Table of Contents

Introduction

Part I

Appendix A

Appendix B

Appendix C

Part II A and B

Part II C and D

Part III

Part IV

References

Summary

Part II—continued

C13. Water Budgets for the Major Kansas Wetlands of Cheyenne Bottoms and Quivira National Wildlife Refuge

The two major Kansas wetlands of Cheyenne Bottoms and Quivira National Wildlife Refuge are shown in figs. II–10 and II–13. Sophocleous and Shapiro (1987) employed the Versatile Soil-moisture Budget (VB) model of Baier et al. (1979) to calculate the water budget of the nonwater-covered area of the Drummond–Tabler soil association (more than 86% of total area of approximately 60 mi²), which encompasses the Cheyenne Bottoms. The water budget was run for two water years (Oct. 1, 1982–Sept. 30, 1984). The vegetation cover of this area consisted of 44% small grains, 40% grasses, and 12% cattails. The average hydrologic budget for that area is shown in table II–21 with an estimated deep drainage of 2 inches/yr. This deep-drainage value is practically equal to ground-water recharge because of the shallow water table in the area.

TABLE II–21—Predevelopment (pre-1960) and present-day (1991–94) annual water budgets for the Rattlesnake Creek basin (values in inches/yr).

TABLE II–22—Average 1983 and 1984 water-year budget for the Drummond–Tabler soil association encompassing the Cheyenne Bottoms.

TABLE II–23—1985–1992 hydrologic budget components for Quivira National Wildlife Refuge (based on Hudson NOAA climatic station).

Sophocleous and Ma (1993) also analyzed the hydrologic budget of a portion of the Rattlesnake Creek watershed encompassing the Quivira National Wildlife Refuge (NWR) in south-central Kansas (145 mi²). They employed the Versatile Soil-moisture Budget (VB) model, supplemented with additional surface runoff routines (Sophocleous and Ma, 1993), to estimate the daily hydrologic budget of the study area during the 8-yr period 1985–1992. The 8-yr average hydrologic budget for the nonwater-covered area pertaining to the Quivira NWR is estimated as shown in table II–23.

Note that the deep drainage is highly variable (high coefficient of variation) and is practically equal to ground-water recharge because of the relatively shallow depth to water table in the Quivira NWR.

C14. Equus Beds Aquifer Modeling

a) Green and Pogge (Green et al., 1973) Study

Green and Pogge (Green et al., 1973; see also Knapp et al; 1975) developed a comprehensive basin hydrology simulator by combining the Kansas Water Budget Model (Smith and Lumb, 1966), a model similar to the Stanford Watershed Model, with a two-dimensional finite difference ground-water model. They field-tested this simulator on the Little Arkansas River basin that encompasses the Equus Beds aquifer for the 1946–1970 period. Ground-water recharge data are only presented for subbasin 5 (within the drainage basin of Little Arkansas River), which includes 139 mi² in portions of townships 23 S. to 25 S. and ranges 1 W. to 2 W., including the towns of Halstead and Sedgwick, Kansas.

Figure II–11 from that study shows the percent of annual precipitation falling on that subbasin that percolates to ground water for the period 1946 to 1970. The amount of percolation recharge is highly dependent on the actual hydrologic conditions that exist during each year. It is clear from that figure, that use of a simple fraction of the precipitation would result in significant errors in the amount of percolation recharge.

FIGURE II–11—Percent of annual precipitation for the period 1946–1970 falling on subbasin 5 within the drainage basin of Little Arkansas River (encompassing Halstead and Sedgwick, Kansas) that percolates to ground water (adapted from Green et al., 1973).

b) Sophocleous et al. (1982) Study

Sophocleous et al. (1982; see also Sophocleous, 1984) employed a finite difference (1-mi² grid cell) parameter-estimation model (Cooley, 1977, 1979) to a 240-mi² area of the Equus Beds aquifer encompassing portions or all of townships 22 S.–25 S. and ranges 1 W.–4 W., including the towns of Burrton, Halstead, and Sedgwick, as well as the Wichita well field area. Steady-state conditions (existing during the early 1940’s) were employed in optimizing model parameters. Two recharge zones were considered in that model, one in the sand dune area north of the town of Burrton, and the other encom-passing the rest of the area. A detailed uncertainty analysis of the model results is presented in that study. The “optimized” recharge estimate for the sand dune region north of Burrton was 6.5 inches/yr, with a standard error of 1.2 inches/yr, whereas the rest of the model region, including the Wichita well field, had a recharge of 1.65 inches/yr with a standard error of 0.5 inch/yr. The ratio of the square root of the error variance in the parameter-estimation model over the difference between the highest and lowest value of head in the region was 3.6 ft/150 ft = 0.02, a relatively small value, indicating that errors in the model were considerably less than the model response as indicated by the maximum head loss of 150 ft. {Sophocleous et al., 1982 (see also Sophocleous, 1984) employed three different solute-transport models to predict the extent and concentration of the brine plume near Burrton vis- a´-vis the Wichita well field.}

