Analyzing a 10-Year Cave Drip Record in James Cave, Virginia: Implications for Storage and Recharge in Shallow Appalachian Karst Systems

Nigel C. Groce-Wright

Thesis submitted to the faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of

Master of Science In Geosciences

Madeline E. Schreiber Ryan Pollyea Ryan Stewart

June 30, 2021 Blacksburg, VA Keywords: hydrology, cavern, groundwater

Analyzing a 10-Year Cave Drip Record in James Cave, Virginia: Implications for Storage and Recharge in Shallow Appalachian Karst Systems

Nigel C. Groce-Wright

ABSTRACT (Academic)

Karst aquifers, characterized by soluble rocks such as limestone and dolostone, provide drinking water to 20-25% of the world’s population and are thus critical global water sources. However, recent work suggests that rapid alteration of karst aquifers due to the impact of climate change on precipitation patterns may affect recharge to these aquifers.

Much of the research on recharge in karst aquifers has relied on using patterns of spring discharge to quantify recharge. Spring outlets allow for continuous monitoring of discharge from karst aquifers, making them easily accessible monitoring sites. However, because springs can integrate multiple flow paths, it is difficult to rely on spring discharge patterns to get information on where and how karst aquifers are receiving recharge. Monitoring closer to the source of recharge through the measurement of cave drips allows for a more accurate analysis of recharge timing and mechanisms.

In this study, I conducted recession analyses on cave drip hydrographs from a 10-year record (2008-2018) of three drip monitoring stations within James Cave (Pulaski Co., VA) to: 1) examine differences in hydrologic characteristics of the epikarst (the zone of soil and weathered bedrock above a karst aquifer); 2) quantify the storage volume of the epikarst and 3) investigate seasonal, and annual trends in recharge.

Results of recession analysis show heterogeneity in epikarst hydrologic characteristics, reflected by calculations of the recession coefficient, α, and storage volume. Calculations of the recession coefficient show subtle differences between the three drip sites, suggestive of spatial heterogeneity in permeability and storage in the overlying epikarst. The storage volume calculations show that during the recharge season (winter- spring), up to 95% of recharge through the unsaturated zone to the cave occurs through rapid pathways (i.e., fractures), and 5% through diffuse pathways (i.e., pores). However, during the recession period (spring-summer), when evapotranspiration is active, recharge through cave drips decreases and occurs predominantly through diffuse flow. Combined, these results underscore the importance of both spatial and temporal characterization of drip rates and other recharge inputs into karst aquifer systems.

Analyzing a 10-Year Cave Drip Record in James Cave, Virginia: Implications for Storage and Recharge in Shallow Appalachian Karst Systems

Nigel C. Groce-Wright

ABSTRACT (General Audience)

Karst aquifers, characterized by soluble rocks such as limestone and dolostone, provide drinking water to 20-25% of the world’s population and are thus critical global water sources. Recent work suggests that climate change may alter how karst aquifers are recharged; however, few studies have addressed this potential impact.

This study expands knowledge of recharge in karst aquifers through analysis of a 10-year record (2008-2018) of three cave drip measuring stations in James Cave (Pulaski Co., VA). I used recession analysis of the cave drip record to investigate temporal trends in recharge and to examine hydrologic characteristics of the epikarst, the zone of soil and weathered bedrock above the cave. Results of this analysis show seasonal patterns in cave drips, with the highest drip rates occurring in the winter and early spring. The analysis also shows spatial differences in hydrologic characteristics of the epikarst. Calculations of the storage volume show during the winter and early spring, up to 95% of recharge to the cave occurs through rapid pathways (i.e., fractures), and 5% occurs through diffuse pathways (pore spaces in the soil and rock).

Results of this study underscore the importance of both temporal and spatial characterization of cave drips and other recharge inputs into karst aquifer systems. The information gained from this study will add the body of knowledge on how karst aquifers receive recharge, which will aid in protection and management of these critical drinking water sources.

ACKNOWLEDGEMENTS

I would like to acknowledge the following people and groups for all their hard work in support of the completion of this research project and thesis. First, Dr. Madeline Schreiber, my advisor, who is one of the most genuine people I have ever met. Her guidance and feedback have been central to my time at Virginia Tech, and I could not have done this project without all of her hard work.

I would like to thank my committee members Dr. Ryan Pollyea and Dr. Ryan Stewart. They were always people I could turn to for technical solutions and professional advice as I navigated the research process. My colleagues Josh Benton and Nicholas Hammond in the hydrogeosciences research group developed R codes that I used to conduct analyses for this thesis. Josh and Nick both took substantial time out of their busy schedules to not only write the codes, but to teach me how to use and write the codes myself. I would also like to acknowledge Wenyu Gao of the VT Department of Statistics for her help with writing and utilizing R-codes for time series analyses.

April Newcomer, the Advising & Enrollment Manager in the VT Department of Geosciences, was also a huge help to me. The challenges I encountered during the pandemic had the potential to delay my completion of the thesis but with April’s help, the process was seamless, and this project would have almost certainly taken longer to complete without her assistance. I would also like to thank the Multicultural Academic Opportunities Program (MAOP) for awarding me the fellowship which allowed me to be a research assistant in Spring 2021 to allow more time to complete this thesis.

I also would like to acknowledge Wil Orndorff, Katarina Ficco and Tom Malabad of the Virginia Department of Conservation and Recreation. Wil, in collaboration with Dr. Benjamin Schwartz, a faculty member at Texas State University, helped set up the James Cave field site in 2007. Since then, Wil, Katarina and Tom have also been instrumental in continuing the monitoring efforts over the years, in addition to former graduate students Jonathan Gerst and Sarah Eagle and countless cavers from the VPI Cave Club. Science is only as good as the data, and without these individuals, I would not have had access to this long-term dataset.

I would like to acknowledge my closest mentors and dearest confidants. Dr. Tamie Jovanelly has been my inspiration since the day we met at Berry College. Her “winners do more” mantra sticks with me today and thanks to Dr. Jovanelly’s support and encouragement, I was able to further my knowledge of geoscience and be truly prepared for what my career brings. The group of friends I made from the mountain biking club at VT have been a constant guiding light during graduate school. These young men are mature beyond their years, and the hours we spend riding our bikes lead to impactful conversations and lasting friendships. My two closest confidants Andrew Issem and Jordan Tunks have been right by my side since I arrived at VT and changed me to a fierce rider and a compassionate role model.

My family, Cheryl, Carson, and Richard Groce-Wright have all been by my side through this entire process. The pandemic took an unforeseeable toll on my mental and emotional health and these people held me close, reassured me constantly, and kept my head raised high to this point, the finish line.

iv

Table of Contents List of Figures ...... vii List of Tables ...... ix 1 INTRODUCTION ...... 1 1.1 Introduction to Karst Aquifers ...... 1 1.2 Recharge in Karst ...... 1 1.3 Recharge through the Epikarst ...... 2 1.4 Previous Research on Recharge in Karst ...... 4 1.4.1 Tracer Methods ...... 4 1.4.2 Geochemical Methods ...... 5 1.4.3 Numerical Modeling ...... 5 1.4.4 Statistical Methods ...... 6 1.4.5 Recession Methods ...... 6 1.4.6 Impact of Climate Change on Recharge in Karst ...... 8 1.4.7 Knowledge Gap of Recharge in Karst ...... 9 1.5 Study Objective ...... 9 1.6 Site Description ...... 9 1.6.1 Location and Geology ...... 9 1.6.2 Climate ...... 11 1.6.3 Site Instrumentation ...... 11 2 METHODS ...... 13 2.1 Processing of Datasets ...... 13 2.2 Recession Analysis ...... 13 2.3 Drip Variability Parameters ...... 15 3 RESULTS ...... 17 3.1 Cave Drip Rate and Precipitation Records ...... 17 3.2 Recession Analysis ...... 20 3.2.1 Correlation Method ...... 20 3.2.2 Matching Strip Method ...... 20 3.2.3 Hydrograph Separation Method ...... 21 3.3 Drip Variability Characteristics ...... 25 4 DISCUSSION ...... 27 4.1 Seasonal and Annual Patterns of Drips ...... 27 4.2 Drip Variability Characteristics ...... 28

v

4.3 Epikarst Storage ...... 29 4.4 Epikarst Hydrogeologic Properties ...... 29 4.5 Diffuse vs. Concentrated Recharge ...... 29 4.6 Comparison of Recession Methods ...... 31 4.7 Conceptual Model of Storage and Recharge in the Epikarst at James Cave ...... 31 4.8 Limitations of the dataset ...... 33 4.9 Limitations of methods...... 33 4.10 Suggestions for Further Work ...... 34 5 SUMMARY ...... 36 REFERENCES ...... 37 APPENDIX A ...... 42 APPENDIX B ...... 46

vi

List of Figures Figure 1: Block diagram of karst groundwater system. Modified from Taylor and Green., 2008.

