An Analysis Of Hydrogeology, Groundwater Discharge,

And Nutrient Input To Clear Lake


Dr. William W. Simpkins, Keri B. Drenner, and Sarah Bocchi



A. Geology And Hydrogeology Of The Clear Lake Region


1.  Introduction.  An understanding of the geology and hydrogeology of the Clear Lake region is needed to understand lake-groundwater interactions.  The following objectives were investigated:


         determine the thickness of Quaternary units underlying the lake and overlying the regional bedrock aquifer;

         estimate hydraulic heads in the regional aquifer and their relationship to the lake elevation and shallow groundwater flow;

         determine the nature and types geologic units affecting flow to and from the lake.


The discussion below provides a summary of the work to date.


            2.  Methods.  We examined nearly 300 well logs from private wells in the region during summer 2000 to determine thickness of Quaternary units below the lake and to estimate hydraulic head (Fig. 1).  The well logs were available on-line at the Iowa Geological Survey Bureau (IGSB) Virtual Geologic Sample Database (GEOSAM) . Based on the logs, we identified the aquifer for each well and estimated hydraulic head from ground-surface elevation and static water level data.  In some cases, ground surface elevation had been estimated by the IGSB.  In others, we estimated the elevation from topographic maps, because field location proved impractical.  In addition to well logs, core was examined from 11 coreholes taken as part of piezometer installations during summer 2000.  Descriptions are given in Appendix 6.  We also examined preliminary maps of the glacial geology of the region provided by IGSB.


Figure 1.  Private well log downloaded from the IGSB’s Virtual GEOSAM database.


            3.  Results and Discussion.  Clear Lake lies within the Algona-Altamont end moraine complex of the Des Moines Lobe (D. Quade, IGSB, verbal communication, 2000).  This is a stagnation moraine built by at least 2 ice advances and characterized by hummocky topography.  The eastern edge of the end moraine is about 1 mile (1.6 km) east of the lake.  East of the end moraine, a late Wisconsinan till plain or Pre-Illinoian surface is encountered.  A large outwash deposit follows the axis of Clear Creek northeast from the lake and merges into an outwash fan complex west of Mason City.  This outwash deposit is important hydrogically, because it provides an effective groundwater drain for the east side of the lake (see Part II under Analytic Element Modeling).  Bettis (1998) speculated that Clear Lake occupies a former subglacial tunnel channel that fed the outwash streams of Clear and Willow Creeks.


            Coring of the east side of the lake basin indicates about 33 ft (10 m) of lake sediment on top of till or outwash (Baker et al., 1992).  Below that Clear Lake is underlain by about 70 ft (21 m) of Quaternary sediment, including till of Wisconsin age and earlier glacial advances, Peoria loess, and paleosol units.  In the uplands on the west, south, and north sides of the lake, till of the Dows Formation, which here consists mostly the supraglacial Morgan Member, is about 66 ft (20 m) thick.  This unit houses the local groundwater flow system that interacts with the lake.  Units directly below the lake (paleosols and older till units) are probably effective confining units with considerably lower hydraulic conductivities that restrict vertical flow in or out of the lake. Coring at the 11 piezometer sites indicated that there are areas of lake sediment (silt and sand) and coarse outwash adjacent to the lake, suggesting that the lake may have been larger in the past.  Peoria loess (Site C) and older till units (Site J) were also encountered during piezometer installation.


            Major aquifers beneath the lake are Devonian in age, and include the Shell Rock (upper Devonian) and Cedar Valley (middle Devonian) Formations (aquifers), which are among the most prolific karst aquifers in Iowa (Appendix 6).  Wells in these units are generally anywhere from 100 to 600 ft (30 to 183 m) in depth.  The city wells of both Ventura and Clear Lake are finished in the Cedar Valley aquifer.  Some wells in the region are finished in the Ordovician, presumably in the St. Peter sandstone.  Approximately 187 wells both within and outside the watershed contained enough information to be able to discern hydraulic head information.  Based on an average lake stage of 1226.2 ft (374 m) during the study, most aquifers show hydraulic heads that are below lake stage (Fig. 2), suggesting a potential for flow out of the lake to underlying aquifers.  Seventeen private wells show heads at or above lake stage, which would suggest a potential for flow from aquifers up into the lake. 


            4.  Conclusions.  Clear Lake sits within a large end moraine complex of the Des Moines Lobe.  Till and associated sediment in the moraine provide the media for local groundwater flow systems that interact with the lake.  Hydraulic head relationships suggest that bedrock aquifers in the region are not hydraulically connected to the lake.  Studies on the geology and hydrogeology surrounding Clear Lake are continuing.

Figure 2.  Hydraulic heads in major aquifers in relation to Clear Lake stage.


B.        Estimation Of Groundwater Discharge To Clear Lake


1.  Introduction.  Estimation of groundwater discharge (or seepage) to lakes is necessary to determine nutrient load, but it is a difficult task and generally involves the extrapolation of small-scale measurements to a much larger lake area.  In cases where the geology beneath the lake is not well known and where discharge may vary, large errors are involved in the measurement and extrapolation steps.  A review of literature on the subject indicates that a number of methods have been used. In this study, groundwater discharge was estimated by three separate and independent methods:


         Direct measurements using seepage meters (Lee, 1977);

         Application of Darcy’s Law using hydraulic head gradient and hydraulic conductivity data (Pennequin and Anderson, 1983; Shaw et al., 1990);

         Use of an analytic element groundwater model (Hunt and Krohelski, 1999).


Appliation of the three methods and their respective results are presented below.