c) Spinazola et al. (1985) Study

Spinazola et al. (1985) employed the three-dimen-sional USGS MODFLOW model to simulate the Equus Beds (upper layer) and Wellington (lower layer) aquifers in Sedgwick, Harvey, Reno, McPherson, and Marion counties using a regular 1-mi² cell grid. The model was run for both steady-state (1940) and transient conditions (1940–1979). Calculated recharge to their model was considered to be a function of 1940 precipitation, the soil type, and thickness of clay in the unsaturated zone. The combination of soil types and unsaturated clay thicknesses were used to define a “recharge factor” for each cell of the model grid, that when multiplied by the average annual precipitation resulted in the recharge estimates employed in the modeling. Figure II–12 displays the 1940 initial-condition simulation recharge estimate, with values ranging from 0.1 to 5.5 inches/yr. The transient simulation consisted of five stress periods between 1940 and 1979 (1940–1952, 1953–58, 1959–1963, 1964–1970, and 1971–79) corresponding to uniform trends in withdrawal from the aquifer, where the well withdrawal was averaged for the length of the stress period. Recharge changed for each stress period as the product of the “recharge factor,” mentioned previously, and average precipitation during the stress period. The simulated water budget for the Equus Beds aquifer for the end of 1940 initial-condition simulation and for the periods 1964–1970 and 1971–79 are shown in table II–24.

FIGURE II–12—Simulated predevelopment (1940) recharge distribution in the Equus Beds aquifer, Kansas (adapted from Spinazola et al., 1985).

TABLE II–24—Simulated water budget for the Equus Beds aquifer (values in inches/yr).

Between 1940 and 1964–1970, withdrawals by wells increased 1,630%. Streamflow gain decreased by 54%, whereas streamflow loss increased by 760%. Between 1964–1970 and 1971–79, withdrawal by wells increased by about 42%. Recharge increased by about 13% during this period; however, the rate of decrease in storage was about 26%, resulting in lower water-table elevations, which in turn resulted in decreased baseflow to streams by about 21%, while loss from streams to the aquifer increased by about 57%. Declining water levels also resulted in a small decrease in ground-water transpiration and boundary flow (Spinazola et al., 1985).

d) Myers et al. (1996) Study

Myers et al. (1996) also employed the three-dimensional USGS MODFLOW model to study interactions between the Arkansas River and the Equus Beds aquifer in parts of Reno, Harvey, and Sedgwick counties, an area of approximately 690 mi², that includes the cites of Hutchinson, Newton, and Wichita. The model area was divided into three layers, and was calibrated for both steady-state (pre-1940) and transient conditions (1940–1989). A variable-spacing model grid was laid out with rows parallel to the Arkansas River, with a smaller grid spacing (1,000 ft x 5,000 ft) near the river. The same recharge distribution developed by Spinazola et al. (1985) for 1940 was assumed valid for the steady-state pre-1940 model (representing 1934–39 conditions). The mean absolute difference between measured hydraulic heads for 235 individual wells and their corresponding middle-layer simulated hydraulic heads was 3.20 ft.

The transient model used the same five stress period as the ones used by Spinazola et al. (1985) plus a sixth stress period (1980–89), characterized by marked fluctuations in the volume of agricultural pumpage. Recharge was based on the mean precipitation at climatic stations at Hutchinson, Mount Hope, and Wichita for each stress period, and was estimated as follows (Myers et al., 1996): 1) the recharge specified for each steady-state model cell was divided by the mean annual precipitation for the pre-1940 period represented in the steady-state model; 2) the resulting quotient for each model cell was then multiplied by the study-area mean annual precipitation for each stress period in the transient model. The mean absolute difference between hydraulic heads for 232 individual wells and their corresponding middle-layer model cell at the end of 1989 was 4.67 ft. Streamflow that was exceeded 70% of the time at each gaging station, assumed to represent baseflow, was compared to model-simulated flow in the stream reach where the gaging station was located. Because streamflows specified for the starting stream reach in the model were held constant for each stress period, the model did not simulate the annual seasonal or short-term variation of measured streamflow (Myers et al., 1996).

