Missouri River Groundwater Study

[Bottom of Page]

Section 1: Introduction

This report is the culmination of a groundwater study of three sites: two levee units and one unleveed area along the Missouri River. Foster Wheeler Environmental Corporation (Foster Wheeler Environmental) and West Consultants performed this study for the U.S. Army Corps of Engineers (Corps) Missouri River Region (MRR). The study sites, in upstream to downstream order, are River Mile 691 (RM691), Levee Unit 488/497 (L488/497), and Tri-County Levee District No. 2 (Tri-County LD 2). The sites are located on the river’s left bank (left side looking downstream) near river miles (RMs) (1960 version) 691, 497, and 95. The L488/497 site is actually two levee districts that were combined into one unit for this study. This study consists of computer modeling of the groundwater hydrology using the U.S. Geological Survey (USGS) MODFLOW program.

Groundwater is an issue that the Corps is addressing as part of the Missouri River Master Manual Review and Update (Master Manual Study). Essentially, this document establishes the guidelines for operating the Corps dams and reservoirs on the Missouri River mainstem. The MRR is responsible for applying these guidelines when making releases from the Corps six mainstem reservoirs. The amount of water released at the Corps’ Gavins Point Dam in South Dakota to some extent influences the river flow and stage in the lower 811 miles of the Missouri River to its mouth at St. Louis. The influence of the dam releases diminishes progressively downstream, as more and more tributaries contribute water to the mainstem.

Because of the nature of the modifications being evaluated, the Corps determined that an environmental impact statement (EIS) would be prepared. The Corps issued the Master Manual Study Draft EIS (DEIS) for public review in 1994. In that DEIS, the Corps identified several alternatives for river operation and the public submitted comments on the DEIS to the Corps. A large number of comments were made, primarily by farmers, regarding the potential for crop damage that could result from one of the alternatives due to high spring river stages.

This alternative, which the DEIS identified as the preferred alternative, proposed greater spring releases and lesser flow later in the season, effectively raising river stages during the spring and lowering river stages during the fall. Farmers who worked land within the levee interior areas expressed concern that the high spring river stages would raise groundwater levels to or near to the ground surface and retard interior drainage and free movement of air and water through the soil root zone. Presumably, this could increase the risk of crop damage or loss. Additionally, the farmers identified a potential impact from groundwater levels dropping excessively from higher spring levels during the early summer. They were concerned that crop damage could result from this condition because roots might grow to shallow depths during the spring due to higher than normal groundwater levels, and the stunted roots might dry out during the summer when groundwater could be much lower.

The intent of this groundwater study is to model the groundwater process at selected sites along the river and to predict the depth to groundwater in the interior area on a daily basis for two alternative river operations: the Current Water Control Plan (CWCP) and an Alternative Plan. For the leveed sites, interior refers to the land area that is protected from mainstream Missouri River floods by a levee. For the RM691 site, interior refers to the floodplain area. The two river operation plans differ significantly in annual flow characteristics (the Alternative Plan has similar flows to that of the DEIS preferred alternative). MRR selected the three study sites as a representative sample of agricultural areas within the floodplain in the lower 811 miles of the river. A MODFLOW computer model was constructed for each of the three levee units. The models were used to simulate the groundwater levels for a 10-year period starting in October 1970 and ending in September 1979. The results of this study will be entered into a MRR modified project benefits accomplishment (HEC-PBA) model to estimate and compare cost impacts associated with various alternative river operation scenarios.

The objective of this report is to demonstrate the technical soundness of the MODFLOW models as applied to the two operational scenarios and three study sites. Conclusions regarding the preference of one alternative versus another will be presented in a revised DEIS (RDEIS). Based on the progression of future RDEIS work, this report may also be updated and included wholly or in part as an attachment to the RDEIS.