2.  Methods: Seepage Meters.  Lee (1977) developed the seepage meter for groundwater discharge estimation in lakes. “Meters” consist of cut-off 55-gallon (204 L) metal barrel that is advanced into the lake bottom.  As groundwater moves into the barrel, lake water is displaced into a plastic bag on the top of the barrel and a volume per time (discharge) can be recorded.  A Darcy flux (L/T) is calculated by dividing discharge by the cross-sectional area of the meter.  Lee (1977) also terms this, albeit incorrectly, a “macroscopic seepage velocity.” Along with hydraulic head measurements, seepage meters can also be used to estimate hydraulic conductivity of lakebed sediments using Darcy’s Law relationships.  Total discharge to the lake is estimated by then applying individual flux measurements to the area of the lake.  Lee (1977) envisioned these meters to have their greatest use in “high to moderately permeable material” presumably where groundwater discharge would be uniformly distributed.  He suggested that measurements in fine-grained, low permeability sediments may be difficult to obtain because of lower discharge and preferential flow out of or into the lake bottom.  Studies have corroborated problems with using meters in fine-grained lake sediments (Pennequin and Anderson, 1983). 


            Measurements (N=341) were made in July-August and November 1999 and again in May-June 2000 at 21 sites in the lake (Fig. 3).  Ventura Marsh was excluded from seepage measurements because of the extremely mucky bottom and the difficulty encountered emplacing the meters.  Original plans called for more or less permanent emplacement of some meters at some sites close to the piezometer nests around the lake, so that repeated measurements could be made at the same place.  This plan was abandoned primarily because of vandalism and heavy boat and beach traffic.  As a result, seepage meters were generally emplaced just prior to use, removed and then moved to another measurement site.  Most measurements made within 3 to 4 m of shore, although some were made at 100 m offshore.  Time and personnel constraints limited our attempt to emplace seepage meters in deep water.  Measurements of volume were done in a 4-L, thin polyethylene bag approximately one hour after emplacement.  All bags were filled with an initial 200 mL of lake water to preclude problems with bag infilling or induced flow (see Shaw and Prepas, 1989). Sites 1 to 4, 6 to 9, and 12 to 21 were visited in 1999, with sites 7, 9, 12, and 13 visited twice (171 measurements).  Sites 3 to 5, 7 to 10, 14, 15, and 22 were visited in 2000, with sites 3 to 5, 7, and 8 visited twice (170 measurements).  Included in that total are multiple sequential measurements at sites 8, 9, 10 and 13 in 2000.  These were done to estimate the intrasite variabilty.  Sites 3, 4, 7, 8, 9, 14 and 15 were visited both in 1999 and 2000 to estimate interannual variation.  Discharge estimates were computed in units of mL/min and converted to flux measurements of mm/s for this report.  Hydraulic head gradient measurements were also made adjacent to many seepage meters using a potentiomanometer device as described by Winter et al. (1988). 



Figure 3.  Map of Clear Lake watershed showing location of seepage meter measurements in 1999 and 2000.  North is to the top of the map.


3.  Results: Seepage Meters.  The mean flux value for all 341 measurements during the 2-year sampling period is 0.45 mm/s (Table 1)(note: 1 mm/s is equal to 1 cm3 m-2 s-1 or

1 ml m-2 s-1).  The range of values is similar to those observed by Lee (1977) for Lake Sallie in Minnesota, Krabbenhoft and Anderson (1986) for Trout Lake in Wisconsin, and Shaw et al. (1990) for Narrow Lake in Alberta.  Most measurements at Clear Lake indicate inflow to the lake.  Anecdotal evidence, based on influx of significantly colder water on our feet during emplacement, corroborates inflow at Sites 3, 9 and 10.  Out of the 341 measurements, four showed zero net flux and four showed negative values.  The lack of many negative values is at odds with hydraulic head data that indicates definite zones of outflow on the east side of the lake (see later discussion).  Histograms indicate a positive skewness in the total data set and in 1999 (0.9) and 2000 (1.9) individually.  Transformations of the data did not improve the normality of the distribution.  Because the data did not appear to meet assumptions of normality, nonparametric statistical techniques are relied on in this analysis.  Boxplots indicate that many of the very large values are outliers.  


            Data from the seepage meters appear to be repeatable at an individual site.  Repeat measurements made in consecutive 1-hour readings at sites 8, 9, and 10 in 2000 indicate no significant difference (p = 0.05) in median seepage rates.  At site 13 (City Beach), however, median seepage rates increased from 0.13 to 0.60 to 0.88 mm/s in consecutive 1-hr measurements.  The reason for this phenomenon, particularly at what is probably a groundwater outflow zone, is unknown.


            Individual sites show interesting differences among them. There are no general trends of seepage around the lake, except that many of the higher flux values occur on the eastern side of the lake.  For example, sites 5 and 6 show high flux values into the lake, despite the fact that these areas are likely ones where groundwater is flowing out of the lake (see later discussion).  This anomaly suggests that other factors may cause the seepage bags to fill at these locations.  Site 10 at Ventura Heights and Site 14 at the beach at McIntosh Woods State Park both show very low fluxes into the lake.  Both of these sites are on very sandy points or spits.  This would appear to be consistent with studies that suggest that groundwater flow is concentrated in embayments and not at points between embayments in lakes.



Table 1.  Statistics for original flux measurements.  Values in mm/s.








Std. Dev.

Std. Error







95% C.I.

of Mean

































            Significant differences, as measured by a Mann-Whitney test, exist in fluxes measured in 1999 and 2000, with 1999 showing larger median fluxes.  This could simply be due to the fact that we did not visit the same sites in the two-year period and that only the most permeable sites were visited in 1999.  However, at sites duplicated in 1999 and 2000, four out of seven sites showed significant differences in their median flux values (p = 0.05) between the two years.  Precipitation totals in nearby Mason City were significantly different in 1999 (44.19 in; 1122 mm) and 2000 (30.59 in; 777 mm) and this likely explains the difference in measurements.  The mean annual precipitation is about 33 in (838 mm).  Higher hydraulic heads in the upland areas in 1999 would have steepened the hydraulic gradient and induced more flow to the lake.  A 12.08-inch (307 mm) rainfall in July 1999 unfortunately occurred during and after many of the measurements.  Downing and Peterka (1978) and Boyle (1994) have shown the effects of rainfall on seepage meter flux in lakes.  Accordingly, the 1999 seepage data were reduced by 31 percent so that data from both years could be used in one data set (Table 2).  Even with these adjustments, there are significant differences between the medians calculated for each site (Fig. 4), as shown by a Kruskal-Wallis test of significant differences.