Numbers refer to the following: 1 is diffuse flow through soils and unconsolidated material, 2 is

concentrated flow through enlarged sinkhole drains, 3 is diffuse infiltration through vertical

fractures and 4 is diffuse infiltration through permeable rock matrix...... 2

Figure 2: Epikarst schematic diagram. Blue arrows reflect precipitation, green arrows represent evapotranspiration from vegetation. Epikarst, consisting of soil and weathered bedrock, transmits flow through fractures, conduits, and pores within the matrix. Flow through epikarst enters the cave as drips through speleothems. Cave drips are measured at drip monitoring stations...... 4

Figure 3: Geology of Pulaski County. Geologic information obtained from Virginia Department

of Mines, Minerals, and Energy...... 10

Figure 4: James Cave Survey, produced by Tom Malabad, MS = drip monitoring station...... 12

Figure 5: Drip monitoring station MS2 (see location in Figure 4). Left photo of drip tarp assembly, right photo of Sarah Eagle and rain gauge data collection assembly. Photo credits: Wil

Orndorff, Virginia Department of Conservation and Recreation...... 12

Figure 6: Time series of drip discharge for drip sites (top) MS1, MS2, and MS3 (mL/min), with

hourly precipitation (bottom) from CDO database for New River Valley Airport, Dublin, VA. 18

Figure 7: Recharge season indicated by blue shading, recession season indicated by green

shading, dry season indicated by white cells, missing data (gray shading) due to instrument

failure. Information from 2008 to 2013 from Eagle et al. (2015)...... 19

Figure 8: Drip rate (mL/min) vs. lag drip rate (mL/min) for baseflow (Q < 50 ml/min) plotted

with data correlated by 1 day lag time. Envelope line plotted by least squares regression model,

fitted for minimum residuals...... 20

vii

Figure 9: Master recession curve (MRC) for each drip site, produced using MRC Generation

Excel spreadsheet (Posavec, 2006). Drip data are shown in blue; MRC fit shown in red...... 21

Figure 10: Individual hydrographs for drip sites MS1, 2 and 3. Log discharge shown to evaluate

K\GURJUDSKVORSHFKDQJHVVHJPHQWVFKRVHQEDVHGRQVORSHFKDQJH5HGVHJPHQWVKRZVĮ1 and

JUHHQVHJPHQWVKRZVĮ2 ...... 24

Figure 11: Maximum measured discharge plotted against coefficient of variation of drip

discharge (SCVP). Fields originally denoted by Smart and Friedrich (1986)...... 26

Figure 12: Conceptual model for James Cave epikarst. Recession analysis results suggest that

MS1 and MS2 have similar water sources and more contribution of diffuse flow than MS3, which has greater influence of concentrated flow...... 32

viii

List of Tables Table 1: Annual precipitation from CDO database for New River Valley Airport, Dublin, VA. 19

Table 2: Correlation method results for baseflow of MS1, MS2, and MS3, including recession coefficient (k) and R2 for the beVWILWRQWKHHQYHORSH&RQYHUVLRQWRĮ GD\ ZDVGRQHXVLQJ

Equation 6...... 20

Table 3: Master recession curve results using the matching strip method for optimal fit over full

drip record (2008-2018), Q0=discharge at the beginning of recession...... 21

7DEOH9ROXPHVDQGĮYDOXHVFDOFXODWHGIRUGULSK\GURJUDSKVIRU06-MS3. Also shown is the

% concentrated flow and % diffuse flow for each hydrograph, using the assumptions outlined in the text. Shaded regions reflect periods of recharge and recession (Figure 7)...... 25

Table 5: James Cave drip dataset variability characteristics (2008-2018); Qmax is maximum

discharge, Qmin is minimum discharge, Qav is mean discharge, Q10 = discharge exceeded by 10%

of the discharge data, Q90 is discharge exceeded by 90% of the data, Iv is the index of variability,

V is the spring variability index (%), SVC is the spring variability, SCVP is the spring

coefficient of variability parameter...... 26

Table 6&RPSDULVRQRIĮ PLQ IRU0606DQG06XVLQJWKHPDWFKLQJVWULSPHWKRG

(average flow), the correlation method (base/diffuse flow) and the individual hydrograph

separation method (rapid/concentrated flow and base/diffuse flow) ...... 30

ix

1 INTRODUCTION

1.1 Introduction to Karst Aquifers

Karst aquifers, characterized by soluble rocks such as limestone and dolostone, provide drinking water to 20-25% of the world’s population (Ford et al., 2007). Karst aquifers initially develop in pore-controlled matrices (Kaufmann et al., 2000). Dissolved carbon dioxide introduced with recharge (Equations 1 and 2) promotes dissolution of calcite (Equation 3), creating fractures and eventually, conduits, which provide low resistance pathways for groundwater flow (White et al., 2002). The result is a highly heterogeneous aquifer with triple porosity: matrix, fracture, and conduit (Ford and Williams et al., 2007). Because karst aquifers have similarities in flow dynamics to both surface water and groundwater systems (White et al., 2002), investigation of karst aquifers requires an understanding of surface and ground water flow dynamics. In addition, due to the complexity of karst systems, robust investigative techniques are needed to assess these aquifers (White et al., 2002, Taylor and Green, 2008).

( ) + (1) 0 2 2 2 3 𝐶𝐶𝐶𝐶 𝑔𝑔𝑔𝑔𝑔𝑔 𝐻𝐻 𝑂𝑂 → 𝐻𝐻+𝐶𝐶𝐶𝐶 (2) 0 − + 𝐻𝐻2𝐶𝐶𝐶𝐶3 → 𝐻𝐻𝐻𝐻𝑂𝑂3 𝐻𝐻 ( ) + ( ) + ( ) + 2 (3) 2+ − 3 2 2 3 1.2 Recharge in Karst𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐶𝐶𝐶𝐶 𝑔𝑔𝑔𝑔𝑔𝑔 𝐻𝐻 𝑂𝑂 𝑎𝑎𝑎𝑎 → 𝐶𝐶𝐶𝐶 𝐻𝐻𝐻𝐻𝐻𝐻

Recharge is defined as water that infiltrates at the land surface, flows through the unsaturated zone, and crosses the water table to enter the groundwater system (Anderson et al., 2015). A key component of characterizing karst aquifers is the quantification of recharge, as effective management of karst aquifers for drinking water supply requires information on the rate of replenishment of water to the aquifer (Scanlon et al., 2002). Estimating recharge in karst is particularly challenging due to the heterogeneity of flow paths created by rapid dissolution of limestone. Recharge to karst aquifers occurs from both diffuse and concentrated flow (Figure 1). Diffuse recharge is derived from precipitation that infiltrates into the pores of the soil and bedrock above a karst aquifer, which can then recharge the underlying aquifer through cave drips

1

(Eagle et al., 2015, Fairchild et al, 2000, Kaufmann et al., 2000, Liu et al., 2016, White et al., 2002). Concentrated recharge can occur through larger fractures, sinkholes or sinking streams, providing rapid and highly focused recharge. The relative proportion of concentrated to diffuse recharge reflects the distribution and connection of conduits and the timing of water fluxes, which significantly effects the variability in spring discharge and water chemistry (Ford and Williams, 1989, Taylor and Green et al., 2008).

Figure 1: Block diagram of karst groundwater system. Modified from Taylor and Green., 2008. Numbers refer to the following: 1 is diffuse flow through soils and unconsolidated material, 2 is concentrated flow through enlarged sinkhole drains, 3 is diffuse infiltration through vertical fractures and 4 is diffuse infiltration through permeable rock matrix. 1.3 Recharge through the Epikarst

Epikarst, also called the “subcutaneous zone” and defined as the uppermost portion of the vadose zone in a carbonate rock formation (Figure 2), exerts an important control over recharge as precipitation infiltrates through this zone prior to recharging the underlying karst aquifer. The

2

epikarst consists of soil and weathered limestone near the surface with high secondary porosity, including fractures and conduits (Ford and Williams et al., 2007, Taylor et al., 2008). The fractures (50 – 500 µm) are enlarged through dissolution until they become conduits (>1 cm in diameter). Distribution of fractures and conduits develop within the epikarst matrix and cause the water flow within the epikarst to vary spatially, particularly with depth (Kaufman et al., 2000, Taylor and Green, 2008). The upper zone of the epikarst exhibits high rates of limestone dissolution because of the proximity to the primary source of carbon dioxide production in the atmosphere and soil (Kaufman et al., 2000). Limestone dissolution also decreases with depth as water becomes

saturated with respect to CaCO3 (see Equations 1- 3 ). These patterns create a contrast in hydraulic conductivity between the near surface zone and the deeper zone of the epikarst (Williams et al., 1983). Infiltration of precipitation into the epikarst is influenced by this contrast in hydraulic conductivity between the shallow and deeper epikarst. The shallow epikarst will transmit water first through the high hydraulic conductivity fractures, but transmission of water downward slows. The epikarst matrix (Figure 2) exhibits a bottleneck where pressure from stored water transmits recharge to the karst aquifer via the “piston effect” (Padilla et al., 1994, White et al., 2002, Williams et al., 1983). Increases in hydraulic head from infiltration through the shallow epikarst drive concentrated recharge through fractures and conduits and subsequently into the epikarst matrix. Recharge from the epikarst to the underlying karst aquifer has elements of diffuse (base) flow and concentrated (rapid) flow, with diffuse flow through the matrix generating baseflow and flow through fractures and conduits transmitting concentrated (rapid) flow (Atkinson et al., 1977, Padilla et al., 1994, Williams et al., 1983). Quantifying the rate of flow and volume of water which passes through the epikarst and into the cave passage yields insights into the hydrologic characteristics of the overlying epikarst.

3

Figure 2: Epikarst schematic diagram. Blue arrows reflect precipitation, green arrows represent evapotranspiration from vegetation. Epikarst, consisting of soil and weathered bedrock, transmits flow through fractures, conduits, and pores within the matrix. Flow through epikarst enters the cave as drips through speleothems. Cave drips are measured at drip monitoring stations. 1.4 Previous Research on Recharge in Karst

A variety of methods has been used to evaluate recharge in karst including tracer methods, geochemical methods, numerical modeling, statistical methods, and recession methods, described in more detail below. Recharge in karst systems is generally investigated using measurements of discharge and water chemistry of springs, due to their accessibility. Spring outlets on the surface, originating from karst aquifers, are natural outlets for water discharging from conduit networks (Taylor and Green, 2008) and these sites often are located where regional or local ground-water boundaries lie. Analysis of springs has also been used to define the extent of and spatial heterogeneity of conduit networks in epikarst (Taylor and Green, 2008), which is important for delineating recharge areas.

