Stream capture from below: formation of the Sinkhole Plain, Kentucky, through the lens of relict topography

Rachel Bosch and Dylan Ward

Figure 1. Geologic map of field area in Central Kentucky Karst. Large red and purple dots indicate geochronology sampling locations. Yellow dots indicate notable landforms. Blue lines follow inferred subterranean routes from dye-trace studies. Orange line is match line for cross-section in Fig. 2.

Abstract

This work extends the established geochronology of the Mammoth Cave region, Kentucky, USA, spatially and temporally, to infer evolution of the karst landscape and to consider the shifting drainage basins of the Barren River and the Green River in relation to regional drainage rearrangements. Previous studies have focused on the Mammoth Cave System and used cosmogenic radionuclide dating to link the incision history of the Green River and the Cave as far back as 3.25 Ma. Before that period, drainage likely consisted of stream networks over clastic bedrock. During early karstification in this region, caves developed in the Girkin limestone (elevation >250 m ASL) in the area that is now ~10 km south of the Green River and ~30 km east of the Barren River. We hypothesize that this occurred about 1–10 Ma before the development of the oldest passages in Mammoth Cave, and that the paleo cave system was part of a karst drainage feeding the Barren River to the west. We used burial and exposure cosmogenic dating of sediments collected from Crystal Onyx Cave in Prewitts Knob and from Little Sinking Creek in the Sinkhole Plain to constrain the age of this stage of karst development, to provide an estimate of the long-term erosion rate of the Sinkhole Plain. These analyses provided dates to determine the abandonment age of Crystal Onyx Cave and provide indication of the average modern rate of Sinkhole plain erosion. These results will place the development of Crystal Onyx Cave and the denudation of the Sinkhole Plain in the context of the Mammoth Cave System geochronology. This has implications for revealing timescales of the evolution of karst systems as well as the interaction of hydrological flow paths in karst areas with the transience of drainage networks. A major research aim in geomorphology is the development of a comprehensive, mechanistic theory of landscape evolution that can account for the earth’s primary topographic forms and their response to forcing factors such as climate change and human activity. Karst landscapes represent a key gap in that knowledge base, and one that impedes our ability to fully understand, interpret, and conserve karst landscapes. By taking advantage of cutting edge geochronological techniques and computational modeling, this project will help fill that gap.

Figure 2. Geologic cross-section along orange match-line in Fig. 1.

Scientific questions

Mammoth Cave and other regional caves consist of a series of discrete cave passage levels, which correlate to fluvial terraces of the Green River. (Palmer, 1989). A key study by Granger et al. (2001) used the differential decay of cosmogenic 10Be and 26Al to date clastic sediments in the different levels of Mammoth Cave, quantitatively constraining the downcutting rate of the Green River over the last 3.5 Ma. Their work placed the highest two levels of Mammoth Cave as having formed >3 Ma. These levels underwent at least two more aggradation events prior to downcutting from level B to level C at 1.92 ± 0.10 Ma. Incision from C to D occurred at 1.39 ± 0.14 Ma. Incision below level D began at 1.24 ± 0.06 Ma, however level D is not considered to be abandoned and has evidence of aggradation timing to 0.7 – 0.8 Ma. Before exhumation of the St. Genevieve and St. Louis Limestones, Central Kentucky was dominated by a fluvial landscape with the dominant drainage basin being toward the Barren River. As the Green River incised, it lowered base level for the Barren River, which served as a connection between the Green and the Pennyroyal Plateau on the other side of the Chester Cuesta. The Big Clifty became breached further and further downstream (Quinlan and Ewers, 1989), soluble limestone layers were encountered, and drainage from the surface by karst development to springs on the Green became a more efficient flow path. With the development of the Turnhole Spring, the third largest karst spring in Kentucky, extensive conduit development was enabled, more surface streams were diverted to underground paths, and sinking streams with surface trends that point them toward the Barren River were pirated by sinkholes to drain into the Green River at Turnhole Spring.

We hypothesize that the trunk passages that are referred to as “knob caves” were previously connected in a now-disrupted ancestral cave system, and that they were cave levels higher in elevation and older than the modern-day level-A passages described by Palmer (1989) in the Mammoth Cave System. This work addresses these scientific questions: How old are the passages in Crystal Onyx Cave and what is their chronological and hydrological relationship to the existing cave levels of the Mammoth Cave System? Are the passages in Crystal Onyx Cave older than Collins Avenue in Mammoth Cave, supporting our hypothesis that Crystal Onyx Cave represents a previous phase of karstification? Or was the development of Crystal Onyx Cave contemporaneous to that of Collins Avenue, and therefore other level-A cave passages? Did the water that flowed through Crystal Onyx Cave drain to the north into the Green River, or to the west into the Barren River? In the process of this karst landscape evolution, how much material was removed from above the modern Sinkhole Plain, and how long did that denudation take? Dating sediments from Crystal Onyx Cave and the Sinkhole Plain to inform our numerical modeling will help to answer these questions.

Methods

To test the hypothesis that the trunk passages of Crystal Onyx Cave represent a relict portion of the ancestral Sinkhole Plain Cave System, and that its formational waters may have fed the Barren River, we have performed analysis of digital elevation models and collected sediment samples for cosmogenic absolute dating to quantitatively constrain our topographic analyses. By digitally reconstructing the paleo landscape and using our geochronology results to constrain this model, we will produce a more quantitatively robust model of the paleo Central Kentucky Karst landscape. We will relate modern erosion rates based on data from Little Sinking Creek (Fig. 3) in comparison with the lowering rate based on the cave age data from Crystal Onyx Cave (Fig. 4).

