An Iconoclast's view of hydrogeologic studies

Ground water investigators love to drill test holes and are never satisfied no matter how many there might be. Once a well is drilled, the hydrologist will probably measure the static level, collect a water sample, and perhaps even conduct an aquifer test, unless he or she wants to test the pump, in which case a pump test is conducted.

Each additional measurement or task creates a new series of problems, which in turn give rise to the ever increasing need for additional test drilling. For example, the well screen may be open to several water-bearing zones, each of which is characterized by a different quality and head. Therefore, the original water level measurement and water sample reflect a composite. Moreover, each water-bearing zone will likely have a permeability that differs from the others, which in turn calls for more observation wells in order to conduct more aquifer tests, which will show that the entire system is as i leaky as a sieve.

Then, of course, we cannot be satisfied with the water sample because either its analysis did not conform to our preconceived notions-we could not explain the presence or absence or concentration of some constituents (obviously the chemist messed up)-or it was not collected, shipped, stored, . or analyzed in the correct manner in the correct time limit (someone else's fault). Perhaps the well was not pumped long enough or was pumped too

long, or perhaps it was not pumped at all (blame this one on the pump). But whatever the reason, it hardly matters anyway because the sample does not actually represent the geochemical conditions that exist in the subsurface.

Obviously, with all of these problems, additional drilling is warranted. Next we need to examine the water level that was originally measured.
First, since it reflects a summation of all the water levels of all the water bearing zones exposed to the screen (assuming no leakage along the casing, which can easily justify at least four more new wells), it is obvious that the measurement is meaningless, worthless, confusing, and inaccurate; and therefore, more wells and measurements are required. After all, if one is going to estimate permeability or ground water recharge, one must have accurate measurements of water levels over a long period (unless the esti mate is based on a computer simulation, in which case one needs no information at all).

It is a well-known fact that the cost of many hydrologic studies, particularly those funded by federal agencies, can get rather out of hand because of a lack of confidence, experience, or competence, or because of an abundance of ignorance or stupidity, or because of the desire for an apparent degree of accuracy that exceeds all boundaries of realism, or because we want to protect ourselves from legal action or from sharp-tongued attorneys representing "the other side," or because of the need to provide enough work to last until retirement so that the final report can be written by some one else, or because the project must be dragged out long enough to permit the investigator to stay in the field until the kids grow up and leave home.

There is no doubt but that test drilling is required in nearly every hydrogeologic study, but it is equally obvious that costs are rising so quickly and to such an extent that other sampling methods and techniques to evaluate the subsurface must be considered, developed, and used. Furthermore, for every project or investigation there should be developed, as a first step, an adequate experimental design that incorporates all available data and is based on fundamental hydrogeologic principles. As a matter of fact, it might be a good idea to prohibit all drilling until the final report is prepared; the drilling budget could be used to prove the validity of the already written final report. If the subsurface data indicate that parts of the report are incorrect, then modification and reinterpretation are required. In many cases this actually can be done.

Site-Specific versus regional studies

Various techniques are available for development of at least a general under standing of the hydrogeologic framework, but their usefulness depends on our perspective. The methods can be thought of as point source (test drilling) or non-point source (regional or areal examination) techniques. In far too  many investigations point source data are used to the total or nearly total exclusion of non-point data. The usual justification which is valid to some extent, is based on site size, with the smaller site receiving the greater detail.
For example, in a examination of a 100 mi2 area, 10 test oles might be considered adequate because the thickness and permeability of the earth materials are expected to range within fairly broad limits and the water level elevation and gradient are predictable. Conversely, in an examination of a 120-acre potential municipal landfill site, 10 test holes might be considered entirely inadequate because great detail may be required to obtain a permit and to overcome the objections voiced by local landowners, the permiting agency, or environmental groups. In either case, the purpose of drilling is to evaluate or determine permeability, thickness, areal extent of selected rock units, water levels, and water quality in orden to predict what will happen when certain stresses are placed on the system.

On the other hand, how much permeability information does a test hole really provide considering the sample return, drilling method, and the exceedingly smal area that the hole represents? For example, increasing evidence clearly shows that macropores and fractures play a major role in ground water recharge and quality in addition to wel-yield, and this type of secondary permeability is not likely to be evident from test drilling alone.