Figure 4.  Boxplots showing seepage measurements at the 21 sites (see Fig. 3).


Table 2.  Descriptive statistics using adjusted 1999 flux measurements.  Values in mm/s.








Std. Dev.

Std. Error







95% C.I.

of Mean
































4.  Calculation of Whole-Lake Discharge.  Extrapolation of individual seepage meter measurements to an entire lake is difficult.  Methods have generally concentrated on defining a relationship between flux and either distance from shore or water depth.  Early studies on lake-groundwater interaction by McBride and Pfannkuch (1975) indicated that the log of discharge decreased away from the shoreline.  Subsequent flow net analysis by Lee (1977) in his original paper also suggested this relationship.  His flow net analysis suggested that most flow would occur within 15 m of shore, although he found that that substantial influx was still occurring at 60 to 80 m offshore.  He presented the relationship: 


Flux = 0.381 * (969)distance


with an r2 = 0.56 (significant at p=0.05).  Brock et al. (1982) also suggested a relationship between water depth and flux, such that (r2 = 0.96).


ln Flux = 1.0719 – 1.5797 * (water depth)


These relationships suggested that fluxes should approach zero at some distance from shore or at some water depth.  Hence, whole-lake estimates of groundwater discharge could be derived from seepage meter measurements made only in the near shore or littoral zone environment.  Cherkauer and Zager (1989) and Boyle (1994) have noted similar relationships.  In contrast, other studies have indicated that flux may increase from shore (Woessner and Sullivan, 1984) or show variability due to the geology under the lakebed (Krabbenhoft and Anderson, 1986).  Studies have also shown that groundwater flux in or out of the lake does occur outside the nearshore littoral zone in non-littoral zone (Belanger and Mikutel, 1985), although fluxes in the latter may be one to two orders of magnitude less (Boyle, 1994). 


            Data in this study showed no statistically significant relationship between seepage flux and distance from shore or water depth.  Thus, it was not possible to delineate a zone in the lake where seepage was zero or could be ignored.  In addition, there are significant differences in fluxes at different locations and, in some cases, the highest values are at documented outflow areas.  Thus, the reliability of the seepage measurements is questionable at those sites.  Rather than try to extrapolate data from each location to some distance into the lake, we chose a more simplified approach.  Groundwater discharge to the entire lake was calculated using the adjusted median flux value and applying it to the percentage of the lake area to which that applies (Table 3).  The maximum discharge based on seepage meter data is 5.4 x 105 m3/d, a value that is disturbingly large given the volume of water in the lake.  It could be less if smaller percentages of the lake, such as nearshore zones where we measured, contribute all the discharge.


Table 3.  Estimates of groundwater discharge to Clear Lake based on seepage meter data.


Area of lake (m2)

Adjusted median flux (mm/s)


95% CI of median





1% area (m3/d)


25% area (m3/d)


50% area (m3/d)









5.  Methods: Darcy’s Law.  An alternative method for estimating groundwater discharge involves the use of the familiar Darcy’s Law equation:


Q = -K I A


where Q is discharge, K is hydraulic conductivity, I is hydraulic gradient (negative because it decreases in the direction of flow) and A is the cross-sectional area through which flow occurs.  The method has been used by Pennequin and Anderson (1983) for Lake Wingra in Wisconsin and by Shaw et al. (1990) for Narrow Lake in Alberta. 


            For this analysis, piezometer nests were installed at 11 sites on the perimeter of the lake in 2000 (Figs. 5 and 6) to intercept groundwater flowlines just prior to entering the lake.  Piezometers consisted of 1.25 in (3.2-cm) id, PVC standpipes with 20-slot, factory slotted screens of 2 to 3 ft (0.61 to 0.91 m) in length.  They were installed at depths between 3 and 31 ft (0.9 and 9.5 m) using hollow-stem augers.  Coordinates (X-Y) and absolute elevation of the standpipes (and the USGS lake stage gage) to within 1 cm were obtained from a professional surveyor using GPS and total station equipment.  Cores were taken from the boreholes and described using standard methods.  Geologic materials encountered include till (Dows Formation, Morgan and Alden Members), outwash, lake sediment, Peoria loess, and fill (see Appendix A).  The variety of materials encountered attest to the heterogeneity present at the lake interface.  Falling and rising-head slug tests were performed in the shallowest piezometers at each nest in January 2001 and analyzed using the Hvorslev (1951) method to determine K.  These piezometers, which are at or near the water table, were used to provide a maximum K value for flow into the lake. 



Figure 5.  Cross-sectional diagram showing piezometers in a nested configuration.



Figure 6.  Topographic map showing location of piezometer nests (blue dots) and boundaries of cross-sections (lines) for discharge calculations by Darcy’s Law.  USGS gaging station is shown by the green dot.  North is to the top of the map.  Sections are 1 mi (1.6 km) square.

            Hydraulic heads in the piezometers were measured from September 2000 through February 2001.  Vertical hydraulic gradients were analyzed for the January 5, 2001, measurements.  A mean horizontal hydraulic gradient was calculated by averaging the difference between lake stage and the water table from the shallow well at each measurement time between September 2000 and February 2001.  Finally, Clear Lake was divided into 11 groundwater discharge sections, based on K and I values from the nearest piezometer nest and the expected geology along the section (Fig. 3).  Distances along these sections were calculated using ArcView.  A uniform depth of 10 m was used for each section to complete the cross-sectional area of flow.  This value is generally greater than the depth of the lake; however, because upward hydraulic gradients are shown by the piezometers at 10 m, the potential exists for groundwater flow into the lake from these depths.


