to measure the effects of brush clearing on runoff. This project includes the monitoring of lateral subsurface flow using a trench (2.5 m deep x 8 m long) located at the downhill end of the plot. Two tracer tests with fluorescent dyes were conducted using simulated rainfall to assess the discrete flow paths discharging into the trench. Uranine was applied in a band at the uphill end of the plot, eosine was applied around the base of a tree in the center of the plot, and Phloxine B was applied in a narrow band at the downhill end of the plot. During the tracer tests, samples were collected at regular intervals from sixteen locations along the trench face where water flowed into the trench. A fluorescence analysis of water samples using a luminescence-spectrometer produced concentration values for each dye. An analysis of the tracer test results revealed: (1) not all areas in the plot contributed to the shallow subsurface flow into the trench, (2) subsurface flow paths displayed a high degree of interconnection, even over short distances, (3) conduits showed strong linkages to multiple plot regions, and (4) conduits showed strong connection with the area surrounding juniper vegetation, with rapid water flow from this area. These results demonstrated that the high variability in flow paths and interconnection encountered in the Edwards Aquifer exists even at the plot scale and highlights the need for greater research into vegetation-flow path interactions in shallow epikarst.

Few studies have quantified recharge into caves from overlying fractures. In areas such as San Antonio, Texas that rely primarily on karst (limestone) aquifers for potable water, the quantity and quality of recharge through these features could have significant implications for evaluating and managing groundwater. Vegetation cover also has the ability to influence recharge characteristics, when the vegetation is located directly above caves and inside the surface water and groundwater drainage basins. These studies in the Edwards Aquifer recharge zone north of San Antonio, evaluate the effects of woody cover on the water budget of two shallow caves. Use of large-scale rainfall simulation equipment above the caves allows re-creation of rainfall events where amount, rate, and duration of previously occurring events are simulated. One cave's footprint is instrumented with throughfall collectors and rain gauges to estimate canopy interception. Four trees are instrumented with stemflow collectors and transpiration measurement devices that record data, later scaled up to account for stemflow over the entire plot. Inside both caves, drip collectors constructed of PVC pipe and clear polyurethane plastic sheeting collect and route water to tipping buckets that digitally record recharge volumes onto a datalogger. In order to evaluate recharge rates and responses typical of the landscape, the collectors capture dripwater falling from the caves' ceiling, which travels into the caves through fractures, and not as focused recharge through sinkholes and cave entrances. Continuous automatic recording of recharge allows us to evaluate the effects of simulated and natural rainfall events on recharge rates, volumes, and durations. Preliminary analysis of natural and simulated data has shown that fractures can move large volumes of recharge quickly into the caves.

Small subtle karst features are ubiquitous in the uplands of the Edwards aquifer recharge zone of south-central Texas. The maintained surface expression of sinkholes, e.g. depressed area, of a number of sinkholes suggests these features are actively undergoing karst processes of soil piping or soil sapping. Much of the urban development around San Antonio and Austin is currently occurring in these upland areas; therefore, it is important to determine the hydrologic function of these features to minimize there potential impact on structures as well as to mitigate water quality concerns. Several artificial recharge experiments were conducted at typical small, soil-lined, upland sinkholes. Artificial recharge experiments using a constant-head ring infiltrometer indicated the infiltration rates at these sinkholes are not significantly higher; but are actually somewhat lower, than control plots. Both rates are within the range of infiltration for the 30% to 40% clay soils in the area. Volumetrically, these features likely recharge more than background due to microtopography. In order to further discover what impacts modification of these features might have on water quality in the Edwards aquifer, several follow-up tests were conducted. Repeat GPR tests were used to determine subsurface geometry of soil and rock and area of wetting distribution. Dyed water was ponded and allowed to infiltrate the features, which were subsequently excavated to determine the distribution of dye in the soil to locate preferential pathways. Once complete, the excavated features were subjected to additional constant head infiltration experiments to determine the infiltration rates of the underlying epikarst. The results of the follow-up experiments were compared with previous experiments to determine epikarst storage and the role soil has on infiltration rates of features with modified soil cover.

Although many people in the world live over karstic terrain, in the karst areas of the Appalachian Mountains they live over the remains of the largest mountains (above the ocean), which were thrown up in massive plate movement, and the accompanying volcanoes. The limestone rock, shale, coal, silt and clay layers laid down millennia ago have been thrust up into a tangled and contorted mess. In many places the strike of the rock is contorted, folded, and may go past vertical. When you drill into this mess, you encounter randomly placed hard rock, soft soil, aggregate layers, and complete voids. For the contractor who is attempting to build, he will find areas where it may take 15 or 20 feet of drilling in order to reach competent rock or hard cemented sand on which to found your structure. And that encounter may take place 50 or 100 feet underneath the surface. Occasionally you may never reach hard rock or competent soil. Drill holes may be wet or dry. Perched water tables and cave-sized voids are common. The designer of projects in karst areas needs to remain aware of the possible difficulties. Rembco does about 2/3 of its work in Appalachia. Decisions sometimes need to be made hourly, or even by the minute, about what action needs to be taken to prevent the loss of drill bits, the loss of casing, and the breakage of equipment, and about how to keep costs within reason. In this paper, we describe a number of projects which illustrate ground rules we have developed to overcome the variables. These projects may sound anecdotal, but when anecdotal evidence constitutes over half the company's work, it is an excellent foundation for approaching such problems. The illustrations come from drilled mini-piles, from soil nailed and anchored retaining walls, and from underground grouting carried out for many purposes including stopping and re-routing water.

By law, certain property damage attributed to sinkhole activity is covered by homeowners' property insurance in Florida. Engineers must devise and implement proper remedial techniques that include the stabilization of the house against future settlement. The remedial efforts most commonly used include underpinning, pressure grouting, or a combination of these two methods. The engineer usually has to provide a cost estimate for these repairs, along with detailed specifications for the contractor. Madrid Engineering Group, Inc. reviewed over 20 houses that were remediated by void filling and densification of raveled soils using low mobility grout, and determined that there was a wide variability in the estimated grout versus the actual grout. In order to obtain better estimates, the grout data was examined and, after normalizing from the base of the grout pin instead of the ground surface, indicated that three distinct injection regimes were encountered. The lower regime is located within the weathered limestone, and grout takes were generally high. A transition zone with moderate grout takes was noted. The remainder of the hole was generally in soft to firm clays or loose sands, and grout takes were much less. The data were used to provide a more accurate estimate of grout takes for future remediation cost estimates.

The Samanalawewa reservoir, formed by a 100m high embankment dam, is one of the largest reservoirs in Sri Lanka built in the recent times. It is located 160km south east of the capitol, Colombo, and is the key element of the 120-million-watt Samanalawewa Hydroelectric Project. It is constructed on one of the main rivers in Sri Lanka, the Walawe River. Geologically the project is in an area of karst within the Highland series of the Sri Lankan Precambrian complex comprised of crystalline metamorphic rocks. The area has been subjected to extensive folding, faulting and hydrothermal reactions. Thus, the project area is geologically complex. On several occasions a number of agencies have studied the project site, conducting various detailed geological, hydrogeological and geophysical investigations. Signs of a possible leak through the right bank appeared when poor geologic conditions were encountered during construction of the dam. Immediately, more investigations were carried out. During the initial trial impoundment, a small spring appeared downstream of the partially filled reservoir, and impoundment was suspended. Additionally, up to a distance of 2.5 km on the right bank, a flat water table was observed responding to the river levels. As a remedial measure, a 100m-deep, 1300m-long grout curtain was constructed, using 13,640 tons of cement, and impoundment resumed. The leakage continued, increased and finally resulted in a land slide. Impoundment was again suspended, and more detailed investigations were called for. The next remediation effort consisted of installing a liner of clay over the suspected ingress zones. Even after using 50,000m³ of clay, the leakage could not be stopped, although it did help to reduce groundwater pressure in the right bank. The reservoir is presently operating with a continuous leakage of 1800 litres per second. Recently, a number of sinkholes were found in an area far from the areas under consideration. Preliminary studies indicate a connection between these sinkholes and the main leakage. Detailed studies are underway to investigate the extent of the ingress areas as an aid in planning suitable remedial measures. Because of the complex geological conditions caused by karst in the area of the site, the exact mechanism of leakage cannot be established, preventing successful remediation.