1.4.1 Tracer Methods

Tracer studies in karst aquifers are used to track the timing of recharge components. These investigations utilize different types of tracers, including fluorescent dyes, injected into an 4

inlet, such as a sinkhole, at a known concentration, after which point water samples are collected at spring outflow points and the samples analyzed for dye concentration (Atkinson et al., 1977, Taylor et al., 2008). Concentrations of tracer dyes are measured in the spring over time. Tracer data are then analyzed using a dye-breakthrough curve which allows for estimates of hydraulic properties. Estimates of flow velocity and discharge can be made by measuring the characteristics of the breakthrough curve. In addition, characteristics which can be calculated using dye-tracer data are first arrival of the leading edge of the dye pulse, time to peak concentration, and time of passage of dye pulse. These measurements give insight into the characteristics of the flow paths (e.g., hydraulic capacity) for recharge within an aquifer (Arbel et al., 2010, Massei et al., 2006, Taylor and Green, 2008).

1.4.2 Geochemical Methods

Geochemical analysis of spring water has been used to infer properties of recharge in karst systems, including the function of the epikarst for storing recharge (Aquilina et al., 2006, Frank et al., 2019). For example, investigation of limestone calcite dissolution rates by way of mass balance for calcite in spring waters can be used to evaluate variation in chemical composition of recharge waters. High alkalinity (HCO3 + CO3) concentration in karst waters indicates rapid dissolution of calcite within the epikarst. The increase in dissolution of calcite increases the size of conduits within the epikarst, leading to concentrated recharge to the saturated aquifer. (Taylor and Green, 2008). Geochemical analysis of springs and other discharge waters for environmental tracers, such as isotopes of water, can be used to determine water age, which can be used to infer recharge. During recharge, water stored in the epikarst will mix with precipitation. The mixing of stored water and infiltrated precipitation provides a signal that can be used as a tracer. Isotopes or solutes present in outflow waters, after the mixing of infiltration and stored water, are tracked for concentrations over time. These concentration data provide a basis for tracking the timing of concentrated recharge and baseflow components of recharge for karst aquifers (Falcone et al., 2008, Li, 2010, White et al., 2002).

1.4.3 Numerical Modeling

5

Diffuse recharge and concentrated recharge can be differentiated by simulating the spring response to precipitation inputs using numerical models (Yager et al., 2013). A time-continuous model is a tool to make estimates for inflow into the conduit system of a karst aquifer. The numerical models can be used to quantify the flux of water into and out of the conduit system of the karst aquifer (Geyer et al., 2008). Numerical models can also be used to differentiate water ages using chemical tracers (Yager et al., 2013), allowing for investigation of the rate at which recharge is translated to the saturated aquifer. From a study of environmental tracers, Yager et al. (2013) shows that inflow occurs rapidly through high conductivity zones initially where it is then mixed with older waters from deeper flow paths. The variation in tracer concentrations of discharge water shows the order of water ages as they are discharged to outflow points. The temporal variation in tracer concentrations in karst recharge waters suggests that shallow flow paths are more numerous and complex than previously suggested (Yager et al., 2013).

1.4.4 Statistical Methods

Statistical analyses, including time series analysis and single-event analysis, have been used to examine recharge in karst. In analysis of time-series data, autocorrelation is a statistical method for determining the relatedness of individual measurements, such as relating discharge measurements from a single karst spring outlet for a defined observation period (Herman et al., 2009). This relatedness can be quantified with a correlation coefficient that, when compared to other datasets, provides insight into the relationship between different hydrologic responses, such as precipitation and spring discharge. Autocorrelation can also be applied to the analysis of single events. By relating subsequent discharge measurements during a single event, the length of influence of the event to the unsaturated zone can be estimated. For example, high flow in the unsaturated zone during a single event influences the response of the karst system, defined as the memory effect, which can be used to characterize the conduit response to high flow events (Herman et al., 2009, Mangin et, 1984, Mayaud et al., 2014).

1.4.5 Recession Methods

6

In hydrology, stream discharge can be analyzed by investigating the receding limb of hydrographs. Recession analyses yield information about the hydrologic system, including distribution of recharge (temporal and spatial) and storage properties of the streambank and surrounding aquifer (Scanlon et al., 2010, Nathan and McMahon, 1990, Fiorillo et al., 2014, Amit et al., 2002). The recession curve is delineated from the maximum discharge to the start of the next increase in discharge (Amit et al., 2002). The recession equation was derived by Boussinesq and others around 1910 (Geyer et al., 2008, Liu et al., 2016, Amit et al., 2002, Nathan and McMahon, 1990) and is expressed as: = (4) 3 -1 −𝛼𝛼𝛼𝛼 3 -1 where Q0 = the initial discharge (L T ), Qt 𝑄𝑄= 𝑡𝑡discharge𝑄𝑄0𝑒𝑒 at time t (L T ), α = the recession constant (T-1), and t = time (T). Equation 4 can also be written as (Brutsaert et al., 1977): = (5) -α 𝑡𝑡 where k is the recession coefficient (= e ), and𝑄𝑄 t is𝑄𝑄 0the𝑘𝑘 reciprocal time unit, as written below:

= = (6) 1� −𝛼𝛼 𝑄𝑄 𝑡𝑡 For the sake of simplicity, the remainder𝑘𝑘 of this𝑒𝑒 thesis�𝑄𝑄,0 �I will focus on the recession constant α and will transform values of k to α using Equation 6. Barnes (1939) suggested that plots of log Q over time can be used to distinguish different components of stream discharge. Shifts in slope, which reflect changes in α, have been used to interpret changes in storage or permeability. Initially, steep slopes reflect rapid flow; as the recession progresses, the slopes often become shallower, reflecting slower flow. In streams, these transitions are interpreted to reflect shifts from surface runoff to interflow to baseflow (Nathan and McMahon 1990). In karst springs, those transitions are suggested to reflect shifts from concentrated flow through fractures/conduits to diffuse flow through pores (Amit et al 2002). The physical meaning of α is represented by:

= (7) 𝐾𝐾 𝛼𝛼 where K is the hydraulic conductivity of an aquifer𝑆𝑆𝐿𝐿𝑐𝑐 with length scale (Lc) and storage coefficient (S) (Fairley, 2017). As the storage and length scale of the aquifer should remain constant, an increase in α should be associated with an increase in hydraulic conductivity. Conversely, as α

7

decreases, hydraulic conductivity decreases. Changes in α thus reflect a physical switching from flow paths of varying hydraulic conductivity (Fairley, 2017). Recession analysis has also been used to evaluate different watershed properties. Baseflow recession characteristics have been used to estimate basin evapotranspiration (Szilagyi et al., 2007), groundwater- surface water interactions (Clark et al., 2011), and water table depths within a land surface watershed model (Lo et al., 2010). Tallaksen (1995) and Hall (1968) described the application of recession analysis and its use in distinguishing the hydrologic properties of a watershed, as it pertains to baseflow. These studies have shown that watershed properties vary across geologic settings and under different anthropogenic influences (Wang et al., 2009). Recession analysis has primarily been applied to surface water systems (Arnold et al. 1995, Chapman et al., 1999) but more recently, this method has been applied to hydrographs of karst springs to examine aquifer storage characteristics and temporal distribution of recharge, including both diffuse and concentrated components (Fiorillo et al., 2010, Amit et al., 2002, Geyer et al., 2008, Padilla et al 1994, Giacopetti et al., 2017, Bonacci, 1993). One recent study by Liu et al. (2016) applied recession analysis to cave drips to derive storage parameters for a karst aquifer.

1.4.6 Impact of Climate Change on Recharge in Karst

Because the primary source of recharge to karst aquifers is precipitation, analysis of recharge must also consider spatial and temporal precipitation patterns. Of key concern for water supply is the impact of climate change on precipitation, as shifts in annual precipitation influence recharge and storage dynamics of all aquifers, including those in karst. For example, simulations of the response of the North Atlantic Oscillation (NAO) to increases in water vapor caused by the warming climate have shown an increase in annual precipitation for the North Atlantic region (Ning et al., 2012). Sea surface temperature has been shown to have a great effect on the cycling of water vapor throughout the atmosphere (Bishop et al., 2018); simulations suggest that increases in sea surface temperature due to climate change will increase the amount of precipitation distributed by winter storms to the Southeast U.S. (Bishop et al., 2018). As limestone and dolostone rocks underlie much of the Southeast U.S., changes to the input of precipitation will likely have impacts on recharge to the underlying karst aquifers in this region.

8

1.4.7 Knowledge Gap of Recharge in Karst

The complexity of karst aquifers demands both qualitative and quantitative analysis on the impacts of changes in annual precipitation in the past 100 years, predicted to occur over the remainder of this century, to these critical aquifers (Bishop et al., 2018). The existing body of research on karst hydrology utilizes hydrologic datasets on time scales which range from hours to months. However, there is a lack of research over longer time periods due to data collection and access issues. Using data sets on the decadal scale or longer are beneficial in that they allow for long term trends to be understood, particularly in sensitive settings such as karst terranes. Additionally, long term data sets allow for comparison across spatial and temporal scales which serves to identify sources of change (Geyer et al., 2008, Herman et al., 2009, Li et al., 2010).

1.5 Study Objective

The overall goal of this study is to utilize a 10-year dataset on cave drips and precipitation at James Cave, Virginia, to examine the relationships between drip discharge and time to evaluate hydrologic characteristics of the epikarst and calculate storage volume in the epikarst. Results of this study will provide insights into changes in recharge within the cave system based on the 10-year period of record, which will inform the broader goal of improving effective management of water supplies in karst aquifers.

1.6 Site Description

1.6.1 Location and Geology

The study site, James Cave, is in Pulaski County, Virginia, within the Valley and Ridge physiographic province. The Pulaski Fault is the dominating feature impacting geologic structure in the area (Eagle, 2013). Bartholomew (1987) details the Pulaski Fault system as a series of Alleghenian thrust faults responsible for westward displacement of the Pulaski thrust sheet units as young as Mississippian in age (Eagle, 2013). James Cave has formed in the limestone and dolomite units exposed at the surface, due to the fault system (Gerst, 2010, Eagle, 2013). Figure

9

3 shows the location of Pulaski County in Virginia (inset), the geology within the county boundaries, and the location of James Cave.

Figure 3: Geology of Pulaski County. Geologic information obtained from Virginia Department of Mines, Minerals, and Energy.

