3.4 Variance in Surface-Water and Spring Discharge in Karst Aquifers
Spring discharge monitoring is one of the most insightful tools used by karst hydrogeologists. Monitoring of variability of spring flow (discharge) and selected water-quality characteristics such as water temperature, pH, specific conductance, and turbidity, provides much of the most useful and important information about the hydrologic characteristics and internal functioning of karst aquifers that can be acquired by field studies (Taylor and Greene, 2008). Modern technological advancements in electronic sensing probes and data-logging equipment have greatly aided and simplified the collection of continuous spring discharge data (Figure 18). Analysis of continuous, high-frequency (short-intervals between individual measurements) spring discharge and water-quality data provides amazing insight into karst aquifer recharge, storage, and discharge functions, and how these change under short-term specific storm events. Long-term continuous monitoring and analysis is used for understanding seasonal changes in weather and annual-to-decadal climate change. A great variety of analytical methods have been devised and used in the analysis of hydrograph plots of spring discharge characteristics. Analyses of peak flow and discharge recession has long been used to assess water-supply potential and sustainability, and to attempt to evaluate the relative contributions of water stored in karst matrix or fracture networks versus water transmitted through conduits. Because of the number and diversity of methods and publications that have been devoted to these topics over many decades, a more detailed and comprehensive review of spring discharge analysis is beyond the scope of this introductory book. Summaries provided by Taylor and Greene (2008) and Goldscheider and Drew (2007), and textbooks by Kresic (2007), Milanovic (1981), and Stevanovic (2015) contain more detailed information.
Figure 18 – Deployment of continuous multi-parameter probes and data-logging equipment at karst spring sites is one of the most useful investigative tools available in the study of karst aquifers. a) Charles Taylor inspecting a mulit-parameter water quality sonde at a karst spring (photo provided by Taylor, Kentucky Geological Survey, 2021). b) Example of hydrograph plotting of continous discharge and water-quality data collected from a karst spring, showing trends and relationships between discharge or flow (Q), specific conductance (SEC), water temperature (T), turbidity, total organic carbon (TOC), and bacterial counts (E. coli), following a storm event (denoted by precipitation (P). Modified from Hartmann and others (2014).
Variability in discharge and water-quality hydrograph plots from continuously monitored springs generally falls between two end members described by the terms “diffuse or slow flow” and “conduit or rapid flow”. Rapid, high-amplitude changes, sometimes described as “flashiness”, in discharge and water chemistry parameters such as turbidity, pH, temperature, or specific conductance have been interpreted as representing a karst aquifer system dominated by conduit flow, whereas gradual, buffered or muted patterns of change are interpreted as indicating dominance of diffuse, non-conduit flow. Most karst aquifers possess a combination of slow-velocity (diffuse) and rapid velocity (conduit) permeability components. Consequently, patterns in hydrochemistry and/or discharge hydrographs, and hysteresis plots sampled across storm events (during rise, peak, and recession periods) are best interpreted as reflecting temporal changes caused by changes in timing, proportions, and mixing of recharge contributed from matrix, fracture, and conduit sources. The water fluxes that generate these changes are head dependent and often driven by recharge events (Figure 19).
Figure 19 – Generic example of the spring discharge behavior from a non karst spring and a mature karst spring. Non karst springs have slower rise and fall in discharge from rain events. A mature karst spring is flashy owing to the rapid movement of rainwater underground through large dissolution features to the spring and into surface streams.
Groundwater and surface-water are almost inseparable in a well-developed karst watershed. The streamflow gaging station for the Orangeville Rise in Indiana, USA, provides an example of a streamflow hydrograph that represents a stream predominantly fed by karst conduits (Figure 20). With the 15-minute unit-value stream discharge data plotted with the daily precipitation the very short time lag between storm event and the peak discharge indicates rapid flow to the gage through conduits. Additionally, for some storm events, a double peak indicates two distinct sources of discharge with differing lag times from the storm to the peak discharge. These double peaks show that the two sources of rapid flow from a storm event are out of phase. These out of phase peaks require further investigation as they could be the result of uneven rain distribution over various sub watersheds, differing antecedent conditions resulting in water moving in different elevation conduit systems, or water moving from sub watersheds with different times of travel. The lag times for the peaks varies from storm to storm and may also be influenced by antecedent conditions, such as, activation of a higher elevation conduit system (overflow conduits). For the hydrograph of Figure 19, it appears that the velocity of diffuse flow (water moving slowly through the system) is approximately 9 to 20 ft3/s (cubic feet per second) (~0.3 to 0.6 m3/s) based on the horizontal portion of the hydrograph between storm events. The streamflow is also dependent on antecedent conditions.