Geotechnical engineering efforts must address the geological findings from site characterization, as exampled in the roadway widening and frontage road improvements along Interstate 70 (I-70), Frederick County, Maryland. The project site traverses a karst area that exhibits an aggressive sinkhole occurrence rate that averages roughly 8 sinkholes per year. Approximately 138 sinkholes have been reported in 2001 within an 8± km² area. A sinkhole risk assessment identified potentially hazardous areas of sinkhole occurrence and predicted the risk levels of sinkhole formation within the project corridor. Since completion of the sinkhole risk assessment analysis, approximately 63 percent of the new sinkholes developed in areas identified as moderate to high risk for sinkhole occurrence. A significant task in the geotechnical engineering efforts involves the control of stormwater runoff utilizing a geosynthetic membrane along drainage ditches and within three stormwater management ponds; creating the largest known impermeable stormwater facility in the State of Maryland. Two pumpstations pump the stormwater through a 1.8± km long force-main system and into an existing creek north of the project site. The pumpstations require a deep foundation system utilizing micro-piles. Proven installation techniques and innovative grout mix make micro-piles well suited for the anticipated cutter and pinnacle subsurface conditions. Geophysical and test-boring data lead to the development of subsurface contours along the top of epikarst and top of rock. The bedrock topography and the characteristics in the epikarst zone provide information on the groundwater flow patterns. Based on this information, the merits of grouting versus an inverse graded aggregate filter system that allows for the movement of water and minimizes the transport of fine materials are compared. Although based on proven geotechnical engineering principles, permitting restrictions prohibited the advancement of select treatments. As such, roadways traversing high-risk areas are grouted. Time Domain Reflectometry (TDR) systems will be used to monitor potential post-construction movement beneath the embankments. The instrumentation system will automatically notify appropriate personnel when a threshold value is exceeded.

In the past 20 years, new towns development in Hong Kong resulted in the discovery of areas of buried marble and karstic cavities. The buried karst has caused serious problems which significantly impact on foundation design and construction. Carbonate rocks include Carboniferous marble, Permian limestone and. Jurassic marble clast-bearing igneous rocks. These different carbonate rock types show various degrees of karst development. Based on a detailed ground investigation, the most intense karst development and the largest cavities occur in massive pure marble, in which the karst zone can extend to depths of more than –127 mPD. The degree of karst development of impure marble is less than that of pure marble. Marble clast-bearing igneous rocks show the least karst development, with large cavities being rare and smaller-scale honeycomb weathering being more typical. The influence factors of buried karst on foundation engineering are also recognized. These factors include geological structure, rock types, hydrogeology, paleogeomorphology and geomechanics. Among these factors, correct classification of rock types is very important in the design of ground investigation and construction works. The misinterpretation may increase property development costs substantially.

In January 2004 a sinkhole formed under a pier of the three span bridge carrying northbound SR 33 traffic over Bushkill Creek. The bridge was located in a highly active sinkhole area. A limestone quarry is located just upstream of SR 33. Attempts to fill the sinkhole and halt settlement of the pier were unsuccessful, and necessitated closing and eventual replacement of both the northbound and southbound structures. The geology includes the Epler and Jacksonburg Formations, which are intensely folded and faulted with numerous joints and fractures. Weathering of these deformed units produce deeply developed karst topography. Subsurface exploration for the new structure included a combination of borings and geophysical methods. Geophysical methods included resistivity and seismic techniques. Rock encountered consisted of limestone and dolomite. Rock was highly voided with soil filled seams, and generally low recovery. The replacement bridges consisted of single span prestressed concrete I-beam structures supported on micro-pile foundations. The single span option had the advantage of construction time and cost savings, and the benefit of no substructure units down in or near the creek. Staying out of the creek had positive implications for environmental and stream health concerns, and major safety improvements of minimizing worker exposure to the most unstable areas of the project site, relative to sinkhole activity. Sinkhole development remained high in and around the site, during construction operations.

A massive inflow of water developed within the limestone bedrock of a quarry for a cement producing facility in Cape Girardeau, Missouri. The water inflow increased rapidly with time from an initial flow of approximately 2,500 gpm to a maximum estimated flow of approximately 40,000 gpm. Karstic infill material was removed from the formation and carried with the water into the quarry creating sinkhole concerns and potential damage to structures. The loss of sediment needed to be stopped in order to reduce the potential for sinkhole development. Therefore, the water needed to be stopped. The use of a deep grout curtain was recommended to intercept the deep voids/fractures that were present and to reduce the water inflow to the quarry. The location of the curtain was ultimately based on the results of the anomalies identified by geophysical testing, and the location of existing sinkholes, quarry seepage, the projection of a fracture within the quarry and confirmation of karstic features during exploratory drilling. During drilling, two major void zones were discovered in the area where the grout curtain was ultimately constructed. These void zones generally extended from 230 to 275 feet below the surface and 300 to 325 feet below the surface. The majority of the void zones were grouted successfully with conventional cement based grouts. Both high mobility grouts (HMG) and low mobility grouts (LMG) were used. However, the water flow through one area located within an approximately 20 foot wide portion of the upper void zone of the grout curtain, referred to as the "hot zone" could not be stopped using conventional cement grouts. Dye traces conducted during the drilling and grouting confirmed the direct connection between the hot zone and the water outflow into the quarry. The use of specialized hot bitumen grouting in conjunction with HMG and LMG was performed to plug the hot zone. The grouting performed ultimately stopped the previously discussed flow of water and therefore the sediment being removed from the formation and deposited into the quarry. To date the grout curtain is performing well, and based on observations of the quarry and piezometers, no indications of a deterioration of the hydraulic effectiveness of the curtain since its implementation have been observed.

Three sinkhole case studies are discussed to show the precarious nature of determining if a sinkhole is present based on commonly used investigative techniques in west-central Florida. Conclusions reached in sinkhole investigations are generally based on site conditions determined from study of essentially: (1) site geologic history, (2) GPR (Ground Penetrating Radar) survey, (3) SPT (Standard Penetration Test) borings, (4) an evaluation of building distress and (5) shallow test pits excavated around the building. Three case studies are presented that show how sensitive the data obtained are to the location of the tests and the interpretation of the data. In most studies different investigators have reached different conclusions at the same site.

The new Sanita factory is a 2 story factory warehouse covering an area of 10,000 m² located in Zouk-Mosbeh, Lebanon. The warehouse is composed of 140 reinforced concrete columns supporting a one way pre-stressed concrete slab designed to carry 25 kN/m² of industrial live load plus a future steel structure. The soil substratum consists of 1 to 4 m of silt on top of the bedrock. The bedrock is hard karstic limestone with cavities and fissures. Because of the karstic nature of the rock, the original design called for a strip foundation on top of the rock with a maximum allowable bearing pressure of 400 kPa. Combined Micropiles and Footing Foundation, CMFF, is a new foundation system that was created by the International Institute for Geotechnics and Materials, iIGM, and implemented as an alternative solution for this particular project. This new system was designed utilizing a finite element model. There were two major concerns regarding this solution. The first concern was with respect to the load-settlement performance of the micropile in the karstic terrain. The second concern was to prove the advantages of the CMFF system considering the particular conditions of the project. Two static load tests were performed on the micropiles with different methods. The results of the tests permitted modeling the soil/structure interactive foundation system. CMFF in karst offered many technical advantages over classical methods. A cost saving of about 40% was also accomplished in the foundation system without compromising safety. This paper summarizes the results of the load tests and the advantages of CMFF in this particular project. The paper also presents the different phases of the foundation system by covering the design details, construction aspects, performance evaluation, and cost analysis.