Remote sensing

It has long been known that water wells drilled on or adjacent to fractures produce higher yields than do other nearby wells. Some date indicate that production of hydrocarbons is also greater adjacent to fracture traces. It is strongly suspected that fracture systems exert a major though subtle control on hydrogeology as they influence permeability, hydraulic gradients, velocity, recharge, and the influx of chemicals to the ground water system. Fracture systems range tens of miles in length at one extreme to a microscale at the other.

As a first step in mapping fracture systems, satellite imagery can be used since long linear features are readily evident and a map of them provides a framework for determining major trends. Next, low altitude photography can be utilized to key on a specific site. Fields checks may be necessary.
Commonly, fracture trend and density can be examined in considerable detail on the ground, particularly in surface mines such as quarries.

Stream hydrographs

Another useful method for evaluating regional hydrogeology is analysis of streamflow data, which are abundant, widely available, and easy to acquire.

Since a stream is nothing more than an extremely long, very shallow, horizontal well, analysis of its flow will provide clues to the permeability, chemical quality, and potential yield of at least shallow aquifers or those that provide ground water runoff to a stream.

Several stream hydrograph separation techniques are available, including some that are computer-based (Pettyjohn and Henning, 1979). Effective ground water recharge rates, both monthly and annually, can be estimated by relating the volume of ground water runoff to the size of the drainage basin. Granted, the ratios obtained reflect regional conditions, but these may be more realistic than data obtained from rain simulators, infiltration rings, or analyses of ground water hydrographs, all of which are highly site-specific.

A computer-separated stream hydrograph is shown in Figure 21.1. This stream, which lies in a low relief till plain in south central Ohio, has a drainage area of 228 mi2. Although not evident from the surface, several relatively thin deposits of sand and gravel are incorporated within the till and provide ground water runoff to the tributaries and mainstream of Deer Creek. The calculated effective ground water recharge rate in the basin above the gaging station during a year of normal precipitation (1967) was 290,000 gpd/miz (0.01 gpd/ft2) or 6.10 in.; thus ground water accounted for 46.4% of the stream's total flow.

Effective ground water recharge rates in different hydrogeologic terrains in Ohio, as indicated by ground water runoff, range from 123,000 to 291,000 gpd/miz (average 291,000) in till covered areas; from 310,000 to 406,000 gpd/miz (average 352,000 in outwash covered areas); and from 160,000 to 198,000 gpd/mi2 (average 179,000) in the moderate to high relief unglaciated part of Ohio, where alternating layers of sandstone, shale, coal, and limestone crop out.

Regional data of the type described above can be very useful, but they are not site-specific since they represent all of the cause-and-effect relations that occur upbasin. Nonetheless, when used in conjunction with other regional techniques, such as flow-duration curves and ratios and seepage measurements, they prove a good first approximation of hydrogeologic conditions, provide good information on regional differences in permeability, and could significantly reduce drilling costs.

Flow-Duration Curves

As pointed out by Cross and Hedges (1959), flow-duration curves are widely used, and there is considerable technical literature on the subject. From a hydrogeologic viewpoint they are most useful for comparing the flow characteristics of different streams because the shape of the curve is an index of the natural storage within a basin. When plotted on logarithmic probability paper, the more nearly horizontal the curve, the greater is the effect of ground water storage.

During dry weather, the flow of a stream is almost entirely from ground water sources. The lower ends of duration curves therefore indicate in a general way the characteristics of the shallow ground water bodies in the drainage basin above the gaging station. Duration curves thus are useful guides in locating possible sources of ground water (Cross and Hedges, 1959, p. 5).

Despite the fact that changes in the drainage basin, such as regulation, diversion, and variable discharges of effluent, will alter the shape of the flow duration curve of some streams, records of even these streams can, by judicious comparison with unaffected nearby streams, provide useful ap proximation of the hydrogeologic system.

A flow-duration curve shows the frequency of occurrence of various rates of flow. It is a cumulative frequency curve prepared by arranging all discharges of record in order of magnitude and subdividing them according to the percentages of time during which specific flows were equaled or exceeded; all chronologic order or sequence is lost (Cross and Hedges, 1959). Flow-duration curves may be plotted on either probability or semilogarithmic paper.