The simulated steady-state and transient ground-water budgets for the periods 1964–1970, 1971–79, and 1980–89 are shown in table II–25.

TABLE II–25—Simulated steady-state and transient ground-water budgets for all three model layers of the Equus Beds aquifer (values in inches/yr).

In general, from 1940 to 1989 boundary inflow, streamflow loss, and well pumpage increased appreciably; and boundary outflow, streamflow gain, and ground-water evapotranspiration decreased. In response to the declin-ing ground-water levels, streamflow gains decreased in the Arkansas River within the model area (from 21 cfs in 1940 to a simulated baseflow loss of about 52 cfs by the end of 1989) and in the Little Arkansas River (from 67 cfs in 1940 within the model area to about 27 cfs by the end of 1989). During 1940-1989, the quantity of chloride discharged from the Arkansas River to the Equus Beds aquifer increased in direct proportion to the volume of water loss from the river (Myers et al., 1996). On the basis of simulated streamflow and assuming that the chloride concentration in river water that moves into the aquifer is 630 mg/L (which is the median concentration in Arkansas River water collected during 1988-1991), the chloride-load discharge from the river to the aquifer was estimated to be about 21 tons/day in 1940 and about 100 tons/day by 1989.

C15. Other County-scale Recharge Studies

a) Grant and Stanton Counties, Southwest Kansas

Fader et al. (1964) estimated recharge from precipitation from a 1939–1942 water-level contour map of Grant and Stanton counties, assuming negligible well pumpage at that time period. By carefully selecting a study area of 160 square miles between the towns of Johnson and Ulysses, where it could be assumed that leakage inflow from the underlying sandstone aquifers was negligible, they estimated the recharge rate to be about 0.3 inch/yr, which is about 2% of the annual precipitation. This recharge rate was applied to the rest of the Grant and Stanton unconsolidated aquifers in their study. Fader et al. (1964) also estimated the recharge of the unconsolidated aquifers in the area from upward leakage from the underlying sandstone aquifers to be of approximately the same magnitude as the precipitation recharge.

b) Seward County

Byrne and McLaughlin (1948) estimated recharge from precipitation in Seward County based on 1941–44 well hydrograph data from three upland wells that showed an average rise of 0.22 ft/yr and assuming an aquifer specific yield of 0.15 to come up with a recharge estimate of about 0.4 inch/yr.

c) Meade County

Frye (1942) made an inventory of water discharged from the artesian water-bearing beds of Meade County by upward leakage through the confining beds, along faults, through springs and flowing and nonflowing wells, and determined that the total annual discharge of artesian water was about 10,000 acre-feet. He concluded that nearly all the recharge to the artesian water-bearing beds occurs to the west and north of the county in southwestern Gray County, parts of Haskell County, northeastern Seward County, and the southernmost part of Finney County, an area of approximately 685 square miles. Assuming equilibrium conditions, where the recharge is assumed to be equal to the discharge, Frye (1942) estimated an average recharge of about 0.27 inch/yr, which represents about 1.5% of average annual precipitation in the area (considered as 18 inches/yr). According to Frye (1942), in the area of sand dunes in southern Finney County and northern Haskell County that percentage is probably much greater, and in certain other parts of the area it is probably much less, and locally there may be none at all.

d) McPherson Moratorium Area, Equus Beds Aquifer

McElwee et al. (1979) conducted a water-budget analysis of the McPherson moratorium area (56 mi2) of the Equus Beds aquifer based on January 1978 water-table levels. Recharge was estimated as 2 inches/yr (table II–26).

TABLE II–26—1978 water-budget analysis of the McPherson moratorium area, Equus Beds aquifer.

D. Field-based Experimental Recharge Studies

Field-experimental ground-water recharge studies in Kansas are highlighted in this section.

D1. Movement of Moisture in the Unsaturated Zone in a Dune Area, Southwestern Kansas (Prill, 1968)

Prill (1968, 1977) conducted a study to investigate the requirements necessary for deep percolation under three types of vegetative conditions: 1) a barren area, 2) a sagebrush-grass community, and 3) a grass community, over a period from the fall of 1964 through 1966. This period depicts maximum moisture changes because 1965 was a year when precipitation was one of the highest on record (29.07 inches in Garden City) and was preceded and succeeded by years when precipitation was below normal (12.23 inches in 1964 and 12.04 inches in 1966).