Understanding the engineering behavior of karst is different from the theory and practice that applies to residual soils or weathered rock formations. Weathered limestone in karst terrane does not behave similar to weathered granite, for example. In the last twenty years foundation problems in karst have kept pace with the rate of development in the Southwest Florida region. Recent problems with the settlement of piers on the Cross-Town Expressway in Tampa have highlighted the need for better subsurface exploration and design methods for foundations in karst terrane. The cost of retrofitting the caissons on this project has added about 30% to the original cost of the project. This experience has proved that in-situ geotechnical testing methods do not adequately assess the geological conditions and in-situ geophysical tests do not adequately define the engineering properties. Foundation design for light structures in karst varies dramatically from those for heavy structures. The differences between the elements of designing foundations for light and heavy structures are not a matter of scale but of understanding the engineering properties of karst. This paper illustrates with examples the essential elements of estimating engineering behavior for large infrastructure projects and heavily loaded structures. At the least, all of the following elements appear essential for heavily loaded structures: a geophysical survey; in-situ geotechnical sampling and strength tests; proper classification of the soil samples, limestone and the weathered limestone; identification of regional hydrogeological and geological features to evaluate the potential for the development of dolines; a reliable method of estimating foundation behavior with specific reference to the location of the foundations; reliable means of evaluating the design of the foundations, and monitoring the foundation behavior. In contrast, the design of lightly loaded structures in karst requires two major components—settlement control and the potential for the development of dolines. Foundation design essentially requires a proper evaluation of the soils and weathered limestone formations. This paper defends the use of the Standard Penetration Test and Cone Penetrometer tests as being adequate for estimating the engineering properties of light structures and recommends the use of Electrical Resistivity or Multiple Analysis of Surface Waves as the most appropriate tests for evaluating the behavior of karst. The paper concludes that expensive errors could be avoided by establishing minimum standards of tests for large projects. Such a protocols do not exist in the private or public sector projects. The author proposes adopting the minimum standards for specific projects such as highways, bridges and tall buildings.

Adapted for use as a mechanical soil exploration device in the 1930's in Holland, the cone penetrometer has evolved into an electric instrument capable of collecting essentially continuous data in the soil profile in the forms of the tip resistance, sleeve friction and pore water pressure. It provides an essentially continuous profile of information and is capable of identifying abrupt changes in soil stiffness and strength. The production of a typical cone rig makes the level of information comparable to that obtained by geophysical methods, but has the advantages of direct exploration. Although it has advantages over the widely accepted standard penetration test and geophysical methods, the cone penetration test has not been widely used for geotechnical characterization in clay-mantled karst. This paper discusses the relative advantages and disadvantages in the application of the cone penetration test to geotechnical site characterization in karst. It presents cone penetration test data collected at two project sites in deeply weathered, clay-mantled karst of east Tennessee and examines the karst conditions as interpreted from the cone penetration data and the anticipated impact of those conditions on project design and construction.

Compaction grouting is the placement of very stiff, low-mobility grout under relatively high pressures in strategic locations so as to consolidate soils and fill voids. Normally a hole is predrilled and grout is either pumped top down or bottom up through a tight-fitting steel casing. The Valley Creek Sewer Relief Tunnel in Bessemer, AL was to be constructed with a tunnel boring machine (TBM) designed for rock, but poor ground conditions consisting of karst limestone formations, which produced tremendous amounts of loose soils, mud, boulders and water prevented adequate advancement. Compaction grouting was employed to improve poor soils along the tunnel alignment ahead of boring operations for about 3,500 feet. The treatment zone was designed to improve a soil block in excess of 20 feet wide from a depth of 30 feet to 65 feet below grade. Grouting was also performed around drop shafts to prevent water inflow. This paper will discuss the design and implementation of a full-scale production-grouting program that began as a test-grouting program. A program, which spanned three construction seasons and saw the placement of approximately 35,000 yards of compaction grout in about 1,000-drilled holes. As a caveat to the grouting program, the tunnel came in close proximity to several sensitive structures, which required underpinning with micropiles and/or conventional shoring. These contractor-designed solutions will also be examined.

The proposed widening of six-mile long section of a major highway in southeastern Pennsylvania involves up to 24 feet deep cuts in karst geology. About $25 million dollars were spent in repairing sinkholes during the widening and interchange improvement of an adjacent section of the same highway during 2000–01. This paper discusses the subsurface investigation and design approaches taken for proposed widening in the cut areas.

The City of Johnson City constructed a new raw water intake structure on the banks of Watauga River near Johnson City, TN. The structure consists of an above-grade electrical room and pump house connected to a T-shaped clearwell that extends out into the river approximately 15 ft. below normal water level. The site is in the Valley and Ridge Physiographic Province of Northeast Tennessee and is in a region of valleys and hills formed on the less-resistant limestones, dolomites and shales. The bedrock at the site is a karstic limestone rock from the Knox Group. The original foundation design for the intake structure was a structural mat with wall footings on bedrock. Uplift resistance was to be provided by a series of rock anchors. However, when the Contractor mobilized to the site and drilled a series of probe holes and excavated test pits around the perimeter of the clearwell to determine the top of rock, they did not indicate the presence of any competent rock. Instead, the Contractor extracted numerous large boulders of limestone rock found within a matrix of residual soil. The boulders were highly irregular in shape and exhibited signs of weathering on all sides. The Owner then contacted Mueser, Rutledge Consulting Engineers (MRCE) to evaluate the conditions and to recommend a foundation system suitable for the difficult site. Additional borings were made at the site and MRCE ultimately recommended using mini-piles to support the new structure, a slurry wall cofferdam around the perimeter to support the excavation and to limit lateral seepage, and an extensive grouting program to help limit the under-seepage into the excavation. The work was subsequently performed by a specialty sub-contractor, and MRCE provided oversight during construction. This paper describes MRCE's investigations, findings and construction methodologies that led to the successful completion of the project.

Recent hydrogeological work, including dye tracing, provides a basis for revising the conceptual site model for groundwater and dense nonaqueous-phase liquid (DNAPL) movement at the site of a historical DNAPL release. This recent work also illustrates some of the problems associated with applying traditional, well-based characterization approaches to karst aquifers. Groundwater and DNAPL movement have been studied at this site for more than 20 years. As the scope of the investigations increased, the conceptual site model for fluid movement through the rock has been periodically updated. Relevant findings based on the latest round of fieldwork include the following: (1) The absence of nearby topographic karst features and large solution voids in boreholes should not be construed to mean that none exist or that conduit flow is not relevant. (2) Pumping tests that produce low yields, homogeneous zones of influence, and drawdown data that fit analytical diffuse-aquifer solutions should not be used to assume diffuse-flow dominance. (3) At the typical site scale, groundwater flow directions and rates estimated using hydraulic head data from monitoring wells can be misleading. (4) DNAPL-migration pathways and the extent of DNAPL in the karst aquifer beneath the site cannot be mapped with the accuracy routinely required by regulators at non-karst sites. (5) Tracer investigations reliably indicate where the groundwater is going, how fast, and in what type of aquifer permeability.

The Shaanbei Energy Base is in northern Shaanxi Province, China. It encompasses the Maowusu Desert and the Shaanbei Loess Plateau. The base is rich in coal and other minerals such as petroleum, natural gas and salt, all of which favor economic development. Water resources are relatively scarce because the area is in arid and semi-arid climatic zones. The limited availability of water constrains economic development. In order to overcome the water-shortage problem, a hydrogeological investigation was conducted. A karst groundwater system was discovered under the eastern side of the base. The water yield from this aquifer was estimated to be approximately 12–17m³/s, and the water quality meets the national drinking water standards. However, whether this valuable water resource should be used for economic development is currently a question in debate. Compared with other water sources in the area, such as mine drainage and surface water in the Yellow River, the karst groundwater has distinct advantages in both quantity and quality. Based on analysis of the hydrogeological conditions of the karst groundwater system, the authors proposed an optimal utilization scheme in which the karst water is used to serve the local economy while the karst system is managed to prevent any adverse impacts from occurring.

In many alpine karst regions, bacteriological pollution resulting from agriculture is the most important problem in groundwater protection. Colloidal tracers like fluorescent polystyrene microspheres are used to evaluate the hygienic risk for groundwater resources. Different sizes of colloids may be used to simulate the fate and transport of different types of micro-organisms. For the field tracer tests, three different types of karst systems within the alpine fold belt have been selected: karst aquifers in carbonate conglomerates and limestones in southern Germany, and unsaturated fissured carbonates in Slovenia. Large parts of Tertiary gravel fans in the foreland of European alpine orogens are made of coarse-grained conglomerates, which mainly consist of carbonate components in a carbonatic matrix. These rocks are predisposed to karstification and show a large variety of karst phenomena like dolines, karren, karst springs and estavelles. Tracer tests with fluorescent dyes and particles show high flow velocities and travel distances of up to 8 km. The flow orientation reflects the geological structure. This demonstrates the need for special groundwater protection schemes that take into consideration the heterogeneity and anisotropy of this aquifer type. The Helvetic nappes form parts of he Northern Alps and mainly consist of Jurassic and Cretaceous carbonate rocks. During low-flow conditions, a comparative tracer test with fluorescent-dyes and microspheres was carried out in the Hölloch, Germany's longest cave (9,343 m). The Association of Tracer Hydrologists (ATH) conducted a comparative tracer test in the Slovenian Dinarids in order to evaluate the behaviour of different tracers in the unsaturated zone of the fissured karst aquifer in Sinji Vrh. The behaviour of the tracers shows important differences, especially for solute and particle tracers. Therefore, the authors strongly recommend using particle tracers for the evaluation of hygienic risks to groundwater.