10

James Cave is located within the Conococheague Formation and is underlain and surrounded by the Elbrook Formation due to a synclinal feature (Figure 3) (Eagle, 2013). The Conococheague formation has been described as a fine-grained bluish-gray limestone with interbeds of shale and varying amounts of dolomite (Hergenroder, 1957). The Elbrook has been described as a thin-bedded dolomite that also contains dolomitic to pure calcareous limestone as well as some shale interbeds (Hergenroder, 1957, Eagle, 2013).

1.6.2 Climate

Pulaski County is in a temperate climate zone with an approximate mean annual temperature of 12.8 °C (The Southeast Regional Climate Center 2021). The county receives on average annual precipitation of 41.0 inches (1042 mm) (The Southeast Regional Climate Center 2021). Analysis of precipitation records from 2008 to 2013 show little seasonality to precipitation (Eagle, 2013). Evapotranspiration (ET), estimated using the Penman-Monteith equation, is highest in the summer months (Gerst, 2010, Eagle, 2013).

1.6.3 Site Instrumentation

A portion of James Cave was instrumented in 2007-2008 with three cave drip monitoring sites (Figure 4). The three sites were assembled to measure drips from the cave ceiling (Figure 5). The three drip sites were established in locations where the cave ceiling lies approximately 15 meters beneath the land surface (Eagle et al., 2015), which corresponds to the range of epikarst depth in many karst regions (Klimchouk, 2003). The drip tarps were assembled beneath locations with 10-15 stalactites actively dripping water (see Figure 5) (Eagle et al., 2015). PVC (polyvinyl chloride) piping overlain with a plastic tarp make up the frame, with a funnel attached to direct water to the recording equipment. The recording equipment collected drip rate data with smart sensor tipping bucket rain gauges (Onset HOBO RGB-M002) attached to microstation data loggers (Onset HOBO H21-002) (Eagle et al., 2015). Due to the high amount of moisture in the cave passage, corrosion of rain gauge sensors in the cave environment led to instrument failure prior to 2012. This issue was addressed by retrofitting of the rain gauges with reed switches and pulse adapters during July 2012 (Eagle 2013). The drip monitoring equipment was removed from James Cave during the fall of 2019.

11

Figure 4: James Cave Survey, produced by Tom Malabad, MS = drip monitoring station.

Figure 5: Drip monitoring station MS2 (see location in Figure 4). Left photo of drip tarp assembly, right photo of Sarah Eagle and rain gauge data collection assembly. Photo credits: Wil Orndorff, Virginia Department of Conservation and Recreation.

12

2 METHODS

2.1 Processing of Datasets

Precipitation data were collected from the National Climatic Data Center (NCDC) archives stored in the Climate Data Online (CDO) database. The New River Valley Airport (Dublin, VA) is the closest weather station to the entrance of James Cave (7.7 miles). The station location and the length of the precipitation record (1968-2021) were the factors which influenced the choice for precipitation data source. The data from CDO included hourly precipitation, daily precipitation, wind speeds, wind direction, hourly snowfall, daily snowfall, relative humidity, and others. The drip dataset from James Cave consists of cave drip readings collected every 10 minutes from drip sites MS1 and MS2, which began in September 2007, and drip site MS3, which was installed in February 2008. The drips were recorded in 10-minute time intervals from the beginning of instrumentation until removal in September 2019. Continuous time series data were offloaded from sites on monthly and bimonthly bases until 2012 (Gerst, 2010, Eagle, 2013), and bimonthly to seasonally until the loggers were removed in 2019. The data were combined using Aquarius Workstation (until 2014) and an R code (after 2014). Prior to use in this study, drips (mm per 10 min) were converted to discharge (mL/min) using a conversion of 3.7 mL to 0.2 mm of precipitation (https://www.onsetcomp.com).

2.2 Recession Analysis

In this study, recession analysis was conducted on the James Cave drip hydrograph dataset to evaluate hydrologic characteristics of the epikarst and calculate storage volume in the epikarst. As the datasets were large (MS1: 427,486 observations, MS2: 459,248 observations, MS3: 434,839 observations), the analyses were conducted using R statistical software. Two approaches were selected for conducting recession analysis of the full cave drip datasets: a correlation method and a matching strip method. These methods have been used to analyze recession data for spring discharge (Fiorillo et al., 2014, Amit et al., 2002) and for cave drips (Liu et al., 2016).

13

The correlation method, based on Equation 6, uses a plot of the current discharge against the discharge at a fixed lag time. This method plots an enveloping line of the correlated discharge observations where the slope defines the recession coefficient (k) for different lag times, t (Nathan and McMahon, 1990). For each drip site, the correlation curve is output with the slope of the envelope line defined as the recession coefficient k for lag time of 1 day. Choosing the optimal lag time depends on the length of the recession periods. Nathan and McMahon (1990) suggest selecting as long a period as possible that allows for a recession period to be analyzed. Because this method is focused on baseflow (Nathan and McMahon, 1990), I conducted the analysis using subsets of the hydrographs with drip discharge values below 50 mL/min. For this study, the optimal lag time was modeled by least squares regression to best fit the enveloping line for the curve. The matching strip method involves plotting a full dataset of recession events in terms of time relative to the start of each event, and then fitting the data with an exponential equation y=be-ax (Posavec et al., 2006). Recession events are ranked and sorted by the magnitude of decrease, after which point the dates are converted into relative time measurements in minutes from the beginning of recession. The segments (or strips) are adjusted horizontally in relative time to yield a master recession curve (MRC), using a VBA program in Microsoft Excel (Posavec et al., 2006). This spreadsheet output an MRC for the drip record at all drip sites. This method can also be used to solve for the recession constant, α (Giacopetti et al., 2017). In addition to these two methods for analysis of the larger dataset, a third method was used to analyze multiple segments on individual drip hydrographs (hydrograph separation method), following the approach of Amit et al. (2002) and Nathan and McMahon (1990). For this analysis, a select number of complete hydrographs were chosen for each drip site. Only events with a hydrograph maximum value >50mL/min and a minimum value of <10mL/min were selected. For the selected hydrographs, semi-log plots of Q vs. time were generated, and the slope of the line, reflecting α, was determined. If there was more than one slope identified in the semi-log hydrograph, the hydrograph was separated into two segments, generating an α for each slope. Following the methods of Amit et al (2002), each hydrograph was separated into two segments where there was a visible change in slope. Those individual sections were then plotted

14

as the slope of the natural log of discharge versus relative time. The α1 and α2 values were also used to calculate storage volume using Equation 8:

( ) = ( ) = , (1 ) + , (1 ) (8) 𝑡𝑡 𝑄𝑄0 1 𝛼𝛼1𝑡𝑡 𝑄𝑄0 2 𝛼𝛼2𝑡𝑡 𝑉𝑉 𝑡𝑡 � 𝑄𝑄 𝑡𝑡 𝑑𝑑 𝑑𝑑 − 𝑒𝑒 − 𝑒𝑒 0 st 1 2 3 -1 where, Q0,1 is the initial flow of the 1 segment𝛼𝛼 of the recession𝛼𝛼 curve (L T ), α1 is the recession

st -1 nd constant of the 1 segment (T ), Q0,2 is the initial flow of the 2 segment of the recession curve

3 -1 nd -1 (L T ), α2 is the recession constant of the 2 segment (T ), and t is time (T).

Assuming that α1 (and V1) reflects concentrated or rapid flow and that α2 (and V2) reflects diffuse flow (Amit et al., 2002), the proportion of concentrated flow for the total volume calculated (V1+V2) for a drip event was calculated as 100*[V1/(V1+V2)]. Conversely, the proportion of diffuse flow for the total volume for a drip event was calculated as

100*[V2/(V1+V2)].

2.3 Drip Variability Parameters

Other useful characteristics of the drip dataset were calculated with a suite of statistical parameters developed by previous authors to express variability in discharge (Flora, 2004,

Kresic, 2007, Giacopetti et al., 2017) including: the index of variability (Iv) (Equation 9), the variability index (V) (Equation 10), the variability coefficient (SVC) (Equation 11), and the coefficient of variation parameter (SCVP) (Equation 12) (Kresic, 2007). Giacopetti et al. (2017) used these parameters to identify highly variable springs (Iv >10; V > 100%; SVC > 10; SVCP >

200) from those that are more constant (Iv < 2; V < 25%, 1< SVC < 2.5; SCVP < 49).

= (9) 𝑄𝑄𝑚𝑚𝑚𝑚𝑚𝑚 𝐼𝐼𝑉𝑉 𝑄𝑄𝑚𝑚𝑚𝑚𝑚𝑚 = × 100 (10) 𝑄𝑄𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑄𝑄𝑚𝑚𝑚𝑚𝑚𝑚 𝑉𝑉 =𝑄𝑄𝑎𝑎𝑎𝑎 (11) 𝑄𝑄10 𝑆𝑆𝑆𝑆𝑆𝑆= 𝑄𝑄90 100 (12) 𝜎𝜎 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑄𝑄𝑎𝑎𝑎𝑎 ∗

15

where Qmax is maximum discharge, Qmin is minimum discharge, Qa is mean discharge, Q10 = discharge exceeded by 10% of the discharge data, Q90 is discharge exceeded by 90% of the data, is the standard deviation of the discharge dataset. (Flora 2004, Giacopetti et al., 2017).

𝛔𝛔 Another method of evaluating variability of flow inlets within caves was developed by Smart and Friedrich (1986) by using the coefficient of variation of discharge and the maximum discharge to define groups of different flow inlets, including seepage flow, percolation flow, vadose flow, shaft flow, and epikarst (subcutaneous) flow. The categories are loosely related to either high or low average discharge, and to either high or low variability. These characteristics are applied to the drip discharge measurements from James Cave, where the variability of the drip discharge and the average discharge characterize the average type of flow inlets to the cave.