            6.  Results: Darcy’s Law.  Vertical and horizontal hydraulic gradients provide evidence for groundwater discharge in and out of Clear Lake.  Sites A, B, C, I, J, and K suggested groundwater inflow at all times.  Sites E, F, and G, showed groundwater outflow at all times.  Sites D and H show both groundwater inflow and outflow during the monitoring period and may change on a seasonal basis.  Vertical hydraulic gradients are present, but they are very small.  Hence, gradients at the water table were used to calculate discharge. Values of K varied from 1.6 x 10-6 m/s in till to 1 x 10-4 m/s in outwash sand.  These values are similar to those seen in the lake from potentiomanometer measurements.  Values for groundwater discharge show variability among sites due to the differences in K values and hydraulic gradients.  However, there is roughly a balance between inflow and outflow, although outflow is the larger of the two values.  Presumably, precipitation and surface water inflow will comprise the difference and raise inflow to equal the outflow values.



Table 4.  Values of groundwater discharge estimated from Darcy’s Law.








grad h


























































































7.  Methods: Analytic Element Model.  Groundwater modeling has proven to be a valuable tool in understanding groundwater-lake interaction (Anderson and Munter, 1981; Krabbenhoft, et al. 1990a; Hunt et al., 1998).  In the past, grid-based, finite-difference models such as MODFLOW (McDonald and Harbaugh, 1988) have been used.  Recent research has shown that the Analytic Element (AE) method (Haitjema, 1995) is applicable to groundwater-lake systems (Hunt and Krohelski, 1996; Hunt et al., 2000).  The AE method can be used as a screening tool at the regional scale to identify boundary conditions, which can improve solutions generated later by smaller-scale and more complex finite-difference simulations (Hunt et al., 1998).


            A Windows-based, AE model, GFLOW 2000, was used to simulate the groundwater flow system and to calculate groundwater discharge into the lake.  This modeling approach is two dimensional, steady state (hydraulic head does not vary with time), and utilizes Dupuit-Forcheimer approximations that assume that groundwater flow is mostly horizontal.  AE methods are relatively new in their application and are based on superposition (i.e., addition or subtraction) of analytic functions, each representing a feature of the aquifer (Strack, 1989; Haitjema, 1995).  Wells, line sinks (streams and tile drains) and inhomogeneities (areas with differing K, porosity, or recharge values) can be addressed in the model.  AE assumes that the aquifer is infinite in extent, with boundaries consisting of rivers, creeks, and lakes that can be easily seen on topographic maps.  A unique feature of the model, the “flux inspector,” allows flux (i.e., discharge) to be calculated across any line drawn in the problem domain.  Hence, discharge to a lake can be estimated by drawing “flux inspection” lines around the lake.


            An analytic element model was constructed for Clear Lake and its surrounding watersheds. Input data included elevations of stream and drainage tiles, K values and groundwater recharge.  Because the model does not explicitly incorporate lakes, Clear Lake was given a very high K value to simulate that effect, as suggested by Hunt et al. (2000).  Both Ventura and Lekwa Marsh were included in the area of the lake, because early model simulations indicated problems with closely spaced line sinks.  An unpublished glacial geologic map (based on digital soil survey data) from the IGSB was used to identify geological units that could affect flow in the vicinity of the lake.  Although Clear Lake is set within the Algona-Altamont Moraine Complex, a large outwash deposit occurs on the eastern edge of the lake.  It is approximately aligned with the Clear Creek outlet and leads into what appears to be an outwash fan east of the City of Clear Lake.  This deposit and the elevation drop east of the City of Clear Lake provide an effective drive for eastward groundwater flow out of Clear Lake.


            8.  Results: Analytic Element Model.  The model was run multiple times until reasonable agreement was reached with the hydraulic heads measured in the piezometers around the lake.  Early model runs omitted the outwash deposit at the east end of the lake.  As a result, the K values for till in the model were 2 x 10-4 m/s.  In later runs, the addition of a more conductive zone on the east end allowed K values to be reduced in the watershed (Fig. 7).  Thus, assigning a K value of 3.5 x 10-3 m/s to the outwash allowed K values in the till to be lowered 5.3 x 10-5 m/s.  Although this is still a high value for till, it is in keeping with K values estimated from slug tests at the lake.  Using a recharge value of 81.3 mm/yr (3.2 in/yr or 10 percent of mean annual precipitation), the final model produced heads similar to those observed in the field (Fig. 8) and produced inflow and outflow in the areas indicated by field data.  The maximum departure from the field data was 0.55 m and the mean absolute difference was 0.27 m (Fig. 8).


Figure 7. Simulation of groundwater flow in the vicinity of Clear Lake using the AE model GFLOW 2000.  Contour interval is 5 ft.  Outwash deposit is outlined in orange.  Calibration points (piezometers) are shown with green diamonds (see Fig. 6 for locations).





















Figure 8. Calibration curve for the AE simulation shown in Figure 7. 

            Perhaps the most unique aspect of this AE model is its ability to calculate fluxes across any line sink, and, in this case, to Clear Lake.  Flux inspection lines were drawn along areas that the model shows to be inflow and outflow zones.  The results suggest that groundwater inflow (discharge) to Clear Lake is approximately 7.9 x 103 m3/d, while the groundwater outflow is approximately –8.9 x 103 m3/d.  Although these values are higher than those suggested by the Darcy’s Law analysis, they are within the same order of magnitude.  Of that total, Ventura Marsh (including areas outside the present water surface on Fig. 6) accounts for about 3.3 x 103 m3/d of groundwater discharge to the lake.


Table 5.  Comparison of groundwater discharge values estimated by the three methods.


GW Inflow (m3/d)

GW outflow (m3/d)

Net (m3/d)

Seepage meters




Darcy’s Law




AE GW Model





            9.  Conclusions.  Estimation of groundwater discharge to Clear Lake was investigated by three independent methods. The Darcy’s Law and analytic element model methods both recognized areas of inflow to and outflow from the lake and corroborated field measurements of hydraulic head.  They produced values within the same order of magnitude (Table 5).  In contrast, seepage meters failed to indicate any areas of groundwater outflow, even though these areas must occur to balance the input of water to the lake.  In addition, values of inflow estimated by the seepage meters were nearly 2 orders of magnitude larger than values calculated by the other two methods and were comparable to the volume occupied by the entire lake.  The large discrepancy could be due to measurements taken in years of different rainfall, heterogeneity in the lake bottom, and/or the inability of these devices to work at Clear Lake.  Assuming some relationship of discharge vs. distance from shore exists, the discharge overestimate could be due to the applying one median flux value over the entire lake.  A more reasonable value might be obtained if the seepage meter data were valid for a smaller portion of the littoral zone, say 1 percent of the area of the lake.  The determination of such an “effective” discharge zone could not be made in this study, however, and may be the subject of future work.  Because of the number of assumptions involved in the Darcy’s Law calculation and the size of the lake, the value from the AE model is probably the best estimate of groundwater discharge to Clear Lake.