A GIS map showing the approximate ground-water flow routes within the karst aquifer in the vicinity of Bowling Green, Kentucky was prepared based upon dye tracing, cave mapping, cave locating by microgravity and electrical resistivity, and water table measurements. The research was performed by the faculty, professional staff and students of the Center for Cave and Karst Studies from 1976 to the present and was directed by Nicholas Crawford. The numerous dye traces are shown as red lines with arrows on the map. Water table elevation contours are shown as blue lines. They were estimated from measurements taken at water wells and storm water drainage wells that are only cased to bedrock and then extend as open holes below the water table, cave streams, springs, and perennial surface streams. The approximate ground-water basin catchment boundaries, based upon dye traces and water table elevations, are shown as dashed green lines. The map identifies the approximate ground-water flow routes to springs and the approximate ground-water basin catchment areas for: (1) Lost River Rise, (2) Graham Springs, (3) Hobson Grove Springs, (4) Double Springs, (5) Mt. Ayr Blue Hole and Bluff Springs, and (6) Harris Spring. Included on the map are dye traces and water table contours northeast of Barren River performed by James Quinlan and Joe Ray (1981, revised 1989). This compilation of karst ground-water data into one map was prepared to assist the City of Bowling Green in meeting the US Environmental Protection Agency Phase II Storm Water Management Plan requirements and to assist Bowling Green with emergency response to any future spills of hazardous liquids.

Fillmore County has been the focal point of dye tracing efforts in Minnesota for several decades. In 1995, a 1:100,000, county-scale springshed map was published. At the county scale, we primarily delineated springsheds greater than 500 hectares. After that project was completed, dye-tracing efforts continued and focused on the South Branch Root River (SBRR) springshed in the western part of the county. At the scale of the SBRR springshed (about 20% of the county), we expanded the boundaries of the large springsheds but also located and refined the boundaries of springsheds that are less than 200 hectares. We found one previously unmapped small springshed (Meyers springshed), documented complex flow boundaries between a large springshed (11. Canfield) and an adjacent small springshed (26. Rainy), and expanded the boundaries of four other springsheds. The small springsheds present a new opportunity for karst watershed research, monitoring and management. The small springsheds are often dominated by one or a few land uses, which permit simpler, more evident identification of the causes of water quality degradation.

The twenty largest perennial springs of Kentucky were identified and ranked over a ten-year period. Since most large springs are not shown on topographic or geologic maps, ranked springs were primarily located from previous hydrogeologic surveys, field reconnaissance, and literature review. Spring flows were ranked by minimum annual discharge, which ranged from 0.15–0.68 m³/s. These springs are classified as 3rd Magnitude, based on the Meinzer (1923) discharge scale. Unit base flow (the ratio of minimum discharge to basin area) revealed diverse hydrogeologic yield of the karst spring group, ranging from 0.22–12.27 L/s/km², suggesting significant unattributed losses and gains. Most large springs are derived from classic fluviokarst basins draining well-developed karst of Mississippian-age limestones. However, one-third result from short stream or meander cutoffs of less than 5 km, which are not necessarily related to well-developed karst terrane. An index of karst basin development can be obtained by the ratio of subsurface flow length to total basin length.

The "Alta Murgia" region (Apulia, Southern Italy) is a karstic plateau, which constitutes the main recharge area of the Murgia karstic aquifer. The natural morphology of the plateau is defined by widespread surface karst features: karst, karren fields, layers of outcropping rock with discontinuous and thin soils. In general, these soils, relatively thin with sparse vegetation, are primarily used as pastures. Recently, agricultural activities have caused dramatic changes. In particular, ploughing and crushing of the karst surface has led to a conversion of the landscape to fields growing wheat and other crops. Stone clearing practice are carried out by means of the repeated grinding of the carbonate rocks with modern technologies, intensive use of bulldozers and machinery, besides destroying and removing rocks from the fields. Practically, the natural karstic soils were replaced with a layer of homogenous soil, useful for cultivation. This cultivated soil has increased the evapotranspiration and significantly reduced the infiltration thus affecting the water balance and the aquifer recharge. In the present study we analyze the impact of changes in soil texture and crop growth on the water balance in the "Alta Murgia" area by analyzing the features of the newly created soil. The characterization of the modified soil involved both field and laboratory measurements. The HYDRUS 2D model was used to simulate the infiltration process in the unsaturated zone. The objective of this paper is to elucidate the effect of stone removal by presenting a water balance for new soils. The simulations demonstrate that the available infiltration rate at the top layer varies from 60% of the overall amount of rainfall for the bare soil to only 25% in the presence of vegetation. Knowledge of the interactions between vegetation, soil and climate is essential to quantify the vadose zone hydrology and, hence the hydrogeology of the underlying karst system.

Measurements of tritium concentration in spring water samples from the lower Pecos valley are used to estimate groundwater residence time in the Roswell Artesian Basin, southeast New Mexico. Natural groundwater discharge occurs from a series of springs, seeps and sinkhole lakes at Bitter Lakes National Wildlife Refuge, located on the Pecos River floodplain east of Roswell, NM. The springs and sinkholes are formed in gypsum bedrock that serves as a leaky confining unit for an artesian aquifer in the underlying San Andres Limestone, and are fed by upward seepage of groundwater from the aquifer. Because wetlands on the Refuge provide habitat for a number of threatened and endangered species, Refuge managers have expressed concern about the potential for contamination by oil and gas drilling and other anthropogenic activity in the aquifer recharge area. Estimates of the time required for groundwater to travel through the artesian aquifer are poorly constrained, ranging from as little as four to greater than 100 years, mainly because of uncertainties regarding the role that karst conduit flow plays in subsurface transport. A better understanding of groundwater residence time is thus needed to make informed decisions about management of water resources at Bitter Lakes. Preliminary results of tritium measurements suggest that groundwater flow rates through the aquifer may be slower than previous workers had estimated; and that water discharging into some of the sinkhole lakes is a mixture of recent and pre-modern groundwater.

Fluorescent dyes have become an integral part of the study and management of ground water in karst environments. Researchers have striven to reduce detection limits and analyze multiple dyes in a single sample while minimizing dye concentrations for environmental, aesthetic and health reasons. The unambiguous separation from background, identification and quantification of fluorescent tracer dyes has increasingly taken on legal implications. Synchronous fluorescence spectroscopy and curve fitting software represent a major advances in the quantitative analysis of low levels of tracer dyes against naturally occurring background fluorescence. Determination of levels of detection (LOD) and levels of quantification (LOQ) are an important part of dye trace design and implementation. Factors that impact LOD and LOQ include levels of natural fluorescent compounds, absolute fluorescence of the specific dyes, the presence of multiple dyes with overlapping peaks and instrumental noise. Characterization of the spectral shapes and concentration dependences of the natural fluorescence background and applied tracer dyes are importanto the determination of a positive dye trace result. Rather than representing noise, the natural fluorophores contain information about the flow environment. Spectral deconvolution with curve fitting software is an important tool in the karst researcher's toolbox.

An investigation of the sandstone karst at Askov, Pine County, Minnesota was conducted utilizing fluorescent dye tracing techniques. Connections were documented between the effluent from WWTF lagoons discharged into a sinking stream and local residential water supply wells southwest of the stream sink, over kilometer-scale distances. The apparent velocity of dye from the steam sink to the residential well was 57 m/day, much faster than would be expected for porous-media flow in a sandstone but slower than expected for well-developed conduit karst in carbonate rock. The traces also document stage-dependent, divergent flow to the northeast and west from recharge to a sinkhole adjacent to the lagoons. A well-developed conduit system extends a few hundred meters under the northern portion of the lagoons with flow velocities greater than 128 m/day to the northeast. A slower flow system extends for at least a kilometer to the west and northwest with a range of apparent flow velocities that average about 9 m/day. The slower flow system reached a broad "fan" of residential wells to the west but was not detected in other wells in the same direction.

