Abstract

The Groundwater Project (GWP) is a global, volunteer-based, nongovernmental organization (NGO), initiated in 2017. It was started to help address the global freshwater crisis identified by both the United Nations and the United Nations Educational, Scientific and Cultural Organization (UNESCO). Sustainable use of groundwater may help address the crisis, especially in developing countries. The major goal of the GWP is to produce free online educational materials for many different audiences (lay people and students at multiple educational levels) and on many different groundwater-related topics. As of the end of 2023 there were 44 online educational publications on GWP web pages.

Karst aquifers serve as vital water resources for large populations, and the large springs typical of karst aquifers often support unique ecosystems. The complexity of karst aquifers is well known to the scientific community. There are currently (as of July 2024) three GWP online publications related to karst. Introduction to Karst Aquifers (Kuniansky and others, 2022) was the first karst related textbook published by the GWP. The audience for Kuniansky and others (2022) is upper-level undergraduate science and engineering students (for example, students in geology, earth science, hydrology, hydrogeology, water resources management, or civil and environmental engineering). The Edwards Aquifer (Sharp and Green, 2022) was the second book published about a very important karst aquifer system in Texas, United States of America. This online book describes how the Edwards aquifer functions as a hydrogeologic system by covering almost every topic in hydrogeology as it relates to this specific karst aquifer, including ecological and water resource management and regulation. The third book is titled Karst: Environment and Management of Aquifers, written by Zoran Stevanović, John Gunn, Nico Goldscheider, and Nataša Ravbar (Stevanović and others, 2024). The environment and management ebook is intended for a broad audience including readers without prior knowledge of groundwater science. The intent is to provide readers with a descriptive and comprehensive understanding of the complexity of karst and why specialized engineering is required for water resource management, prevention of aquifer contamination, conservation of ecosystems, and construction on karst terrain. This ebook also has an appendix that includes over 80 photographs of karst around the globe.

These first three karst related GWP books were shepherded through the publication process by Amanda Sills, Eileen Poeter, and John Cherry. If you are reading this abstract and interested in volunteering to publish karst related educational material for the GWP, feel free to contact Eve Kuniansky at elkunian@gmail.com.

Abstract

Sinkholes are a characteristic landform of karst; thus, mapping of karst areas may include documenting the degree of sinkhole formation within the karst landscape. The ability to capture these closed depressions is tied to the availability of accurate digital elevation models (DEMs) for deriving topography. Using the national coverage of DEMs at 1/3 arc-second resolution (approximately 10 meters) and the development of computational methods to extract closed depressions from these elevation models, a national inventory of closed depressions in karst and pseudokarst regions was produced (Doctor and others, 2020; Jones and others, 2021). These data were then combined with nationally consistent data for factors related to geology, soils, precipitation extremes, and urban development to create a heuristic additive model of sinkhole susceptibility at 6-kilometer grid cell resolution within karst regions of the conterminous United States (Wood and others, 2023).

The resulting maps identify potential sinkhole hotspots based on current (2019) conditions and an estimated 50 years (2070–2079) into the future based on projected climate change and urban development scenarios (fig. 1). Areas characterized as having either high or very high sinkhole susceptibility contain 94–99 percent of already known or probable sinkhole locations from state databases for Tennessee, Missouri, and Kentucky. States and counties with the highest amounts and percentages of land in zones of highest sinkhole susceptibility were identified (fig. 2). These results provide a uniform index of sinkhole potential as a starting point for national-scale land use planning, in contrast to more localized assessments produced through various methods within individual States or smaller areas. Projected changes in extreme precipitation and development did not substantially change the locations of current hotspots of highest sinkhole susceptibility. Land use and human disturbance of natural hydrologic conditions are exacerbating factors to the natural conditions for sinkhole development that evolve over geologic time. These influences will likely have more impact on future sinkhole occurrence than hydrologic impacts resulting from projected climate change.

Abstract

“²Karst terrains are characteristic of much of the eastern two-thirds of Tennessee. The occurrence of karst features in Tennessee affects property development, infrastructure, water supply, contaminant transport, and flood and drought planning in Middle and East Tennessee. Karst aquifers in Tennessee provided close to 40 million gallons per day to public water systems in 2015 with the carbonate formations in the Valley and Ridge province of East Tennessee being the second most productive aquifer in the state. The interconnection between surface water and karst systems results in offstream flooding in Tennessee. Sinkhole collapse and the potential for sinkhole collapse have affected subdivisions in several regions in the State. The importance of karst resources to the hydrology and ecology of Tennessee has only been fully defined in relatively small areas. Additional work is needed to further evaluate karst features relative to public water supplies and susceptibility of the systems to contamination, the karst hydrology and ecology along the Cumberland Plateau escarpment, and impacts and controls of sinkhole flooding.”

²Extended abstract published in Bradley (2021).

Abstract

Karst landscapes can often create challenging environments for the planning and design of infrastructure because they are often both particularly susceptible to contamination from surface activities and serve as habitat for unique and sensitive biota. Because of these conditions, it is necessary to understand the characteristics of a karst environment in order to better preserve water quality and prevent degradation of subterranean ecosystems. Wear Cove is a carbonate fenster (window) located in eastern Tennessee along the northwestern boundary of Great Smoky Mountains National Park. The Wear Cove fenster is similar to other nearby fensters at Cades and Tuckaleechee Coves and was created when the hanging wall of the Great Smoky Fault was thrust over the Ordovician strata of the Knox Group. The geologic setting has created a rolling, lower relief cove floor composed of Ordovician Jonesboro Limestone and Blockhouse Shale, surrounded by steep-sided mountains composed of metamorphic and meta-sedimentary strata. In 2021, the U.S. Geological Survey (USGS) in collaboration with the National Park Service (NPS) conducted seepage investigations along Cove Creek and its major tributaries to quantify any streamflow gains or losses, and to determine the impact of the karst on Wear Cove streamflow. Two streamflow surveys were conducted by USGS staff in September 2021 and December 2021. Both streamflow surveys occurred over the span of 2 days. During these surveys, nearly 80 sites were visited for either discharge measurements or zero flow observations. Results from these surveys indicate that Cove Creek, the mainstream in Wear Cove, is largely a gaining stream as it passes through the cove. However, surveys in Cove Creek tributaries underlain by the Jonesboro Limestone all showed losing stream behavior, and in many cases, the tributaries sank entirely into the subsurface. Tributaries in the eastern portion of Wear Cove that are underlain by Blockhouse Shale appeared to largely gain streamflow. These surveys will help both current and future NPS personnel in the planning of any infrastructure in Wear Cove and will be used to help alleviate or mitigate any potential impacts to the underlying karst aquifer.

Abstract

The karstic Edwards and Trinity aquifers are classified as major sources of water in south-central Texas by the Texas Water Development Board. During 2019–23, the U.S. Geological Survey in cooperation with the Edwards Aquifer Authority, mapped and described the geology and hydrostratigraphy of the rocks composing the Edwards and Trinity aquifers in parts of Bandera and Kendall Counties from field observations of the rock outcrops. The thicknesses of the mapped lithostratigraphic members and hydrostratigraphic units were also estimated from field observations.

The Cretaceous rocks in the study area are part of the Trinity Group and Edwards Group. The groups, formations, and members are composed primarily of layers of marls, shales, and limestones. The limestones are composed of mudstone through grainstone, framestone, and boundstone; dolomite; and argillaceous and evaporitic rocks.

The principal structural feature in southern Bandera and Kendall Counties is the Balcones fault zone. The Balcones fault zone is the result of late Oligocene and early Miocene extensional faulting and fracturing which was a result of the eastern Edwards Plateau uplift. In the Balcones fault zone, most of the faults in the study area are high-angle to vertical, en echelon, normal faults that are predominantly downthrown to the southeast.

Hydrostratigraphically, the rocks exposed in the study area, listed in descending order from land surface, are the Edwards aquifer, the upper zone of the Trinity aquifer, and the middle zone of the Trinity aquifer. Descriptions of the hydrostratigraphic units, thicknesses, hydrologic function, porosity type, and field identification are provided and are described further in Clark and others (2023), except for the Bandera and Love Creek hydrostratigraphic units of the Edwards aquifer, which were identified from the mapping for this study (Clark and others, 2023).

Abstract

Monroe County, West Virginia, contains the type localities of the Mississippian Greenbrier Group carbonate rocks that host world-class karst in the central Appalachian region. The Greenbrier Group carbonates in Monroe County occupy a lowland interior plateau with rolling, moderate relief of up to 500 feet (152 meters) that is pock-marked with sinkholes and fringed by ridges of uplifted, more resistant siliciclastic rocks. The Greenbrier Group karst in the county has been extensively studied, with decades of cave exploration leading to significant insights into the karst development (White, 2018); however, the individual formations within the Greenbrier Group and their structural relations were not previously mapped in the original county geologic report by Reger and Price (1926). Therefore, new geologic mapping was conducted in conjunction with a hydrogeologic study of the county (Kozar and others, 2023) to map the Greenbrier Group at the formation level (fig. 1). The new geologic mapping divided the Greenbrier Group into four map units at a scale of 1:40,000, which include, from oldest to youngest, (1) the Hillsdale Limestone, (2) the Denmar and Taggard Formations, undivided, (3) the Pickaway Limestone, and (4) the Union Limestone, Greenville Shale, and Alderson Limestone, undivided.

In addition, a belt of Ordovician-age carbonate rocks with prominent karst development occurs along the northwest flank of Peters Mountain and extends across the entire southeastern border of the county along the state line between West Virginia and Virginia. These Ordovician carbonates were also mapped in more detail in the recent study by Kozar and others (2023) than in Reger and Price (1926). The new mapping used modern stratigraphic nomenclature and divided the Ordovician carbonates into six units including, from oldest to youngest, (1) the Beekmantown Formation, (2) the New Market and Lincolnshire Limestones, undivided, (3) the Big Valley Formation, (4) the Moccasin Formation, (5) the Eggleston Formation, and (6) the Reedsville Shale.

The geologic mapping was enabled by the acquisition of a lidar-derived elevation model for the entire county (Cox and Doctor, 2021a). Using the elevation data, an inventory of karstic closed depressions for the entire county was created, along with a map of the estimated density of dolines and cave entrance locations (Cox and Doctor, 2021a, b, c). The degree of karst development in each of the map units is illustrated on figure 2 as represented by the number of features in each unit.

Because of the differences in structural, stratigraphic, and topographic settings, the styles of karst development between the Mississippian and Ordovician rocks are markedly different. Within the Greenbrier Group carbonates, sinkhole development is most evident within low to moderately dipping units of the Pickaway Limestone, Denmar Formation, and Union Limestone. The highest density of closed depressions occurs north of the town of Union in the center of the county (fig. 3).

The strike of bedding within folds and along faults is largely responsible for controlling regional groundwater movement. For example, within the Mississippian Greenbrier Group in the central portion of the county, two large caves, Union Cave and Hurricane Ridge cave, extend laterally, parallel to newly mapped minor thrust faults. Key to the understanding of karst development in the Greenbrier Group carbonate strata is the recognition of trunk-passage cave development at the contact between the laterally continuous mudstones, limestones, and shales of the Maccrady Formation and the overlying Hillsdale Limestone (Balfour, 2018). A dye trace conducted for the study by Kozar and others (2023) demonstrated surface streamflow sinking at this contact along Taggart Branch south of Gates traveled over 9 miles (14.5 kilometers) to the north across the county and arrived at Dickson Spring 29 days after injection with approximately 8.7 percent dye recovery by mass. The Taggard Formation also contains shale, but it is thin and discontinuous and, therefore, was not found to have a major influence on karst development at its contact with the underlying Denmar Formation, or with the overlying Pickaway Limestone.

Karst development formed within the Ordovician rocks on the northwestern flank of Peters Mountain is geologically and hydrologically distinct from karst development in the Mississippian rocks. The Ordovician karst is similar to a scarp-slope type of karst development in the Greenbrier Group Mississippian rocks on Powell Mountain in southwestern Virginia, first described by Schwartz and Orndorff (2009). Surface streams on the scarp slope of Peters Mountain begin as springs primarily within the Reedsville Shale and the colluvium and ancient alluvial fan deposits of Silurian quartz sandstones that cover the slopes. Epigenetic karstification occurs where these streams sink as they encounter the lithologic contact with older Ordovician limestone units downslope; however, these sinking streams invade a pre-existing paleo-phreatic network. The paleo-phreatic cave passages extend vertically along joints and across bedding within structural folds, then continue along strike and down gradient toward a local base level stream. The deeper phreatic conduit formation may have been influenced by upwelling of warmer water along the St. Clair thrust fault at depth. The presence of deep-seated, long residence time water is evidenced by the spring temperatures and geochemistry. For example, a series of warm springs emerge at Sweet Springs with temperatures up to 23.3 degrees Celsius (Vesper and others, 2019). Additionally, the stable carbon isotope compositions (δ¹³C) of dissolved inorganic carbon in the thermal spring waters are particularly elevated, in the range of −3 to −5 per mil (Sack and Sharma, 2014; Vesper and others, 2022), indicative of substantial isotopic exchange with the host carbonate bedrock. Associated with these springs are extensive tufa deposits, and the modern springflow continues to precipitate calcium carbonate and some gypsum (Vesper and others, 2019). Moreover, studies of the microbiology from a nearby cave influenced by thermal waters showed evidence of carbonate dissolution occurring as a result of sulfuric acid produced by sulfur-cycling bacterial colonies in the cave (Summers Engel and others, 2001).

The presence of a deep, hypogenic karst flow system within the Ordovician carbonates along the St. Clair thrust zone, as evidenced by warm water sulfidic springs, contrasts with the modern fluviokarst within the Mississippian carbonates of the interior plateau setting of Monroe County. These two karst systems that occur in proximity within the same county warrant additional study to further elucidate the controls on such different styles of karst development.

Abstract

Flynn Creek crater is considered the birthplace of impact speleology, a term coined by Milam and Deane (2006) to describe the study of caves within impact craters. Flynn Creek crater was formed in a shallow sea environment approximately 360 million years ago and is currently the only known impact structure to host a cave within its central uplift. This unique and scientifically rich location was the focus of a multiyear drilling program that occurred from the late 1960s to the late 1970s and resulted in the collection of over 3,000 meters of continuous drill core acquired from this impact crater. These samples are now part of the U.S. Geological Survey (USGS) Astrogeology Science Center (ASC) Terrestrial Analog Sample Collections (TASC) and are available to the science community for research and teaching purposes. Available online at: https://www.usgs.gov/centers/astrogeology-science-center/science/terrestrial-analog-sample-collections.