Several flow-duration curves for Ohio streams are shown in Figure 21.2. During low-flow conditions (the flow equaled or exceeded 90% of the time), the curves for several of the streams-such as the Mad, Hocking, and Scioto Rivers and Little Beaver Creek-trend toward the horizontal, while those for Grand River and White Oak and Home Creeks all remain very steep.

Mad River flows through a broad valley filled with very permeable sand and gravel. The basin has a large ground water storage capacity and, consequently, the river maintains a high sustained flow. The Hocking River valley also contains outwash in and along its flood plain, particularly in the upper reaches, which provides a substantial amount of ground water runoff. Above Columbus, the Scioto River crosses glacial till and thin layers of limestone that crop out along the stream channel; ground water runoff in this reach is relatively small. Immediately south of Columbus, however, the Scioto River valley is filled with coarse outwash, and during low flow the discharge increases substantially at succeeding downstream gages.The reason that Mad River has a higher low-flow index than the Scioto River is that the former receives ground water runoff through its length, while ground water runoff to the Scioto River increases significantly only in the area of outwash south of Columbus.

White Oak and Home Creeks originate in bedrock areas where either relatively thin alternating layers of shale and limestone are covered by till (White Oak Creek) or sandstone, shale, and limestone crop out along hillsides (Home Creek). The low permeability of the strata and the greater relief in these basins preclude the storage of large amounts of ground water, and consequently the low flows characteristic of these streams are far less than those of the streams filled or partly filled with outwash.

Flow ratios

Walton (1970) reported that grain-size frequency-distribution curves are somewhat analogous to flow-duration curves in that their shapes are indicative of water-yielding properties of rocks. He pointed out that a measure of the degree to which all of the grains approach one size (the slope of the grainsize frequency-distribution curve) is the sorting. One parameter of sorting is obtained by the ratio (D₂₅/D₇₅)^1/2, where D₂₅ is the grain-size such that 25%  is smaller.   Walton modified this equation by replacing the 25 and 75% grain-size diameters with the 25 and 75% flow (Q₂₅/Q₇₅)^1/2,. In this case a low ratio is indicative of a permeable basin that has a large storage capacity. This technique provides another simple and quick method for hydrologic evaluations of drainage basins.

The Q₂₅ and Q₇₅ data are easily obtainable from flow-duration curves. The data from Figure 21.2 show that the Mad River has a flow ratio of 1.59 and the Scioto River a ratio of 2.64, while Home Creek, typical of a basin of low permeability, has the highest ratio of 5.09.

 Seepage or dry-weather measurements

Seepage or dry-weather measurements consist of flow determination made at several locations along a stream during a short time interval when there is no surface runoff. Many investigators prefer to initiate seepage runs during the stream's 90% flow.

It is not always possible to conduct an actual seepage survey due to time, manpower, or financial constraints. In these cases, flow-duration curves may serve as valuable substitutes.

Seepage measurements permit a quick evaluation of ground water runoff-how much there is and where it originates-and provide clues to the geology of the basin as well. The flow of some streams increases substantially within a short distance. Under natural conditions the increase is most likely related to increased ground water runoff originating in deposits or zones of high permeability in or adjacent to the stream channel. These gaining reaches may consist of deposits of sand and gravel, fracture zones, solution openings in limestone, or merely local facies changes. In addition, ground water may also discharge through a series of springs or seeps along valley walls or in the stream channel.

In areas where the geology and ground water systems are not well known, streamflow data can provide a means of testing estimates of the ground water system. If the stream-flow data do not conform to the estimates, then the geology must be more closely examined (LaSala, 1968). For example, the northwest corner of Ohio is crossed by the Wabash and Fort Wayne moraines, between which lies the St. Joseph River. As indicated by the Glacial Map of Ohio (Goldthwait et al., 1961), the St. Joseph Basin consists mainly of till. However, low-flow measurements show that the discharge of the river increases more than 14 cfs along its short reach in Ohio, indicating that the basin contains a considerable amount of outwash, which in this case is covered by a relatively thin layer of till.