This paper presents a comprehensive review of commercially available multi-level well completion technologies to simplify the selection of a reliable and cost effective, multi-level completion method for specific karst sites. Karst formations present substantial monitoring well completion challenges especially when a single core hole spans several distinct zones of secondary porosity including micro-scale fracture networks and macro-scale voids, cavities, and caves. Designing a cost effective well completion strategy that isolates and reliably samples these features is often an important element of environmental investigations of contaminated karst. A number of technologies are commercially available that permit the installation of multiple sampling intervals within a single borehole. These technologies were initially designed for application in solid rock or stable boreholes; however, they are also applicable for karst formations. During an environmental investigation of groundwater impacted with chlorinated solvents in the Knox Group dolomite of the Valley and Ridge Province of Eastern Tennessee, several core holes were advanced to depths ranging from 66 to 176 feet (20.1 to 53.6 meters) below ground surface. The identification of numerous distinct voids and fractured zones, which intersected groundwater impacted with chlorinated solvents, required a vertically discrete monitoring network; however, the cost and complexity of core hole advancement prohibited drilling of separate core holes for each monitoring point. Hence, this review. Technologies considered included the Westbay, Solinst Waterloo multi-level, BarCad III, Solinst Continuous Multi-channel Tubing (CMT), and Water FLUTe systems as well as the more traditional nested small diameter wells. Each technology was evaluated against a number of criteria including the ability to isolate multiple sampling zones, ease of installation, requirements for sampling, sample quality, durability/longevity, material and installation costs, and sampling cost and complexity. The water FLUTe system provided the best combination of features to address the specific conditions encountered in the karst dolomite at the site.

Many of the flooding problems in Bowling Green, Kentucky are associated with sinkholes. Many existing sinkhole flood zones are approximate and do not conform to surface topography. An effort to update flood maps and calculate flood elevations for four large sinkholes was begun to assist the City of Bowling Green. Problems with the mapping were encountered for two of the sinkholes that lie along Nashville Road. For these two sinkholes, flood elevations based on topographically delineated watersheds were far below observed elevations. Sources of flood elevations higher than predicted could arise from two sources, (1) flooding and spillover from uphill sinkholes, and (2) back flooding from the subterranean Lost River. This project addressed the first of these two causes. The larger of the two sinkholes, Nashville Road North had direct runoff from 11 hectares, but when spillover was considered, the drainage area grew to more than 520 hectares. GIS was used to determine volumes of sinkholes at the elevation of spill points. It was found that the volume held had little correlation with the catchment draining directly to the sinkhole. The EPA Storm Water Management Model (SWMM 5.0) was used to calculate runoff and hydraulics for this complex of sinkholes. To calibrate the model with a large storm that occurred in April 1998, it was necessary to increase the losses from the runoff to a level significantly higher than recommended. This is justifiable for Bowling Green because of numerous stormwater injection wells that receive part of the runoff.

This paper applies environmental isotope techniques to interpret recharge mechanisms and estimate transit time of a covered karst aquifer in semiarid NW China. The aquifer is of major importance to the water supply of the region, in particular in meeting the increasing demand on water resources and in sustainability assessment in the future. Deuterium, oxygen-18 and tritium of 459 samples were analyzed from precipitation, surface water and groundwater during one hydrological year. Based on the isotope analysis, the Ordovician carbonate recharge to the aquifer is mainly from direct infiltration of atmospheric water in carbonate outcrops in Sandao and the Dacha valley. Stable isotopes show that karst groundwater in the Dacha valley seems to have recharged in a catchment area with a mean elevation of 1685–1854 m. The distinct independent isotope composition of water in the limestone aquifer in the Dacha valley from that in Anguo suggests no interconnection between them. However, the temporal evolution of the stable isotopes concentrations, the tritium contents and the hydrochemistry show the existence of an interconnection between Dacha and Sando Ordovician limestone aquifers. The isotope compositions also show that the groundwater in the Ordovician limestone is a mixture of old groundwater with modern recharge from local infiltration. The maximum percentage of modern water is 71%, which was usually found at the intensively fractured areas. Adopting a model with exponent and piston time distribution, the mean turnover time of groundwater in the Ordovician carbonate rocks was evaluated to be 36 years. The size of the groundwater reservoir is estimated to be 0.134 billion m³ and the mean storage coefficient is 0.0073.

Jim Woodruff Dam was constructed across the Apalachicola River on the Florida-Georgia border to create Lake Seminole. Polk Lake Spring is located about 244 m (800 ft) downstream, and a large boil, created by water rising from a hole in the river bed, is located in the Apalachicola River about 549 m (1,800 ft) downstream. This research was initiated by the Mobile District U.S. Army Corps of Engineers to determine if water from the spring and/or boil is leakage from the reservoir, groundwater, or both. Five potential leakage sites were identified on the bed of Lake Seminole. The most distant hole identified by a lake bed survey was approximately 434 m (1,425 ft) upstream from the dam. A dye tracer investigation was conducted to determine if leakage from these sites was contributing to the discharge at the spring and/or boil located downstream from the dam. Three fluorescent dyes were used for the tracer tests. The investigation indicates that lake water sinking into holes in the lakebed is flowing through old solution conduits in the limestone bedrock that existed previous to dam construction. All the leakage confirmed by dye tracing is resurging at the boil located in the river downstream from the dam. However, some of the leakage first resurges at Polk Lake Spring and then sinks at Polk Lake Sink to finally resurge at the boil. The dam was built upon the lower Tampa Limestone (Chattahoochee), a semi-confining unit for the Upper Floridan Aquifer. The geologic dip in the vicinity is downstream perpendicular to the dam. The Suwannee Limestone, the uppermost geologic unit of the Upper Floridan Aquifer, outcrops under Lake Seminole but dips below the Tampa semiconfining unit in the vicinity of the dam. The actual route taken by the sinking lake water to the boil is unknown. However, two hypotheses are presented. First hypothesis: leakage is occurring through reactivated sinkholes in the lake where the Tampa semiconfining layer is thin. This leakage is then flowing through the Suwannee Limestone (Upper Floridan Aquifer) down dip to resurge as an artesian spring at the boil located about 549 m (1,800 ft) downstream from the dam. Second hypothesis: leakage is flowing through the Tampa Limestone directly to Polk Lake Spring and the boil.

The appropriate approach to dye tracing in karst areas depends upon the objective and context of the trace. Dye tracing in karst areas is undertaken to address geographical, hydrogeological and contaminant problems at particular spatial and temporal resolution in the context of prior knowledge, available resources, social and legal expectations and environmental constraints. The value of a trace is improved if the objective can be formalized into a rational hypothesis and where the signal is demonstrably distinct from error. This requires sampling and analysis as much to define error as to detect the signal. The tolerable error depends on the dye trace objective and context, and scales with the sophistication of both, becoming increasingly critical and challenging as higher level interpretations are made. The appropriate technique for a particular trace therefore depends not only on the problem and context, but also upon the necessity of defining and correcting errors. Simpler problems such as establishment of underground connections can often be usefully tackled with simple techniques. Variable background fluorescence is a particularly difficult systematic error in dye tracing that can be reduced by supplementary sampling and control. This approach is illustrated for fluorescence spectra and in situ filter fluorometry. To extract a signal from spectra a statistical correction has been developed allowing compositional and concentration corrections to highlight anomalous samples. Supplementary sampling is required to provide the background statistics necessary for such an approach. The strong spectral coherence of background allows concurrent green fluorescence measurements to define variable background fluorescence during a red dye trace. The relationship between red and green fluorescence in un-dyed samples can be used to model background behavior in the presence of the red dye.

When attempting to regulate land use to protect groundwater quality and to prevent karst hazards, regulatory approaches most often form the focus of the efforts. For example, a zoning overlay for water supply protection may serve as the cornerstone for local source water protection programs. However, zoning and local ordinances fail to address many of the indirect effects of urbanization in karst regions. State statutes, limits on local authority and local political considerations often constrain the effectiveness of these regulatory approaches. Public outreach, voluntary incentives and non-regulatory approaches promise more effective control of land use and deleterious effects of uncontrolled growth in sensitive karst areas. For example, both conservation easements and nonpoint source control/agricultural assistance programs generally center on the improvement of riparian vegetation and open space protection. However, these tools prove just as effective in helping achieve the goals of source water protection and hazard mitigation in karst areas. Reorienting emergency response and training exercises towards these goals require only minimal adjustments to existing programs and promise cost effective results. This paper reviews proven non-regulatory tools including conservation easements, purchase of development rights programs, transfer of development rights programs, low impact development and conservation subdivision design. The authors then show examples of how some of these tools may be used to achieve source water protection and karst hazard mitigation goals. Finally, the paper suggests ways that local governments may achieve the level of communication and cooperation at the local and regional levels needed to successfully implement these non-regulatory tools. The authors conclude that in many settings these non-regulatory tools promise better results for less monetary and political cost than the more often used regulatory tools.