Introduction

Flynn Creek crater is a 360-million-year-old impact crater situated in the Highland Rim physiographic providence of present day north-central Tennessee (N 36°17', W 85°40') (Roddy, 1968). The crater is about 3.8 kilometers in diameter, more than 200 meters (m) deep, has a flat floor, terraced rim, and a central uplift (Roddy, 1977a, b; Wilson and Roddy, 1990; Evenick, 2006). Flynn Creek crater is one of the original six structures on Earth to be confirmed as having an impact origin (the others being Meteor Crater, Arizona, United States; Reis Crater, Germany; Waba, Australia; Hollifard, Canada; and Bosumtwi, Ghana), as well as the first impact crater recognized to have formed in a marine environment (Jaret and King, 2018).

At the time of impact, this area of Tennessee was part of a shallow Late Devonian sea, underlain by Ordovician-age carbonates (Roddy, 1977a; Klapper, 1997; Tucker and others, 1998; Schieber and Over, 2005). When the impactor struck, marine sediment was excavated from the rapidly forming transient crater and deposited as ejecta. Following this initial displacement of sediment was the movement and volumetric rebound of rock beneath the crater floor (Ulrich and others, 1977), causing deep-seated rocks to be violently forced upward, resulting in a central peak inside the crater (Melosh, 1982). Post-impact erosion and deposition on the original transient crater floor were soon inundated by coarse breccias, grading into finer-grained breccias (Roddy, 1977b). Eventually, the Late Devonian sea breached the crater’s rim, leading to marine resurgence and the deposition of black, silty muds that later lithified into the Chattanooga Shale (Roddy, 1968, 1979, 1980).

Structural Deformation

Within the last 100 years, impact cratering studies have made leaps and bounds in our understanding of not only the surface evolution of our Moon and other planetary bodies, but also how meteorite impacts have influenced Earth’s geologic history (French, 1998). Depending on the size of the impactor and resulting transient crater, different degrees of modification influence the final shape of a crater (Melosh, 1989). The smallest impact craters are in the form of a simple bowl shape (fig. 1A). As crater size increases and a crater’s morphology becomes more complex, terraced rims and a central uplifted peak form (fig. 1B). Increasing in complexity from there, peak ring structures are formed when the original uplifted peak collapses due to overextension, resulting in an interior ring on the floor of a crater (fig. 1C), followed by large impact basins with multiple rings (fig. 1D). Most complex of all are massive flat-floored basins, such as the South Pole-Aitken basin (2,500 kilometers in diameter) on the Moon (Baldwin, 1974; Gault and others, 1975; Pike, 1976; Roddy, 1979; Melosh, 1989; French, 1998).

Increases in structural deformation occur in conjunction with morphological complexities. Characterizing the extent and influence of deformation associated with an impact crater and the resultant morphology are important factors in understanding the evolution of a planet’s surface (Roddy, 1979). The ability to carry out such work on a planetary body beyond Earth is very limited, restricted mostly to satellite or orbiter imagery and only studying what is exposed on the surface. Terrestrial impact craters are incredibly valuable to the development of our understanding of impact-related processes. Having direct access to structurally deformed rocks that experienced the intense heat and pressure that only an impact could generate gives us the ability to characterize the different stages of impact crater morphologies on Earth and then extrapolate that to other planetary bodies (Roddy, 1979).

Flynn Creek Crater Drilling and Preliminary Findings

Detailing the extent of brecciation, depth of deformation, and the level of shock metamorphism for complex craters can be achieved through studying terrestrial impact craters. Dr. David Roddy (U.S. Geological Survey [USGS]) sought to acquire such information at Flynn Creek crater through a multi-year drilling program (Roddy, 1977a, b; 1979). Drilling was carried out in two phases: phase I occurred in 1968 and resulted in six drill holes and about 760 m of continuous core; phase II occurred in 1978–1979, resulting in an additional 12 drill holes and about 3,060 m of continuous core (fig. 2; Roddy, 1980).

Evaluating the cores collected from the drilling program led Roddy (1968, 1977a, 1980) to further determine that Flynn Creek crater has an average depth of about 90 m below the pre-impact surface, its breccia lens is only about 40 m thick, and at a depth of about 100 m below the breccia lens, deformation is no longer visible. These findings infer that the impactor penetrated and excavated a wide, yet very shallow crater (fig. 3; Roddy, 1968, 1977a, b; 1980). Cores drilled into the central uplifted peak and on its flanks showed deep-seated target rocks were excavated from a depth of about 450 m, and brecciated rocks were observed down to a depth of about 770 m, which shallowed to about 450 m near the edges of the uplift (Roddy, 1977b, 1980).

Curation and Preservation of the Flynn Creek Crater Cores

Over 2,000 core boxes from Dr. David Roddy’s drilling program were shipped to Flagstaff, Arizona, where it was intended to make the collection available for public access (Hagerty and others, 2013). Unfortunately, such plans were halted when Dr. Roddy passed away unexpectedly in 2002. A decade later, a core recovery and preservation effort was implemented by Dr. Justin Hagerty (USGS). This resulted in transporting the core boxes to large secure shipping containers (fig. 4) located at the USGS Flagstaff Science Campus, and the recovery and preservation of damaged core boxes were carried out alongside a detailed inventory process, in which information on each core box was recorded in a digital database (Hagerty and others, 2013; Gaither and others, 2015, 2017, 2021).

USGS ASC Flynn Creek Crater Sample Collection

This collection, as well as other sample collections are housed by the USGS Astrogeology Science Center (ASC) Terrestrial Analog Sample Collections (TASC) program (Gaither and others, 2015, 2021). The TASC provides the scientific community with a range of samples, including but not limited to, impact crater drill cores and cuttings, hand samples, and rock powders, all of which are available for use in scientific research.

Available on the USGS ASC Flynn Creek Crater Sample Collection website are high-resolution images of each cataloged drill core within the core library (fig. 5). Each core box can be evaluated virtually on the website, and samples of interest can be requested by emailing the author (agullikson@usgs.gov). Instructions on obtaining data from the Flynn Creek crater and other sites are currently available from the website (U.S. Geological Survey, 2024).

Summary

The USGS ASC Terrestrial Analog Sample Collections currently house three individual collections: Flynn Creek Crater, Meteor Crater, and the Shoemaker Collection. These collections preserve the scientific legacies of pioneer planetary scientists, Dr. Eugene Shoemaker and Dr. David Roddy, and are available to the scientific community for research and teaching purposes.

The Flynn Creek Crater Sample Collection houses over 2,000 boxes of cores from 18 separate boreholes within the impact crater. All cores in this collection have been imaged. In the coming months, the Flynn Creek Crater Sample Collection digital database will be published and registered with both ReSciColl and SESAR² to increase discoverability. If interested in conducting research using these samples, please see U.S. Geological Survey (2024).

Abstract

Understanding spatial distributions and temporal variability of groundwater flow and chemistry in karst aquifers is a complex task using fluorescent dye tracing, exploration and mapping of cave streams, geology and potentiometric surface mapping, and other methods. In the 1970s and 1980s, Jim Quinlan and colleagues completed hundreds of dye traces in and around Mammoth Cave National Park and mapped more than 75 kilometers of cave passages including many cave streams. A classic 1981 map defined the major karst drainage basins. Updated in 1989 (Quinlan and Ray, 1989), these and subsequent data are now archived by the Kentucky Geological Survey as the Karst Atlas of Kentucky: Karst Groundwater Basin Maps (Kentucky Geological Survey, 2024).

Great Onyx (GO) Spring in the Park’s Hilly Country was shown but not labeled in the Karst Atlas, and little was known until early 2000s dye traces by Joe Meiman led to the first basin map of about 4 square kilometers (Kentucky Geological Survey, 2024). A high-resolution groundwater flow investigation was started in 2017 to better define the GO basin boundaries and the aquifer’s three-dimensional (3D) internal plumbing. This represents the next generation of fine-scale karst groundwater flow characterization, supporting design of a long-term monitoring system to quantify carbon, nutrient, and sediment fate and transport. Remarkable access to the basin’s essentially pristine contemporary underground flow system is made through more than 20 kilometers of cave passages in GO and Mammoth Caves. The two caves have been hydrologically connected, and dye tracing has shown that GO Spring is fed by at least a third-order stream. A model was also developed to explain an evolutionary decoupling of surface topography and groundwater flow in the ravines of the Hilly Country (Kentucky Geological Survey, 2024).

Abstract

Alternative water supply technologies include aquifer storage and recovery, and brackish groundwater desalination. Aquifer storage and recovery extends the duration water supplies are available by storing them underground for later retrieval. Most desalination projects in Texas use reverse osmosis membranes to remove minerals and contaminants from brackish water to create less saline water supplies. In addition to traditional groundwater production wells, these alternative water supply technologies may also rely on aquifers. Aquifer storage and recovery projects use aquifers as storage zones for recoverable water and may also use groundwater as a source of water to inject and store. Some desalination facilities use aquifers as a source of brackish water and may also use deep, confined saline aquifers as storage zones for concentrate disposal.

Of Texas’ 31 major and minor aquifers, 11 contain karstic zones. The most widespread karst aquifers in Texas are the Cretaceous Edwards and Trinity aquifers. Four of the 5 existing aquifer storage and recovery projects and 8 of the 39 existing brackish groundwater facilities in Texas utilize these aquifers (fig. 1). Other karst and limestone aquifers utilized by existing alternative water supply facilities include the Bone Spring–Victorio Peak aquifer, which provides brackish groundwater for desalination and an unconfined storage zone for recharge via infiltration, and the Fusselman and Montoya Formations, which are used to dispose of desalination concentrate.

The Texas State Water Plan is updated on a 5-year cycle to provide a water supply road map. During this update cycle (2023–2027), the plan compares projected water demands with projected water supplies to estimate needs for five decades into the future. Mostly driven by population growth along the Interstate 35 corridor, projected growing water demands in this region are leading to more planned water supply strategies involving the underlying Edwards and Trinity aquifers. In addition to new traditional groundwater production projects, communities and water suppliers plan to use alternative water supply technologies to meet anticipated needs. In the 2022 State Water Plan (Texas Water Development Board, 2022), alternative water supply projects that would utilize karst aquifers in Texas include 14 of 34 recommended aquifer storage and recovery projects and 5 of the 33 recommended brackish groundwater desalination projects (fig. 1).

To support these planned alternative water supply projects, the benefits and challenges identified during the development and operation of existing alternative water supply projects in the karst aquifers of Texas have been reviewed. These topics include rapidly responsive water table changes in the high transmissivity portions of the aquifers, fast water velocities related to steep hydraulic gradients, large groundwater production volumes supported by voids and fractures, geochemical complexities from high mineral content and water mixing, and habitat conservation for groundwater species.

Abstract

The Eocene- to Miocene-age Gambier Limestone in the Gambier Basin, situated in the southeast corner of the state of South Australia, hosts more than 50 recorded caves and cenotes (Webb and others, 2010). Most of these caves and sinkholes extend below the water table, providing some of the best cave diving opportunities in the world as well as significant opportunities for studying palaeoenvironmental archives. The most extensive underwater cave system recorded in the region, with more than 10 kilometers of surveyed passages, is the Green Waterhole-Tank Cave Complex; a connection between these two caves was established in 2018. This system largely sits under land owned by the Cave Divers Association of Australia (CDAA) and is a South Australian State Heritage Listed site nominated for its outstanding values in palaeontology, speleology, and geology. Access to the Green Waterhole-Tank Cave Complex is an Advanced Cave rated site and is accessible only through the CDAA with access points on CDAA and Department of Environment and Water (DEW) managed properties. The correct level of training and restricted access to the site is imperative to ensure safety of the divers and protection of the cave.

Despite divers exploring the Gambier Limestone caves since the 1950s, and the known potential of karst systems to preserve rich environmental records, underwater cave deposits in this region have remained relatively untapped as sources of palaeoenvironmental information. This is due to the challenges associated with safe scientific diving in underwater caves and the limited number of scientifically trained cave divers (with less than six active scientific advanced cave divers in Australia). In the Gambier Limestone, palaeoenvironmental archives previously recovered from underwater caves consisted of late Quaternary vertebrate fossils and associated sediments, and freshwater stromatolites (Newton, 1988; Kelly, 1998; Mather and others, 2024). Additional palaeoenvironmental samples that were collected between 2022 and 2024, as well as the diving considerations associated with their collection and documentation, are described herein.

Although samples have been collected from Gouldens Hole, Engelbrecht Cave East, and Green Waterhole (also known as Fossil Cave), all in the Mount Gambier region, the focus of the most recent investigations has been in a chamber of the Green Waterhole-Tank Cave Complex called the Bone Room. Primary access is through the Tank Cave entrance; this requires an approximately 45-minute, 950-meter swim before accessing the chamber through a series of tight constrictions. Alternative access can be achieved through Green Waterhole, and although this presents a shorter access, the extreme constriction between Green Waterhole and Tank Cave makes this route more difficult and hazardous. In our fieldwork the Tank Cave entry was used, with the Fossil Cave passages considered a strategic access point to be used only in emergencies. The length of access to the Bone Room required us to critically evaluate the following diving considerations: (1) open circuit versus closed circuit rebreathers; (2) number and positions of stage tanks; (3) mitigation of decompression through the use of Nitrox; (4) level of training and experience of the cave divers; (5) the number of divers in the system at any one time; and (6) the number of dives per day impacting ‘silt-out’ (sediment settling time required for visibility), due to the cave system having no significant flow.

By virtue of their underwater setting and difficult access, palaeoenvironmental deposits within these phreatic caves remain largely pristine and undisturbed (Louys, 2018). We have been able to recover the following palaeoenvironmental archives from underwater settings: (1) micro and macro vertebrate fossils including extinct megafauna; (2) historical bone deposits associated with the earliest European colonization of the region; (3) samples associated with palaeontological and zooarchaeological remains for dating, including samples for optically stimulated luminescence and electron spin resonance dating; (4) sediment cores for pollen analysis; (5) environmental and ancient DNA samples from water and sediment; (6) speleothem deposits, particularly calcite rafts; and (7) freshwater stromatolite deposits. Although analysis of these samples is still ongoing, this project demonstrates the significant scientific value of a wide range of environmental archives and proxies hosted in underwater cave deposits and highlights ways that these can be successfully extracted and examined. This project is ongoing, and methods and techniques are still being refined in collaboration with the CDAA, scientific divers, and associated researchers.