16

3 RESULTS

3.1 Cave Drip Rate and Precipitation Records

The cave drip discharge data (2008 to 2018) collected in James Cave (Figure 6), along with the precipitation record for the same period (Figure 6; also see summary in Table 1), show seasonal trends that are generally consistent from year to year. The seasonal trends include a period of increased drips (recharge, ~ winter-spring), decreasing drips (recession, spring- summer) and no drips (dry period ~fall-winter), as summarized in Figure 7. For this analysis, the recharge period is defined as a period with increased drip rates (> 10 mL/min). The recharge period at the three drip sites typically starts in the early winter (January/February) but can start as late as March/April (see Figure 7, 2011 season for example). In mid to late spring, the drip discharge goes into a period of recession, typically from April to August, where drips decrease exponentially. The dry period is defined as the period where drip discharge goes below 10 mL/min. This typically occurs in fall and winter, at which point the recharge season starts again.

17

MS1

MS2

MS3

Precip (mm)

Figure 6: Time series of drip discharge for drip sites (top) MS1, MS2, and MS3 (mL/min), with hourly precipitation (bottom) from CDO database for New River Valley Airport, Dublin, VA.

18

Figure 7: Recharge season indicated by blue shading, recession season indicated by green shading, dry season indicated by white cells, missing data (gray shading) due to instrument failure. Information from 2008 to 2013 from Eagle et al. (2015).

Table 1: Annual precipitation from CDO database for New River Valley Airport, Dublin, VA.

Year Precip (mm)

2008 537.7

2009 674.1 2010 619.8

2011 587.8

2012 389.6 2013 399.03

2014 1257.3

2015 1593.3 2016 1236.2

2017 1217.4

2018 1441.7

19

3.2 Recession Analysis

3.2.1 Correlation Method Figure 8 shows the results from the correlation of drip measurements to the previous drip measurement (lag drip rate). The slope of the envelope line fitted to the correlation curve is the recession coefficient, k, which was converted into α using Equation 6. Fitting of the envelope line was conducted to minimize residuals between the drip curves and the line. Results of the linear regression analysis are shown in Table 2. As observed in Figure 8, although there are slight differences in the correlation plots between the three drip sites, the resulting k (and thus, α) values for baseflow are similar.

MS1 MS2 MS3

Figure 8: Drip rate (mL/min) vs. lag drip rate (mL/min) for baseflow (Q < 50 ml/min) plotted with data correlated by 1 day lag time. Envelope line plotted by least squares regression model, fitted for minimum residuals.

Table 2: Correlation method results for baseflow of MS1, MS2, and MS3, including recession coefficient (k) and R2 for the best fit on the envelope. Conversion to α (1/day) was done using Equation 6.

Drip site k Calculated α (1/day) Calculated α (1/min) R2

MS1 0.925 7.80E-02 5.41E-05 0.92

MS2 0.90 1.05E-01 7.32E-05 0.96

MS3 0.92 8.34E-02 5.79E-05 0.98

3.2.2 Matching Strip Method

Figure 9 shows the results for the master recession curve (MRC) generated using the matching strip method (Excel spreadsheet provided by Posavec et al., 2006). Equation 13 was

used to model each recession to solve for discharge (Qt), the initial Q at the start of the recession

(Q0), α (1/day), and x as relative time (day).

20

ିఈ௫ ܳ௧ = ܳ଴݁ (13)

The fits of the exponential equation (Figure 9) to the dataset have overall high R2 values; however, there are deviations of the data from the exponential equation, especially at the higher flows for MS2. Summary results of the analysis are shown in Table 3.

Figure 9: Master recession curve (MRC) for each drip site, produced using MRC Generation Excel spreadsheet (Posavec, 2006). Drip data are shown in blue; MRC fit shown in red.

Table 3: Master recession curve results using the matching strip method for optimal fit over full drip record (2008-2018), Q0=discharge at the beginning of recession.

Q0 (mL/min) Į (1/day) Į PLQXWH MS1 147.26 1.73E+00 1.20E-03 MS2 223.67 7.05E-01 4.88E-04 MS3 702.8 3.41E+00 2.37E-03

3.2.3 Hydrograph Separation Method

21

Figure 10 shows results of the hydrograph separation method to determine α1, and α2 for

individual hydrographs for each drip site. Note that the fits for α1 are generally better than for α2; at lower flows, there is lower accuracy of drip discharge measurement.

MS1: Separated Log drip rate (mL/min) α1 (red) plot α2 (green) plot Hydrograph vs. time

MS2: Separated Log drip rate (mL/min) α1 (red) plot α2 (green) plot Hydrograph vs. time

22

MS3: Separated Log drip rate (mL/min) Į1 (red) plot Į2 (green) plot Hydrograph vs. time

23

Figure 10: Individual hydrographs for drip sites MS1, 2 and 3. Log discharge shown to evaluate hydrograph slope changes, segments chosen based on slope change. Red segment shows Į1 and green segment shows Į2

Using the values of Į1 DQGĮ2, the storage volume was calculated using Equation 8 for each segment (Table 4). Į1 estimates were RQDYHUDJHWLPHVODUJHUWKDQĮ2 for drip site MS1,

IRU06Į1 LVRQDYHUDJHWLPHVODUJHUWKDQĮ2, and for MS3, Į1 is on average 18.5 times

larger WKDQĮ27KHYDULDWLRQRIĮ1 DQGĮ2 is greatest for drip site MS3 and lowest for drip site

MS1. Results presented in Table 4 show that on average V1 is 1.87 times larger than V2 at drip site MS1, V1 is on average 3 times larger than V2 for drip site MS2, and V1 is on average 5.97

times larger than V2 for drip site MS3. The variation of V1 and V2 is greatest for drip site MS3 and lowest for drip site MS1.

24

Table 4: Volumes and α values calculated for drip hydrographs for MS1-MS3. Also shown is the % concentrated flow and % diffuse flow for each hydrograph, using the assumptions outlined in the text. Shaded regions reflect periods of recharge and recession (Figure 7).

Drip Date V1 (L) V2 (L) V1 +V2 α1 α2 % % Diffuse site (L) (1/min) (1/min) Concentrated flow flow

MS1 4/1/08-4/28/08 364.89 88.6 453.49 2.00E-04 6.21E-05 80.5 19.5 (recharge)

7/8/13-8/30/13 207.5 1013.88 1221.38 7.71E-05 3.50E-05 17.0 83.0 (late recession)

5/15/18- 74.15 106.9 181.06 2.30E-04 4.34E-05 41.0 59.0 6/15/18 (early recession)

MS2 4/1/08-4/28/08 163.96 36.46 200.42 6.30E-04 5.69E-05 81.8 18.2 (recharge)

4/1/09-4/20/09 837.63 198.31 1035.94 1.00E-04 5.27E-05 80.9 19.1 (recharge)

7/1/13-8/10/13 728.44 419.3 1147.74 2.00E-04 6.00E-05 63.5 36.5 (late recession)

4/25/14-6/1/14 868.89 208.32 1077.21 1.00E-04 2.21E-05 80.7 19.3 (early recession)

4/20/15- 485.87 182.35 668.21 1.20E-04 2.45E-05 72.7 27.3 6/15/15 (early recession)

5/25/18-7/1/18 82.69 10.14 92.83 2.00E-04 6.19E-06 89.1 10.9 (early recession)

MS3 5/1/09-5/25/09 369.18 19.52 388.69 1.70E-03 3.79E-05 95.0 5.0 (recharge)

4/25/14-6/1/14 168.1 75.02 243.12 1.20E-03 1.30E-04 84.0 16.0 (early recession)

5/20/17-6/1/17 105.51 6.73 112.24 5.00E-04 2.75E-05 69.1 30.9 (early recession)

3.3 Drip Variability Characteristics

Results of the drip dataset variability calculations using the drip discharge record for sites MS1, MS2, and MS3 are presented in Table 5. Comparison of these parameters for our three sites reveals several differences. First, using the variability characteristics outlined by Giacopetti

25

et al. (2017), all three drip sites would be characterized as highly variable. Second, MS1 and MS2 show similar values for Iv, V, SVC and SCVP, suggesting similarities in the water source that feeds the drips at these sites. In contrast, MS3 has lower values of Iv and SVC and higher values of V and SCVP, suggesting that there are likely differences in the sources of variability between MS1/2 and MS3.

Table 5: James Cave drip dataset variability characteristics (2008-2018); Qmax is maximum discharge, Qmin is minimum discharge, Qav is mean discharge, Q10 = discharge exceeded by 10% of the discharge data, Q90 is discharge exceeded by 90% of the data, Iv is the index of variability, V is the spring variability index (%), SVC is the spring variability, SCVP is the spring coefficient of variability parameter.

Qmax Qmin Qavg Iv V (%) Q10 Q90 SVC SCVP (mL/min) (mL/ (mL/min) (mL/ (mL/min) min) min) MS1 506.9 0.04 25.26 13700.0 2006.6 59.2 0.4 155.8 198.5 MS2 507.0 0.04 20.9 13703.5 2425.8 62.2 0.4 168.0 169.2 MS3 725.1 0.18 13.76 3919.2 5268.0 34.8 0.4 94.0 273.8

Drip variability was also examined using the conceptual model of Smart and Friederich (1986) to identify the type of drip flow (Figure 11). Smart and Friederich (1986) denoted 5 types of “inlet” flow into caves: percolation, shaft, vadose, seepage, and epikarst (subcutaneous) flow, differentiating these using maximum drip discharge and the coefficient of variation of drip discharge. All three drip sites fall in the epikarst flow category.

Figure 11: Maximum measured discharge plotted against coefficient of variation of drip discharge (SCVP). Fields originally denoted by Smart and Friedrich (1986).

26

4 DISCUSSION

4.1 Seasonal and Annual Patterns of Drips

The drip discharge recorded at the three drip sites in James Cave varies both seasonally and annually. During the recharge period, the variation in drip discharge is associated with precipitation events (Figure 6). However, as the growing season begins in the spring, the cave drips go into recession, after which drips effectively stop until the start of the next recharge period (see Figures 6 & 7). These patterns were observed for a shorter period of the James Cave dataset by Eagle (2013) and Eagle et al. (2015) and have also been documented in the literature (Doctor et al., 2006, Eagle 2013, Fairchild et al., 2000, Herman et al., 2009, Taylor et al., 2005) for shallow, Mid-Atlantic karst terranes.

