A complete understanding of geochemistry is not a requirement for evaluating water-quality data to facilitate understanding karst aquifers. The main characteristic of karst aquifers is that they are composed of carbonate and evaporite rocks (rocks that can dissolve over time). Limestone is predominantly composed of calcite (CaCO₃). Sand, clay and silt are commonly found in limestone as impurities but are not common in dolomite. If a limestone contains a lot of clay size particles or clay minerals, it is sometimes called a mudstone and generally forms a semi-confining unit with K < 0.1 m/d. Dolomite (CaMg[CO₃]₂) can be formed by dolomitization of limestone, where the calcite is recrystallized and magnesium (Mg) replaces some of the calcium (Ca). The exact processes of dolomitization remains an area of research and is not completely understood. Most carbonates and evaporites are deposited in marine environments. One exception is tufa that forms when calcite precipitates out of freshwater often near mineral springs. Evaporites are a non-clastic sedimentary rock composed primarily of minerals produced from a saline solution as a result of extensive or complete evaporation of the water and often occur as layers of gypsum (CaSO₄·2H₂O), anhydrite (CaSO₄) and halite (NaCl) within a sequence of carbonate rocks. Evaporite minerals dissolve faster than limestone and thus, most evaporite rocks are found in more arid environments or at depth within carbonate sequences where freshwater has not flushed out the original seawater. Saline or hyper-saline water or SO₄ within an aquifer system may indicate the presence of evaporite layers.

Rainwater can be slightly acidic, because carbon dioxide and water in the air react to form carbonic acid, a weak acid. As this acidic rainwater moves through carbonates and evaporites, it dissolves rocks until the dissolved rock minerals reach their saturation concentration in water. This process is called chemical weathering in many textbooks. Thus, dissolution begins in fractures and joints. It occurs at greater rates closer to land surface, or at the current base level of rivers, or at the base of sea cliffs. In confined zones, after large conduits or preferential flow layers have formed, new chemically under-saturated water may move through the system and dissolution will continue. In general, the longer time water remains in the carbonate rock, the higher the total dissolved solids. Thus, the collection of specific conductance data can be useful in karst aquifer studies. These data are presented as a chemograph and may indicate the presence of an evaporite layer at depth, seawater intrusion near the coast, or zones of slower diffuse flow versus faster conduit flow.

Basic Water Quality Data

If basic water-quality data are available in a study area, contour maps of individual constituents can indicate the location of potable water. The simplest water-quality data to obtain is specific conductance, which can be used to estimate total dissolved solids (Hem, 1985). Often for drinking water supply studies, water-well samples are collected, and these can be compiled for the study area. Common standard inexpensive laboratory water quality analyses include acidity (pH), electrical conductance (EC), total dissolved solids (TDS), cations [calcium (Ca²⁺), magnesium (Mg²⁺), sodium (Na⁺) and potassium (K⁺)] and anions [bicarbonate (HCO³⁻), chloride (Cl⁻), sulfate (SO₄²⁻), nitrate (NO₃⁻) phosphate (PO₄²⁻) or orthophosphate (PO₄³⁻)]. Common field parameters measured are pH, temperature, and EC (a surrogate for TDS).

If laboratory analyses are available, then more complex graphical water quality analyses will provide insight. The most common water quality diagrams used to evaluate the anion/cation chemistry are Stiff diagrams (Stiff, 1951) and trilinear diagrams (Piper, 1944).

A Stiff diagram is a graph with four parallel horizontal axes extending on each side of a vertical zero axis. Concentrations of four cations and four anions are plotted on the left and right side of the vertical axis respectively. They are all plotted in the same sequence with concentration in milliequivalents per liter at the same scale. The plotted points are connected to create an irregular polygonal shape (Hem, 1985, page 175). Often the different water types in an aquifer show up as similar shaped polygons. Stiff patterns can be plotted on a map (Figure 58). The color coding on this map reflects the hydrogeologic unit the water sample was taken from. The Stiff diagrams within a unit do not all have a similar polygon shape indicating that the water quality depends on influences other than the formation chemistry.