C.        Estimation Of Nutrient Load From Groundwater


            1.  Introduction.  A geochemical investigation of groundwater was undertaken in order to understand the presence and absence of nutrients and contaminants in groundwater and their potential to enter Clear Lake.  Groundwater samples from the 32 out of 33 piezometers were analyzed for Total P, Total N, Si, alkalinity, electrical conductivity, and pH. Additional parameters (major cations and anions, trace elements, dissolved O2, dissolved organic carbon) were measured in order to understand the geochemical environment in which the nutrients occur.  Geochemical speciation models and soil P measurements were used to determine potential sources of P.  Selected samples were analyzed for fecal coliform bacteria and caffeine, in order to determine potential sources of nutrients and Cl.  A radioactive isotope of hydrogen, tritium (3H), was used to determine the relative age of the groundwater.  Nutrient and contaminant loads from groundwater to Clear Lake were calculated from estimates of groundwater inflow and outflow and estimates of the concentrations of nutrients (primarily P, N, and Si) and Cl in groundwater.  Nutrient load per time was calculated by multiplying discharge (L3/T) times concentration (M/L3).  Because of Clear Lake’s nature as a flow-through lake, nutrients will be added to the lake in areas of inflow and lost from the lake in areas of outflow.


            2.  Methods.  Groundwater samples were taken from the 32 piezometers on 7 dates between September 2000 and February 2001 (9/15/00, 9/30/00, 10/31/00, 11/14/00, 1/5/01, 1/15/01, and 2/24/01).  Samples of lake water and sediment porewater (groundwater?) were also taken from the potentiomanometer device on various dates from May through October.  These samples were analyzed primarily for Total P, Total N, Si, alkalinity, pH, and electrical conductivity (results will be reported at a later date).  Samples for total geochemistry (cations, anions, trace metals), organic carbon, NO3-N, and NH4-N were taken on 10/31/00.  Samples for tritium (3H), fecal coliform, and caffeine were taken on 1/15/01.  Although the piezometers installations were completed in July, sufficient time was allowed to develop them and clean out non-formation water that may have been introduced during the drilling and installation process.  This is particularly critical for sampling of P, because turbid water, which is often produced by the type of material encountered in this investigation, may contain sorbed P.  Sampling was accomplished using a peristaltic pump and dedicated polyethylene sampling tube in each piezometer.  The sampling tube was pulled up from the bottom during sampling to avoid introduction of sediment into the sample.  Some samples, however, always contained sediment, due in part to less than ideal completions.  One well volume was purged at an interval of one to several days prior to sampling. 


            Samples for cations, anions, trace metals, and dissolved organic carbon (DOC), were filtered with a 0.45 mm filter prior to entering the sample bottle.  Cation and trace metal samples were preserved with 1 mL of 4.5 N HNO3.  Samples for NO3-N and NH4-N were preserved with 1 mL of 8N H2SO4.  Samples for DOC were preserved with 1N HCL.  All samples were kept at nearly 4 °C prior to analysis.


            Samples for Total P, Total N, Si, alkalinity, electrical conductivity, and pH in groundwater were analyzed in the Limnology Laboratory in the Department of Animal Ecology.  Total P, Total N, and Si samples were run in triplicate and the mean is reported here.  Total P was determined using persulfate digestion and ascorbic acid using Hach PhosVer3 powder pillows (EPA 365.2/Standard Method 4500-P-E).  Total N was determined using second derivative spectroscopy (Crumpton et al., 1992).  Silica (SiO2) was determined using the molybdate-reactive method.  Major cations, anions, and trace metals (including P) were analyzed by inductively coupled plasma mass spectrometry (ICP-MS) at the Soil and Plant Analysis Laboratory, University of Wisconsin-Madison.  Alkalinity and pH were determined using a Hach digital titrator and pH meter, respectively, in the Hydrogeology Laboratory within 2 days of collection.  Cl, SO4, F and NO3-N were analyzed by ion chromatography in the Department of Geological and Atmospheric Sciences. Analyses were speciated and charge balances checked to within 10 percent with the geochemical model NETPATH v. 2.0 (Plummer et al., 1994). DOC was analyzed using persulfate oxidation in the Department of Agronomy.  Dissolved O2 was measured in the field by the modified Winkler-azide method.  Tritium (3H) was analyzed using a scintillation counter at the Environmental Isotope Laboratory at the University of Waterloo (Ontario, Canada).  The National Soil Tilth Laboratory analyzed samples for NO3-N and NH4-N using copper cadmium reduction on a Lachat apparatus.  Caffeine (EPA 3510) and fecal coliform bacteria (SM18-9222D) were analyzed at the University of Iowa Hygienic Laboratory.  Bacteria were analyzed within 1 day of sample collection.  Samples from continuous core at Sites B, F and J (Fig. 6) were analyzed for pH, organic carbon and soil P at the Soil Analysis Laboratory in the Department of Agronomy at Iowa State University.  Soil P was analyzed using both the Bray and Olsen methods.     