Abstract

The Upper Floridan Aquifer (UFA), one of the world’s most productive freshwater aquifers, is largely made up of Eocene- to Oligocene-age carbonate rocks that comprise the Suwannee and Ocala Limestones (Williams and Kuniansky, 2016). Though the Ocala Limestone consists almost entirely of pure carbonate rock, the Suwannee Limestone may contain quartz sand and impurities. Overlying the Suwannee and Ocala Limestones and Avon Park Formation is the Miocene-age Hawthorn Group. This group consists predominantly of siliciclastic clays and sands with basal limestone deposits, some dolomite, and large deposits of phosphate minerals (Scott, 1990; Williams and Kuniansky, 2016).

The UFA’s karst topography facilitates dynamic water-rock interactions, allowing for the formation of biofilms that serve as localized hotspots for biological activity intimately connected to rock and mineral surfaces (Jones and Bennett, 2014; Casar and others, 2021). Prior work in the UFA has characterized the groundwater microbial communities, revealing diverse microbial life discharging from the aquifer despite low cell biomass (Malki and others, 2020; Scharping and Garey, 2021; Qi and others, 2023; Barry-Sosa and others, 2024). In aquifer environments, microbial biomass is estimated to be dominated by communities attached to surfaces (Flemming and Wuertz, 2019). However, these communities have rarely been characterized in detail. Furthermore, the contribution of rocks and mineral-associated microbial communities to the overall microbial diversity of aquifer caves and conduits remains poorly understood.

In this study, eight diver-accessible water-filled caves of springs in north and north-central Florida were used to investigate the microbial communities in the subsurface. Sampling locations chosen for this work consisted of both first (more than or equal to 2.83 cubic meters per second [m³/s]) and second (0.28–2.83 m³/s) magnitude springs, with each spring connected to extensive cave networks. Most locations are situated in the unconfined portion of the aquifer, providing access to rocks and minerals of the Suwannee and Ocala Limestones. Two sites are within the semi-confined aquifer and provide access to rocks and minerals of the Hawthorn Group.

Utilizing scanning electron microscopy and electron dispersive spectroscopy (SEM-EDX), microbial biofilms on sediment grains, carbonate, and iron-oxyhydroxide minerals were documented. Micrographs showed diverse biofilm architectures across all samples. Biofilms from carbonate cave walls showed homogenous structures of helical stalks whereas micrographs from silica quartz sands showed clustered biofilms with eukaryotic diatoms encased in calcium deposits and biogenic silica. Iron-rich clays revealed sparse extracellular polymeric substance (EPS) and little biofilm formation.

These results show that microbial biofilms are present at macro and microscopic scales and provide visual evidence of diverse microbial biofilms in association with various mineral substrates. Future studies will expand on this SEM characterization as well as explore chemoautotrophy in biofilm communities to elucidate subsurface contributions to carbon cycling in the aquifer, providing a holistic view of groundwater ecosystem processes occurring in the UFA.

Abstract

Blind valleys are recognized features in karst terranes (for example, Jennings, 1985, p. 95–99, and Ford and Williams, 2007, p. 359–361). In the downstream direction, blind valleys typically end at an abrupt bedrock headwall. The surface stream sinks at one or more stream sinks, swallets, or sinkholes at the base of the headwall. Perennial or intermittent surface streamflow simply ends at the sink points. A blind valley is a closed basin and if deeper than the contour interval of the map, is shown as closed contour lines with hachures. Because they divert surface runoff into active groundwater flow systems, blind valleys are critical features in understanding the integrated surface-subsurface character of karst terranes. Blind valleys introduce surface water directly into rapid groundwater flow systems.

Four blind valleys are shown correctly on the 1960s paper versions of the Cherry Grove, Minnesota-Iowa, and Wykoff, Minnesota 7.5-minute U.S. Geological Survey (USGS) topographic maps. Decades of field observations, multiple groundwater traces using fluorescent dyes, and recent lidar-derived elevation mapping have documented their existence unambiguously and in detail. However, these blind valleys have “disappeared” from the recent (2013, 2016, and 2019) digital versions of the Cherry Grove and Wykoff topographic maps.

On the 1967 Cherry Grove map, the perennial upper Canfield Creek sinks in the York Blind Valley in section 21 of York Township. Repeated groundwater traces (MGTD, 2024) show that the water sinking in the York Blind Valley resurges in Odessa Spring on the Upper Iowa River about 11 miles east-southeast of the York Blind Valley. The York Blind Valley not only pirates 8.9 square miles of the Root River basin to the Upper Iowa River basin, but the York Blind Valley creates an interstate water transfer of surface water from Minnesota to Iowa.

On the Cherry Grove map, the perennial upper Canfield Creek is shown correctly as a continuous blue-line feature entering the York Blind Valley and flowing across the blind valley until the stream sinks in the terminal swallet; the blue line ends at the terminal sinking point. The York Blind Valley is shown as closed, hachured 1290- and 1300-foot contour lines. From the end of the blind valley for approximately 3,000 feet southeast downstream, in a surface swale, no blue line indicative of streamflow is shown; intermittent surface flow is then shown as a dashed blue line.

On the recent digital versions of the Cherry Grove map, the York Blind Valley is not shown as closed, hachured contour lines. A “phantom” (non-existent) straight channel is shown connecting the 1290- and 1300-foot contour lines around the blind valley, with those contour lines farther down the valley. Canfield Creek is shown (incorrectly) flowing through the blind valley, and the phantom channel is shown as a dashed blue line, indicating intermittent streamflow. The analogous modifications are true for three other blind valleys on the Cherry Grove and Wykoff topographic maps. Surface streams that are not present are now shown as blue-line features through and downstream from the blind valleys, while hachures that once indicated the blind valleys are now gone.

These are not accidental changes or based on updated information. The USGS National Geospatial Program has confirmed that it was an “automated cartographic process issue” (Mitch Bergeson, National Map Liaison—IA, MN, WI, written commun., 2023) and that errors reside in the current USGS National Hydrography Dataset (NHD) (Joel Skalet, National Geospatial Technical Operations Center, written commun., 2023). All these changes are contradicted by lidar data and field observations.

These four examples raise a fundamental question: Do analogous, incorrect changes exist on other recent digital versions of USGS topographic maps available online in karst areas around the United States? Such errors would present potentially serious environmental management problems. Environmental management decisions typically are based on “the most recent available maps and information” from authoritative sources such as the USGS. The individuals and agencies making such decisions may be unaware of the existence of karst features in their areas that are correctly shown on the 1960s maps. Management decisions based on incorrect information on the digital topographic maps could have major, unintended consequences.

Abstract

All historical and modern spring data available to or collected by the author were compiled into a single database that was used to document the ranges of physical and chemical spring characteristics throughout Virginia, and for the first time, investigate geologic factors that may influence their occurrence in the region (Maynard, 2023a, b). The spring database was examined based on (1) the geologic unit from which the springs discharged; (2) proximity to major fault systems, geologic contacts, dikes, and fold axes; and (3) distribution across a major river basin. Data were processed from a variety of sources to remove duplication, improve locational accuracy, and evaluate available water quality sampling results. The current database includes 1,638 spring locations, 5,913 unique field measurement events, and 2,916 laboratory water quality sampling events (Maynard, 2023b). Coverage spans all five physiographic provinces, but the bulk of field and sampling results are currently in the State’s northern Valley and Ridge province. Spring locations were improved when possible, through fieldwork and the use of high-resolution orthoimagery and LiDAR (Light Detection and Ranging) data. Sampling and laboratory analysis errors were evaluated for a subset of springs by calculating anion and (or) cation charge balance values when possible.

The largest geochemical differences occurred between springs in different rock type groups. Springs in igneous and metamorphic rock areas of the Blue Ridge province exhibited the coldest mean water temperatures, lowest dissolved mineral content, and highest dissolved oxygen content. Most of these low-yielding springs occurred at higher elevations and were believed to be heavily influenced by recent precipitation stored and transmitted primarily through boulder-dominated regolith present in relatively small recharge areas. Because of the lack of soluble material in these systems, the springs discharged weakly acidic “soft” water. Springs found in the unlithified sedimentary units of Virginia’s Coastal Plain physiographic province also tended to discharge weakly acidic groundwater with low dissolved mineral content but with warmer mean spring water temperatures and higher sodium and chloride concentrations than springs in any other part of the state. Although little work has been done on these springs to date, their characteristically warmer water temperatures reflect the higher mean annual air temperatures of this low-elevation region, and higher sodium and chloride concentrations were attributed to the infiltration of aeolian sea salts through silica-rich gravels and sands of the province’s unconfined surficial aquifers.

The hydrogeologic unit (HGU) with the greatest number of springs cataloged in the database was the Cambro-Ordovician Carbonate (CO Carb) HGU with the Conococheague Formation being the formation-level unit with the most springs. However, when considered in the context of outcrop area, the Ordovician Edinburg-Lincolnshire-New Market (O-ELN) HGU emerged as having the greatest number of springs per square mile of calculated outcrop area.

Springs that occurred in the O-ELN HGU stood apart from other sedimentary rock HGUs in several physical and chemical characteristics. On average, springs in this unit were found to occur at lower elevations and discharge greater volumes of water than other sedimentary rock HGUs. The O-ELN springs were also found to discharge groundwater that had a higher mean temperature, specific conductance, dissolved-solids concentration, and hardness, as well as higher calcium, magnesium, bicarbonate, sulfate, chloride, iron, and nitrate concentrations than other sedimentary rock HGUs. These factors indicate that groundwater discharging from O-ELN rocks tends to be more chemically mature and influenced by longer and deeper flow paths. Rocks of this HGU tend to occupy topographically low, valley-floor positions, often at slightly higher elevations than less transmissive shale units of the Martinsburg Formation (part of the O-S Siliciclastics HGU in this study). This contrast in transmissivity probably acts as a barrier to groundwater flow and forces deep groundwater flow paths from O-ELN and other upland HGUs to the surface in or near O-ELN rock units.

No obvious associations between mapped faults or dikes and spring occurrences were detected using a recent statewide 1:500,000-scale structural compilation. Similar results were calculated where 1:24,000-scale digital compilations of fault, dike, and fold axes were available for a 15-quadrangle area of relatively dense spring data coverage. This is believed to mainly reflect the incomplete nature of the current spring database but also may reflect the fact that many, if not most, faults and hydrologically significant fractures remain hidden in Virginia due to dense vegetative cover and overburden. Future comparisons will benefit from the enhanced topographic resolution offered by the growing availability of LiDAR data, which is likely to increase the recognition of the surficial expression of bedrock fracture patterns across parts of the state.

The distribution of available spring data was examined within the Shenandoah River watershed. A zone of influence appeared to extend from the edges of mountainous highlands bordering the basin up to 5 miles into valley-floor carbonate settings. This zone was marked by decreased mean spring water temperatures and specific-conductance values, increased dissolved oxygen content, and greater ranges in flow, temperature, and pH variability than carbonate springs more distal to the mountains. Springs with elevated temperature and conductance values were found to be prevalent in the central valley setting, often near the contact with a downslope fine-grained siliciclastic unit. These relations were more pronounced in the southern and eastern portions of the watershed and are believed to show the physiochemical influence of allogenic recharge from non-carbonate highlands on the predominantly carbonate valley-floor aquifer system proximal to the mountains that border the watershed. The presence of springs with elevated mean temperature and specific-conductance values in the central portions of the watershed indicate that these areas may be points of discharge for deeper and longer flow paths through the carbonate-dominated valley floor aquifer system. This pattern may be governed by the interaction of convergent subregional flow paths with lithologic or structural barriers to groundwater flow.

Abstract

Geotechnical and environmental projects in karst regions typically have a higher geo-hazard risk compared to sites in non-karst regions. Karst features that require consideration in these projects include variable bedrock depth with shallow bedrock pinnacles adjacent to deep soil-filled cutters, floating boulders, and overhanging ledge rock; variable soil and rock conditions including very soft clays and highly weathered zones at the soil-rock interface (for example, epikarst); and features that may collapse or settle that often originate with voids in rock that may be filled with soil, water, or air. These features result in sudden or slow ground subsidence such as sinkholes and closed depressions, and unpredictable hydrogeologic pathways, and can lead to complications with foundation element installation and effectiveness. These, in turn, can result in differing condition claims, schedule delays, material overruns, and safety concerns.

Projects in karst regions require a more thorough site characterization because engineers and geologists cannot rely on interpolating subsurface conditions between test borings as is typically done in non-karst regions. Site characterization methods in karst may include a thorough desk study, including fracture-trace analysis and review of available sinkhole maps and historic aerial photographs; site reconnaissance; geophysical methods; and more extensive intrusive investigation methods including test borings, test pits, and air-track probes than would be conducted on a typical non-karst site.

Geophysical investigations can be designed to gain knowledge about site conditions for a variety of site investigations. These may range from a screening-level survey across a potential development site as part of due diligence and (or) to assist with site civil layouts, to a high-resolution survey focusing on a particular site feature such as a recurring sinkhole or specific structure footprint to add information for the final design of foundation elements. Common geophysical methods utilized in karst regions for civil engineering projects include electrical resistivity imaging (ERI), multi-channel analysis of surface waves (MASW), and electromagnetics (EM).

This presentation will focus on several case histories of geotechnical and environmental projects in karst regions in Pennsylvania, Delaware, Virginia, and Maryland where geophysical investigations were conducted, along with other site characterization methods, to better characterize the site. One site is a roadway underlain by dolomite with multiple frequent and recurrent sinkholes where MASW was used to estimate the extent of the karst zone for designing a grouting program. The MASW survey was also repeated after grouting to estimate the effectiveness of the grouting. At another site, ERI and MASW were used to estimate the bedrock surface below disintegrated marble in an area where environmental sampling showed gaps in Dense Nonaqueous Phase Liquid (DNAPL) contaminant movement. The geophysics helped determine the locations and depths of well screens to delineate the contaminant. Two other sites in dolomite and limestone bedrock regions were screened for site selection purposes using a desk study along with EM and ERI to understand the general subsurface site conditions.