This report is organized by study site. General methodologies are described later in this chapter, and a summary of results follows the last study site section. One section is devoted to each study site. The site sections are presented in upstream to downstream order, starting with RM691. The site sections include subsections on geographic setting; data collection; development of a conceptual model of the groundwater hydrology; MODFLOW model calibration, validation, and sensitivity; and 10-year simulation results. The summary section discusses the more important considerations and findings of the modeling process. The final section provides references.

The appendices that follow the main report include key data, supporting reports, and data analyses. Soil boring logs for at least the USGS monitoring wells and each site’s exploratory boring are in Appendix A. The data from the falling-head permeability tests are in Appendix B. Appendices C and D contain aquifer test reports. Soil boring logs for Tri-County LD 2 appear in the Appendix of the Tri-County Levee Unit 2 Aquifer Test Report in Appendix D. Appendix E shows detailed plots of the model simulation computed heads. The results from pre-simulation testing are in Appendix F.

Each study site was investigated in a consistent manner: the MODFLOW computer program was applied in each case, and the data collection and input and output processing were conducted uniformly. The approach was general, but the data were predominantly site-specific. The study teams collected most of the data for building the models, but the USGS provided groundwater level data for model calibration and MRR provided the river stages for model simulations. This section presents a description of the general modeling methodology and issues that were commonly addressed for each levee unit study.

The application of the MODFLOW program followed a typical modeling process: data collection, model construction, calibration, and simulation. The input required by the computer modeling program defined the type of data to be collected. MODFLOW input data are organized into modules, and each module represents a unique object of hydrologic importance (e.g., precipitation and river stage). During data collection, a conceptual model was incrementally developed for the hydrogeologic system of each site. The conceptual model is a description of the local hydrogeologic system and how this system is represented by the data modules and mathematical equations that compose the MODFLOW computer program. Using the conceptual model as a guide, a MODFLOW model was built for each study site. The models were calibrated to the groundwater level observations taken by the USGS from November 1995 to January 1996. This was a period when the Missouri River stages were declining. Calibration was validated using USGS groundwater level observations taken during the January and February of 1996 following the calibration period. Model calibration and validation lends credence to the simulation results attained by running the model for the 10-year period of record.

In order to guide the data collection process as well as to prepare the MODFLOW input files, it was necessary to develop a conceptual model of the groundwater at each site. The conceptual model is basically an organized, quantitative understanding of the physical factors that determine groundwater flow at the sites, including:

• geologic framework
• hydrologic framework
• hydraulic properties
• sources and sinks of groundwater.

The geologic framework includes site-specific data on the physical location and stratigraphy of underlying aquifers and aquitards (confining units). Stratigraphic information includes strata thickness, continuity, lithology, geologic structure, and depth to and nature of bedrock. The hydrologic framework includes groundwater flow directions, boundaries, potentiometric elevations, time-varying river stages, and vertical and horizontal gradients. Hydraulic properties include river bed conductance; aquifer transmissivity; hydraulic conductivity, storativity, and specific yield; homogeneity and spatial variations; and anisotropy. Sources and sinks of groundwater include processes through which groundwater enters or leaves the modeled aquifer. These processes involve direct recharge (precipitation, evapotranspiration (ET), and infiltration) and inflows from upland recharge areas, drains and pumping, and hydraulic connectivity with the Missouri River.

The development of the conceptual models was an iterative process. As more information became available, and after the models were first built, the behavior of the groundwater system became clearer. Potentiometric elevations were monitored in USGS wells in each study site throughout the early stages of model development. Before model calibration began, the frequency and magnitude of groundwater response to time variable river stages were analyzed to estimate an appropriate model simulation time step. Additional conceptual model information, such as regional aquifer discharges, was collected during the calibration process to investigate reasons for differences between computed and observed heads.