            3.  Results and Discussion. 

            a.  Geochemical Environment in Groundwater.  The geochemical environment in groundwater is one dominated by Ca and HCO3 ions and pH values between 7 and 8 – a typical system in carbonate terrain (Appendix 7).  In addition, the system is reducing and alternate electron acceptors are clearly being used to oxidize organic carbon (Appendix 7).  Samples taken in October 2000 show that dissolved O2 is largely absent and NO3-N concentrations are < 1 mg/L.  Nitrogen, when present, occurs as NH4-N, and dissolved Fe concentrations reach 15 mg/L.  In addition, most groundwater samples smell of H2S and some may produce CH4 gas.  Dissolved organic carbon (DOC) was present in all samples.  Mean and median values were 4.5 and 4.3 mg/L, respectively and the highest concentration of 11.7 mg/L was shown at Site G (Appendix 7).  As such, the system is similar to that described by Simpkins and Parkin (1993) for till in central Iowa; however, wetlands surrounding the lake may also contribute DOC.  Fe2+ will be the dominant Fe species; thus, limiting the role of ferric oxides and sesquioxides to sorb P and immobilize it, as is commonly the case in lake sediments (Nriagu and Dell, 1974; Williams et al., 1976).  While the P contribution to Clear Lake under these redox conditions may be large, the contribution of NO3-N is probably negligible.  Silica is generally conservative at near-neutral pH and is relatively unaffected by low redox conditions.


Table 6.  Statistics for Total P, Total N and SiO2 concentrations in groundwater from 32 piezometers.  Total P concentrations in mg/L.  Total N and SiO2 concentrations in mg/L.





Std. Dev.

SE Mean




Total P






< 0.01



Total N



















            b.  Phosphorus (P) in Groundwater.  Total P concentrations in the 32 wells (219 samples) showed a mean value of 237.7 mg/L, with a standard deviation of 232.9 mg/L (Table 6 ).  Because some very high concentrations skew the distribution, the median value of 172.9 mg/L may be a better measure of the central tendency.  Although there are some differences in the concentrations during the 7 sampling periods, they are generally not significant (Fig. 9).  Low concentrations reported on 10/31/00 were due to insufficient shaking of the samples (and dislodging P from the bottles and sediment) prior to analysis.  Consequently, we used the ICP-MS value of Total P in the statistical analysis.  Some differences are apparent among samples taken from the different piezometers (Fig. 10), most noticeably in samples taken from shallow piezometers.  A trend exists for higher concentrations at shallow depths, which suggests a near-surface source for P (Fig. 11).  Preliminary geochemical modeling suggests that the system is still undersaturated with respect to vivianite, a ferrous phosphate (Fe3(PO4)2 ·8H2O) mineral known to control P concentrations in groundwater.  Clear Lake wells #2 and #3 showed Total P concentrations of 8.4 and 10.5 mg/L, respectively.

Figure 9.  Boxplots showing Total P concentrations in groundwater.                        

Figure 10.  Boxplots showing concentrations of Total P at each piezometer.  Differences in concentrations exist between piezometers.




Figure 11.  Vertical profile of Total P concentration in groundwater at all sites.  Highest values occur at shallow depths, but there is no consistent trend with depth.


            Most of these P concentrations would be considered high in comparison to previous studies of P in groundwater (i.e. Robertson et al., 1998).  However, concentrations of P in drainage tile water have been shown to range from 7 to 900 mg/L in Canada (Bolton et al., 1980) and the midwestern United Statess (Baker et al., 1975; Schwab et al., 1980; Bottcher et al., 1981). Macropore flow at shallow depths has been shown to promote leaching of particulate P in similar soils and presumably geologic units containing fractures (Beauchemin et al., 1998).  In any case, these P values are within the concentration range to contribute to lake eutrophication. 


            c.  Phosphorus (P) in Geologic Materials.  Till, lake and outwash sediments that surround Clear Lake also contain extractable (soil) P in the core (Fig. 12).  Whether that P occurs naturally in these materials or is the result of anthropogenic activity cannot be determined at this time.   Highest concentrations occur in the solum and decrease below, where concentrations appear similar despite different materials.  This suggests that P is either released at shallow depth or the presence of anthropogenic P.


Figure 12.  Variation of Soil P (Olsen method) with depth in core.  Soil pH > 7.5.


            d.  Nitrogen (N) in Groundwater.  In contrast to P, concentrations of Total N are more consistent (Fig. 13) and considerably less (Table 6). The mean, standard deviation and median values are 1.1, 1.57 and 0.8 mg/L, respectively.  Data from Total N analyses are corroborated by the individual NO3-N and NH4-N concentrations (Appendix 7).  The lack of NO3-N in these samples is consistent with the lack of dissolved O2 and use of alternate electron acceptors in the system.  Interestingly, many Total N concentrations >1 mg/L at shallow depths may be due to NH4-N.  Samples taken from piezometer C-30, consistently show high NH4-N concentrations, which are responsible for the outliers in Figure 13.  Production of NH4-N in loess is consistent with earlier studies in central Iowa (Simpkins and Parkin, 1993).  It is unclear at this time whether groundwater supplies much N as NH4-N to Clear Lake.  Groundwater at Sites F and G, in lake-outflow areas, show NH4-N concentrations that may reflect concentrations in the lake itself (Appendix 7).


Figure 13.  Total N concentrations in groundwater. Outliers are due to high NH4-N concentrations in piezometer C-30 finished in loess.


            e.  Silica (SiO2) in Groundwater.  Concentrations of SiO2 were generally consistent among the 7 sampling dates and are within the range common for groundwater systems at near-neutral pH in these materials (Table 6; Fig. 14).  Mean, standard deviation and median concentrations were 40.4, 14.3 and 37.3 mg/L, respectively.  There is some variability in samples from different piezometers; however, the differences are not significantly different (Fig. 15).  Some of the higher concentrations are found in the shallowest piezometers. 

Figure 14.  SiO2 concentrations in groundwater. 