Figure 58 – Map of Stiff diagrams as presented with Google Earth. The colors on the diagram indicate samples from different hydrogeologic units. Note how the relative shape of the stiff diagrams from the same unit are fairly similar, indicating that there are water quality differences in different hydrogeologic units and demonstrating the usefulness of plotting the individual stiff diagrams on a map. Image created by Keith J. Halford and copied from free software available for Piper and Stiff diagrams.

Trilinear plotting systems have been in use since the early 1900s. All cation and anion analyses are computed in terms of milliequivalents per liter and some are added together, such that the total number of lumped anion groups is three and cation groups is three. Then the values are calculated as a percentage of the total cation milliequivalents or anion milliequivalents. The simplest diagrams are two equilateral triangles of the same size with each side scaled from 0 to 100; one for the cation groupings and one for the anion groupings. Then a diamond shaped plotting area floats above and in the center of the cation and anion equilateral triangle. The axes of the diamond are parallel to the triangles for cations (lower left) and anions (lower right), and are the same as the equilateral triangles. The upper left of the diamond is usually SO₄ plus Cl and upper right Ca plus Mg (Figure 59). The diagram helps identify chemical facies and water types. All the sample analyses from wells in the study area are generally plotted together. On some trilinear diagrams the relative magnitude of total dissolved solids is shown by using different diameter circles (Hem, 1985, Figure 37). On others, the spatial nature is indicated by using a different symbol for each sample and having another map that indicates the location of each sample; still others use a number for each sample rather than symbol on both the trilinear diagram the map. The number can be used to link the full laboratory analysis provided in a database to the sample location.

Figure 59 – The Piper diagram can be separated in hydrochemical facies. The letter and number labels are as follows: A – Calcium type; B – No dominant type; C – Magnesium type; D – Sodium and Potassium type; E – Bicarbonate type; F – Sulphate type; G – Chloride type; 1 – Alkaline earths exceed alkali metals; 2 – Alkali metals exceed alkaline earths; 3 – Weak acids exceed strong acids; 4 – Strong acids exceed weak acids; 5 – Magnesium bicarbonate type; 6 – Calcium chloride type; 7 – Sodium chloride type; 8 – Sodium bicarbonate type; 9 – Mixed type. Image by Kvgunten, 2017, Hydrochemical facies in the Piper diagram licensed under CC BY-SA 4.0.

Plotting analyses of samples from wells that are successively down gradient from each other may reveal a linear chemical trend that indicates a mixing process. The graphs may lead to development of hypotheses that require more advanced research using various mixing models. The diagrams are a staple for evaluating natural-water chemistry of both surface and groundwater (Hem, 1985). The pros and cons of various diagrams used in geochemistry are discussed in Dauda and Habib (2015).

This section has not discussed concepts from basic inorganic chemistry or geochemistry. For readers who would like to know more about natural water chemistry, Hem (1985) provides an excellent free book (a link is provided in the reference section).

Exercise 21 invites the reader to consider the difference between alkaline earths and alkali metals.

Use of End Member Mixing to Estimate Groundwater Contribution to Surface Water

Water quality data is often used in groundwater studies and particularly in karst systems for: understanding groundwater surface-water interactions; investigation of flow paths at contaminated sites; understanding the origins of groundwater and the formations the water has moved through; determining water age; and identifying mixture of waters of different age or from different formations.

Groundwater investigations for water supply in karst aquifers often need to characterize the groundwater discharge at springs or within a river reach (groundwater/surface-water interaction). If there is a chemical, an isotope, or a physical property present in the groundwater; but not present in the surface water (or is present at substantially different concentration or magnitude) that information can be used to develop a mixing model. The mixing model is used to determine the amount of groundwater in the surface water.