The initiation, and extent, of the recharge, recession and dry periods varies between years (Figure 7). Recharge can start as early as December and as late as March, a four-month difference. Similarly, the start of the recession period (end of recharge) varies year to year, with some years starting in April and other years starting as late as early July. The dry season generally starts in August/September, but sometimes can begin as late as October. These variations relate to the balance between precipitation and potential evapotranspiration. When precipitation exceeds potential evapotranspiration, the excess water is available for recharge as cave drips (termed “effective recharge”, see Eagle et al. 2015).

Although the interannual trends between the sites are similar, there are some differences. For example, in 2014 the recharge season begins in December for drip site MS1, in January for MS2, and in February for MS3. Another example is in 2017, when drips descend into the dry period during September for MS1, and during October for MS3 (dataset for MS2 is incomplete in 2017). These differences likely relate to differences in epikarst properties above each drip site, discussed in more detail below.

Drip events occur quickly after precipitation events in the early spring, during the recharge period. For example, in April (4/1-4/28) of 2008, there is a short lag time between the start of the precipitation event and the start of drip increase for drip sites MS1 (4 hours) and MS2 (6 hours). Drip events occurring in the late spring exhibit longer lag times between precipitation and drips. For example, at drip site MS1, the drip event occurring 5/15/2018-6/15/2018 occurred

27

13 hours after the precipitation event. The lag time increases during the recession period. For example, at MS2 during the recession period of 2018 (7/1-8/10), the lag time between precipitation and drip start was 24 days. This delay in response is likely related to the increasing potential evapotranspiration during the summer months, as plants take up precipitation, leaving less to saturate the epikarst.

4.2 Drip Variability Characteristics

Discharge variability parameters (lv, V, SVC) indicate highly variable discharge for all drip sites, which is not surprising considering the large variations in drip discharge due to seasonal patterns of recharge, recession, and dryness that we observe in the dataset. However, the results for all parameters (Iv, V, SVC, and SCVP) show marked similarity between drip sites MS1 and MS2, with MS3 being dissimilar. These results suggest that the flow paths and recharge mechanisms in the epikarst above MS1 and MS2 are similar to each other, which makes sense because drip sites MS1 and MS2 are located close together, about 100 m (see Figure 4). Although also highly variable, MS3 has different values for the variability characteristics, suggesting different water sources and pathways from MS1 and MS2.

Smart and Friederich (1986) provide a graphic representation of variability of epikarst waters, plotting Qmax against the coefficient of variation of discharge. Based on Figure 11, all three James Cave drip sites fall within the range of “epikarst flow” defined by Smart and Friederich (1986) as pathways characterized by “moderate to high discharge with very high coefficients of variation due to their intermittent operation”. Combined with the discharge variability characteristics, these results suggest heterogeneity in hydrologic characteristics in the epikarst above James Cave.

It is important to note that in contrast to the spring measurements, where a spring outlet can be fully gauged, the drip discharge measured in James Cave is influenced by the tarp area. Thus, analysis and comparison of absolute values of Q are less useful for drip discharge than they are for springs.

28

4.3 Epikarst Storage

In addition to yielding information on the recession coefficient, recession analysis also allows for calculation of the volume drained from the epikarst during drip events (Amit et al.,

2002). Our results (Table 4) show volume estimates for both the concentrated flow (V1) and the

diffuse flow (V2) components of recharge for the three sites for different time periods.

Concentrated flow (V1) estimates are highly variable, ranging from 6.73 to 1013.88 L.

It is interesting to note that the volume estimates change within each site over time, suggesting that infiltration of precipitation into the epikarst can potentially access new storage reservoirs, depending on antecedent conditions. Additional work would be needed to test this idea. These volume calculations also give us useful information on the volume of epikarst storage that drains during recession. For example, in MS1, during recession in 2013 (7/8/13 to

8/30/13), the baseflow volume (V2) was estimated to be over 1000L, the highest observed in this study. But, once that drains, the epikarst is depleted for the rest of the season.

4.4 Epikarst Hydrogeologic Properties

The hydraulic conductivity and storage properties of the epikarst can be characterized by

the differences observed between α1 and α2. From Equation 7, α represents the ratio of hydraulic

conductivity (K) and storage (S). In this study, two α parameters for concentrated/rapid (α1) and diffuse/base flow (α2) were differentiated using the hydrograph separation method. As α1 is greater than α2, the ratio of hydraulic conductivity to storage is higher for α1 than for α2. Since storage is a physical property of the epikarst, during individual events, the average storage

should remain constant during the event. For this reason, fluctuations in α likely reflect fluctuations in hydraulic conductivity of different portions of the epikarst (i.e., matrix vs. fractures) as the epikarst fills and drains. 4.5 Diffuse vs. Concentrated Recharge

The three methods of recession analysis applied in this thesis can be used to address different types of flow (diffuse/base flow vs. concentrated/rapid flow) contributing to the overall drip discharge. The correlation method of hydrograph analysis focuses on baseflow (Nathan and McMahon, 1990). We associate this baseflow with diffuse recharge, where flow is occurring

29

mostly through pore spaces. The hydrograph separation method allows for separating components of the hydrograph into rapid/concentrated flow (represented by α1) and diffuse/base flow (represented by α2). In contrast, the matching strip method used to develop the master recession curve results in average characteristics of the hydrograph, encompassing both rapid/concentrated flow and diffuse/baseflow.

Comparison of recession analysis results (Table 6) show that the three drip sites have overall similar hydrologic characteristics, as reflected by similarities in the average α (10-3 to 10-4 1/min) as derived from the matching strip method. Rapid/concentrated flow α values (1/min) are in the 10-3 to 10-4 range, while diffuse/base flow are 1 to 2 orders of magnitude lower (10-5 range). At the broad scale, these values make sense in relation to one another.

Table 6: Comparison of α (1/min) for MS1, MS2, and MS3, using the matching strip method (average flow), the correlation method (base/diffuse flow) and the individual hydrograph separation method (rapid/concentrated flow and base/diffuse flow)

Matching Strip Individual (α1) Individual (α2) Correlation (average) (rapid/concentrated) (base/diffuse) (base/diffuse) MS1 1.20E-03 1.70E-04 4.67E-05 5.41E-05 MS2 4.88E-04 2.25E-04 3.67E-05 7.32E-05 MS3 2.37E-03 1.27E-03 4.67E-05 5.79E-05

The relative proportion of rapid/concentrated flow to base/diffuse flow (see Table 4) suggests that during the recharge period, the dominant flow is rapid/concentrated (>80%). For example, during the recharge period of 2009 at MS3 (5/1-5/25) the proportion of rapid flow is 95% of the total with base flow making up 5%. In addition, an event during the recharge period of 2009 at MS2 (4/1-4/20) showed 81% of flow was rapid flow with 19% as base flow. These results suggests that during recharge, concentrated/rapid flow dominates the drip discharge, likely through fractures.

In contrast, during drip recession, the dominant flow shifts to base/diffuse flow. For example, during an event at MS1 early in the recession period of 2018 (5/15-6/15), the relative proportion of rapid to base flow is 41% to 59%. Another event in the recession period of MS2 in 2013 (7/1-8/10) shows a relative proportion of rapid flow to base flow as 63.5% and 36.5%, respectively. This diffuse flow is likely through pore spaces in the soil and weathered bedrock of the epikarst.

30

4.6 Comparison of Recession Methods

Table 6 shows the comparison of α values estimated using the three methods: correlation, matching strip, and hydrograph separation. The individual hydrograph separation yielded

numbers strikingly close to those of the multi-event correlation analysis. The α1 values representing concentrated/rapid flow fall in the range of 1.7x10-4 to 1.27x10-3 (1/min), which are

up to two orders of magnitude greater than the α values for diffuse/base flow. The average characteristics of the recession curves as determined by the matching strip method are represented by α values ranging from 4.88x10-4 to 2.37x10-3 (1/min). Each of the recession analysis methods utilized in this study has advantages and disadvantages. The generation of a master recession curve using the matching strip method is advantageous as it evaluates the entire dataset, and its results reveal overall characteristics of the dataset. The disadvantage is that the exponential model does not necessarily fit the entire dataset (see MS2, Figure 9). The correlation method is advantageous because the entire dataset can be analyzed, but the method is best used for baseflow. The advantage of using the separation of

hydrographs (α1 and α2) is the method allows for examination of concentrated vs. diffuse flow for individual hydrographs. Disadvantages of the separation method include 1) the parameter estimates are for single hydrographs and not the full dataset and 2) the hydrographs often show more than one change in slope, necessitating additional separation.

4.7 Conceptual Model of Storage and Recharge in the Epikarst at James Cave

The results from recession analysis and the variability characteristics of the cave drip discharge at James Cave from 2008 to 2018 suggests that the epikarst above James Cave contains a network of recharge pathways and storage zones that occur in pore spaces and fractures (Figure 12). During recharge events in the winter and early spring, precipitation infiltrates the soil and weathered bedrock and starts to fill fractures and subsequently, pore spaces. When the pore spaces connect and fractures “wet up”, infiltration can flow through these pathways to create the drip hydrographs that we see in the recharge period. Drip sites MS1 and MS2 are similar in terms of the α values from recession analysis and variability characteristics, but there are some subtle differences. In Table 4, comparison of storage volumes during the same recession event (7/1/2013 to 8/10/2013) shows that MS1 had a

31

higher % of diffuse/baseflow (83%) than MS2 (37%), suggesting that MS1 may have greater storage reserves than MS2.