Figure 15.  SiO2 concentrations in groundwater among the piezometers.  Many shallow piezometers (e.g., C8, D6, I6) show the highest concentrations.


            f.  Chloride (Cl) in Groundwater.  Chloride concentrations in groundwater ranged from 1.0 to 73.1 mg/L, with a mean of 17.1 mg/L.  The mean Cl concentration in Clear Lake is about 16 mg/L.  Because there is no known natural source of Cl in the glacial sediment, its origin must be anthropogenic.  Potential sources include agricultural fertilizers, septic systems and road de-icing salts and solutions.  Data from this study indicate that concentrations generally decrease with depth, which suggests that activities near the surface are adding Cl to the groundwater.  It is common to find Cl concentrations of this magnitude in agricultural areas based on other studies in similar materials (Simpkins and Parkin, 1993).  However, concentrations of 73.1 mg/L are unusual.  We tested the hypothesis that septic system effluent was influencing shallow groundwater by analyzing samples for fecal coliform bacteria and caffeine.  Tests of groundwater in all the shallow piezometers were negative (<10 CFU/100 mL).  Caffeine was analyzed in groundwater samples from B10, C8, and D6, and was found to be less than the 40 ng/L quantitation limit. Thus, high Cl concentrations are not directly influenced by septic tank effluent at these sites.  However, plots of Na versus Cl revealed that many higher concentrations found in shallow piezometers help define a 1:1 meq relationship with an r2=0.59 (Fig. 16).  This suggests that road de-icing activities around the perimeter of the lake may be adding Cl to the groundwater. 



Figure 16.   Relationship of Na and Cl concentrations (meq/L) in groundwater.  Alignment along a 1:1 line suggests that the source of Cl is NaCl.


            g.  Tritium (3H) in Groundwater.  The radioactive isotope of hydrogen, tritium (3H), was analyzed in groundwater samples taken from each piezometer (see Appendix 7).  The profile of 3H with depth is shown in Figure 17.   Interpretation of the data is somewhat clouded by the precision of ±8 TU; however, cost considerations precluded use of higher precision analyses.  Nevertheless, there is an interesting lack of a trend with depth.  Piezometers B-30 and C-30 showed no 3H activity, suggesting groundwater recharge prior to 1953.  Many samples throughout the profile down to 30 ft (10 m) show essentially modern 3H activities, indicating recent recharge.  However, a majority of the samples show 3H activities in the 20 TU range that would indicate groundwater that was recharged during the 1970s.  Alternatively, these higher than modern 3H activities could be associated with Gulf moisture sources in summer thunderstorms when most groundwater recharge occurs in Iowa (Simpkins, 1995).  The 3H composition of the lake water is also not known.  Interpretation of the data may be helped by the analysis of stable isotopes d18O and d2H in lake water, precipitation, and groundwater, which will hopefully occur in summer 2001.




Figure 17. Profile of 3H (TU) in groundwater in the piezometers.  Shaded area is the approximate modern 3H input in Iowa (after Simpkins, 1995).


Table 7.  Summary calculations of nutrient and contaminant load to Clear Lake.

Flow direction

Q (m3/d)

from Part I

Nutrient or contaminant


conc. (mg/L)





Total P




Total N




Silica (SiO2)









Total P




Total N




Silica (SiO2)









Total P




Total N




Silica (SiO2)





            h.  Nutrient and Contaminant Load to Clear Lake.  Given an understanding of nutrient and Cl concentrations in groundwater adjacent to the lake, and assuming that these concentrations represent what would be likely be transported by groundwater to the lake, a calculation of the nutrient/contaminant load may be made.  For this simplified analysis, we will use the model-calculated discharges (Table 5) and use the median concentrations separated by inflow and outflow areas.  Estimates indicate that about 1.37 and 6.32 kg/d of Total P and N, respectively, are added to the lake via groundwater (Table 7).  The form of P (e.g., sorbed, dissolved, organic) entering the lake is not known at this time, but it is likely that N enters the lake as either NH4-N or organic N.  Calculations indicate that nearly 300 kg/d of SiO2 enter the entire lake via groundwater.  Estimates of Total P, Total N, and SiO2 to Ventura Marsh alone are 0.57, 2.64, and 123.1 kg/d, respectively, using the total median values.  Losses of these components from the lake in Table 7 are generally higher because additions to the lake from other sources are not included in the analysis.


            Chloride in groundwater, probably from road de-icing activity, adds about 113 kg/d to the lake. It is interesting to note that the mean outflow concentration of Cl (15.4 mg/L) is similar to the mean Cl concentration of the lake of about 16 mg/L.


            4.  Conclusions.  The geochemical environment of groundwater near Clear Lake was investigated and found to contain reducing conditions that may suppress NO3-N concentrations while liberating P.  Using analyses from 7 separate sampling dates, we found that groundwater adds about 1.37, 6.32, and 294.7 kg/d of P, N, and SiO2 to the lake.  Road de-icing activity, and undoubtedly some septic systems and fertilizers, adds about 113 kg/d of Cl to the lake.  Clear Lake itself loses nutrients and Cl via groundwater outflow.  The long-term future and health of Clear Lake may depend on identifying the sources and reducing inputs of these components to groundwater.



Anderson, M.P. and J.A. Munter. 1981. Seasonal reversals of groundwater flow around lakes and the relevance to stagnation points and lake budgets.  Water Res. Res. 17(4): 1139-1150.


Baker, J.L., K.L. Campbell, H.P. Johnson, and J.J. Hanway. 1975. Nitrate, phosphorus, and sulfate in subsurface drainage water.  J. of Env. Quality 4:406-412.


Baker, R.G., L.J. Maher, C.A. Chumbly, and K.L. VanZant. 1992.  Patterns of Holocene environmental change in the Midwestern United Statess. Quat. Res. 37:379-389.


Beauchemin, S., R.R. Simard, and D. Cluis. 1998. Forms and concentration of phosphorus in drainage water of twenty-seven tile-drained soils.  J. of Env. Qual. 27:721-728.


Belanger, T.V. and D.F. Mikutel. 1985. On the use of seepage meters to estimate groundwater nutrient loading to lakes. Water Res. Bull. 21(2): 265-272.


Bettis, E.A. 1998. Stop 3: Lunch at Clear Lake – Lake History in Anderson, R.R. and Bunker, B. J. (eds.) Fossil shells, glacial swells, piggy smells, and drainage wells: the geology of the Mason City, Iowa area. Geological Society of Iowa Guidebook 65, p. 27.