Radon-222 (²²²Rn) is one radioactive isotope of the inert radon gas that is used to directly estimate the groundwater discharge to a surface water. Radon-222 is a radioactive decay product from Uranium-Thorium minerals, which are within many rocks in varying amounts (Klepper and Wyant, 1957; René, 2017). Radon-222 is the longest-lived of the Rn isotopes with a half-life of 3.82 days, and is the most-studied isotope of Rn. Natural radon concentrations in the atmosphere are so low that surface waters in contact with the atmosphere will continually lose radon by volatilization so the ²²²Rn concentration in surface water is negligible. Consequently, where ²²²Rn is present in rocks of an area, groundwater has a higher concentration of ²²²Rn than surface water. Kraemer and Genereux (1998) provide a detailed discussion of ²²²Rn mixing models and the use of ²²²Rn to determine areas of groundwater discharge to streams. The use of ²²²Rn for this purpose has been applied to all aquifer types including karst aquifers.

If the concentration of a constituent in groundwater is constant and in large contrast to the concentration in surface runoff from precipitation and overland flow, then data describing continuous discharge of a stream and continuous concentration of the constituent of interest in the stream can be collected and groundwater discharge to the surface water body can be estimated. In short, one must know the two end member concentrations. In the case of ²²²Rn, it is assumed that the concentration in the surface runoff is zero, so all ²²²Rn measured in surface water is from local groundwater discharge to the surface water.

Assumptions/simplifications involved in estimating the groundwater discharge are:

• 100 percent surface runoff (precipitation plus overland flow) occurs after the largest storm event peaks:
• the constituent end member concentration of surface runoff, C_r, can be determined by measuring C of stream water after extremely wet periods (note in the case of Radon-222 the surface water end member is zero); and,
• the groundwater end member, C_g, can be determined from multiple groundwater samples.

Even if the groundwater concentration changes a small amount through time due to recharge events, if that concentration is measured and the surface-water end member concentration (C_r) is stable, an estimate of the groundwater discharge can be made using the chemical mass balance of Equation 7 and the streamflow mass balance of Equation 8.

	CQ = CgQg + CrQr	(7)
	Q = Qg + Qr	(8)

where:

C	=	concentration in the stream water (ML−3)
Cg	=	concentration in 100% groundwater (ML−3)
Cr	=	concentration in 100% surface runoff water (ML−3)
Q	=	total discharge of the stream (L3T−1)
Qg	=	groundwater portion of the total discharge of the stream (L3T−1)
Qr	=	surface runoff portion of the total discharge of the stream (L3T−1)

Figure 60 and Equation 9 show the relationship for calculating groundwater discharge to the stream (Q_g).

(9)

When the constituent is thought to only occur in groundwater, for example in the case of ²²²Rn, the concentration in surface runoff water C_r can be taken as zero, then Equation 9 simplifies to Equation 10.

(10)

Figure 60 – Given the end-member concentration of groundwater (C_g) and surface runoff (C_r), the groundwater component (Q_g) of total streamflow (Q) is computed from the streamflow concentration (C) time series.

An end-member mixing model can be used to calibrate traditional graphical hydrograph separation and thus, better estimate the groundwater component of stream flow. This component is called base flow. The procedure was developed by Stewart and others (2007) and applied to the Floridan aquifer system near Tampa by Kish and others (2010). Stewart and others (2007) used two years of continuous specific conductance data (a surrogate for total dissolved solids) and discharge measurements collected at 10 stream gages operated by the US Geological Survey to adjust the parameters of the automated hydrograph separation program HYSEP (Slotto and Krause, 1996) in order to calibrate the hydrograph-separation model to the conductance mass balance equation. The end member values for conductance were determined by measuring conductance of water samples from wells in the aquifer and conductance of stream water during dry periods to determine C_g, and measuring conductance of stream water after extremely wet periods to determine C_r. Calibration of the value of the 2N* parameter of HYSEP reduced the 10-station average error between HYSEP and conductance-mass-balance-derived cumulative base flow from 40 percent to 5 percent (Stewart et al., 2007). Calibration is the process of adjusting the value of model parameters until the values simulated by the model match the measured field observations. These calibrated HYSEP parameters can be used with historical daily streamflow records to estimate the groundwater component of total streamflow more accurately for that site with HYSEP.