Foster Wheeler Environmental conducted two main data collection tasks: office and field. Office data (hydrologic and geotechnical reports, levee design and operation and maintenance (O&M) manuals, maps, electronic data files, etc.) were collected from the Corps’ Omaha and Kansas City Districts; the USGS Iowa City and Independence Districts; the Iowa Department of Natural Resources (DNR); the University of Missouri at Kansas City and Columbia; the City of Columbia, Missouri Water Department; and the Midwestern Climate Center. Field data were collected at each study site. Water levels from on-site monitoring wells were obtained from the USGS. Foster Wheeler Environmental also drilled and logged one exploratory boring to refusal and performed one to two falling head permeability tests at each study site. Additionally, Foster Wheeler Environmental performed two pump tests: one at Tri-County LD 2 and another at Levee Unit L550 in the vicinity of the L488/497 site. MRR provided the Missouri River stages data for the model simulations.

Calibration is the process of adjusting initial estimates of model input parameter values until the computed groundwater elevations for a single point in time, or for drawdowns over a period of time, closely match actual observations. The observed elevations and drawdowns are referred to as calibration and validation targets. The models were calibrated to steady-state conditions using elevation targets, and to transient conditions using drawdown targets. Water levels observed in USGS monitoring wells from November 1995 through February 1996 were used for the calibration and validation targets for all of the models. The differences between computed and target elevations and drawdowns (i.e., range of errors) had to be within a specified tolerance and statistical distribution for the model to be considered calibrated. The models were calibrated to the early part of the USGS monitoring well record, and validated to the later part. The calibration tolerances were established in the Quality Control (QC) plan that was developed for this project.

The statistical distribution of error between computed and target calibration parameter values is summarized in a CALSTATS table for each model. This error is referred to as residual, and the mean, variance, and root-mean-square (RMS) for the residuals are the primary statistics of interest. The goals for calibration of the model were to: reach a mean residual close to zero, minimize the absolute mean residual, and maintain an RMS equal to or less than the maximum range of observed heads. In addition to CALSTATS, the spatial distribution of the residuals was assessed to detect any systematic error in one or between model layers. The spatial distribution of residuals was qualitatively checked for randomness.

Steady- state models were built and calibrated to the groundwater elevations observed in the USGS wells at a single point in time. A period of constant river stage and precipitation is required for calibration of a steady-state model. The summer and fall of 1995 qualified because the Missouri River was running at a fairly constant stage, and the weather was consistently hot and dry. Steady-state calibration targets were selected from late November 1995 observations.

A transient model was built after steady-state calibration was achieved. The transient model was then calibrated and validated to drawdowns observed over several months following the steady-state calibration date. The transient calibration used the computed heads from the calibrated steady-state model for initial conditions from which drawdowns were computed. Transient calibration often resulted in a calibration parameter set that was different from the steady-state model set. When this occurred, the steady-state model had to be rerun with the transient parameter set to confirm that the newly computed steady-state heads remained within calibration tolerances. Sometimes several iterations of this steady-state and transient calibration cycle were made until a balance between steady-state and transient calibration was obtained.

Model validation was conducted after achieving steady-state and transient calibration. Ideally, groundwater elevations and drawdowns from an independent steady-state target date and transient target period would be available for model validation. The calibrated model would be run for the validation date(s) and the computed heads and drawdowns would be compared to the calibration tolerances; however, the available USGS monitoring well data indicated that there were no suitable steady-state conditions to validate the steady-state model. Steady-state validation was not conducted for this reason. Sufficient USGS monitoring well data were available to conduct a transient validation. Data from the period following the end of transient calibration were used for model validation.

The transient calibration period was assumed to be independent of the validation period even though their end and start dates, respectively, were the same. The transient calibration period was selected because the Missouri River stage was declining throughout December 1995 after about a 6-month period of continuous high discharges from Gavins Point Dam. The transient calibration period ended when the Missouri River decline was complete. The river stages during the following validation period probably responded more to tributary inflows, local precipitation, and temperature than to releases from Gavins Point Dam. This change in the river stage control was the basis for assuming that the validation period was independent of the calibration period.