The study site is located in the extensive area of dune sand immediately south of the Arkansas River near Garden City in southwestern Kansas. At each site, neutron probe access holes containing 2-inch aluminum tubing for moisture-content logging were drilled through the dune sand into the underlying alluvial deposits. The water table in this area was about 30 ft below the top of the alluvial deposits.

Even though the period of study included a year when precipitation was nearly the highest on record, built-up moisture under a sagebrush-grass community penetrated to a depth of only 14 ft, whereas the zone of evapotranspiration extended to at least 17 ft (Prill, 1968). Under a grass community where the zone of evapotranspiration extended to about 11 ft only a small amount of moisture (2 inches) moved as deep percolation. Under a barren area, where most of the loss by evaporation occurred in the upper 1 ft, large quantities of moisture moved as deep percolation.

Prill (1968) estimated that the average annual recharge in the vegetated area is about 0.5 inch. Prill (1968) pointed out that the periods when conditions are favorable for recharge are few and usually occur when precipitation is considerably above average during the nongrowing season and the early part of the succeeding growing season. The high rate of evapotranspiration all but eliminates the possibility of recharge during the summer months.

D2. Pilot Recharge Assessment at Two Sites in South-central Kansas (Great Bend Prairie and Equus Beds Aquifers) (Sophocleous and Perry, 1985)

Sophocleous and Perry (1984, 1985, 1987) experimentally assessed ground-water recharge at two instrumented sites in south-central Kansas (one site near Zenith in Reno County (GMD5), and one near Burrton in Harvey County (GMD2) over an approximately 19-month period during 1982 and 1983. Although the two instrumented sites were located in sand-dune environments in areas characterized by shallow water table and subhumid continental climate, a significant difference was observed in the estimated effective recharge. The estimates ranged from less than 0.1 inch at the Zenith site to approximately 6.1 inches at the Burrton site during the major recharge period from February to June 1983. The main reasons for this large difference in recharge estimates were the greater thickness of the unsaturated zone and the lower moisture content in that zone resulting from lower precipitation and higher potential evapotranspiration for the Zenith site. Effective recharge took place only during late winter and spring. No summer or fall recharge was observed at either site during the observation period of this study.

D3. Recharge Assessment for the Great Bend Prairie Aquifer in GMD5 (Sophocleous, 1992, 2000c)

Recharge-related variables were monitored in the field on a year-round basis at 10 sites distributed throughout the GMD5 area (fig. II–13; Sophocleous, 1991, 1992, 1993a, 1993b, Sophocleous and Stern, 1993). The methodology used in quantifying recharge for the region consisted of combining the hydrologic or soil-water balance on a storm-by-storm year-round basis with the resulting water-table rises. Each recharge-assessment site was equipped with a weighing and recording rain gage, a neutron-probe access tube for measuring the soil-profile water content, a water-table well with a water-level recorder, and two deeper piezometers. Two of the sites also were equipped with weather stations that recorded solar radiation, air temperature, relative humidity, barometric pressure, and wind speed. Using the data collected at these sites and detailed weather data from the Sandyland Experiment Station, just south of St. John (fig. II–13), the soil-water balance for each recharge-producing storm period was calculated. By associating the result with the consequent water-table rise, which was tied to specific precipitation events, reliable effective recharge values for different storm periods were obtained (Sophocleous, 1991).

FIGURE II–13—Groundwater Management District No. 5 boundary (heavy gray line) with ground-water recharge assessment sites (triangles). River basins are outlined with black lines.

Table II–27 gives the average values of measured precipitation, depth to water table, and estimated annual recharge for all the monitored sites in GMD5. For the original recharge sites 1 through 5, data have been collected from 1985 to 1992, whereas for sites 6 through 10 data were collected during the 1988–1992 period. The unusually high recharge estimates for site 4 in Reno County, which received the highest precipitation among all sites, were due to the site being located on the streambank of a tributary to Wolf Creek where the depth to the water table is very shallow, approximately 2–4 ft. Sites 8, 9, and 10 received no detectable recharge during the period 1988–1992.

TABLE II–27—1985–1992 site-specific ground-water recharge estimates for GMD5.

During the flood year of 1993, average precipitation in Stafford County was approximately 36.6 inches (which is approximately double the long-term average) and estimated average recharge (based on sites 1, 2, 3, and 5) was 5.9 inches. This amount of estimated recharge caused an average county-wide water-table rise of 5.4 ft (Sophocleous et al., 1996).