Abstract

“⁶A multimethod geoelectric survey was implemented between January and March 2022 along a 220-meter-long reach of the bedrock-lined streambed of East Fork Poplar Creek in Oak Ridge, Tennessee, to identify locations of surface-water and groundwater exchange and to characterize the subsurface flow paths that convey water between the stream and its flood plain. A waterborne self-potential (WaSP) survey was completed in January 2022 to measure the electric streaming-potential field in the stream. Electric resistivity tomography (ERT) was performed in March 2022 on the flood plain adjacent to the WaSP survey reach to map the electric resistivity distribution and characterize the hydrogeology and subsurface flow paths that facilitate surface-water and groundwater exchange in the bedrock-lined stream. The combination of WaSP and ERT data support the qualitative interpretation that surface-water and groundwater exchange likely occur along fractures in outcropping bedrock and along two fault lines that intersect the limestone creek bed.”

⁶This abstract is from a complete article by Ikard and others (2023).

Abstract

Karst features in western New York commonly are subtle due to the relatively recent onset of karstification processes, or obscured by thick glacial overburden, and thus can be difficult to assess. Five likely karst drainage features were examined using multiple geophysical techniques, including frequency-domain electromagnetics, ground-penetrating radar, electrical-resistivity tomography (ERT), passive seismic, and seismic refraction, to better determine the nature of the subsurface fill material, bedrock topography, and karst dissolution features. All five sites had sufficient physical property contrast to reveal subsurface features, and features such as bedrock lows, voids, fractures, and variation in fill properties were identified. Although ERT methods provided the most useful information relating to karst dissolution features, the combination of methods yields greater insight and support for characterization of the subsurface.

Introduction

Western New York contains several karst units of Ordovician-Silurian age that include limestones, dolostones, and evaporites. Karst development in the region was influenced by the numerous glacial advances and retreats of the Laurentide ice sheet; during advances, karst dissolution may be limited, but new bedrock fractures may be created by ice loading, which can later be solutionally widened and lead to further karst development. Most karstification in western New York is immature to submature, consisting mostly of solutionally widened fractures with no development of caverns or significant surface features typical of mature karst terrains. The Middle Ordovician Onondaga Limestone is the unit most prone to development of dissolution features (Kappel and others, 2020), and solutionally widened fractures, sinkholes, and sinking streams are observed. These features commonly are subtly expressed and can be difficult to distinguish from glacial topographic features, and the nature and thickness of unconsolidated glacial sediments overlying the Onondaga Limestone are highly variable. Thus, the presence of karst features can be overlooked in the region, which may lead to contamination of groundwater in karst aquifer units from surface spills and agriculture.

Genesee County in western New York State, is a predominantly rural and agricultural area that largely relies on groundwater resources for municipal water supplies, drawing from unconsolidated glacial sediments underlying the county seat of Batavia, NY. However, there are also more than 1,000 domestic water-supply wells across the county that are primarily drawing water from karstic bedrock aquifers. Over the past several decades, Genesee County has experienced a high number of contamination incidents in these domestic wells (Reddy and Kappel, 2010; Kappel and others, 2020). However, the groundwater-flow paths in the area, and relation to karst features, are understudied.

The purpose this pilot study is to evaluate shallow surface-geophysical methods to image karst drainage features in Genesee County, New York, and to characterize the nature of the subsurface and extent of karst development. This study employs multiple geophysical techniques at five karst drainage features to characterize depth to bedrock and bedrock topography, identify possible fractures or voids, characterize fill material, and locate possible connections between surface and bedrock. Results of this study will be used to evaluate whether these drainage features may be used as injection sites for future dye-trace studies. This study also provides additional validation of modeled closed depressions documented by Sporleder and others (2021) as part of a set of regional karst mapping data products.

Methods

Site Selection

An inventory of potential karst drainage features in Genesee County was created using the closed depression polygons, bedrock maps, and groundwater depths from Sporleder and others (2021). Closed depression features that coincided with verified karst features from Reddy and Kappel (2010) were prioritized. Using these data and other available information from state resources, a matrix of site characteristics, including a site description, proximity to discharge points, proximity to known contamination sites, ease of accessibility, depth to bedrock, and depth to groundwater was developed for 13 sites. The sites were ranked based on these criteria and the top five sites were chosen for the pilot study (fig. 1). All five sites are surface depressions within the limits of the Onondaga Limestone, with variable information available on depth to bedrock and groundwater.

Results and Discussion

N Bennett Heights

The sinkhole at N Bennett Heights is a small, subtle depression approximately 0.15 acre in area that has been identified as a patterned sinkhole by Reddy and Kappel (2010) (fig. 2). The site is at the southern end of a farm field in a vegetation buffer zone between the field and a row of houses along N Bennett Heights. The site is less than 1 mile from known sites of contamination (Reddy and Kappel, 2010; Kappel and others, 2020).

This site was the only site where all tested geophysical methods were employed. FDEM surveys were conducted with both the DualEM and GEM-2 FDEM instruments, and results indicate near-surface materials with EC in the 3–5 milliSiemens per meter (mS/m) range (fig. 2A and 2B), which likely reflects unsaturated and/or coarse material with low specific conductance. No significant trends in EC were observed between the depression and the surrounding field, although some higher values were observed with the DualEM along the southern edge, likely due to proximity to a chain-link fence.

ERT showed an increase in the resistivity around a depth of 10 feet (ft) that likely indicates transition to bedrock (fig. 2C). Some areas of higher resistivity are observed within the bedrock that may indicate karst features such as voids or weathered zones.

GPR data from the location of the ERT profile lines show a possible bedrock reflector around a depth of 6 ft (EM velocity of 0.08 meter per nanosecond (m/ns) assumed based on diffraction hyperbola fitting) inside the depression (fig. 2D). Given the shallower depth observed compared to the ERT results, this reflector may also be attributed to soil structure. Seismic refraction results also indicated an overburden thickness of 8.5 ft, with a P-wave velocity of 600 meters/second (consistent with unconsolidated material) overlying bedrock with a P-wave velocity of about 1,200 meters/second (consistent with limestone). A single HVSR measurement of moderate quality indicates bedrock around a depth of 8.5 ft. Differing bedrock depths between methods may reflect an epikarst or weathered zone at the top of the bedrock.

Genesee County Community College

This site is on the campus of the Genesee County Community College and is a broad closed depression with an area of approximately 3.5 acres, with a sharp, approximately 5 to 6-foot-tall scarp along the eastern edge. The central and deepest part of the depression is forested, while the southern and northern sections extend into open fields. Bedrock is close to the surface around the campus, and the sinkhole has been used by campus facilities management as a disposal site for waste rock from construction projects. FDEM, GPR, and ERT methods were employed to characterize this site.

FDEM surveys were conducted with the GEM-2 instrument, and results indicate near-surface material in the 0–3 mS/m range, with little to no measurable signal in the central part of the sink (fig. 3A). This lack of signal may be due to bedrock being very near or at the surface in the center of the sinkhole, as a higher signal was obtained around the eastern edge of the sink where the thicker soil scarp is present. GPR results indicate numerous sub-horizontal reflectors in the upper 10 ft data (EM velocity of 0.07 m/ns) that may indicate layers of bedrock and/or soil, as well as an area of lower reflection amplitudes observed in the center of the sinkhole (fig. 3C). ERT results correspond with the other observations, indicating thin soils and bedrock around a depth of 5–7 ft. An area of high resistivity in the center of the line that corresponds to the low amplitude zone observed in the GPR data, may indicate the presence of a significant karstic void near the deepest part of the sinkhole (fig. 3B).

Genesee Valley Agri-Business Park

The Genesee Valley Agri-Business Park depression is the largest of the sites, with an approximate area of 8 acres (fig. 4A). The depression has relatively steep sides to the north and south, with a broad flat base with trees and brushy undergrowth. A small stream enters the depression from the east, and the site floods during high rainfall events. The easternmost section of the depression is fenced and was not accessible during the study. Kappel and others (2020) and Richards and Boehm (2012) report exposed bedrock in the area; however, this was not observed in the western section of the depression covered by this study and may be limited to the eastern section. FDEM, GPR, ERT, and HVSR data were collected at this site.

Seven HVSR measurements were made at the site, four in the central axis of the depression and three on the southern rim of the depression. One measurement was taken in duplicate with consistent results; measurements were generally of moderate to excellent quality. Depth to bedrock in the center of the depression was between 7–9 ft deep, with one measurement of 19 ft deep in the eastern part of the depression. Depth to bedrock on the rim ranged from 21–39 ft deep.

FDEM surveys were conducted with the DualEM instrument. Results indicate near-surface materials in the 2–5 mS/m range, indicating unsaturated and/or coarse sediments with low specific conductance, with a zone of high EC in the eastern part of the depression (fig. 4A). EC was higher in the sinkhole than on the ridges; some anomalously high values were recorded due to proximity to the metal fencing along the eastern edge. The high EC zone, which corresponds to both the deeper HVSR measurement of 19 ft and the location of the drainage stream, was targeted for the ERT survey. The southern section of the ERT profile at the top of the break in slope (about 30 ft distance in the profile) indicates a vertical high resistivity structure, potentially indicating a fracture related to the formation of the depression (fig. 4C). ERT results from within the depression indicate a gradual transition to bedrock, with a zone of more conductive soils corresponding to the high EC zone in the FDEM survey. This area likely has a higher moisture content and finer-grained sediments than other parts of the sink due to drainage from the stream. A small dip in the section around the 150 ft point corresponds to the 19 ft HVSR measurement, possibly indicating a bedrock low.

GPR results show areas with strong shallow (less than 5 ft, based on fit EM wave velocity of 0.09 m/ns) sub-horizontal reflections most prominently visible in the western part of the depression (fig. 4B, 0–700 ft along the profile). Such reflections were not visible along the ERT line (no interpretable GPR information was available along this line). These reflections may be indicative of soil structure or may indicate the bedrock surface, albeit at somewhat shallower depths than estimated by HVSR. Possibly, areas lacking these reflections are zones where the bedrock is disturbed (for example, collapsed or weathered), such as in the eastern part of the depression.

Quinlan Road

The sinkhole at Quinlan Road is a tree-lined, steep-sided swallet in the center of a very large, closed depression located in a farm field. It is the primary drainage outlet for a massive compound sink, with a man-made drainage ditch connected to the swallet to help drain the larger closed depression feature during high runoff events. Richards and Boehm (2012) interpret this as a solution sinkhole and found through use of pressure transducers that the sinkhole is subject to groundwater flooding. The Onondaga Limestone is visible in the sides of the sinkhole, starting at approximately 10 ft of depth from the surface. Due to the steep sides, hard, dry soil and exposed rock, and the narrow swallet, ERT measurements were not obtainable at this site. HVSR measurements also were not taken at this site, as bedrock is exposed in the sinkhole sides. FDEM and GPR measurements were made across the swallet and surrounding field.

Both the DualEM and GEM-2 instruments were used to collect FDEM data. The results indicate near-surface materials in the 4–10 mS/m range, with low EC values in the sinkhole and higher EC values in the fields and along the drainage ditch (figs. 5A–5C). The lower values in the sinkhole correspond to the thin dry soil and exposed bedrock, which appears to be limited to the tree-lined sinkhole area, with increasing EC values where soils are likely thicker and have higher moisture content on the fields.

GPR results surrounding the swallet (fig. 5D) indicate variable sub-horizontal reflections in the upper 6–10 ft (based on fit EM velocity of 0.1 m/ns), indicating shallow soil structure or possibly the bedrock. Increases in depth and/or absence of these reflections may indicate underlying bedrock collapse/dissolution (for example, figure 5D at 328 ft along the profile, corresponding to where the drainage ditch enters the swallet). Deeper structures in the GPR data likely represent air-reflections from the side of the GPR unit (for example, the feature at −13 ft elevation at the beginning and end of the profile on figure 5D probably corresponds to an aboveground observed metal structure). GPR profiles crossing the swallet feature itself were uninterpretable due to these air-reflections and positional inaccuracies.

Mud Creek

The Mud Creek sinkhole is a broad, relatively steep-sided depression in the streambed of Mud Creek, an intermittently flowing losing stream, just north of a stone and concrete culvert through a now-defunct railroad embankment. The sinkhole is typically water filled at normal groundwater levels, even when other parts of the streambed are dry; during the time of data collection, the region was experiencing drought and the sinkhole was also dry. The Onondaga Limestone crops out downstream, and may be exposed under the railroad culvert, but the presence of the railroad bed construction material (also Onondaga Limestone from local quarries) makes determination of this difficult. The sinkhole is filled with very wet, organic-rich mud, along with some railroad debris like old wooden ties and broken pieces of limestone and concrete. Data were collected at this site using FDEM, GPR, ERT, and HVSR instruments.

FDEM data were collected using the DualEM instrument and indicate near-surface materials in the 3–14 mS/m range (fig. 6A). In general, the sinkhole area shows elevated EC values compared to the surrounding areas upgradient and along the railroad embankment. Readings in some areas where the FDEM transect crosses over itself are not consistent, which may be due to the steep topography affecting the level of the sensor with respect to the ground surface. Higher EC in the sinkhole center may reflect the high moisture content of the organic-rich mud infill. GPR results (fig. 6C) show near-surface soil structure as well as a concave reflector consistent with the sinkhole shape, with a maximum depth of approximately 20 ft (based on fit EM velocity of 0.08 m/ns).

The ERT results show higher resistivity values in the sinkhole compared to the upgradient rim (fig. 6B). A near-vertical, very high resistivity anomaly is observed in the center of the sinkhole and is observed in both the raw data and the inverse result, indicating the presence of a significant void space in the center of the sinkhole, potentially related to an underlying solution-enhanced fracture. Closer evaluation of GPR results from this area somewhat supports this in terms of fewer visible reflections (fig. 6D). Two duplicate HVSR measurements of moderate to excellent quality were taken within the sinkhole and provided consistent results of approximately 42 ft of depth to bedrock, which is deeper than indicated by nearby quarries and outcrops downstream. These measurements may be aligned with a near-vertical feature interpreted in the ERT data where the depth to bedrock may be locally deeper than in surrounding areas.