Figure 20 – Daily precipitation (vertical bars) and 15-minute unit-value stream discharge data for U.S. Geological Survey streamflow-gaging station 03373550 (Orangeville Rise at Orangeville, Indiana, USA), showing double-peak nature of storm peaks and the lag time between rainfall and discharge from the spring. Modified from Bayless and others (2014).
Hydrograph separation is a method by which overland runoff is separated from stream base flow. Base flow can be dominated by groundwater discharge; however, watersheds vary, and base flow can have contributions from bank storage, or slow snow melt, or in flat terrain very slow surface-water drainage, thus may not be predominantly groundwater discharge. Continuous monitoring of discharge at the spring itself is best but not always possible. At the Orangeville Rise gage, daily streamflow hydrographs were processed using an automated hydrograph separation program with the assumption that most of the base flow is groundwater discharge from the karst basin (Figure 21). The daily streamflow hydrograph at Orangeville Rise does not exhibit double peaks as are visible in the 15-minute unit-value discharge data. Use of graphical hydrograph separation methods to estimate groundwater discharge is greatly improved by using water quality data as demonstrated in Section 5.5 of this book. The Orangeville Rise stream gage is within a well-developed karst area where base flow represents between 53 to 98 percent of the flow as estimated using monthly data for station 03373550 (Orangeville Rise at Orangeville, Indiana), November 1, 2011 to February 28, 2013 (Bayless et al., 2014, Table 8).
Figure 21 – Estimated base flow, following Sloto and Crouse (1996, page 4), for U.S. Geological Survey streamflow-gaging station 03373550 (Orangeville Rise at Orangeville, Indiana), November 1, 2011 to February 28, 2013. From Bayless and others (2014).
The Indiana study site provides an example of hysteresis. For this example, a plot of water level in a well in the basin versus 15-minute discharge data for a storm event at the Orangeville Rise gage indicates a looping pattern (Figure 22). The hysteresis plot is created by plotting the pairs of unit values and connecting these as a line between adjacent points in time and noting the rising limb and falling limb of the stream gage hydrograph. As streamflow increases on the rising limb, the groundwater level also rises. The graph line created by this time series loops over itself and the falling limb of the graph line is above the rising limb line with groundwater level on the y-axis and stream discharge on the x-axis (both linear axes). This indicates that the groundwater level peak lags behind the streamflow peak. The analysis of data for Orangeville Rise gage, shows how examination of 15-minute discharge data indicates possible networks of conduits at different elevations and the flashiness of base flow indicates multiple porosity within the karst aquifer.
Figure 22 – Fifteen-minute unit-value water levels in the Marshall Farm Well, Indiana (383840086301101), versus 15-minute unit-value streamflow from the Orangeville Rise, Indiana gage station (03373550), showing a counterclockwise hysteresis that is indicative of the groundwater peak lagging slightly behind the surface-water peak for a storm peak starting on September 7, 2012. From Bayless and others (2014).
Wong and others (2012) report that telogenetic karst systems tend to reflect flow through the conduit networks and fractures with flashy spring discharge as there is little flow or storage in the rock matrix. Whereas in eogenetic karst systems, which tend to have some storage and interconnected voids in the rock matrix, the spring discharge can be dampened by this storage. Florea and Vacher (2006) note several significant differences in apparent hydraulic/hydrologic behavior between eogenetic and telogenetic karst aquifer types:
- the flashiness or ratio of maximum to mean (Qmax/Qmean) discharge is smaller in springs of eogenetic karst than springs of telogenetic karst;
- aquifer inertia (system memory) is larger in eogenetic karst because:
- eogenetic karst aquifers have a buffered or longer response time to recharge inputs; and,
- high-frequency storm events affect discharge less in eogenetic karst, basically reflecting differences in interaction between the matrix and conduits because of differences in matrix porosity and permeability between eogenetic and telogenetic karsts.