During calibration, model input parameter values were varied to determine their effect on model output. A detailed sensitivity test of the calibrated steady-state and transient model was conducted on the parameters that were applicable to the calibration process.

The calibrated L488/497 model was used to investigate the short-term response of groundwater to river stage changes and precipitation events. Test runs were made for a simulation period of 1 month and 1 year. Both the CWCP and the Alternative Plan river stages were simulated for these periods for a total of four test runs. The results of the tests were used to determine the simulation time step length and to estimate the level of effort required to run long-term simulations.

The initial conditions were created by running the calibrated steady-state model with the river stage for the date of the first pre-simulation time step. An initial condition was developed for both the CWCP and the Alternative Plan. The start date of the 1-month test was June 21, 1986, and for the 1-year test the start date was October 1, 1984. Appendix F contains figures that show the test simulation results.

The results of the tests provided additional information on initial conditions and their influence on the simulations. The initial conditions are an estimate of the heads at the start of simulations. The initial conditions consist of a head for each cell in each layer of the model. The modeled conditions immediately following a model startup are usually not accurate because the initial conditions are only estimates of the actual hydrologic conditions and have not achieved dynamic equilibrium. Simulations must be run until a dynamic equilibrium for the overall model is reached, or in other words, until the start-up effects of the initial conditions have passed. The 1-month and 1-year test runs showed that the start-up effects due to initial conditions were persistent and that dynamic equilibrium was not reached during the simulation periods. The results are discussed in greater detail in Section 3.6, but it is useful to highlight the important results here because the approach to simulation was based on the pre-simulation test results.

The 1-month test showed that the difference between the Alternative Plan and CWCP initial conditions propagated through to the end of the simulation. The influence of the difference in river stages throughout the month was overshadowed by the difference in the initial conditions. The results showed no evidence of the model reaching a dynamic equilibrium within a month.

The 1-year test showed that the model did not reach dynamic equilibrium within a year at cells located further away from the river. Model cells close to the river responded quickly to river stage changes, and they reached dynamic equilibrium within the year. The head at cells further away from the river (mid-way between the river and the bluffs) steadily decreased throughout the year, indicating that the model was recovering from initial conditions at model startup and that an equilibrium period longer than one year would be required for the model to completely reached dynamic equilibrium.

The initial condition effect occurs mainly because we do not know conditions in every model cell at model startup and because the hydrologic system is never truly in a steady state. In other words, the steady-state condition is a condition that remains constant forever; this condition never actually occurs within the study sites (or any site along the Missouri River). The conditions described by a steady-state model show an aquifer that is in a static equilibrium, which is much different than an aquifer in a dynamic equilibrium. An aquifer takes forever to reach true static equilibrium, whereas dynamic equilibrium may be achieved after several cycles of the most slowly varying component of river stage and aquifer recharge have occurred. The main difference between the two equilibria for the test runs is that the heads at locations far away from the river are higher for the static than the dynamic equilibrium case because static equilibrium was not reached throughout the model and too much water was stored in the aquifer at model startup.

The transition from a steady-state initial condition at model start up to a dynamic equilibrium as the simulation progressed is shown in the Appendix F figures of the water year 1986 simulation. The decreasing trend in the heads shown in the L488/497 figures indicates that excess water was stored in the aquifer at the initial condition and it slowly drained during the year. These results also indicate that the transition from static to dynamic equilibrium was not complete after 1 year. These tests showed that it was necessary to run the long-term simulations for over 1 year before results were independent of the initial conditions.

The figures showing the detailed simulation results (Appendix E) illustrate the effect of initial conditions. Each study site had a different initial condition with respect to the dynamic equilibrium. The magnitude and duration of the effect as a function of distance away from the river also differed from site to site.