The mainstream of the Auglaize River in northwestern Ohio rises from a mass of outwash that lies along the front of the Wabash moraine. The southwest-flowing river breaches the moraine near Wapakoneta and then flows generally north to its confluence with the Maumee River at Defiance. A gaging station is near Ft. Jennings in a till plain area slightly above a reservoir on the Auglaize. This gage measures the flow resulting as an endproduct of all causative hydrologic factors upbasin (ground water runoff, surface runoff, slope, precipitation, use patterns, etc.)-it merely shows inflow to the reservoir. Low-flow measurements, however, indicate that nearly all of the baseflow is derived from a small deposit of outwash along the distal side of the Wabash moraine; there is no gain across the wide till plain downstream, which makes up most of the stream's basin.

A number of discharge measurements have been made in the Scioto River basin in central Ohio. The flow measurements in themselves are important because they show the actual discharge-in millions of gallons per day, in this case-at about 90% flow (Figure 21.3). Note that the discharge at succeeding downstream sites on the Scioto River is greater than the flow immediately upstream. This shows that the river is gaining and that water is being added to it by ground water runoff originating largely from the adjacent outwash deposits.

A particularly useful method for evaluating streamflow consists of relating the discharge to the size of the drainage basin (cfs or mgd/miz of drainage basin). A cursory examination of Figure 21.3 shows that it is convenient (and totally arbitrary) to separate the flow into three distinct units: Unit 1 falls in the range of 0.010 to 0.020 mgd/mi2, Unit 2 includes 0.021 to 0.035 mgd/mi2, and Unit 3 0.036 to 0.050 mgd/miz. The Olentangy River and Alum and Big Walnut Creeks fall into Unit 1, Big Darby and Deer Creeks into Unit 2, and the Scioto River, Walnut Creek, and the lower part of the Big Walnut Creek into Unit 3. Even though the latter watercourses fall into Unit 3, their actual discharges vary widely, from 3.07 to 181 mgd.

Logs of wells drilled along the streams of Unit 1 show a preponderance of fine-grained material that contains only a few layers of sand and gravel; all these wells yield less than 25 gpm and commonly no more than 5 gpm. Logs of wells and test holes along Big Darby and Deer Creek, however, indicate that several feet of sand and gravel underlie 5 to more than 25 ft of finegrained alluvial material. Adequately designed and constructed wells that tap these buried outwash deposits produce as much as 500 gpm. Glacial outwash, much of it coarse-grained, forms an extensive deposit through which Unit 3 streams and rivers flow.

The outwash extends from the surface to depths that in places exceed 150 ft. Industrial wells constructed in these deposits, most of which rely on induced infiltration, can produce more than 1000 gpm. Thus, it is evident that by combining seepage data and well yields with a geolopic map, it is possible to develop a potential well yield map. The potential ground water yield map relies heavily on streamflow measurement as well as good judgment, buy nonetheless provides, with some geologic data, a good first approximation of ground water availability.

A water quality-duration curve is similar to its counterpart, the flow-duration curve, and is prepared in the same manner except that the stream's concentration of selected chemical constituents replaces discharge data. The quality-duration curve can be used for three purposes. First, it shows the stream's range in concentration of any substance examined; this is useful for water treatment plant designs and dilution studies. Secondly, it can provide a good approximation of the chemical quality of ground water in the zone of active circulation. Finally, in certain instances it can be used to indicate areas of contaminated surface water.

As discussed previously, during a stream's period of low flow, all of the water in the channel consists of ground water runoff, unless the stream receives effluent. Therefore, within limits, at these times a stream's quality closely approximates the chemical quality of shallow ground water in the basin. (This assumption is not entirely correct, however, because some chemical changes may occur at or near the ground water-surface water interface.)

Although more work needs to be done along these lines, it appears that a stream's 10% concentration (the concentration equaled or exceeded 10% of the time) provides a reasonable estimate of the chemical quality of shallow ground water if the stream is not contaminated. It is imperative, however, that values obtained from quality-duration curves be compared with analyses of well water. A quality-duration curve for Ohio's Paint Creek based on data from 1967 is shown in Figure 21.4. This is a typical curve for a stream uncontaminated with respect to chloride. The 10% concentration is about 21 mg/L; shallow well data indicate that the natural concentration of chloride in the basin generally ranges from 3 to 28 mg/L. Thus the estimate of ground water quality based on the quality-duration curve is in close agreement with actual well data.