Several dramatic and costly sinkhole occurrences within a few years prompted a review of sinkhole policy in Pennsylvania. A committee was created to determine how the state Department of Environmental Protection (PADEP) addresses sinkhole issues throughout all its program areas in order to formulate a comprehensive policy regarding sinkhole prevention and response. Pennsylvania does not have a formal policy or procedure to address sinkhole problems of any size even if they constitute a public emergency. From 1997 to 2005, the Bureau of District Mining Operations participated in investigations of major sinkholes occurrences relating to quarry operations. This resulted in the preparation of basic repair plans and alerted the Department, in general, to the growing problem of sinkholes associated with water withdrawal, urbanization and storm water handling. Reporting sinkholes, authorizing repairs, providing information or assistance to the public and private entities, and requirements for permit applications were not consistent across program areas. The committee was able to identify many items that could improve response to sinkhole problems and even address the potential for sinkhole collapse related to permit authorizations in order to prevent costly repairs in the future. Many steps were recommended that could be implemented gradually over all applicable programs. Identified action items included education of staff, determination of best management practices, and increased interagency coordination. Additionally, it was determined there is a need to provide accurate information on sinkhole prevention and remediation to the general public as well as to municipal governments and those involved with geotechnical, residential, and commercial activities. Finally, a response plan must be formulated and funding must be identified in cases of sinkhole situations that pose an imminent threat to health and safety. This approach can serve as a model for other state government agencies to improve their efforts to effectively deal with sinkhole situations and better serve public needs.

Most often in the United States, regulation of land use falls to local governments. Similarly, local governments form the local focus of efforts to impose regulations to minimize karst hazards. In addition, the federal government and the states impose very little regulation on karst landscapes. However, the structure of the United States' federalist form of government often limits the authority of local governments to carry out this function. The powers of the federal government, as well as those of the states, are laid out in the United States Constitution. The Tenth Amendment of the United States Constitution provides that, "[t]he powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people." This provision at once greatly limits the powers of the federal government and places a great deal of power with the states. In addition to the federal and state governments, nearly 40,000 local governments exist in the U.S. However, the United States Constitution makes no mention of the power, if any, that should be given to these local governments. Consequently, settled law provides that local governments are mere creatures of, and under the total control of, state legislatures and state constitutions. Local governments must derive power from state constitutions, charters or enabling legislation passed by state legislative bodies. In this context, a struggle arose during the middle-1800's as to whether local governments should be vested with inherent broad powers or should be limited in authority. Judge John Dillon, an Iowa state court judge and municipal law expert, proved to be the most influential force in deciding this issues. Judge Dillon derived Dillon's Rule, which states that local governments only possess those powers expressly granted or inherently necessary to the local government. If any reasonable doubt exists, the courts resolve the doubt against the local government. The resulting constraints prove to drastically impact the scope of local government regulation in karst terrain. This paper reviews the use of Dillon's Rule across the country, finding that 39 states impose the rule, at least with respect to some local governments. The author then outlines the impact of limited local government on the regulation of karst hazards. For example, the Standard Zoning Enabling Act arguably fails to empower local governments to regulate land uses to mitigate and minimize karst hazards, include groundwater contamination. The paper concludes that many local governments lack authority to adequately regulate karst areas to protect water quality and minimize hazards. States must examine and expand existing enabling authority to authorize effective tools to manage these hazards in karst terrains.

Synthetic Aperture Radar (SAR) interferometry, a newly progressed technique capable of detecting very small ground surface elevation changes, was proposed to detect and monitor sinkholes subsidence susceptibility over large karstic areas. This technique, which was proposed by researchers in the area of geophysical studies in 1989, uses (SAR) images to measure very small (1cm or less) surface motion over large areas with good spatial resolution (tens of meters). The advanced method has been used in the studies of earthquakes and other natural phenomenon and has shown promising results. The effectiveness of the remotely sensed technique in determining the damage caused by sinkholes is assessed by using a simplified model approximated by the normal probability (Gaussian) distribution theory proposed by researchers in 1969 to study the mechanism of the resulting ground surface subsidence, estimate the order of magnitude of such settlement and finally determine whether these settlements are sufficiently large to be detected by InSAR. A specific purpose database, including both collapse and subsidence sinkholes, was developed to store and organize all the important information relative to different case histories. This information was used to further investigate the relevance of SAR interferometry in studying the phenomenon. The relevance of the proposed technique was further investigated by utilizing three European Remote Sensing (ERS) Synthetic Aperture Radar (SAR) images encompassing an almost 5-year period of observation (1992–1997), three successful interferograms with short perpendicular baseline (>100 meters) from a descending orbital trajectory have been produced. Analysis of these interferograms indicate a relative regional subsidence of up to 0.5 cm/year within the Devil's Throat Sinkhole area, located in the Gold Butte district of Clark County in the Southeast corner of Nevada. Observations of deformation along profiles constructed on the produced interferograms also show a steep gradient in displacement rates occurring very close to the area where the Devil's Throat Sinkhole exists. The demonstration of the technique in the Devil's Throat area, applied using data from the European Space Agency's ERS systems, has shown promising and unique results. These results demonstrate the effectiveness of the technique in studying small-scale ground deformations over large karstic areas.

Sinkholes are commonly utilized for agriculture, and are thus subject to application of fertilizers, including animal manures and biosolids. Because sinkholes are hydrologically connected to underlying aquifers, fertilizer application within sinkholes may adversely impact groundwater quality. Few scientific data are available to support effective management of fertilizer application that will minimize karst groundwater contamination. Our research aims to characterize the hydrogeology and recharge mechanisms in agricultural sinkholes. Initial work includes physical characterization of several sinkholes at the Virginia Tech Kentland Experimental Farm, Whitethorne, VA. Nearly 100 2-D Electrical Resistivity Tomography (ERT) profiles, topographic mapping, and physical observations were used to generate 3-D sinkhole models of the shallow subsurface. Preliminary results generally show a clear resolution of changes in soil composition, and in some cases, the bedrock-soil interface. We conducted an experiment utilizing 2-D Differential ERT to monitor vadose water movement and to delineate preferential flowpaths during a recharge event. Repeated measurements in the same location before, during, and after a large storm, show significant resistivity differences interpreted to be related to changes in soil moisture due to infiltration. Similar experiments using 3-D Differential ERT are underway that will attempt to model infiltration and delineate preferential flowpaths over larger areas. 3-D Differential ERT results will be compared with simultaneous soil moisture measurements collected using down-hole Time Domain Reflectometry (TDR). If modeled ERT values can be correlated with the TDR soil moisture values, this technique has the potential to provide high resolution 3-D soil moisture data that can be used to greatly improve the visualization of vadose flow in shallow sinkholes.

Sinkholes are one of the well known geological hazards in the state of Florida. In Alachua County Florida, sinkholes have become problematic in highway retention ponds, especially along State Road 26 west of I-75. As part of a Florida Department of Transportation effort to minimize the effects of sinkholes in retention ponds, multiple geophysical surveys and closely spaced boring were used to characterize a retention pond that had experienced previous sinkhole activity. Within the approximately 0.6 hectare retention pond, a survey grid consisting of lines and stations, both with 1.52 meter spacing was established. One of the survey lines was aligned with 12 geotechnical borings that were previously conducted to establish the site stratigraphy. The borings were placed approximately 3 to 7.6 meters apart. The depth of the borings ranged between approximately 9 to 12 meters. Five different geophysical techniques, ground penetrating radar (GPR), electrical resistivity (ER), gravity survey, multielectrode resistivity (MER), and capacitively coupled resistivity (CCR) were used to characterize the subsurface stratigraphy in the vicinity of the borings. As expected in a karst region, the stratigraphy determined from the closely spaced borings was very erratic; the limestone bedrock surface varied between depths ranging from the ground surface to greater than 9 meters, within a lateral distance of 15 meters. Open voids within the limestone, one over 6 meters in length, were also detected in the borings. The results of the geophysical surveys compared favorably with the stratigraphy determined through borings, and multiple geophysical techniques were able to detect anomalies in the same vicinity at the same depths. Based on the work conducted at this site, both traditional intrusive exploration techniques and geophysical surveys were required to fully characterize the subsurface. Two recommendations for retention pond site characterization were derived from this work: geotechnical borings must be conducted to determine depth of bedrock, and at least two geophysical methods measuring different subsurface properties and have approximately the same depth of investigation with the depth of investigation being at minimum the soil/bedrock interface.