A 10-year simulation period was selected for water years 1970 through 1979. This was originally planned to be a 45-year period, but the run-time for a simulation period of this duration was prohibitive. The 10-year period was selected to include a range of wet to dry (precipitation) years, high to low water years (river stage), and years having pronounced differences between CWCP and Alternative Plan river stages. The years preceding the simulation period were drought years, and the following years had significant flooding. The simulation period was a transition period with the range of desired hydrologic characteristics. The 10-year period was assumed to be a representative sample of the 45-year period because of its variety of river stages and differences between plans.

The 10-year simulation period was preceded by a 4-year simulation to produce an initial condition that approached a dynamic equilibrium. The simulations actually started October 1, 1965, and the first 4 years were discarded to ensure that they were not influenced by the initial conditions estimated for the start-up date. The calibrated steady-state model was run to provide the initial conditions for the 14-year simulations. The Alternative Plan and CWCP simulations used the same initial conditions. Appendix E contains the time series of computed heads within cross-sections at each study site for the entire 14-year period. The decreasing trend in head at locations far from the river, in the L488/497 cross-sections, is shown in the figures in Appendix E. This is the same situation encountered in the 1-year test run. The transition period from steady-state equilibrium to dynamic equilibrium was about 4 years at the RM691 and L488/497 sites, and about 1.5 years at Tri-County LD 2. Computed heads from tThe end of the first 4 water years (1966 through 1969) of simulations were used for the initial conditions of the 10-year simulations for all three study sites.

The time steps used in the simulations were selected to reduce the computation time for the simulation while still providing adequate accuracy. Short, 2-day time steps were used during the growing season (April through September) to provide accuracy and detail during this important period. The time steps were longer during the rest of the year. Semi-monthly time steps were used during the winter months (December through February). In the months immediately preceding and following the growing season (October and March), a time step of one week was used to transition between short to long time steps. This time stepping scheme resulted in 108 time steps per year.

A time series of computed groundwater elevations (heads) at locations along cross-sections through the study site were plotted for each simulation. There were up to three cross-sections in each site. The cross-sections start at the Missouri River and run nearly perpendicular to the river across the adjacent plain. In most cases, the cross-sections follow a line defined by the USGS monitoring wells. The time series of heads along the cross-sections demonstrates the variation with distance from the river in the response of groundwater to river stage changes.

In order to provide MRR with input for their modified HEC-PBA model (used to estimate potential crop damages), Foster Wheeler Environmental post-processed the simulation data to calculate the percentage of the total model area that had heads within 2, 3, or 4 feet of the ground surface. The time series of these percentages were plotted for each river operation plan, and compared by subtracting the CWCP percentage from the Alternative Plan percentage for every time step. This post-processing is referred to as the "How Close" analysis. For the study sites with levees (L488/497 and Tri-County LD 2), How Close was run for the part of the model that represented the protected (interior) area and for the unprotected (exterior) area.

Foster Wheeler Environmental developed a QC plan at the beginning of the project and revised it as necessary throughout the project execution. The plan established a framework for model documentation, calibration tolerances, sensitivity analysis, and an independent peer review procedure. A Foster Wheeler Environmental independent peer reviewer conducted QC reviews at two points during this study. The reviewer evaluated the draft aquifer test reports of L550 and Tri-County LD 2 pump tests and the model documentation after model simulations were completed. MRR also participated in project QC. MRR held an in-progress meeting at the completion of field data collection and at the beginning of model calibration, and reviewed the draft aquifer test reports.

A standard set of tables and figures are presented for each model to describe the key model input, and results of model calibration, validation, sensitivity, and simulation. A base map showing the study area boundaries, surface water features, principal towns and lines of transportation, USGS monitoring wells, the exploratory boring, and the levees and outlet structures was produced for each study site. Input and output data were plotted on these maps. The input maps include data locations, site topography, model grid layout, recharge zones, and aquifer stratigraphic geometry and hydraulic properties. The output maps include steady-state head contours for the calibrated model and for sensitivity runs made for recharge and river stage.