Summary

Five karst drainage features in western New York were examined with geophysical techniques, including FDEM, GPR, ERT, HVSR, and seismic refraction, to determine the nature of the subsurface with respect to bedrock topography and glacial overburden. All five sites had sufficient physical property contrast to reveal geophysical anomalies related to karst development. FDEM methods were the most sensitive to soil property contrasts, while GPR showed detailed subsurface complexity at most sites that need additional methods to support interpretation. HVSR data were of varying quality, depending on the depth to bedrock and nature of the fill material. ERT was the most useful method for providing interpretable information about karst features, revealing potential bedrock voids and fractures. These geophysical data are available in a USGS data release. Future studies in the region, such as tracer tests or aquifer models, would benefit from multiple geophysical techniques to evaluate these features and gain insight into the nature of karstification.

Abstract

Kīlauea volcano hosts numerous pit craters that are inferred to have formed in competent bedrock (lava flows with minor tephra and other sediments), including Wood Valley Pit Crater. The Wood Valley Pit Crater is a 50-meter-deep, nearly circular pit that includes access to a cave entrance, which provides an opportunity to monitor cave climate throughout a cave that is ordinarily inaccessible. Cave climate observations in this volcanic pseudokarst area included cold trapping, cave breathing, possible effects from geothermal heating, and possible atmospheric thermal tide-induced cave fog.

Introduction

Kīlauea volcano hosts numerous pit craters that are inferred to have formed in competent bedrock (lava flows with minor tephra and other sediments), and some of these pits did not form along large normal faults or grabens. These pits are located along zones of ground cracking, and patterns of these ground cracks were used by Okubo and Martel (1998) to demonstrate the presence of large opening-mode fractures below the pits. These zones of ground cracking exhibit predominantly horizontal extensional displacements, with little, if any, vertical displacement characteristics of normal faults. The pit craters are also located within the volcano’s rift zones and at the summit region, both areas where magma flow is known to occur in dikes or larger conduits (Neal and others, 2019).

Fourteen pit craters and pit crater complexes (clusters of multiple overlapping pit craters) are known to occur along Kīlauea’s East Rift Zone (ERZ), with two of these no longer visible due to infilling by lava flows associated with the 1969–1974 Mauna ‘Ulu eruption (U.S. Geological Survey, 2023). Kīlauea’s summit region experienced multiple episodes of pit crater formation during the recent (2008–2018) activity (Patrick and others, 2020; Poland and others, 2021). The Southwest Rift Zone (SWRZ) contains approximately 15 pit craters and pit crater complexes, the majority of which are located both up-rift and down-rift of the Great Crack, an eruptive fissure that was active in 1823 (Stearns, 1926; Stone, 1926). The total number of pit craters along the SWRZ is approximate because several of the pit crater complexes adjacent to the Great Crack coalesce with large aperture (greater than 10 meters (m) wide), non-eruptive, talus-filled fissures, and individual pits become difficult to distinguish from collapsed sections of those fissures.

Pit craters of Kīlauea’s SWRZ appear to have formed in competent bedrock and were not associated with large normal faults; most of these pit craters are known to contain caves at the pit bottoms (Okubo and Martel, 1998; Halliday, 2002; Favre, 2014; Coons, 2016; Cushing, 2016). Only two ERZ pit craters, Devils Throat (Jaggar, 1947) and Lua Niʻi (Macdonald and Eaton, 1964), are known to have contained caves at their bottoms. However, the caves within these two pits are no longer visible because their entrances have been buried by talus from the pit walls.

Wood Valley Pit Crater Cave

The floor of Wood Valley Pit Crater (WVPC) can only be accessed via rappelling. At ground level, WVPC is approximately 35 m x 28 m in diameter and is elongate in the northeast-southwest direction. Inside the pit, the base is approximately 60 m x 40 m in diameter and is elongate in the north-south direction. The pit’s shape is that of an elliptic conical frustum and is approximately 41 to 47 m deep.

The walls of WVPC comprise a sequence of fractured basalts, primarily ‘a‘a flows. The floor of the pit consists of poorly sorted, angular blocks of basalt that are about 2 m in length or less. These blocks appear consistent with the ‘a‘a flows exposed in the pit walls and therefore are interpreted to have been derived through collapse of those walls.

The WVPC and the entire length of its known cave system was first described by Favre (1993) and re-surveyed by Favre (2014) using a handheld laser rangefinder and magnetic compass. These surveys revealed that the entrance to this cave is located at the base of the northeast wall of the pit. The cave is roughly linear in map view and trends northeast-southwest, which is parallel with the local orientation of the SWRZ. The horizontal extent of the cave is approximately 475 m, and at its deepest point is approximately 60 m below the floor of WVPC.

Favre (1993, 2014) revealed an uncommon origin for the WVPC cave. They noted that the walls and ceiling of the cave were coated in lava dripstone, as is often observed in lava tubes and eruptive fissures. Further, the largest room in the cave, informally named “Salle du Four” translated as the “Oven Room,” is shown to have an ovate (Favre, 1993) to tabular (Favre, 2014) shape in cross section. Favre (1993, 2014) interpreted this cave as being the top of a partially drained dike, with the glassy dripstone and ‘a‘a, and pahoehoe floor composing the basalt (magma) that drove the dike’s formation. Although Walker (1988), Okubo and Martel (1998) and others have interpreted Kīlauea’s pit craters to have formed above partially drained dikes, examples where such dikes are known are uncommon. Lua Nii (Macdonald and Eaton, 1964) and “Pit H” along the Great Crack (Okubo and Martel, 1998; Halliday, 2002; Coons 2016) are the only other known examples of pit crater/cave systems that formed above and within partially drained dikes.

Methodology and Data Collection

Both cave geometry and climate data were collected in stages over a period of several years (Cushing and others, 2024). Each visit to the cave was typically 7–10 days. During each visit, climate data were extracted from the pressure, temperature, and humidity data, while part of the cave was surveyed using a three-dimensional mapping system. A description of the approach for the collection of both types of data follows.

A tripod-mounted three-dimensional mapping system was used to map the shapes of WVPC and two of the largest rooms in its cave, as well as to map the locations of the pressure, temperature, and humidity sensors therein. The mapping system consists of an infrared laser rangefinder with an integrated inclinometer and an optical azimuth encoder (Impulse 200 and TruAngle from Laser Technology, Inc.) mounted on top of a surveying tripod. This system has the key advantage of being able to be broken down into small, lightweight components that can be secured in moderately sized hard cases for transport to the bottom of WVPC and through the tight squeezes of the cave. The laser rangefinder records target locations in spherical coordinates (range, azimuth, inclination) relative to the focal point of its receiver, which are subsequently converted to Cartesian coordinates. Using this system, the entirety of WVPC was remapped and a total of about 75 m of the approximately 475-meter length (Favre, 2014) of its cave.

Temperature (rock and air), relative humidity, and barometric pressure data were collected at the cave entrance (located at the bottom of the pit) and throughout the cave network (fig. 1; table 1). An entire year of data were collected for the cave walls surrounding one of the cave entrances and the upper portion of the cave, and about 6 months of data for the lower portion of the cave, which included the horizontal basaltic dike. Unfortunately, wind data from either inside the cave or near the entrance were not collected.

Temperature and humidity data can be separated by two date ranges: (1) 10/19/2012–4/18/2013, and (2) 4/18/2013–11/30/2013. The first dataset range only included pit and upper cave observations, while the second dataset range included the pit and the entire cave network. Pressure data, acquired at 10-minute intervals, were available from 10/22/2012–8/5/2014. The pressure sensor was occasionally relocated. A second pressure sensor acquired data at 30-second intervals from near the cave entrance from 4/20/2013–9/19/2013.

Temperature: Rock and air temperatures were concurrently recorded using a HOBO data logger U23-003s, with two external probes. One probe was inserted into a small hole drilled into the rock (channel 1), while the other probe (channel 2) was left exposed to the air. These sensors have an absolute accuracy of 0.2 degree Celsius (°C) and a relative accuracy of 0.02 °C. The digitization level as measured from the data was determined to be 0.023 °C, which is comparable to the stated relative accuracy.

Humidity: Humidity was measured throughout the cave complex using a HOBO data logger U23-001s, where air temperature and humidity are measured with internal sensors. These sensors were usually colocated with one of the U23-003 data loggers. These sensors are not accurate at relative humidities above 95 percent, otherwise they have an accuracy of 2.5 percent.

Pressure: Barometric pressure was measured near the cave entrance and at the back of the cave using HOBO S-BPA-CM10 pressure sensors. The sensor accuracy is plus or minus 3.0 millibars (mbar) over the full pressure range at 25 °C. The data resolution (digitization error) is 0.1 mbar, which is consistent with the data collected.

Temperature Data and Trends

Four sensors were placed around the east entrance to the cave. Each sensor has two external probes, so one was inserted into a drill hole in a rock while the other was placed to better record air temperature. Figure 2 shows that the rocks exceeded 20 °C during the winter months when they received direct sunlight. Air temperatures showed that nighttime temperatures, especially during the winter (and rainy) months, often were cooler than the rocks. A comparison to figure 3 shows that cold trapping likely occurs as the cave cools with depth.

Relative Humidity Data and Trends

The cave air spends more time saturated as one descends deeper underground. By the time one gets to the horizontal dike, the air is consistently saturated (fig. 4).

Barometric Pressure Data and Trends

Barometric pressure was measured about 10 m below the west cave entrance and at the end of the horizontal dike. These data are shown on figure 5. The pressure offset is consistent with an elevation change of about 40 m. The phase difference between the two sets of measurements was negligible.

Atmospheric Thermal Tides

Fourier transform analysis of the barometric pressure data showed the influence of atmospheric thermal (or solar) tides (fig. 6).

Cave Fog

Part of the data collection in the cave was the use of an infrared laser rangefinder. Survey data of the “Oven Room” from the rangefinder often produced a scattering of distances instead of clean consistent distances to the room’s wall and ceiling. It was realized that the laser beam was likely reflecting off water droplets suspended in the cave air (cave fog), and that the number or density of droplets varied throughout the day. This hypothesis was anecdotally supported by images (fig. 7) taken by Tim Titus of Chris Okubo at two different times of day, one where the air is relatively clear and the second where there appear to be suspended droplets. The hypothesis is further supported by comparing the observed air temperature to expected temperature changes based on pressure and the ideal gas law (fig. 8) if no phase change occurred. The apparent stability of the observed temperature when compared to the pressure-derived temperature suggests that latent heat was exchanged. This is possible as the air was always saturated. Further analysis is needed, but it’s possible that cave fog was observed 100 m underground, driven by solar tides.

Summary

Wood Valley Pit Crater Cave provided an opportunity to monitor cave climate throughout a cave that is ordinarily inaccessible (access to a horizontal dike that is about 100 meters underground). The entrance of the cave is in the bottom of a 50-meter-deep nearly circular pit, likely formed from collapse. The cave passage continued 50 meters farther underground from the entrance, providing access to a horizontal dike that extends for hundreds of meters.

The cave climate appeared to be mainly driven by external pressure changes from weather systems and solar tides, the flow of cooler air into the cave (cold trapping), percolation events, and possible effects from geothermal heating.

Cave “fog” was observed to form and dissipate in the lower part of the cave—the horizontal basaltic dike—due to pressure-induced adiabatic expansion/compression and the air remaining saturated with water vapor year-round.

Abstract

The purpose of this project was to monitor the cave climate of Bonito Flow Ice Cave in Sunset Crater National Monument, Arizona. The main purpose of the climate monitoring was to determine if the Ice Cave was still an “ice cave,” in other words, a cave that contains perennial ice. The data acquired from March 2, 2021, to December 13, 2022, consisted of air temperature, relative humidity, and barometric pressure throughout the cave and included temperature and humidity measurements outside the cave, but near the entrance. This cave is considered a sacred spot by at least three southwestern Tribes because it was a historical source of ice and continues to be a place for pilgrimages and ceremonies.

Sensors were distributed throughout the cave with a focus on a large ice puddle that forms on the cave floor during the winter months. Air temperatures on the cave floor often showed decreases that corresponded to times when the outside air temperatures were below that of the cave air temperature, indicating the flow of colder air into the cave. The cave air was saturated with water vapor most of the time. An air temperature gradient existed between the cave floor and ceiling; this gradient was greatest during the summer. Hoarfrost was observed on the cave walls and ceiling, indicating that the rock was at or below freezing temperatures.

Sunset Crater Bonito Flow Ice Cave may still be an ice cave, but additional monitoring and analysis are needed. The visible presence of ice at the ice puddle occurred for about 11 out of 12 months each year, but perennial ice may still exist just below the rocky cave floor. The temperature of the floor never rose above a few degrees Celsius, and once freezing cold air from the surface flowed back into the cave, it dropped to below freezing. Therefore, these data indicate that this cave is a static ice cave.

Introduction

Ice caves (rock-hosted caves that contain perennial ice or snow) are threatened by global warming. The issue of global climate change has been a topic of public debate and scientific discussion over the last few decades. The impact of climate change on cave climate is complex. How a cave climate responds to changing surface conditions is multivariate and depends on a complex interplay among outside air temperature (especially minimum diurnal and seasonal air temperatures); surface temperatures (this includes effects caused by changes in vegetation); the timing, amount, and deposition mode (rain vs. snow) of precipitation; and the cave structure (configuration) itself. For example, warming daytime air temperatures may have minimal effect on cave temperature if nighttime and winter above-ground air temperatures remain sufficiently cold and continue to flow into the cave (pushing warmer air out), keeping the cave temperature at or below freezing (in other words, cold trapping; Williams and McKay, 2015).

One such cave was initially formed by the Bonito Lava Flow in Sunset Crater National Monument during the Sunset Crater volcanic eruption that occurred about 1066 (Houk and Scott, 1995). This cave became a source for perennial ice and was used by local indigenous populations (Stoffle and Van Vlack, 2022), and later, by settlers of European descent (Houk and Scott, 1995). Because the cave was a reliable source for ice, some southwestern tribes (Hopi, Zuni, Southern Paiute) considered the cave a sacred spot, and it continues to be a destination for spiritual pilgrimages and ceremonies (Hopi, Zuni) and celebrated in migration songs (Zuni) (Stoffle and Van Vlack, 2022). To protect the cave, the entrance has a locked gate and is only accessible with a permit from Sunset Crater Monument staff and permission from the Hopi and Zuni tribes. Figure 1 shows the local terrain, with the cave entrance in the background.