Even apparently simple problems in karst terrain can require a great deal more effort to identify correctly, than can be at first realized. As an example, a private property owner in Bexar County, Texas noticed a suspicious depression next to his driveway. This was eventually followed by the formation of a small hole in the ground, which he was able to fill temporarily, while a geotechnical investigation was carried out to determine the possibility of a threat to the house structure. Traditional drilling did not encounter any voids in the ground, so the use of non-invasive electrical tomography was then employed. Two-dimensional resistivity lines were conducted across the property. These were able to detect a higher-resistivity zone close to the surface, but only of the order of a few hundred ohm.meters, which would not normally correspond to a true void, and was later shown to correspond to an outcrop of dense limestone above the water table. Since there were prima facie indications of a karstic void, a three-dimensional survey was carried out at the site, in a snake-like pattern across the back yard. Special software was used to integrate the results into a three-dimensional spatial model, which then showed the existence of a very-high resistivity zone (in excess of a thousand ohm.meters) indicative of a true void. This was calculated to be about 4 cubic meters in volume, but sufficiently far below the surface (and in an off-center location) that it was determined not to be a threat to the house foundations. The void was sufficiently localized that the probability of detection by conventional drilling was very small, and even the probability of detection by two-dimensional surveys could be shown to be significantly less than 50%.

The Marshall Space Flight Center is a NASA facility located in a karst prone area of northern Alabama. The geology consists of Mississippian age Tuscumbia Limestone underlying clay residuum and an epikarst zone of variable thickness. Groundwater flow is affected by karst conduits, fractures, and structural trends through the site. In an effort to characterize karst features on a local and regional basis, geophysical measurements were integrated with existing well and boring data along with regional geologic information. Geophysical measurements included microgravity, multichannel analysis of surface waves (MASW), and geophysical logging. The use of multiple geophysical methods provided supporting datasets and overlapping depths of investigation. The measurements were focused on 14 contaminant Source Areas to provide a detailed characterization of the top of rock, the epikarst zone, and anomalous zones within the residuum and upper limestone. The data were also used to map the thickness of the weathered zone in the upper limestone that may contain karst conduits for preferential groundwater flow. Additional reconnaissance measurements were made throughout the site to provide a regional assessment of karst conditions. Models of the geophysical data generally correlate well with each other and with existing boring data. The regional data also show a good correlation with geologic structure maps of the area. The integrated geophysical approach to the karst characterization allowed multiple geologic variables to be assessed simultaneously with a greater level of spatial sampling than borings alone. The results of the investigation will be used to efficiently place new borings and develop a more complete understanding of the karst hydrogeology.

CPS Energy^TM of San Antonio constructed three electric transmission lines (Fair Oaks Ranch Area, Stone Gate-Hill Country and Stone Gate to Green Mountain) in the last three years across the boundaries of the Edwards Limestone Aquifer in San Antonio, Texas. The purpose of geophysical studies (2-D resistivity imaging, ground penetrating radar and natural potential) was to determine voids, caves or any other karstic features beneath the power pole locations. Results of 2-D resistivity imaging, ground penetrating radar (GPR) and natural potential (NP) surveys helped to determine locations and depth of voids in the near surface as deep as 40 feet. In this way, we relocated the pole locations, if necessary, to reduce the potential for the construction to impact threatened or endangered species and to locate more solid limestone to seat the pole footings. One hundred fifty-four (154) pole locations were studied along these three transmission lines. During the drilling of these pole locations, which were based on the geophysical data, three significant (>5 feet in diameter) and two insignificant (<5 feet in diameter) voids were encountered. If a cave or void was encountered, an engineering evaluation was conducted to determine if the pole installation can proceed as planned, or if the pole installation design needs to be changed, or if the hole should be abandoned and a new hole drilled. Thus, the purpose of this paper is to discuss the geophysical data obtained over some of the drilled pole locations with significant voids and why the geophysical surveys did not detect them. In addition, we discuss the geophysical signature of a known karstic feature such as Tick'n Delight Cave in the vicinity of these pole locations.

An electrical geophysical survey, utilizing residual potential mapping (RPM) and high resolution resistivity (HRR), was completed at a TCE contaminated site near San Antonio, TX. RPM is a method based on the "mise-a-la-masse" technique, where the subsurface target is imaged using a direct-current transmission source placed at depth, and receivers placed on a regular grid on the surface. The RPM method records the disturbance in the potential field creating a two-dimensional plan view image of the changes in electrical properties. RPM is sensitive to lateral variations in these properties, making it an excellent tool for mapping facture zones and subsurface voids. Conversely, HRR is a direct-current method conducted strictly on the surface along linear transects. A two-dimensional profile of electrical properties is created. Closely spaced and parallel HRR profiles allow a three-dimensional reconstruction of the subsurface. Several transmitting locations, located between 60 and 150 ft below ground surface, were used for the RPM survey. Receivers were placed on the surface within a regularized grid of 310ft by 300ft. The results of the RPM survey showed distinct areas of high and low RPM response, relative to a modeled background potential that assumes homogeneity. A total of 16 parallel HRR profiles were collected over the same RPM grid with a depth of investigation of approximately 60 ft. The electrical resistivity collected with HRR showed a general pattern of high values near the surface and decreasing values with depth. Background removal was conducted on the profiles to enhance subtle features. The background-removed HRR profiles were used to create a three-dimensional plot showing possible interconnected pathways of karst features.

HydroPhysical^TM logging, along with optical and acoustic televiewer logging, was applied in five open bedrock boreholes at the Volunteer Army Ammunition Plant (VOAAP) in Chattanooga, Tennessee to identify water-bearing fractures and conduits. Data obtained from the HydroPhysical^TM (HpL) and geophysical logging was evaluated in real time in the field so that Solinist Continuous Multi-channel Tubing "CMT" wells with multiple sampling chambers could be properly constructed. The fracture zones were identified to evaluate the distribution of explosive contaminants in the karst aquifer wherein primary flow paths are through solution-enlarged fractures and bedding planes. The CMT wells are designed to provide vertical head distribution within fracture zones and vertical contaminant profiles in support of the groundwater corrective measures study (CMS) and long term monitoring points. The HpL logging techniques identified water-bearing fracture zones in each borehole under ambient and stressed conditions with flow rates ranging from 0.007 to 18.5 gpm. The number of water-bearing fracture zones identified ranged from one to eight per borehole. Additionally, optical or acoustic televiewer data was acquired for greater depth resolution of fractures and fracture orientation information. Based on the HydroPhysical^TM data, CMT wells were installed in each borehole with multiple fracture zones with the well screens targeting specific water-bearing fractures. The depths of the screened intervals for the multi-chamber wells ranged from 43 to 145 feet bgs. Contaminant profiles generated from the multi-chamber wells showed concentrations of explosives varied widely among fractures within each well. The HpL and geophysical logging accurately defined preferential flow zones at lower cost and requiring less time than traditional techniques. Once logged, data were evaluated in the field, which allowed the CMT wells to be designed, constructed, and installed by the geologist with minimal delays.

GPR and SPT data from 93 sites in West-Central Florida was analyzed to evaluate the effectiveness of GPR in determining the presence of sinkhole features. SPT results were used as the criteria to determine if the GPR method was effective. The size of the feature being investigated is considered in determining the combined effectiveness of these methods. The effects of near-surface clay on GPR results are discussed as they impact test results. From these factors a candid discussion of the advantages and disadvantages of these methods in accurately assessing geological conditions as they relate to karst activity in West Central Florida is provided.