Drip site MS3 is different from MS1 and MS2, with higher D1 values and discharge variability (see Tables 4 and 6). During drip events, the percentage of flow is mostly rapid (up to 95%, see Table 4) with a lower proportion of baseflow. This can also be seen visually in the shape of the hydrographs with higher flashiness of discharge relative to MS1 and MS2. Overall, these results suggest that cave drips at MS1 and MS2 have recharge contributions from both fracture and pore flow that originate from similar sources. In contrast, the recession analyses reveal that MS3 has a differing balance between concentrated and diffuse flow than MS1 and MS2. In addition, the higher values of D suggest that concentrated flow through fractures is dominant in the epikarst that drains to MS3, which may reflect a more weathered or “karstified” epikarst. The differences in characteristics that we observe between MS1/2 and MS3 also suggest heterogeneity in the permeability and storage characteristics of the soil and weathered bedrock in the epikarst of James Cave.

Figure 12: Conceptual model for James Cave epikarst. Recession analysis results suggest that MS1 and MS2 have similar water sources and more contribution of diffuse flow than MS3, which has greater influence of concentrated flow.

32

4.8 Limitations of the dataset

The dataset used for this analysis has several limitations. First, the datasets are incomplete, as the instruments (rain gauge and datalogger) failed multiple times during the 10 years of data collection. The environmental conditions in the cave passage are marked by high humidity, thus the rain gauges and data loggers used to collect the drip measurements were under constant exposure to conditions leading to corrosion. Due to the frequency of site visits in the early part of the study (Gerst, 2010; Eagle, 2013), the equipment was serviced or replaced to resume data collection within days of the instrument failures. However, after 2013, there were fewer site visits and thus instrument failures were often not recognized for several months. In addition, there are periods of missing data during high drip events, when the drip tarps would fill with silt, clogging the rain gauges. This prevented comparison of drip rates in periods of low precipitation (2008-2013; 400-700 mm annual precipitation) to periods of higher precipitation (2014-2018; >1000 mm annual precipitation) (see Table 1). Due to the missing data during the high drip events, I was not able to quantify the impact of precipitation increase on drip rates, which was one of the early objectives of the study. Another limitation of using the drip data to interpret the physical characteristics of the epikarst is the lack of accessibility to the epikarst. A topographical survey of the James Cave passage was performed and tied to surface elevation, relating the relative orientation of the monitoring stations and to determine the thickness of the epikarst overlying the stations (Gerst 2010). In addition, an Electrical Resistivity Tomography (ERT) survey identified drip stations relative to zones of varying resistivity within the epikarst (Gerst, 2010). Although this is useful information for estimating epikarst thickness, it does not give information on the presence of fractures that influence recharge. Last, the study is limited to only three drip monitoring stations. The monitoring stations had to be carefully selected for this study as to ensure ease of access for data downloads. Thus, inferences made in this study are limited to the data recorded individually at each of the three monitoring stations.

4.9 Limitations of methods

33

There are also limitations of the methods used in this study. First, the recession analysis methods assume homogenous and isotropic media, and Darcian (pore) flow. However, results from this study suggest that permeability varies throughout the epikarst. Thus, although we needed to make these assumptions to do the analysis, we know that the assumptions are likely violated. There are other equations that are used for fracture flow, but there are not equivalent recession analysis methods for these equations. Second, hydrograph separation methods to delineate distinct segments can also be challenging as it can be difficult to decide how to make the delineation. In this study, the delineation was done visually, which has associated error. Attempts were made to delineate different hydrograph segments using the matching strip method of Posavec et al. (2006) but our dataset was too large for the Excel spreadsheet, causing it to crash. Third, correlation of drip rate to lag rate is limited in that it assumes linearity between subsequent calculations. This is a necessary assumption to fit an envelope line to the curve which determines the recession coefficient (Nathan and McMahon 1990). Last, using α to evaluate hydrogeologic characteristics is imperfect, as it assumes that there is a physical meaning to this parameter. As shown in Equation 7, α is a lumped parameter that reflects the ratio of hydraulic conductivity (K) to storage coefficient (S). Thus, using α to interpret K or S individually can be challenging.

4.10 Suggestions for Further Work

The work presented in this thesis can be furthered by additional statistical analyses, re- instrumentation of the cave passage, and/or instrumentation of a different cave within the same geologic unit. Time series analyses could be conducted on the James Cave data set to evaluate the effect of seasonality on the drip data and to model the drip data with and without seasonality effects. Although the drip site assemblies have been removed from James Cave, re-installing the instruments would be feasible. A new dataset could be compiled and compared to the results of recession analysis presented in this thesis. Additional drip records under higher precipitation conditions would support an analysis of the impact of changes in precipitation on drip events, and thus recharge, to karst aquifers in the Southeastern U.S. Last, there are several mapped caves nearby to James Cave that have forest cover instead of pasture. A drip dataset from a cave in the same geologic formation as James Cave but with different land use would yield additional information about the recharge and storage properties

34

of the shallow epikarst of the Conococheague. Comparing drips from differing vegetation, and thus evapotranspiration, also would yield information on the connection between karst recharge and vegetation cover.

35

5 SUMMARY

This study utilized a 10-year cave drip record to evaluate recharge and storage in the epikarst of a local cave (James Cave, Pulaski VA). Three recession analysis methods were utilized: a correlation method, a matching strip method, and a hydrograph recession method. Results of the recession analysis show a cave drip response that reflects both rapid recharge through fractures, and diffuse/base flow due to draining of pores over time. Results show a similarity in hydrologic characteristics, reflected by the recession constant α and drip variability characteristics, at two of the drip sites located with 100 m of each other (MS1 and MS2), which are different than a third site (MS3). Overall, this study suggests that temporal spatial characterization of cave drip rates is important for quantifying recharge to karst aquifers.

36

REFERENCES

Amit, H., Lyakhovsky, V., Katz, A., Starinsky, A., Burg, A., 2002, Interpretation of Spring Recession Curves: Ground Water, v. 40. Anderson, M. P. W., Woessner, M.P., Hunt, R.J., 2015, Simulation of Flow and Advective Transport (2nd Edition), Applied Groundwater Modeling, Elsevier. p. 181-202. Aquilina, L., Ladouche, B., Dorfliger N., 2006, Water storage and transfer in the epikarst of karstic systems during high flow periods: Journal of Hydrology, v. 327. Arbel, Y., Greenbaum, N., Lange, J., Inbar, M., 2010, Infiltration processes and flow rates in developed karst vadose zone using tracers in cave drips: Earth Surface Processes and Landforms, v. 35. Arnold, J., Allen, P.M., 1995, Automated Base-flow Separation and Recession Analysis Techniques: Ground Water, v. 33. Atkinson, T. C., 1977, Diffuse Flow and Conduit Flow in Limestone Terrain in the Mendio Hill, Somerset (Great Britain): Journal of Hydrology, v. 35. Barnes, B. S., 1939, The structure of discharge-recession curves, Eos Transactions: American Geophysical Union, v. 20. Bartholomew, M. J., 1987, Structural evolution of the Pulaski thrust system, southwestern, Virginia: Geological Society of America bulletin, v. 99, no. 491. Bishop, D., 2018, Investigating the Causes of Increased Twentieth-Century Fall Precipitation over the Southeastern United States: American Meteorological Society: Journal of Climate, v. 32. Bonacci, O., 1993, Karst springs hydrographs as indicators of karst aquifers: Hydrological Sciences Journal, v. 38. Brutsaert, W., 1977, Regionalized drought flow hydrographs from a mature glaciated plateau: Water Resources Research, v. 13. Chapman, T., 1999, A comparison of algorithms for stream flow recession and baseflow separation: Hydrological Processes, v. 13. Clark, M., McMillan, H., Collines, D., Kavetski, D., Woods, R., 2011, Hydrological field data from a modeler’s perspective: Part 2: process-based evaluation of model hypotheses: Hydrological Processes, v. 25. Doctor, D., Alexander, E.C., Petric, M., Kogovsek, J., Urbanc, J., Lojen, S., Stichler, W., 2006, Quantification of karst aquifer discharge components during storm events through end-member mixing analysis using natural chemistry and stable isotopes as tracers: Hydrogeology Journal, v. 14.

37

Eagle, S. D., 2013, Analysis of Hydrologic and Geochemical Time Series Data at James Cave, Virginia: Implications for Epikarst Influence on Recharge, M.S. Thesis, Virginia Tech Department of Geosciences, Blacksburg, VA. Eagle, S. D., Orndorff, W., Schwartz B., Doctor, D., Gerst, J., Schreiber, M., 2015, Analysis of hydrologic and geochemical time-series data at James Cave, Virginia: Implications for epikarst influence on recharge in Appalachian karst aquifers: The Geological Society of America Bulletin v. 516. Fairchild, I., Borsato, A., Tooth, A., Frisia, S., Hawkesworth, C., Huang, Y., McDermott, F., Spiro, B., 2000, Controls on trace element (Sr–Mg) compositions of carbonate cave waters: implications for speleothem climatic records: Chemical Geology. v. 166. Fairley, J., 2017, A model of exponential decay, Models and Modeling: An Introduction for Earth and Environmental Scientists, John Wiley & Sons, Ltd., v. 2. Falcone, R. A., Falgiani, A., Parisse, B., Petitta, M., Spizzico, M., Tallini, M., 2008, Chemical and isotopic (d18O&, d2H&, d13C&, 222Rn) multi-tracing for groundwater conceptual model of carbonate aquifer (Gran Sasso INFN underground laboratory – central Italy): Journal of Hydrology, v. 357. Fiorillo, F., Dogliono, A., 2010, The relation between karst spring discharge and rainfall by cross- correlation analysis (Campania, southern Italy): Hydrogeology Journal, v. 18. Fiorillo, F., 2014, The Recession of Spring Hydrographs, Focused on Karst Aquifers: Water Resources Management. v. 28. Flora, S., 2004, Hydrogeological characterization and discharge variability of springs in the Middle Verde River watershed, M.S. Thesis, Northern Arizona University, Flagstaff, AZ. Ford, D., Williams, P., 2007, Karst Hydrology and Geomorphology, West Sussex, England, John Wiley & Sons Ltd. Ford, D. C., Williams, P.W., 1989, Karst Geomorphology and Hydrology, London, UK, Chapman & Hall. Frank, S., Goeppert, N., Ohmer, M., Goldscheider, N., 2019, Sulfate variations as a natural tracer for conduit‐matrix interaction in a complex karst aquifer: Hydrological Processes, v. 33. Gerst, J. D., 2010, Epikarst control on flow and storage at James Cave, VA: an analog for water resource characterization in Shenandoah Valley karst. M.S. Thesis, Virginia Tech Department of Geosciences, Blacksburg, VA. Geyer, T., Birk, S., Liedl, R., Sauter, M., 2008, Quantification of temporal distribution of recharge in karst systems from spring hydrographs: Journal of Hydrology, v. 348.