Our primary purpose for climate monitoring was to determine if Bonito Flow Ice Cave was still an ice cave. Data acquired consisted of air temperature, relative humidity, and barometric pressure throughout the cave, accompanied with temperature and humidity measurements outside the cave, near the entrance.

Data Collection Locations Within the Cave

Data collection was conducted from March 2, 2021, to December 13. 2022, and consisted of air temperature, relative humidity, and barometric pressure measurements. Temperature and humidity were collected outside of the cave (locations 1 and 2 in table 1), and at several places within the cave (locations 3–8 on figure 2 and table 1). Pressure data were collected at three locations within the cave (locations 3–5 on figure 2 and table 1). Additional weather data were collected approximately 800 meters north of the cave site, providing useful context for cave data interpretation (Cushing and others, 2023). Photographs from locations shown on figure 2 are provided as figs. 3–8).

Ice Observed—Stalagmites, Stalactites, Ice Puddle, and Hoarfrost

The presence of icicles (stalactites) indicates that percolated water from surface snowmelt was a significant source of water for Bonito Flow Ice Cave. Ice stalagmites often formed beneath the ice stalactites. The lowest part of the cave typically had a “puddle” of ice (location 4 on figure 2 and table 1; fig. 7) that was visible at least 11 months of each year. The water source was likely from percolated snowmelt that did not immediately freeze or meltwater from the ice stalagmites (fig. 5) and stalactites (fig. 6). Hoarfrost, the result of water vapor condensing directly on the cave walls and ceiling, was also observed seasonally (late winter) (fig. 8).

Climate Data Collected—Temperature, Humidity, and Pressure

Figure 9 shows nearly 2 years of climate data collected in Bonito Flow Ice Cave for 2021–2022. Data gaps occurred due to either sensor failure or from the data logger memory becoming full. The general trends were repeatable between the 2 years. Generally, the amplitude of temperature variations progressively decreased past the cave entrance. The exception was the cave ceiling (location 6, fig. 2 and table 1), which nearly tracks the entrance temperature. Relative humidity variations were observed at the entrance. The remainder of the cave was saturated almost all year. The exceptions were the cave ceiling, and possibly during periods when the temperature of the cave was below freezing. However, relative humidity measurements at temperatures below freezing are suspect due to instrument limitations. The cave is a static ice cave (in other words, no flow-through) (Williams and McKay, 2015), and is only 69 meters in length. As such, the three pressure measurements generally tracked each other.

Entrance Data and Trends

The cave entrance (location 2 on figure 2 and table 1) temperature and humidity data are generally consistent between the 2 years (fig. 10). Near-total saturation (100 percent relative humidity) occurred when the air temperature was near freezing and during the summer period of the monsoons, typically July–August.

Ice-Puddle Data

The ice puddle is location 4 on figure 2 and table 1. Three temperature and humidity sensors were placed near this location, and the multi-year data are shown on figure 11. This is likely the location of any perennial ice, even though there were about 30 days in early autumn when the cave floor appeared free of visible ice and standing water (at least for 2022). Perennial ice could be present just under the rocky cave floor, as floor temperatures return quickly to freezing temperatures.

Mid-Cave Data

Farther into the cave passage, past the ice puddle, is the mid-cave section. Both floor and near-ceiling (fig. 4) data were collected (fig. 12).

Back of the Cave Data

The back section of the cave is separated from the mid-cave section as shown on figure 2. Climate data for this part of the cave are shown on figure 13.

Summary

Climate sensors for measuring air temperature, relative humidity, and barometric pressure were distributed throughout Sunset Crater Bonito Flow Ice Cave for 2021–2022, with a focus on a large ice puddle present on the cave floor during the winter months. Air temperatures on the cave floor often showed decreases that correspond to times when the outside air temperature was below the cave air temperature, indicating the flow of colder air into the cave. Cave air was saturated for most of the year. An air temperature gradient existed between the cave floor and ceiling. This ceiling-to-floor temperature gradient was greatest during the summer.

Sunset Crater Bonito Flow Ice Cave may still be an ice cave, but additional monitoring and analysis are needed. The visible presence of ice at the ice puddle location occurred for about 11 of 12 months in 2021–2022, but perennial ice may still exist just below the rocky cave floor. Floor temperatures never rose above a few degrees Celsius and quickly returned to freezing once cold air from the surface flowed into the cave again.

Abstract

Monitoring of cave environments is an essential process for deciphering records of past climate change preserved in the geochemical composition of speleothems or mineral cave deposits. This study presents data from a multiyear monitoring effort in Titan Cave, Wyoming, a site of interest due to the abundance of speleothems suitable for paleoclimate reconstruction. Titan Cave exhibits annual cave air temperature fluctuations of less than 0.4 degree Celsius, along with consistent relative humidity, drip rate, and partial pressure of carbon dioxide (pCO₂) throughout the year. Small variations in drip rate were noted to be associated with multiseasonal to multiannual regional precipitation trends, such as the widespread western United States drought that lasted from fall 2020 through spring 2022. Stable isotope measurements from drip water (del hydrogen-2 or deuterium [δ²H], del oxygen-18 [δ¹⁸O]) are also relatively constant throughout the year and across different drip sites in the cave, varying by only 2 per mil (‰) in δ²H and less than 0.4‰ in δ¹⁸O. However, stable isotopes (δ¹⁸O, del carbon-13 [δ¹³C]) measured in modern calcite grown on artificial substrates vary spatially and temporally within the cave.

In the Pisa Room of Titan Cave, modern calcite collected from drip sites in the center of the room is more negative in both δ¹⁸O and δ¹³C than modern calcite collected from drip sites along the room’s wall, suggesting differential water flow paths and (or) in-cave disequilibrium effects. The middle of the Pisa Room was identified as the location best suited for future speleothem paleoclimate reconstruction due to the high density of speleothem growth and calcite δ¹⁸O values closer to equilibrium than in other Pisa Room locations. Based on the documented stability of the cave environment and the relative lack of high-resolution paleoclimate data from this region of the northern Rocky Mountains, Titan Cave was found to be a favorable cave for the development of speleothem paleoclimate records.

Introduction

Karst environments hold great potential for preserving records of past terrestrial climate change. Speleothems, or mineral deposits formed in caves over time, can be dated at high precision and preserve a number of climatic signals in their geochemical makeup, revealing aspects of past climate during the speleothem’s growth (Lachniet and others, 2014; Wong and Breecker, 2015; Oster and Kelley, 2016). Speleothems provide the opportunity to reconstruct long-term terrestrial climate change, extending records of past precipitation variability beyond the instrumental and tree ring records (Cheng and others, 2013; Wendt and others, 2018). Geochemical proxies such as ratios of the stable isotopes of oxygen (δ¹⁸O) and carbon (δ¹³C) in the speleothem mineral structure can provide information about past climate variations above a cave. Oxygen isotopes have commonly been used to reconstruct past changes in the δ¹⁸O of precipitation, recording shifts in precipitation intensity, seasonality, temperature, and moisture source region (Bar-Matthews and others, 1997; Tremaine and others, 2011). Carbon isotope ratios are reflective of past shifts in soil respiration, vegetation above the cave, water-rock interactions, prior calcite precipitation, and degassing (Fohlmeister and others, 2020). In many environments, changes in speleothem δ¹³C occur as the result of water availability above the cave, effectively recording wetter vs. drier climate conditions over time (Oster and others, 2020).

Although common climatic and environmental controls on proxies have been documented across landscapes and ecosystems, each cave environment is unique and requires intensive study to effectively translate speleothem geochemical data into useful records of past climate change. Rigorous cave monitoring approaches have proven to be an effective method for understanding location-specific karst processes influencing drip water and speleothem geochemistry (Druhan and others, 2021; Oster and others, 2012, 2021; Sekhon, 2021). These modern monitoring approaches focus on collecting and analyzing cave drip water and calcite precipitated on artificial substrates such as glass plates in the cave, along with recording changes in cave temperature, pCO₂, humidity, and drip rate.

In this study, a comprehensive, multi-year dataset is presented that includes measurements of cave temperature, pCO₂, relative humidity, drip rate, and water and modern calcite stable isotope compositions from Titan Cave (TC), in northern Wyoming. Titan Cave is the first known cave to be monitored in the northern Rocky Mountains and hosts numerous speleothems dated to significant climatic intervals in Earth’s recent past, including the Holocene and Last Interglacial Period. Titan Cave also sits in a region of the Rockies which lacks high-resolution paleoclimate data beyond the tree ring record (Pederson and others, 2011), despite the importance of this information for better understanding long-term climate variability and drought. Results from this study indicate that speleothems from TC are suitable for reconstructing long-term paleoclimate trends due to limited cave ventilation and minimal changes in cave microclimate and drip rate on seasonal timescales. Lastly, intra-cave variations in drip water and modern calcite geochemistry were noted that are anticipated will be present in speleothem paleoclimate reconstructions from TC.

Methods

Site Description

Titan Cave is a wild cave located in northern Wyoming near the border with Montana (fig. 1A). The cave is managed by the Bureau of Land Management (BLM) specifically for scientific research, and the only known previous study at this site consisted of preliminary radon testing (unpublished). The cave is located in the Bighorn River area and is less than 1 kilometer (km) from Natural Trap Cave, a significant paleontological site. Natural Trap Cave hosts a well-documented fossil record spanning the last glacial cycle (Kohn and McKay, 2010, 2012; Meachen and others, 2016), but no speleothem records of past climate change. Titan Cave is situated in the upper section of the Mississippian-age Madison Limestone, which consists of gray limestone and red siltstone paleokarst-breccia and ranges from 330 to 900 feet thick in this region (Sandberg and Klapper, 1967). The entrance to Titan Cave (TC), at 1,427 meters in elevation, sits in a small depression in the Tensleep Sandstone of the overlying Amsden Formation.

Titan Cave receives both cold-season precipitation (predominantly snow) and shoulder season/summer precipitation from convective storms. Moisture in this region is sourced from air masses originating in the Arctic, the Gulf of Mexico, and the Pacific Ocean (Bryson and Hare, 1974). However, high elevation regions can experience significantly different climatologies compared to lower elevation areas. For example, precipitation at high elevation in the Rocky Mountains is generally dominated by Pacific-sourced winter westerly storm systems (Sjostrom and others, 2006), whereas lower elevation areas may receive greater fractions of summertime precipitation due to the influence of the North American Monsoon (Despain, 1987). Precipitation data from Deaver, WY, about 30 kilometers from TC at a similar elevation, indicate that winter snowfall and summer rainfall equally contribute to the annual precipitation budget in this region (Applied Climate Information System, 2024).

Sample collection and cave monitoring at TC takes place in the Pisa Room, roughly 300 meters from the cave entrance (fig. 1B). Because of the geometry of the cave, the Pisa Room is poorly ventilated and sits beneath an estimated 25–50 meters of overlying host rock. The Pisa Room is characterized by thousands of active and dormant speleothem formations (primarily stalactites and stalagmites), ranging from centimeters to over 3 meters in length (fig. 1D). At present day, the cave itself is mostly dry with limited actively dripping sites.

Results

Drip Rate

Drip rate at all monitored sites in the Pisa Room is consistently slow, rarely reaching rates faster than one drip every 2 minutes (fig. 2). Site TC-15 has the slowest drip rate, averaging 0.12 drips per minute (dpm) during the monitoring period, which spans from September 2021 until October 2023, with a short hiatus due to a dead drip logger battery. TC-15 drip rate is nearly constant, excluding a slight positive trend toward faster drip rates from June through August 2023. During the overlapping period in the two records, trends in drip rates are nearly identical at sites TC-1 and TC-2, however, mean drip rate is roughly twice as fast at TC-1 (0.47 dpm) compared to TC-2 (0.20 dpm) (fig. 2). Drip rates at TC-1 and TC-2 show high correlation (Pearson’s R = 0.91, p <0.0005), displaying nearly synchronous changes on hourly to daily timescales. This strong coherence is unsurprising as the sites are only about 2 meters apart. TC-1 drip rate declines steadily during the monitoring period, decreasing by roughly 0.05 dpm per year from 2019 to 2023. TC-1 drip rate begins decreasing less rapidly in Summer 2022, synchronous with increasing summer rainfall in Deaver, WY (fig. 2).

Water Isotopes

Pisa Room drip water δ¹⁸O values range from −20.26 per mil (‰) to −19.9‰, and δ²H values range from −158.94‰ to −157.07‰ (fig. 3). Water collected for δ¹⁸O at 4-day intervals using the autosampler at site TC-2 ranges between only −20.26‰ and −20.16‰ from May 2022 to October 2023. Considering the analytical uncertainty of 0.02% (1σ), the variability in site TC-2 drip water δ¹⁸O appears to be extremely limited (fig. 4). Limited inter-site δ¹⁸O variability within the Pisa Room was also noted. Drip water δ¹⁸O collected instantaneously at sites TC-1 and TC-15 in October 2023 are identical within analytical uncertainty (−20.06‰ and −20.08‰), and instantaneously collected drip water from TC-1 in September 2021 is also similar (−20.02‰). Drip water collected near the cave entrance (site MT drip on fig. 3) in September 2021 and October 2022 is notably more enriched in ¹⁸O, with values of −18.19‰ and −17.75‰ measured, respectively. Two snow samples collected from above TC in October 2023 have an average δ¹⁸O value of −23.90‰ and a δ²H value of −175.54‰.

Modern Calcite Stable Isotopes

Modern plate calcite δ¹³C values range from −6.11‰ to 0.65‰ and δ¹⁸O values range from −17.33‰ to −15.43‰ (figs. 5 and 6; table 2). Within the Pisa Room, systematic variations in modern calcite δ¹³C and δ¹⁸O were noted, based on drip site location (fig. 5), in addition to changes with time (fig. 6). Sites TC-1 and TC-2, located in the center of the Pisa Room, record more negative calcite δ¹³C and δ¹⁸O values compared to sites TC-13 and TC-15 which sit closer to the cave wall.

Table 2.    

Titan Cave plate calcite stable isotope data.

[δ¹³C, del carbon-13; δ¹⁸O, del oxygen-18; ‰ VPDB, per mil Vienna PeeDee Belemnite]

Table 2.    Titan Cave plate calcite stable isotope data.