Karst conditions in the glaciated terrain of north-central Iowa are often difficult to ascertain during typical site reconnaissance. Surface and downhole geophysical techniques are useful to better characterize karst features where they are hidden by a generally featureless topography. An iterative investigative strategy, involving high-resolution surface geophysical techniques, complemented with downhole geophysical logging and rock coring, proved effective in providing sufficient information to adequately characterize bedrock conditions in this difficult setting. An 800-acre site in north-central Iowa was one of several proposed for a coal-fired electric generating facility. Concerns arose over the possibility of karst conditions at the site that could impact facility design and construction. Although several shallow closed topographic lows, described as sinkholes by soil scientists, were mapped on the property, no subsurface investigations had been conducted in the vicinity to document bedrock quality or karst development. An investigation using high-resolution surface geophysical techniques to focus an exploratory drilling program was proposed. The boring program was coupled with downhole geophysics to document the bedrock quality and establish better stratigraphic control and to determine the structural character of the shallow bedrock. Surface geophysical techniques included micro-topography using a differential global positioning system (DGPS) to produce a digital terrain model (DTM) that would identify and locate topographical expressions to a resolution of less than 0.5 foot. A high-resolution electromagnetic (EM) conductivity survey was conducted simultaneously in an attempt to identify areas of higher moisture content that may indicate areas of incipient karst development. The results of these surveys were used to identify anomalies that were then surveyed using ground penetrating radar (GPR) to provide three-dimensional imaging and identify specific features to drill. Finally, a shallow seismic survey that utilized Multichannel Analysis of Surface Waves (MASW) was conducted after drilling to provide better definition of shallow solutional features and to compare this emerging technology with the other methods. Two candidate geophysical anomalies were core-drilled to depths ranging from 50 to 150 ft with particular attention applied to identification of subsurface voids. Boreholes were then investigated using downhole video, caliper, natural gamma, resistivity, and spontaneous potential geophysical logging. The caliper was used to identify the locations of bedrock cavities. The natural gamma was used to identify shale aquitards that may affect the distribution of solutional activity and also serve as stratigraphic marker horizons. Downhole video provided indications of groundwater pathways and verified fabric geometry of the fractures and solutional features. Results of the investigation suggest that the DTM and EM were effective for rapid identification of bedrock surface anomalies and soil compositional anomalies over a large area that served to focus the more expensive invasive tools. As a result of signal loss from fine-grained and relatively higher moisture content soils, these techniques could not identify the location of deeper bedrock voids. GPR penetration was limited in areas of clayey till. However, the GPR successfully detected soil surface anomalies where subsequent drilling discovered silt-filled grikes. MASW identified the of the bedrock surface, and was successful in verifying a previously identified filled grike. In summary, this study illustrates that coupling area-wide surface geophysics with a limited boring program can adequately define the bedrock character across large areas where topographic clues are lacking.

This paper presents ongoing effort to develop a multi-state Karst Feature Database (KFD) using Geographic Information System (GIS) and Database Management System (DBMS) technologies. A regional KFD of the western part of Upper Mississippi Valley Karst (Hedges and Alexander, 1985) including southeastern Minnesota and northeastern Iowa is created by merging Minnesota KFD with available karst features and geological information in Iowa. The ultimate goal of this research is to develop a complete regional Upper Mississippi Valley KFD that includes all karst features in northwestern Illinois, southwestern Wisconsin, northeastern Iowa, and southeastern Minnesota. Existing state, county, and sub-county karst feature datasets in Minnesota and Iowa have been assembled into the KFD, which is capable of visualizing and analyzing the entire data set across state boundary. GIS coverages of the entire datasets were projected based on NAD83 datum. The whole karst region is subdivided into Prairie du Chien Karst, Galena-Maquoketa Karst, and Silurian-Devonian Karst. This KFD will provide data and tools for researchers and land-use planners to access karst hazards and to manage water resources of the Upper Mississippi Valley Karst.

This study evaluated potential interstate alignments traversing Kentucky's Eastern Pennyroyal karst region in Pulaski and Laurel Counties. Data collected during the karst inventory and field assessment was used to evaluate the potential for karst development and determine a karst vulnerability rating for the resident karst ecosystem. The inventory and assessment consisted of three main components: karst terrain features; karst hydrology; and karst (cave) fauna. This paper principally reports on the components and structure of the karst feature database and the use of GIS to develop a vulnerability rating for karst areas within the study bands. The study was conducted in three phases; literature search, field reconnaissance, and vulnerability assessment. The field reconnaissance covered over 25 miles along three 1,000-foot wide bands. Over 1,100 karst features were located by Global Positioning System (GPS), classified, described, and photographed. Sixty-three caves were visited with 114 taxa found as part of the cave fauna studies. Of the cave sites sampled, 29 were inhabited by globally rare species. Both the physical and biological field data gathered as part of the field studies was entered into the database before being incorporated into a 4-step GIS model. The karst feature database, GIS model, and vulnerability rating were adapted based on Karst Inventory Standards and Vulnerability Assessment Procedures for British Columbia (Ministry of Forestry; 2003). The vulnerability potential was based on a qualitative evaluation of the following attributes: epikarst development; surface karst feature density; subsurface karst potential; surficial material character; and bedrock type and proportion. The presence of unique fauna or habitat was used to increase the potential vulnerability rating. The assessment delineated and graphically depicted areas of low, moderate, high, and very high karst vulnerability. The study results will be used to present karst impacts in the Environmental Impact Statement for the project.

Assessment of sinkhole hazard (karst collapse and mining collapse) at a national scale is very important for government's decision-making on land-uses. This paper introduces the analysis and assessment of sinkhole hazard using GIS functions such as buffer analysis, overlay analysis, and spatial analysis. Factors of the risk assessment include types of carbonate rock, the geomorphological settings, hydrogeologic conditions, human activities, and land use. Human activities consist of urbanization, mining, and distributions of highway and railway. All factors were digitized as corresponding GIS coverages and assigned weight values based on the distribution and probability of sinkhole collapses. The Analytic Hierarchy Process (AHP) was used to determine the weight of each factor. The assessment result is divided into five levels: very high risk, high risk, medium risk, low risk, and very low risk. The very high-risk areas are mainly located in Guangxi, Guizhou, Yunnan provinces. A few very high risk areas are sparsely distributed in Hubei, Hunan, Chongqing, Sichuan, and Shaanxi.

Vast deposits of evaporites are present in the vicinity of Carlsbad, New Mexico. While evaluating a section of State Highway 128, sinkholes were discovered near and within the highway right-of-way. Several sinkholes, with dimensions of roughly 0.3 meter, were found in the vicinity of a dip section in the highway. Surface seismic refraction geophysics was used to both identify the presence of and delineate seismic anomalies in the shallow subsurface. A twelve channel seismograph with 36 meter geophone array and sledge hammer energy source was used. Typical seismic refraction results provided subsurface profiles where harder sedimentary rock was lying in a near horizontal orientation, where it dipped significantly, and where it was missing. In addition, anomalous loss of signal between geophones across the seismic lines indicated the presence of subsurface anomalies, including significant fractures. Refraction microtremor surface wave seismic was also performed. Resulting s-wave profiles indicated the presence of lower velocity rock horizons underlying the harder rock, provided a cross-check with seismic refraction interpretations and provided a deeper depth of investigation than the seismic refraction results. Test pits located based on seismic profiles confirmed the presence of shallow gypsiferous solution structures.

Karst and karstic features are a common and well-developed phenomenon that can be seen extensively in the Iranian Plateau, especially in the Zagros Mountains. However, little is known about their occurrence and extension in its evaporitic rocks. One of the very famous evaporitic formations well known in Iran and other countries around the Persian Gulf area is the Gachsaran Formation of Miocene age. Its exposures are largely distributed in the southwestern Khuzestan Province. The study area is located in its central to eastern part and is an area with extensive outcrops of the mentioned formation having some interesting karstic features. Undoutedly, solution processes and tectonic stresses had a key role on their establishment. The area is geologically part of the Zagros simply folded belt that obeys its general NW-SE trend. They are mostly thrusted over the Khuzestan Plain mainly by NW-SE striking thrust faults along the mountain-plain boundary. In the area of study, the formation is composed of anhydrite, red and grey marls, salt and sandy limestone. As a seismotectonically active region, it is significantly affected by few numbers of large and small faults such as Lahbari, Jarreh, Dasht-e Shir and Dasht-e Anar faults. Most of the karstic features such as sink holes, springs and cavities are related to the geologic structures. Direction of the stream courses is also related to them. Due to the incompetency and solubility of the layers, some solution features were formed within small folds. The morphology of the evaporite outcrops reflects the influence of the geologic structure. The important recognized trends are: SW-NE, E-W and NW-SE. This study was the first attempt to define karstification in the Gachsaran Formation with regard to its impact on the quality of surface and ground waters and to its limiting factors. Particularly with regard to the development of dam construction, it is planned that a comprehensive field survey and remote sensing analysis will be done to determine and define all the possible karstic features.