Figure 3. Rachel Bosch collects modern sand from a gravel bar deposit in Little Sinking Creek.

Figure 4. Siliciclastic sediment deposit in Crystal Onyx Cave.

Mapping Results and Discussion

The surface drainage basins for the Upper Green River and the Barren River were determined by routing flow over a DEM using LSDTopoTools (Clubb et al., 2017). When comparing these surface basins with the karst groundwater basins delineated using dye-trace techniques (Quinlan and Ray, 1989; Toomey, 2019), it was observed that the groundwater divide between the Green River and the Barren River is to the west and the south of the surface divide in the vicinity of the headwaters of the Turnhole Bend karst drainage basin (Fig. 5). This discrepancy in drainage basin boundaries correlates well with Chi mapping results (Fig. 6), also performed using LSDTopoTools. Here we see higher chi values in the headwaters of the Barren River surface basin in that same area, indicating that the Green River Basin is pirating water from the Barren Basin. We interpret this as the piracy having begun underground when the Turnhole Basin began capturing water from the Graham Springs Basin. This raises the question: what is the time difference between the expression of this drainage rearrangement underground and the transmission of that signal to the surface? Associated questions: What is the rate of migration of the drainage divide? How do karst drainage networks respond to regional drainage rearrangements? And how do flow transience signals get transmitted through karst networks and to the surface networks that feed them?

Figure 5. Comparison of surface and subsurface drainage basins over the Sinkhole Plain, Kentucky, USA. Blue lines indicate underground flow paths inferred from dye tracing.

Figure 6. Chi map of the Upper Green and Barren River Basins, Central Kentucky, USA.

How has this advanced our thinking?

Crystal Onyx Cave, which sits right on the boundary between these two major drainages (Fig. 6), likely drained westward to the Barren River when it was part of a hydrologically active karst conduit system. We hypothesize that the Barren River Basin was larger than the Upper Green River Basin about ten million years ago. As the Green River incised headward, it intersected a meander bend in the ancestral Barren River, capturing the upstream portion of the Barren drainage. This drastically increased the drainage area of the Green River and decreased the drainage area of the Barren River. Coupling the shift in drainage with the gentle region dip of the bedrock to the northwest, the incision of the Green River accelerated relative to the incision of the Barren River, leading to the breach of the Mississippian carbonate strata, the onset of widespread karstification, and the shift of piracy mechanism from surface-dominated to subsurface-dominated. This hypothesis is illustrated as a sequence of three stages of drainage arrangement (Fig. 7).

Figure 7. Hypothesized sequence of drainage patterns for Kentucky, Illinois, Indiana, and Ohio from about 10 million years ago through present time. After Granger et al. (2001), Teller (1973), and Thornbury (1958).

What's next?

We are waiting on geochronology results--

26Al/10Be cosmogenic burial dating of sediments from Crystal Onyx Cave (purple dot in Fig. 1) and
10Be cosmogenic dating for basin-averaged erosion rate from Little Sinking Creek (red dot in Fig. 1).

These numbers will help us determine the rate of denudation and timing of exhumation for the Sinkhole Plain. They will also help us determine the chronology of incision of the Green River and its relation to Crystal Onyx and other knobs caves. Were they part of an ancestral Mammoth Cave System or did they drain to the west to Barren River? This will help answer our questions about the rate of migration of the Barren River-Green River drainage divide on the Sinkhole Plain.

Sources Cited

Clubb, F. J., Mudd, S. M., Milodowski, D. T., Grieve, S. W. D., & Hurst, M. D. (2017). LSDChannelExtraction v 1.0. Zenodo. https://doi.org/10.5281/zenodo.824198.

Granger, Darryl E, Derek Fabel, and Arthur N Palmer. “Pliocene- Pleistocene Incision of the Green River, Kentucky, Determined from Radioactive Decay of Cosmogenic 26Al and 10Be in Mammoth Cave Sediments.” Geological Society of America Bulletin 113, no. 7 (2001): 825–36.

Palmer, Arthur N. “Geomorphic History of the Mammoth Cave System.” In Karst Hydrology, 317–37. Springer, 1989.

Quinlan, James F, and Ralph O Ewers. “Subsurface Drainage in the Mammoth Cave Area.” In Karst Hydrology, 65–103. Springer, 1989.

Quinlan, James F, and Joseph A Ray. “Groundwater Basins of the Mammoth Cave Region, Kentucky, Showing Springs, Major Caves, Flow Routes, and Potentiometric Surface.” Occasional Publication. Mammoth Cave, Kentucky: Friends of the Karst, 1989.

Teller, James T. “Preglacial (Teays) and Early Glacial Drainage in the Cincinnati Area, Ohio, Kentucky, and Indiana.” Geological Society of America Bulletin 84, no. 11 (1973): 3677–88.

Thornbury, William David. “The Geomorphic History of the Upper Wabash Valley [Indiana].” American Journal of Science 256, no. 7 (1958): 449–69.

Toomey, Rickard. “kgwbm_karstdye toomey modified.” Dye trace data shapefiles for the state of Kentucky. 12 March 2019.