Figure 61 shows a graph of base flow estimates from a storm event using the end-member mass balance approach with different physical and chemical data (Kish et al., 2010). Kish and others (2014) contend that the continuous specific conductance mass balance estimate of groundwater discharge is the most accurate because it is measured at 15-minute intervals as is stream discharge; whereas the estimates from sampled specific conductance, calcium, and magnesium concentration were sampled periodically, not conducted daily. Kish and others (2014) do not speculate why the estimate using magnesium is lower than using calcium. By connecting the groundwater contribution to total flow estimated from periodic sampling points of calcium concentration results in that line above the total discharge line, which is obviously not possible (Figure 61).

Figure 61 – Observed storm hydrograph and groundwater discharge estimated by geochemical mass balance for selected major ions at the SR39 surface water gaging site near Tampa, Florida, USA, March 2004 storm. From Kish and others (2010).

Further information about hydro-geochemistry is discussed by Drever (1988), Clark and Fritz (1997) and Deutsch (1997). Additionally, the natural temperature variation between groundwater and surface water can be used as a tracer (Kurylyk et al., 2017).

Use of Natural Stable Isotopes

Natural stable isotope chemistry has also been used for understanding diffuse versus conduit flow with other mathematical models. Isotopes are atoms of the same element that have different numbers of neutrons. Isotopes have the same number of protons (positive charge) and electrons (negative charge) but differ in molecular weight due to different numbers of neutrons (neutral charge). The previous section discussed Radon-222 (²²²Rn), which is a naturally occurring radioactive isotope of radon gas and considered an unstable isotope owing to radioactive decay and subsequent transformation to a daughter product. There are many naturally occurring stable isotopes that are used in hydrology, such as Hydrogen-2 (²H), Oxygen-16 (¹⁶O), Oxygen-18 (¹⁸O), Carbon-12 (¹²C), and Carbon-13 (¹³C). The hydrogen and oxygen isotopes are frequently analyzed for hydrologic studies as these make up the water molecule. The natural abundance of these isotopes differs over time and climatic conditions. In groundwater studies, stable isotopes are often used to understand recharge. An excellent reference on the use of stable isotopes in hydrology edited by Kendall and McDonnell (1998) has a chapter devoted to the use of isotopes in groundwater hydrology. A Groundwater Project book by Cook (2020) introduces isotopes and environmental tracers as indicators of groundwater flow.

Water Quality Issues for Water Supply

In the United States, private water-well permits are managed at the local or State level, however, minimum drinking water standards are set Nationally by the US Environmental Protection Agency (USEPA, 1991, 2001, 2009). When exploring karst aquifers for water supply its important to collect base line water chemistry data of cations and anions and, for shallow karst systems fed by sinking streams, bacteria.

If evaporite layers were deposited along with the carbonate rocks, often brackish or hypersaline groundwater is present. Evaporite rocks dissolve more readily than pure limestone and create areas of very high total dissolved solids within an aquifer system or deeper units where the original seawater is never replaced through freshwater circulation as in the Floridan aquifer system (Williams and Kuniansky, 2015). However, brackish water can be treated or mixed with fresher water to lower the total dissolved solids to potable levels (Stanton et al., 2017).

Karst aquifers can be especially vulnerable to contamination from human activity as noted in Section 3.7; thus, surface land-use can introduce toxic substances into these aquifers. In some cases, the substances are naturally present or can be mobilized from natural materials by introducing water of a differing chemistry. Aquifer storage and recovery efforts stalled in Florida because the chemistry of injected surplus surface water mobilized arsenic and other trace elements naturally occurring within the limestones (Cowart et al., 1998; Arthur et al., 2001, 2002, 2007; Williams et al., 2002).

Exercise 22 invites the reader to consider land use activities that might result in degradation of the water quality if conducted on the outcrop or near sinking streams of a karst aquifer.