Time series plots were made of input and output data. River stages were plotted with monitoring well observations and computed heads for the calibration and validation periods. For simulations, the computed heads at several points along the site cross-sections were plotted with the river stages for a given alternative. Full 10-year simulation period plots appear in the main report, and 14-water year plots appear in Appendix E for each site and each of the two water control plans.

The cross-section plots were reformatted and included in Section 5.2. Time series of computed heads for both the Alternative Plan and the CWCP were plotted together for selected locations on the cross-sections. The influence of the river on groundwater for the two water control plans was compared using these plots. The plots illustrate the difference between the plans and the different types of river influence on groundwater for the three study sites.

A spatial analysis of the number of days that the depth to groundwater (DTW) was equal to or less than 2 feet and 4 feet was conducted using a geographic information system (GIS). The simulated heads for the spring season of 1973 and 1976 were analyzed. 1973 had the greatest average spring Missouri River stage for the Alternative Plan of all years in the 10-year simulation. 1976 had the greatest average difference between Missouri River stages for the Alternative Plan and the CWCP of all years in the 10-year simulation. For each cell in the model, the number of days with DTWs of 2 and 4 feet or less was counted during the period starting April 1 and ending June 15 (76 days). The number of days with DTWs less than 2 feet was subtracted from the number of days with DTWs less than 4 feet to produce a day count that indicates persistence of high groundwater during the spring. The counts were categorized into three groups representing a low number of days (20 percent or less of the 76- day period), a medium number of days (20 to 80 percent), and a high number of days (80 percent or greater). The groups were assigned a unique shade and their locations were plotted for each study site. Plots were made for each of the two simulation periods and two water control plans. Also, two plots were made for 1973 and 1976 to show the difference between plans. Eight GIS plots were made for each study site. These plots signify the potential for non-recovery from high groundwater problems during the early growing season. Consistently high groundwater areas are identified in these plots for the two seasons in the 10-year simulation period that represent a high stage year. High groundwater areas are also identified for a more typical stage year, but one where the Alternative Plan stage is much larger than the CWCP stage.

The quality of the simulations is demonstrated by the accuracy of the model calibrations and by the analyses of simulations, which show that the hydrologic processes identified in the conceptual model are adequately represented in the MODFLOW models. The model output analyses include evaluations of groundwater level responses to river stage changes as functions of distance from the river and the magnitude and duration of peak river stages. One of the most important findings of the study is that the depth to groundwater in levee interior areas is not greatly affected by changing from the CWCP to the Alternative Plan. The analysis of areas with shallow depth to groundwater, using the simulation results, supports this finding.

The influence of the Missouri River on adjacent groundwater levels diminishes with distance from the river; however, the slope of the ground surface must be considered when evaluating the effect of the river on depth to groundwater. In study sites RM691 and L488/497, only the areas near the river typically had shallow depth to groundwater. In Tri-County LD 2, the shallowest depths to groundwater were typically found in the upland areas. Basically, there are three factors of importance: 1) the stage of the Missouri River, 2) the slope of the groundwater, and 3) the slope of the ground surface. When evaluating the impact of the Missouri River stage increases on depth to groundwater, two basic things must be considered: the magnitude (elevation) and the duration of the rise. The impact due to stage increase seems more obvious. When the stage is high, groundwater will be backed up in the aquifer to the elevation of the river water surface. One qualifier must be added to this case—the river must remain high long enough for the water to flow out of the river to fill the void spaces in the aquifer soils. The steady-state sensitivity analysis showed the case where the river stages would be permanently increased. This is an extreme and unrealistic situation, but it does demonstrate the limits of areas with shallow depth to groundwater. The time factor was quantified in the RM691 study site where, for a 1-day stage rise, only 90 percent of the river stage rise is seen at 550 feet from the river. This range of influence increases with longer duration stage increases, but at a decreasing rate (see Section 2.7).