It is often possible to determine whether a stream is contaminated by means of quality-duration curves, although the method is somewhat subjective and must be based on some prior knowledge. A quality-duration curve of sulfate in Raccoon Creek, a stream contaminated by drainage from coal mines, is shown in Figure 21.5. In this case the 10% concentration of sulfate (250 mg/L) is far greater than the background content (about 75 mg/L), but the concentration of carbonate hardness is within expected limits (220 mg/L). Therefore, despite the fact that a stream is contaminated with respect to one or more constituents, the contamination does not completely invalidate the usefulness of curves depicting other constituents.

Quality seepage measurements

Another method to determine shallow ground water quality is to collect surface water samples during dry-weather conditions from a wide area within a short time interval. In this case emphasis should be placed on small tributaries since they are less likely to be contaminated. These data can be incorporated with seepage measurements, but the latter are not essential for quality evaluations. Once the chemical data are analyzed, concentrations can be plotted on a base map using a color code to represent selected ranges in concentration. In this manner background quality becomes readily evident.

Alum Creek flows southward across a till covered area in central Ohio.
Within de basin  are scores of abandoned and producing oil wells and even more dry holes. Oil-field brine-holding ponds have been used in this area for years, and leakage from many of them has locally contaminated the ground water with high concentrations of chloride. Contaminated ground water eventually flows into the stream, but during much of the year the chloride rich ground water is diluted by surface runoff and ground water runoff from uncontamined areas.

Nearly 100 stream samples were collected during a single day reflecting low-flow conditions. Most of the samples were collected from small tributaries near their confluences with the main stream. These data were augmented with samples from field drainage tile. Most of the samples contained less than 15 mg/L of chloride, which is the background concentration, but others were significantly higher. The small basins that contained higher than background chloride concentrations were assumed to be contaminated (Figure 21.6).

The configuration of each contaminated small basin was delineated on atopographic map, on which were plotted oil and gas wells and tests. Thelatter provided some control on the point sources of contamination. It wasthen possible to estimate the general size of each ground water contaminated area (Figure 21.7).

Using the regional approach described above, it is possible to minimize drilling costs for monitoring wells because uncontaminated areas are readily evident and the investigator can then key on selected sites. Once contaminated areas have been located, additional surface water samples can be collected from the sub-basin, and these should permit a more detailed assessment of the area.

Summary and conclusions

Commonly a research plan is not formulated prior to the initiation of hydrogeologic field studies. Such a plan, if based on sound hydrogeologic principles, should lead the investigator down a logical path toward a solution that minimizes costs and maximizes efficiency.,

Generally hydrogeologists depend on the very expensive and locally seasonal process of drilling test holes and wells to obtain basic data. Such drilling is usually essential, but a variety of techniques are available to assess hydrogeologic conditions from a regional perspective that can substantially reduce field costs. Generally the data needed for regional studies can be easily and inexpensively obtained.

References

Cross, W. P. and Hedges, R. E. (1959). Flow duration of Ohio streams. Ohio Department of Natural Resources, Division of Water, Bulletin 31.

Goldthwait, R. P., White, G. W., and Forayth, J. L. (1961). Glacial map of Ohio. U. S. Geological Survey, Misc. Geol. Inv. Map 1-316.

LaSala, A. M. (1967). New approaches to water-resources investigations in upstate New York. Ground Water 5(4):6-11.

Pettyjohn, W. A. (1975). Chloride contamination in Alum Creek, central Ohio. Ground Water 13(4):332-339.

Pettyjohn, W. A. (1982). Cause and effect of cyclic changes in ground-water quality. Ground Water Monitoring Rev. 2(1):43-49.

Pettyjohn, W. A. and Henning, R. (1979). Preliminary estimate of ground-water recharge rates, related streamflow and water quality in Ohio. Water Resources Center, Ohio State University, Project Completion Rept. No. 552.

Walton, W. C. (1970). Groundwater Resource Evaluation. McGraw-Hill, New York.