Site	Start Date	End Date	δ13C (‰ VPDB)	δ13C uncertainty (1σ)	δ18O (‰ VPDB)	δ18O uncertainty (1σ)	Location in Pisa Room
TC-1	9/1/19	9/15/21	−4.96	0.06	−16.44	0.08	Middle
TC-1	9/15/21	5/27/22	−2.29	0.02	−17.04	0.04	Middle
TC-1	5/27/22	10/13/22	−3.19	0.02	−16.93	0.04	Middle
TC-1	10/13/22	6/5/23	−3.67	0.02	−16.69	0.07	Middle
TC-1	6/5/23	10/26/23	−3.96	0.03	−16.86	0.03	Middle
TC-2	9/1/19	9/15/21	−3.33	0.03	−16.95	0.07	Middle
TC-2	9/15/21	5/27/22	−2.86	0.06	−16.95	0.09	Middle
TC-13	9/15/21	5/27/22	−1.37	0.09	−16.36	0.10	Slope
TC-13	5/27/22	10/13/22	−2.72	0.08	−15.9	0.03	Slope
TC-13	10/13/22	6/5/23	−1.7	0.04	−15.53	0.04	Slope
TC-15	9/15/21	5/27/22	−0.26	0.06	−15.81	0.09	Slope
TC-15	5/27/22	10/13/22	−1.11	0.04	−15.43	0.04	Slope
TC-15	10/13/22	6/5/23	0.65	0.05	−15.65	0.07	Slope
TC-15	6/5/23	10/26/23	−0.63	0.04	−15.82	0.03	Slope
TC-16	5/27/22	10/13/22	−5.03	0.03	−17.33	0.07	Front
TC-16	10/13/22	6/5/23	−5.06	0.09	−16.32	0.07	Front
TC-16	6/5/23	10/26/23	−6.11	0.06	−16.83	0.09	Front

Discussion

The suite of monitoring data collected from Titan Cave across multiple years allows for the systematic and cave-specific interpretation of geochemical variability recorded in drip water and modern plate calcite at this site. In this section, individual monitoring datasets from TC are discussed. These data are then summarized and synthesized as it relates to interpretations of past climate change in the Bighorn region as recorded via TC speleothems.

Conclusions

This study presents data from the first known comprehensive, multi-year cave monitoring campaign in the Rocky Mountains in the western conterminous United States. The dataset includes measurements of cave temperature, relative humidity, pCO₂, drip rate, and water and modern calcite stable isotope compositions from Titan Cave, Wyoming. Monitoring efforts reveal a remarkably stable cave environment, with consistent cave temperature, humidity, and pCO₂ throughout the year. Small fluctuations in drip rate were observed that are consistent with multi-seasonal to annual trends in precipitation, along with slight shifts in drip water and plate calcite stable isotope values on similar timescales. Modern plate calcite δ¹⁸O and δ¹³C values show intra-cave variability along with a systematic trend away from isotopic equilibrium conditions in specific areas of the cave. Thus, the middle of the Pisa Room was identified as the location best suited for future speleothem paleoclimate reconstructions due to the high density of speleothem growth and calcite δ¹⁸O values closer to equilibrium than at other Pisa Room locations. Because of the documented stability of the cave environment and abundant suitable stalagmites, Titan Cave was found to be a favorable cave for the development of speleothem proxy records.

Abstract

On Earth, caves are unique environments at the intersection of geology, climate, and biology. Given that the same terrestrial speleogenetic processes exist throughout the solar system, it would be surprising if caves beyond Earth did not exist. Thousands of potential cave entrances (or subsurface access points) have been identified from Earth’s Moon to Pluto’s moon, Charon. To date, our most comprehensive knowledge of these potential subsurface access points is for the Moon, Mars, and Titan, which collectively contain more than 20,000 features. Missions are either ongoing or planned for these three planetary bodies. One of these missions may ultimately detect a cave and potentially confirm it contains a laterally trending passage.

Introduction

There are multiple definitions of caves in the scientific literature, but for purposes of this abstract caves are defined as a subsurface void that is potentially accessible from the surface. On Earth, caves are unique subsurface environments at the intersection of geology, climate, and biology. The subterranean realm contains critically important habitats for a panoply of lifeforms including bats (Furey and Racey, 2016), subterranean-adapted fauna (Howarth and Wynne, 2022; Deharveng and others, 2024), relict plant species (Wynne and others, 2014; Monro and others, 2018; Ren and others, 2021), and microbes (Tomczyk-Żak and Zielenkiewicz, 2016; Hershey and Barton, 2018). Caves are also time capsules of the past, enabling researchers to reconstruct past geological processes and environmental histories (Wong and Breecker, 2015; Comas-Bru and others, 2019), as well as serving as treasure troves of paleontological (Jass and George, 2010; Schubert and Mead 2012; Berger and others, 2015; Douka and others, 2019) and archaeological (Moyes, 2012; Sponsel, 2015; Brady and Prufer, 2021) materials.

Given that the same fundamental speleogenetic processes on Earth exist throughout the solar system, it would be surprising if caves beyond Earth did not exist. Scientific speculation about caves on the Moon date back to 1966 (Heacock and others, 1966; Halliday, 1966), when images of rilles were suggested to be surface expressions of lava tubes. These papers were followed by additional studies on lunar lava tube formation processes (Oberbeck and others, 1969; Greeley, 1971), and were later expanded to include Mars using Viking data (Carr and others, 1977).

In 2007, with the advent of high spatial resolution imaging of the surface of Mars, seven deep cylindrical pits were discovered in the volcanic region of Arsia Mons (Cushing and others, 2007). Although these features have yet to be confirmed as actual cave entrances (providing access to horizontally extended subsurface voids), they did mark a notable milestone in planetary cave science.

The same year, 2007, was also the advent of a series of planetary cave workshops and conferences that brought together an interdisciplinary and international group of scientists and engineers with the common goal of studying caves on Earth and beyond. The first workshop was the Lava Tubes: Earth–Moon–Mars workshop convened at El Malpais National Monument, Grants, NM. A list of all workshops and conferences since 2007 are provided in table 1.

Planetary Cave Goals and Objectives

In the last Planetary Science Decadal Survey, a white paper focusing on planetary caves (Titus, Wynne, Boston, and others, 2021) proposed a series of goals and objectives for this emerging sub-discipline (table 2). Although these goals and objectives will likely be updated during the next International Planetary Cave Conference, Titus, Wynne, Malaska, and others (2021) provided a viable framework for the path forward. In addition, Wynne, Mylroie, Titus, and others (2022) conducted a survey of the planetary cave research and exploration community to identify the highest priority planetary cave science and engineering questions that should be addressed over the next few decades. We aligned the existing objectives with the appropriate top 15 questions; many of the questions are technology- and capability-based, as opposed to purely focused on science (table 2). This suggests that the expansion of the goals and objectives document should not only include science investigations, but also technology development and testing activities.

Planetary Cave Formation Processes

Our understanding of cave formation processes starts with an understanding of terrestrial caves and their associated speleogenetic processes that initiate and drive their modification. Table 3 describes cave formation processes—most of which occur on Earth. The first step in identifying where caves may exist beyond Earth begins with low-resolution imaging that shows the morphologies of various planetary landscapes are consistent with cave formation processes. As such, this represents the key starting point for any planetary cave inventory.

Exploration Framework: Identify, Characterize, Explore!

To identify potential caves, landscape mapping and characterization images must be assessed to determine whether speleogenetic processes may exist on a given planetary body. This can be accomplished using low-to-medium spatial resolution optical or infrared imagery for bodies with no or relatively transparent atmospheres, and with radar for those bodies with thicker, partially opaque atmospheres. For most rocky and icy planetary bodies, we are at this stage of knowledge. As higher spatial resolution imagery becomes available, as is currently the case for the Moon and Mars, these data could be used to identify potential cave entrances, partially depressed segments, or other surface indicators.

Characterization is the next step in the proposed framework (table 4 and fig. 1). This can range from imaging confirmation that a potential cave entrance visually connects to a lateral subsurface passage to the use of methods, including geophysical surface remote sensing techniques (for example, ground penetrating radar (GPR), electrical resistivity, seismic tomography, or gravity measurements) to map the extent of the feature.

Exploration is the final phase and ranges from robotic entry and examination of the interior of the cave to long-term monitoring of climatic, seismic, and radiation environments. Ultimately, this will include human exploration and use of caves beyond Earth (Boston, 2000, 2010).

Solar System Inventory

Up to this point, the word “cave” has been used to refer to a subsurface void that is potentially accessible from the surface. Wynne, Mylroie, Titus, and others (2022) introduced the term subsurface access points (or SAPs) (fig. 1). The authors defined an SAP as an opening on the surface of a planetary body detectable by remote sensing; this definition is more compatible with the current remote stage of planetary cave research and exploration. This term captures features ranging from deep fissures to atypical pit craters, which may not be immediately perceivable as a fully developed cave but may have undetected lateral passage associated with it.

Cave formation processes exist across the solar system from Mercury to Pluto (Wynne, Mylroie, Titus, and others, 2022). Thousands of potential cave entrances (or SAPs) have been identified from as close as our own Moon to as far away as Pluto’s moon, Charon. To date, our most comprehensive knowledge of SAPs is for the Moon, Mars, and Titan, which support 221, 1,062, and 21,706 features, respectively (fig. 1 and table 5).

Summary

The landscapes observed on most planetary bodies across the solar system are consistent with several speleogenetic processes. As such, planetary caves likely exist from Mercury to Pluto. For planetary bodies such as the Moon and Mars, where very high spatial resolution (meter to sub-meter resolution) is available, numerous compelling SAPs have been identified; many of these features may provide access to lateral subsurface voids. Although sufficiently high spatial resolution imagery is not available for Titan, medium resolution (about 200 meters/pixel) radar imagery has identified tens of thousands of locations where the surface is consistent with speleogenetic processes. The Moon, Mars, and Titan have either active missions or planned launches scheduled for this decade. Although none of these missions will specifically search for and identify caves, they will provide additional remote sensing assets to characterize and monitor the surface. Perhaps one of these missions to the Moon, Mars, or Titan, will find an SAP and confirm that it truly extends into a horizontal subsurface void.

Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Abstract

Karst groundwater basins have historically been delineated by qualitative or quantitative fluorometric tracer tests (Jones, 2019). Dye tracer testing is by far the most common tool applied by karst groundwater scientists not only to assess groundwater flow directions and flow velocities, but also to delineate groundwater basin boundaries between groundwater basins. Although dye tracer testing to delineate contributing areas to karst springs is a commonly used and effective technique, a recent study completed by the U.S. Geological Survey in Monroe County, West Virginia, utilized a hybrid approach for delineating karst groundwater basins (Kozar and others, 2023) that coupled dye tracer tests with hydraulic gradients obtained from a county-wide potentiometric-surface contour map (Kozar and others, 2023).

Previously conducted dye tracer tests provided 20 subsurface connections between injection sites and monitored resurgences and were used to delineate contributing areas for the major karst basins in Monroe County, West Virginia (Jones, 1997). Two new tracer tests were also conducted for the recent Monroe County hydrogeologic study to better delineate the groundwater divides between the northern and southern portions of the Greenbrier Group karst aquifer in Monroe County (Kozar and others, 2023). A potentiometric-surface map was created for Monroe County by coupling the locations of major springs with 260 groundwater-level measurements taken in domestic wells across the study area. Land-surface elevation for the potentiometric-surface map was based on a county-wide digital elevation model (DEM) developed for the study (Cox and Doctor, 2021), and the resulting potentiometric-surface map was constrained by the National Hydrography Dataset (https://www.usgs.gov/national-hydrography/national-hydrography-dataset) and contoured at 50-foot intervals with an accuracy of plus or minus 25 feet.

This hybrid approach, which coupled delineations of karst watersheds using dye tracer test methods with hydraulic gradients and groundwater divides derived from the potentiometric-surface map was used to update karst basins previously delineated by Jones (1997; fig. 1). Dashed lines on figure 1 show the previously delineated basin areas for the major groundwater basins within Monroe County, West Virginia. The dark black lines show the revised groundwater basin delineations based on the additional hydraulic gradient data provided by the potentiometric-surface map developed for Monroe County.

Even though the potentiometric-surface map is based on combined hydraulic heads potentially from 1 or more water-bearing zones from the 260 groundwater-level measurements made for the recent study, more detailed delineations of the karst groundwater basins were made possible by including the hydraulic gradient and groundwater divides represented on the county-wide potentiometric-surface map. There was good agreement among the surface stream network, the dye tracer tests, the potentiometric-surface contours, and the redefined groundwater basins. However, as the area is heavily karstic, even the redefined groundwater basins should be regarded as approximate given the complex subsurface flow paths common in karst aquifers.

Abstract

Karst aquifers are important drinking water resources in Tennessee. Groundwater in karst landscapes is susceptible to contamination due to the high level of surface and groundwater interaction. Communities create Source Water Protection Areas (SWPAs) for drinking water springs that are typically estimated, rather than delineated by a formal study. A partnership between the U.S. Geological Survey (USGS) and Tennessee Department of Conservation (TDEC) was formed to investigate karst springs, particularly those used as sources of drinking water, using fluorescent dye tracing. USGS and TDEC staff identified possible vulnerabilities that may exist within public drinking water systems based upon maturity of karst development, underlying geology, and uncertainties related to estimated recharge areas. Work began in the fall of 2022 in the community of Vanleer, located in Dickson County along the Western Highland Rim (physiographic section) of Middle Tennessee. Mississippian St. Louis and Warsaw limestones are the primary strata in this well-developed karst area. Initial dye injections provided results that highlight the need to improve the current estimated SWPA. Results from work in Vanleer continue to provide a better understanding of groundwater movement and flow path direction and give insight into surface and groundwater interactions.