Sophocleous (1992, 1993a, 2000c), using a combination of statistical (forward stepwise regression) analysis and GIS overlay analysis, identified the portion of the GMD5 area that each recharge site or cluster of sites represents (fig. I–11), and derived an area-weighted average recharge for the GMD5 of 1.4 inches/yr based on the 1985–1990 recharge-site data, as shown in table II–28.

TABLE II–28—Recharge zonation of GMD5 based on GIS overlay analysis.

Additional details on the regionalization methodology for the Great Bend Prairie recharge are presented in Part I, Section 8.2. Additional information on GMD5 recharge can be found in Sophocleous (1992, 1993a, 1993b, 2000c).

D4. Deep Vadose Zone Study of Ground-water Recharge in the High Plains Aquifer of Southwest Kansas (Sophocleous et al., 2002; McMahon et al., 2003)

Recent improvements in technology make it possible to study the deep vadose zone, well below the rooting depths of plants. Such technology has been employed to monitor, on a continuous basis, the deep pore-water fluxes en route to the High Plains aquifer (HPA). This proof-of-concept pilot study provided, for the first time, information on the quantity and pattern of water fluxes in the deep vadose zone that impact the management of both the quantity and quality of the HPA in Kansas. The Sophocleous et al. (2002) preliminary investigation evaluated the use of heat-dissipation sensors and advanced tensiometers for measuring pore-water pressures in deep boreholes at two irrigated land-use sites (Sites 1 and 2 located in southern Finney County, fig. II–7) and one natural grassland site (Site 3 located in the Cimarron National Grassland in Morton County, fig. II–7). Continuous time-series data obtained from the heat-dissipation sensors (installed at 116 ft at the irrigated sites and at 137 ft at the grassland site) revealed constant pore-water pressures with time over the May 2000–September 2001 monitoring period.

The observed time-series patterns of pore-water pressure head in the deep vadose zone of the High Plains imply homogenization at depth of near-surface, temporally varying water fluxes, resulting in much lower intensity but nearly constant (steady) recharging fluxes to the High Plains aquifer.

The measured average hydraulic-head gradients for all sites, together with the estimated recharge fluxes based on Darcy’s law, are shown in table II–29. It is interesting to note that the hydraulic-head gradients between the deepest heat-dissipation sensor and the ground-water level at each site were approximately 0.75 for Sites 1 and 2, whereas for Site 3 it was nearly zero.

TABLE II–29—Measured hydraulic-head gradients and estimated water fluxes in deep, unsaturated High Plains sediments based on heat-dissipation sensors. Site locations are shown in fig. II–7 (adapted from Sophocleous et al., 2002).

Darcian methodology was used to obtain estimates of recharge at all the three investigation sites. Estimated recharge rates for the irrigated land-use sites (Sites 1 and 2) were appreciably higher (0.12 inch/yr and 0.02-0.04 inch/yr, respectively—table II–28) than the estimated recharge rate for the natural grassland site (Site 3; less than 0.01 inch/yr—table II–29). In all cases the estimated annual recharge values were less than one percent of annual precipitation. Although the large uncertainty associated with these estimates and the small number of study sites precludes using these flux estimates alone to draw firm conclusions regarding present-day recharge in the region, the irrigated and natural grassland sites selected are representative of irrigated and grassland areas overlying the High Plains aquifer in southwestern Kansas.

The U.S. Geological Survey conducted several chemical and tracer analyses at the aforementioned sites (McMahon et al., 2003). Chloride profile analysis (see Appendix B of Part I on tracers for recharge estimation) of the grassland site indicated that the recharge flux below the root zone ranged from less than 0.1 to 0.39 inch/yr, with an average of 0.2 inch/yr, and that estimate is considered to be a long-term (on the order of hundreds of years) estimate of recharge at the grassland site (McMahon et al., 2003). Both irrigated sites had bomb tritium (see Appendix B of Part I) and pesticides detected in both the unsaturated and saturated zones (McMahon et al., 2003), which implies a much higher recharge rate at those sites than at the grassland site.

The Darcian-based water-flux estimation aspect of this High Plains aquifer program was a pilot study of recharge assessment. That study showed that deep-vadose zone hydrology, a mostly unexplored frontier due to technological obstacles, can be monitored and analyzed. More instrumented sites, similar to the ones employed here (and taking advantage of the experience gained during this study), in combination with additional methodologies, are needed to assess the deep vadose zone water (and chemical) fluxes reaching the High Plains aquifer.