Bolton, E.F., J.W. Aylesworth, F.R. Hore. 1970. Nutrient losses through tile drains under three cropping systems and two fertility levels on a Brookston clay soil.  J. Soil Sci. 50:275-279


Bottcher, A.B., E.J. Monke, and L.F. Huggins. 1981. Nutrient and sediment loadings from a subsurface drainage system. Trans. ASAE. 24:1221-1226


Boyle, D.R. 1994. Design of a seepage meter for measuring groundwater fluxes in the nonlittoral zones of lakes – evaluation in a boreal forest lake.  Limnol. Oceanogr. 39(3): 670-681.


Brock, T.D., D.R. Lee, D. Janes, and D. Winek. 1982. Groundwater seepage as a nutrient source to a drainage lake; Lake Mendota, Wisconson. Water Res. 16: 1255-1263.


Cherkauer, D.S. and J.P. Zager. 1989.  Groundwater interaction with a kettle-hole lake: Relation of obvservations to digital simulations. J. Hydrol. 109: 167-184.


Crumpton, W.G., T.M. Isenhart, and P.D. Mitchell. 1992. Nitrate and organic N analyses with second-derivative spectroscopy. Limnol. Oceanogr. 37(4): 907-913.


Downing, J.A. and J.J. Peterka. 1978. Relationship of rainfall and lake groundwater seepage. Limnol. Oceanogr. 23(4): 821-825.


Haitjema, H.H. 1995. Analytic element modeling of groundwater. New York, Academic Press. 395 p.


Hunt, R.J. and J.T. Krohelski. 1996.  Application of an analytic element model to investigate groundwater-lake interactions at Pretty Lake, Wisconsin.  J. of Lake and Reservoir Management 12(4): 487-495.


Hunt, R.J., M.P. Anderson, and V.A. Kelson. 1998.  Improving a complex finite-difference ground-water flow model through the use of an analytic element screening model.  Ground Water 36(6): 1011-1017.


Hunt, R.J., Y. Lin, J.T. Krohelski, and P.F. Juckem. 2000. Simulation of the shallow hydrologic system in the vicinity of Middle Genesee Lake, Wisconsin, using analytic elements and parameter estimation.  U.S. Geological Survey, Water Resources Investigations Report 00-4136, 16p.


Hvorslev, M.J. 1951. Time lag and soil permeability in ground-water observation. Bulletin 36, U.S. Army Corps of Engineers Wat. Exp. Station.


Krabbenhoft, D.P., M.P. Anderson, and C.J. Bowser. 1990.  Estimating groundwater exchange with lakes, 2. Calibration of a three-dimensional, solute transport model to a stable isotope plume.  Water Res. Res. 26(10): 2455-2462.


Lee, D.R. 1977. Device for measuring seepage flux in lakes and estuaries.  Limnol. Oceanog.22(1): 231-238.


McBride, M.S. and H.O. Pfannkuch. 1975. Distribution of seepage within lakebeds. J. Res. U.S. Geol. Surv. 3(5): 505-12.


McDonald, M.G. and A.W. Harbaugh. 1988.  A Modular Three-Dimensional Finite Difference Model.  Techniques of Water-Resources Investigations of the U.S. Geological Survey, Book 6, Chapter A-1.


Nriagu, J.O. and C.I. Dell. 1974. Diagenetic formation of iron phosphates in recent lake sediments. American Mineralogist 59: 934-946.


Pennequin, D.F. and M.P. Anderson. 1983.  The groundwater budget of Lake Wingra, Dane County, Wisconsin.  Technical Report WIS WRC 83-01.  43 p.


Plummer, L.N., E.C. Prestemon, and D.L. Parkhurst. 1994. An interactive code (NETPATH) for modeling NET geochemical reactions along a flow PATH - version 2.0. U.S. Geological Survey Water-Resources Investigations Report 94-4169, 130 p.


Robertson, W.D., S.L. Schiff, and C.J. Ptacek. 1998. Review of phosphate mobility and persistence in 10 septic system plumes.  Ground Water 36(6): 1000-1010.


Schwab G.O., N.R. Fausey, and D.E. Kopcak. 1980. Sediment and chemical content of agricultural drainage water.  Transactions of the ASAE 23: 1446-1449


Shaw, R.D. and E.E. Prepas. 1989. Anomalous, short-term influx of water into seepage meters. Limnol. Oceanogr. 34(7): 1343-1351.


Shaw, R.D. and E.E. Prepas. 1990a. Groundwater-lake interactions: I. Accuracy of seepage meter estimates of lake seepage.  J. of Hydrol. 119: 105-120.


Shaw, R.D. and E.E. Prepas. 1990b. Groundwater-lake interactions: II. Nearshore seepage patterns and the contribution of ground water to lakes in central Alberta.  J. of Hydrol. 119: 121-136.


Shaw, R.D., J.F.H. Shaw, H. Fricker, and E.E. Prepas. 1990. An integrated approach to quantify groundwater transport of phosphorus to Narrow Lake, Alberta. Limnol. Oceanogr. 35(4): 870-886.


Simpkins, W.W. and T.B. Parkin. 1993. Hydrogeology and redox geochemistry of CH4 in a late Wisconsinan till and loess sequence in central Iowa.  Water Resources Research 29(11):3643-3657.


Simpkins, W.W. 1995. Isotopic composition of precipitation in central Iowa.  J. Hydrol. 172:185-207.


Strack, O.D.L. 1989.  Groundwater Mechanics: Englewood Cliffs, N.J. Prentice-Hall, 732 p.


Williams, J.D., J.-M. Jaquet, and R.L. Thomas. 1976. Forms of phosphorus in the surficial sediments of Lake Erie.  J. Fish. Res. Can. 33: 413-429.


Winter, T.C., J.W. LaBaugh, and D.O. Rosenberry. 1988. The design and use of a hydraulic potentiomanometer for direct measurement of differences in hydraulic head between groundwater and surface water.  Limnol. Oceanog. 33(5): 1209-1214.


Woessner, W.W. and K.E. Sullivan. 1984. Results of seepage meter and mini-piezometer study, Lake Mead, Nevada. Ground Water 22(5): 561-568.