Abstract

Results of a dual-tracer test conducted at Pah Tempe Hot Springs near Hurricane, Utah, from mid-January 2021 to mid-April 2021, indicate that dye injected down a well discharged from the main spring (pool) and from several other springs along the Virgin River downstream from the main spring. Dye recoveries from activated charcoal placed in the springs were erratic over the course of the monitoring period and appear to have resulted in large part from highly variable adsorption of the dye onto the charcoal due to interference from bacteria and cyanobacteria that thrive in the spring water. The erratic recoveries are also likely influenced by the complex heterogeneity of the groundwater flow system in the vicinity of the Hurricane fault zone. On the basis of the relative fluorescence (intensity) observed from analysis of activated charcoal detectors, most of the dye appears to have discharged at Upper and Lower Aqueduct, Boulder, and PVC springs on the north side of the Virgin River, and Cabin springs on the south side of the river. Although dye was recovered at Pah Tempe Springs (main pool) within the first week after the injection, dye recovery from Upper and Lower Aqueduct springs occurred throughout the entire 3-month monitoring period. Dye was also recovered in all samples collected from Boulder, PVC, and Cabin springs; however, monitoring at these springs only occurred during the latter half of the study. Dye recoveries at selected springs in the downstream parts of the study area indicate that groundwater movement in the vicinity of the well is likely along preferential flow path(s) (solution-enhanced fractures and (or) faults) within the Hurricane Fault damage zone that trend sub-parallel to and likely under the Virgin River.

Continuous water samples also were collected from the main pool of Pah Tempe Springs and downstream in the Virgin River throughout the study period and analyzed for both fluorescein dye and bromide. Results of analysis of samples collected from the main spring pool appeared to be negative for dye and generally within the range of concentration for background samples collected prior to the dye injection, even during the time frame that charcoal detectors from the spring were positive for dye. Results of analysis of bromide samples from the spring also reflected natural background concentrations of bromide in the water. Results of analysis of samples collected from the Virgin River downstream from the springs and analyzed for the dye also indicated very low concentrations overall, but several periods of slightly higher values appear to coincide with activated charcoal samples collected during those same periods and considered positive. Similarly, concentrations of bromide in water samples collected from the river downstream from the springs also reflected natural background concentrations in the first part of the monitoring period. However, a notable increase in bromide concentration in the river above the natural background concentration occurred after about February 10, which may represent in part, bromide that was injected down the well. The lack of distinct breakthrough curves for both tracers at the main spring and downstream in the Virgin River appears to reflect a complex, heterogenous fracture network between the injection well and the springs that influences the movement, direction, and discharge points of tracers from the groundwater system.

Introduction

Pah Tempe Hot Springs (Dixie Hot Springs, LaVerkin Hot Springs) are located along the Virgin River near Hurricane, Utah, and issue from fractured Permian-age limestones of the Toroweap Formation. The springs discharge primarily from fractures in an alcove (main pool) along the south wall of Timpoweap Canyon (fig. 1) and from both banks of the river along a 1,500-foot reach of the channel (fig. 2) on the east side (footwall) of the northeast-trending Hurricane Fault zone (Gardner, 2018; Godwin and others, 2021). Springs also rise from multiple points in the riverbed. The springs discharge from as much as 12 feet above river stage at base flow, and total spring discharge along the reach averages about 11.5 cubic feet per second (ft³/s). Discharge from the springs has an average dissolved-solids concentration of about 9,650 milligrams per liter (Bureau of Reclamation, 1981) with a total dissolved-solids load averaging about 99,000 tons per year (Gerner and Thiros, 2014). Average water temperature of the springs is about 40 degrees Celsius.

In May–July 2020, a reconnaissance dye-tracer test was conducted to determine the discharge point of water moving through a Bureau of Reclamation well along the Virgin River upstream from Pah Tempe Hot Springs (fig. 2). On May 5, 2020, 900 grams of fluorescein (uranine) dye, a commonly used dye for groundwater tracing in karst terrains and shown to be successfully used as a tracer in geothermal systems (Adams and Davis, 1991), were injected down the well to a depth of about 190 feet, a zone of preferential flow into fractures. The dye was pre-mixed with river water in a 5-gallon bucket, then injected down the well using a peristaltic pump at a rate of about 1 liter/70 seconds. During the test, eight springs were monitored along the north and south sides of the Virgin River downstream from the injection well, as well as outflow from a group of pipes near the injection well (fig. 2). The Virgin River was also monitored downstream from all spring inflow. Total discharge from all springs during the dye test was approximately 10 ft³/s, and flow in the Virgin River was about 300 ft³/s.

Results of analysis of activated charcoal detectors collected over the 3-month period after injection showed a recovery of the dye in the main (alcove) and outflow pools of Pah Tempe Hot Springs (fig. 1). Possible or likely recoveries of the dye also were made at six other springs along the north bank. Dye was also detected in the Virgin River downstream from the springs. Straight-line distance to the main pool of the hot springs from the injection well is approximately 1,066 feet (325 meters). Maximum groundwater travel time to the main spring pool was about 9 days based on recovery of the dye from the initial detectors pulled from the spring, indicating that groundwater movement is likely along fractures and (or) conduits (solution-enhanced fractures) within the carbonate-rock aquifer from which the spring discharges.

On the basis of results obtained from the dye-tracing study in May–July 2020, a second follow-up study was initiated in January 2021. The objectives of this study were to (1) repeat the 2020 tracer test by conducting a dual qualitative and quantitative tracer test using fluorescein dye and sodium bromide, (2) determine the fractions of water in the well that discharge at the main pool of Pah Tempe Hot Springs, (3) determine a more precise time of travel between the injection well and the spring, as well as residence time of the tracers in the groundwater system, and (4) verify the discharge points at springs where dye was detected during the 2020 tracer test, and whether other springs along the Virgin River are also discharge points for the tracers.

Approach/Methods

On January 19, 2021, a dual-tracer injection was made in the Bureau of Reclamation well that was used for the May 2020 test. For this test, both sodium bromide (NaBr) and fluorescein (uranine) dye were pre-mixed with water pumped from the Virgin River into a large tank and then gravity-fed down the well through a large-diameter hose into the injection zone at approximately 190 feet from land surface. The sodium bromide was added to the tank as a salt and stirred until it was totally dissolved in the water. The dye powder was pre-mixed in a 5-gallon bucket, then poured into the tank with the bromide where they were injected as a co-mingled solution. The injection was conducted over a period of about 5 hours, but as periodic slugs. Overall, approximately 115 kilograms of sodium bromide were injected down the well in conjunction with 2 kilograms of sodium fluorescein dye.

The tracer test was partitioned into qualitative and quantitative components. For the qualitative component, activated charcoal detectors or packets were used for adsorption of the dye to determine discharge points. For the quantitative component, the fluorescein dye as well as the injected sodium bromide were analyzed in water samples that were collected on a periodic basis from ISCO automatic samplers at the main Pah Tempe Springs rise pool and downstream in the Virgin River below the input of all springflow. Water sample analyses can be used to determine a more accurate time of travel from the well to the main spring and to establish breakthrough curves for the individual tracers that can be compared.

Activated charcoal detectors were placed at most of the same springs used in the 2020 test, with several additional (previously unknown) springs also incorporated that were found later during the monitoring period. In total, 12 springs were monitored along with outflow from a set of pipes that was monitored during the 2020 test. The Virgin River was also monitored downstream from all inflow from the springs on the north and south sides of the channel, as well as springs rising in the bed of the channel that could not be individually monitored. Three days prior to the injection on January 19, charcoal detectors were placed in all springs and in the Virgin River for the purpose of determining the relative intensity of background fluorescence from natural organic materials in the water that can potentially mask detection of the injected dye (Alexander, 2005). This set of background detectors was pulled and replaced with another set of detectors in the morning of the day of the injection. Sites monitored with charcoal detectors were assigned unique names for this study, and from upstream to downstream along the Virgin River (fig. 2) were as follows:

The dye was recovered on 6–12 mesh activated charcoal that was placed into fiberglass screen packets suspended from concrete weights or attached to tent stakes placed in or just downstream from the spring discharge points. In most cases, duplicate packets were employed to serve as back-ups if one of the packets was damaged or lost and also as a verification of the result from its companion packet. The dye packets were left in situ for a period that correlated with the exchange of bottle trays in the autosamplers (refer to section on quantitative component of study) every 3 days initially, extending to longer intervals as the monitoring period progressed. At the end of each time interval, the detectors were exchanged with new ones.

The detectors were transported back to the Utah Water Science Center (UTWSC) laboratory in Salt Lake City where the dye was extracted from the activated charcoal with a 5-percent solution of 70-percent isopropyl alcohol and potassium hydroxide, a commonly used eluent for elution of dyes (particularly fluorescein) from charcoal (Quinlan, 1989). Presence of the dye in some samples was determined by visual observation if the concentration was high enough. Most samples, however, were analyzed on a Shimadzu RF-5301PC scanning spectrofluorometer. Results of analysis were reported qualitatively in terms of negative (no dye was detected), positive (dye was observed instrumentally), very positive (dye was observed visually in the solution) or potentially positive (dye response meets certain criteria but not able to be corroborated). Evaluation of the results of analysis of the scanned samples for the presence of dye is largely based on the protocols used by Crawford Hydrology Laboratory (2019) at Western Kentucky University.

In addition to the use of activated charcoal for the adsorption of dye during the study, two ISCO 3700 series automatic water samplers were employed to collect water samples for both the dye and sodium bromide on a timed basis. These samplers were installed at the Pah Tempe Springs main pool and downstream in the Virgin River at the same location as the charcoal detectors (fig. 2). The autosamplers were programmed to collect 400-milliliter (mL) samples on a 2-hour basis at the start of the test, but this interval was increased later in the monitoring period to 4, 6, and 8 hours until the end of the study. In addition, the autosamplers were set up on an offset schedule to enhance the sampling coverage throughout the monitoring period. Three days prior to the dye injection, samples were also collected on a 2-hr basis to determine the amount of natural fluorescence in the water that could potentially interfere with the detection and analysis of the fluorescence attributed to the dye. Unlike the charcoal detectors which can only be used to establish hydraulic connections from input to output, the purpose of the autosamplers was to obtain a record of the travel time of the dye and the bromide as the tracers moved from the well to the main spring and downstream in the river below the other spring inputs. This would result in breakthrough curves for each of the tracers that could be compared and from which other data could be extracted such as first arrival of the tracer at the sampler, time to peak arrival and concentration, and residence time of the dye and bromide in the groundwater system. Although the adsorption of dye onto activated charcoal is a passive, cumulative process, water samples represent the instantaneous (real-time) concentration of dye in the water.

At the end of each sampling period after 24 samples had been collected, the bottle trays were replaced with a new cleaned bottle tray, and the sampler was re-programmed for the next sampling cycle. The bottles were subsequently moved to an onsite cabin where the samples were split into 250-mL and 125-mL polyethylene bottle subsamples for analysis of the fluorescein dye and bromide, respectively. Samples for dye analysis were transported to the UTWSC laboratory in Salt Lake City where they were analyzed on a Turner Designs TD700 series filter fluorometer. Samples for bromide analysis were transported to the University of Utah Department of Geology and Geophysics laboratory where they were analyzed by ion chromatography.

Results of Analysis and Discussion

Activated charcoal has a strong affinity for the sorption of dyes; however, that sorption can be highly variable, sometimes producing substantially different or inconsistent values that can be challenging to evaluate with respect to whether the sample is positive for dye. These factors include the length of time the charcoal is in the water, velocity of the water, and the location of the charcoal in the spring or stream. Dye recoveries at the monitored sites varied between detectors placed near each other at the same location, and over the period of monitoring, which increased over time. Velocities also ranged from slow (pools) to greater than 1 foot/second. As a result, dye recoveries could not be verified at some sites or at selected times. In most cases dual detectors were placed at each of the monitoring sites, and thus, would be expected to yield similar results with respect to adsorption of the dye. The variability in dye recovery observed throughout the monitoring period, however, is interpreted to be primarily the result of a heavy coating of white, non-photosynthetic filamentous bacteria and cyanobacteria (fig. 1) (Hannah Bonner, Utah Division of Water Quality, written commun., June 12, 2024) on the packet itself, which impeded the sorption of dye onto the charcoal within the packet. The bacterial coating appears to have occurred very rapidly (within hours) of placement of the detectors in the springflows, resulting in a substantially reduced window for adsorption of the dye. Coatings on detectors that were only 1 foot apart (opposite ends of the weighted assembly used to suspend the detectors) also were observed to be variable in some instances, resulting in a substantial difference in values of fluorescence and thus dye adsorption between the two detectors. In addition, positive dye recoveries were obtained intermittently at some sites, which again, may have resulted from variable adsorption resulting from bacterial coatings, or possibly other factors influencing movement of the dye through the fractured network from the well to the springs.

Results of Water Sample Analysis

Results of analysis of water samples collected from the Pah Tempe main spring pool were characterized by dye values that ranged from only 0.01 to 0.06 parts per billion (ppb). Because this range in values was also within the range in values for natural background fluorescence in the spring water prior to the dye injection, no samples could be considered unequivocally positive for dye, even during the first week after injection when charcoal samples were positive. This phenomenon may be explained in part, by the very low concentrations of dye in the water that were not detectable in an instantaneous sample taken every 6 hours, compared to the cumulative adsorption of the low-level concentrations of the dye on activated charcoal over a multi-day period. Results of analysis of water samples collected from the Virgin River downstream from the spring inputs were characterized by values that ranged from 0.03 to 0.13 ppb. Values of background fluorescence in the river water prior to dye injection also overlapped values of samples collected after the injection, generally falling between 0.04 and 0.08 ppb. Nonetheless, values of samples with slightly higher fluorescence (greater than or equal to about 0.12 ppb) appear to coincide with activated charcoal samples collected during the same periods (January 27–29, January 31–February 2, and possibly after March 11) that were positive. Positive recoveries from charcoal collected at Pah Tempe main spring on January 25 are also likely reflected in samples collected at the Virgin River sampler on January 24. Overall, results of analyses of water samples collected from Pah Tempe main spring and to a lesser degree from the Virgin River downstream from the springs, cannot be represented as distinct breakthrough curves, but rather as erratic responses to the dye as it discharged from the groundwater system.

Results of analysis of bromide at Pah Tempe main spring and in the Virgin River yielded similar results. Bromide concentrations in the main spring pool remained essentially at background levels throughout the monitoring period, ranging between 3.1 and 3.3 parts per million (ppm), which represents the natural concentration of bromide in the water discharging from the main spring (fig. 3). Bromide concentrations in the Virgin River below the inflow of all springs also were typically at background levels (1.5 to 1.75 ppm) for the first part of the monitoring period until about February 3, after which concentrations decreased abruptly to less than 1 ppm. Concentrations subsequently rose to levels slightly above previously measured background levels (2.0 ppm), after which concentrations again increased slightly to between 2.2 and 2.3 ppm, then stayed constant for the remainder of the monitoring period (fig. 3).