38

Giacopetti, M., Materazzi, M., Pambiachi, G., Posavec, K., 2017, Analysis of mountain springs discharge time series in the Tennacola stream catchment (central Apennine, Italy): Environ Earth Sci, v. 76. Hall, F. R., 1968, Base-Flow Recessions-A Review: Water Resources Research, v. 4. Hergenroder, J. D., 1957, Geology of the Radford Area, Virginia. Geology. Blacksburg, VA, Virginia Polytechnic Institute and State University. M.S. thesis. Herman, E. K., Toran, L., White, W., 2009, Quantifying the place of karst aquifers in the groundwater to surface water continuum: A time series analysis study of storm behavior in Pennsylvania water resources: Journal of Hydrology, v. 376. Kaufmann, G., Braun, J., 2000, Karst aquifer evolution in fractured, porous rocks: Water Resources Research, v. 36. Klimchouk, A., 2003, Towards defining, delimiting, and classifying epikarst: Its origin, processes, and variants of geomorphic evolution: Speleogenesis and Evolution of Karst Aquifers, v. 2. Kresic, N., 2007, Hydrogeology and groundwater modeling, Boca Raton, CRC Press. Li, Q., Hailong, S., Wang, J. 2010, Hydrochemical Response of Epikarst Spring to Rainfall: Implications of Nutrition Element Loss and Groundwater Pollution: Polish Journal of Environmental Studies, v. 19. Li, X.-D., Liu, C.Q., Harue, M., Li, S.L., Liu, X.L. 2010, The use of environmental isotopic (C, Sr, S) and Hydrochemical tracer to characterize anthropogenic effects on karst groundwater quality: a case study of the Shuicheng Basin, SW China: Applied Geochemistry, v. 25. Liu, A. W., Brancelj, A., Burnet, J., 2016, Interpretation of epikarstic cave drip water recession curves: a case study from Velika Pasica Cave, central Slovenia: Hydrological Sciences Journal, v. 16. Lo, M.-H., Famiglietti, J., Yeh, P. J.-F., Syed, T.H., 2010, Improving parameter estimation and water table depth simulation in a land surface model using GRACE water storage and estimated base flow data: Water Resources Research, v. 46. Mangin, A., 1984, Pour une meilleure connaissance des systemes hydrologiques partir des analyses correlatoire et spectrale: Journal of Hydrology, v. 67. Massei, N., Wang, H.Q., Field, M.S., Dupont, J.P., Bakalowicz, M., Rodet, J., 2006, Interpreting tracer breakthrough tailing in a conduit-dominated karstic aquifer: Hydrogeology Journal, v. 14. Mayaud, C., Wagner, T., Benischke, R., Birk, S., 2014, Single event time series analysis in a binary karst catchment evaluated using a groundwater model (Lurbach system, Austria): Journal of Hydrology, v. 511. Nathan, R. J., McMahon, T.A., 1990, Evaluation of Automated Techniques for Base Flow and Recession Analysis: Water Resources Research, v. 26., 1990, Identification of Homogeneous Regions for the Purposes of Regionalization: Journal of Hydrology, v. 121.

39

Ning, L., Crane, R., Wagener, T., Najjar, R, Singh, R., 2012, Probabilistic Projections of Anthropogenic Climate Change Impacts on Precipitation for the Mid-Atlantic Region of the United States: Journal of Climate, v. 25. Padilla, A., Pulido-Bosch, A., Mangin, A., 1994, Relative Importance of Baseflow and Quick flow from Hydrographs of Karst Spring: Ground Water, v. 32. Posavec, K., Bacani, A., Nakic, Z., 2006, A Visual Basic Spreadsheet Macro for Recession Curve Analysis: Ground Water, v. 44., 2010, Fully Automated Objective-Based Method for Master Recession Curve Separation: Ground Water, v. 48. Scanlon, B., Healy, R., Cook, P., 2002, Choosing appropriate techniques for quantifying groundwater recharge: Hydrogeology Journal, v. 10. Smart, P., and Friederich, H., 1987. Water movement and storage in the unsaturated zone of a maturely karstified carbonate aquifer, Mendip Hills, England; Proceedings of Conference on Environmental Problems in Karst Terranes and their Solutions: National Water Well Association, Dublin, Ohio, p. 59-87. Szilagyi, J., Gribovszki, Z., Kalicz, P., 2007, Estimation of catchment-scale evapotranspiration from baseflow recession data: Numerical model and practical application results: Journal of Hydrology, v. 336. Tallaksen, T. M., 1995, A review of baseflow recession analysis: Journal of Hydrology, v. 165. Taylor, C. J., Nelson, H., Hileman, G., Kaiser, W., 2005, Hydrogeologic-Framework Mapping of Shallow, Conduit-Dominated Karst—Components of a Regional GIS-Based Approach: US Geological Survey Karst Interest Group Proceedings. Rapid City, South Dakota. Taylor, C., Green, E., 2008, Hydrogeologic Characterization and Methods Used in the Investigation of Karst Hydrology, in Field Techniques for Estimating Water Fluxes Between Surface Water and Ground Water, U.S. Geological Survey, Reston VA Wang, D., Cai, X., 2009, Detecting human interferences to low flows through base flow recession analysis: Water Resources Research, v. 45. White, W. B., 2002, Karst hydrology: recent developments and open questions: Engineering Geology, v. 65. Williams, P. W., 1983, The Role of the Subcutaneous Zone in Karst Hydrology: Journal of Hydrology, v. 61. Yager, R., Plummer, N., Kauffman, L., Doctor, D., Nelms, D., Schlasser, P., 2013, Comparison of age distributions estimated from environmental tracers by using binary-dilution and numerical models of fractured and folded karst: Shenandoah Valley of Virginia and West Virginia, USA: Journal of Hydrology, v. 21.

40

Virginia Department of Mines and Mineral Resources, 1978, Geologic Map of Virginia.

41

APPENDIX A

Monthly Total Precipitation (10/2007-6/2019) Year Month Precip total (mm) 2007 10 76.45 2007 11 9.14 2007 12 50.80 2008 1 23.62 2008 2 30.99 2008 3 42.93 2008 4 91.95 2008 5 52.83 2008 6 53.09 2008 7 106.17 2008 8 14.73 2008 9 23.88 2008 10 0.51 2008 11 34.80 2008 12 62.23 2009 1 23.11 2009 2 31.50 2009 3 62.74 2009 4 31.75 2009 5 64.77 2009 6 69.60 2009 7 56.39 2009 8 50.29 2009 9 36.07 2009 10 52.58 2009 11 70.10 2009 12 125.22 2010 1 82.04 2010 2 47.75 2010 3 62.99 2010 4 28.45 2010 5 52.58 2010 6 37.08 2010 7 87.12 2010 8 32.77 2010 9 55.63 2010 10 44.20 2010 11 56.39 2010 12 32.77

42

2011 1 7.87 2011 2 50.04 2011 3 79.50 2011 4 145.54 2011 5 85.09 2011 6 8.89 2011 7 59.69 2011 8 4.57 2011 9 49.02 2011 10 30.23 2011 11 36.58 2011 12 30.73 2012 1 28.19 2012 2 30.48 2012 3 36.07 2012 4 50.04 2012 5 35.81 2012 6 32.51 2012 7 50.04 2012 8 13.97 2012 9 46.48 2012 10 15.24 2012 11 16.51 2012 12 34.29 2013 1 66.80 2013 2 9.40 2013 3 32.51 2013 4 32.77 2013 5 24.89 2013 6 16.51 2013 7 31.75 2013 8 45.47 2013 9 12.45 2013 10 22.10 2013 11 41.91 2013 12 62.48 2014 1 8.64 2014 2 47.75 2014 3 68.33 2014 4 125.73 2014 5 119.38 2014 6 80.77 2014 7 93.98

43

2014 8 266.70 2014 9 61.98 2014 10 221.49 2014 11 86.11 2014 12 76.45 2015 1 30.48 2015 2 54.86 2015 3 84.84 2015 4 102.11 2015 5 168.66 2015 6 113.03 2015 7 285.50 2015 8 180.85 2015 9 123.95 2015 10 90.68 2015 11 188.47 2015 12 169.93 2016 1 66.55 2016 2 129.29 2016 3 72.90 2016 4 61.21 2016 5 120.65 2016 6 176.02 2016 7 159.77 2016 8 175.51 2016 9 90.68 2016 10 49.02 2016 11 56.39 2016 12 78.23 2017 1 85.85 2017 2 41.66 2017 3 94.49 2017 4 148.34 2017 5 175.01 2017 6 106.68 2017 7 135.89 2017 8 165.86 2017 9 41.66 2017 10 184.40 2017 11 20.57 2017 12 17.02 2018 1 41.91 2018 2 118.62

44

2018 3 119.13 2018 4 93.73 2018 5 157.23 2018 6 35.56 2018 7 100.84 2018 8 279.15 2018 9 156.97 2018 10 162.31 2018 11 103.38 2018 12 72.90 2019 1 88.14 2019 2 109.98 2019 3 26.92 2019 4 158.24 2019 5 81.28 2019 6 95.25

45

APPENDIX B

Drip data

The drip datasets will be made available in VTechData.

46