The frequency of river stage changes and their effects on groundwater heads at varying distances for each study site were shown in Sections 2.6, 3.6, and 4.6. The time-series cross-section plots of the heads show the river and cells within a few thousand feet from the river vary directly with the river stage. The more rapid head fluctuations of the river stage become less apparent in the groundwater as distance from the river increases. At RM691, the maximum extent of the influence of the river appears to be at the groundwater divide that runs north-south near the City of Onawa, Iowa. For L488/497, the influence of the river appears to be negligible beyond 5,000 feet. In this case, the runoff from the bluffs near the entrance of creeks into the floodplain has more influence on groundwater heads than the river. In Tri-County LD 2 the floodplain is much more narrow than it is at the other study sites, so the groundwater heads in the entire levee unit area are influenced to a large extent by the river.

The Tri-County LD 2 site exemplifies the role of the slope of the ground surface. Tri-County LD 2 is very flat and narrow compared to the other sites. The morphology of the Missouri River floodplain is the main reason for this difference. The lower river is more incised than the upper river channel. Also, Tri-County LD 2 is the furthest downstream site, and the river valley gradient is less downstream than upstream. Given that the groundwater gradient is towards the river, but the land is flat, the groundwater has the potential to be closer to the ground surface farther away from the river.

The areas with a shallow depth to groundwater cannot be determined by examining only the time-series head cross-sections. These cross sections show the variation with distance from the river of the head response to stage changes. A companion analysis program developed for this study, "How Close," completes the analysis by comparing the elevation of the land surface with the computed groundwater surface at each cell of the model.

The How Close program was run for the protected (leveed) and unprotected (unleveed) areas within Levee Unit L488/497 and Tri-County LD 2. The RM691 site does not have a levee, so the entire area modeled is considered unprotected. The How Close program added up the areas of cells within groups defined by their depth to groundwater. The percent of the total modeled area is output for each group of cells. The modeled percentage for the RM691 site is smaller than the other two leveed sites because the floodplain is much wider at this site, and the effects of the river stage changes diminish as one goes further from the river. Nevertheless, the general pattern observed in the How Close plots is the same for all sites and for both simulation alternatives.

The How Close plots provide the most revealing information for evaluating differences between plans. First, the difference between plans is relatively small in terms of percent of total modeled area. Second, the seasonality of differences between plans is vividly shown in the difference plots. And third, the difference between the unprotected and the protected areas of Levee Units L488/497 and Tri-County LD 2 are significant. The difference in affected areas between plans in the unprotected area is approximately two times greater than the difference in the protected area.

The interior extent of Missouri River influence on groundwater was plotted using the GIS. The figures show that the influence of the high springtime river stages produced by the Alternative Plan extends across the site interior in three different ways. At RM691, the elevated spring stages travel like a wave slowly moving through the aquifer and attenuating with distance from the river. The remnants of this wave can be seen at up to 10,000 feet 1.5 months later. At L488/497, the spring wave does not travel across the interior as far as it does at RM691. The remnants of the wave can be seen at 3,570 feet 1.5 months later. At this site, the runoff from the bluffs keeps groundwater high and attenuates the wave more than at RM691 where bluffs are far from the river. At Tri-County LD 2, the river stage fluctuation is much greater and the floodplain width is much less than at the two upstream sites. At Tri-County LD 2, the groundwater across the entire levee unit rises and falls almost in unison with the river. The figures show that the model describes some interesting hydrologic processes associated with ground and surface water interactions along the Missouri River.

Given that the model results are reasonably representative of the pertinent hydrologic processes, the simulation results can now be used to evaluate potential differences in high groundwater impacts between the CWCP and the Alternative Plan. Tables 5.1 and 5.2 show a comparison between plan maximum areas from the How Close analysis. The maximum area associated with depth to groundwater of 2 feet or less and 4 feet or less and the corresponding percent of total site area are presented. The tables show very small differences between plan maximum affected areas for all three sites. The maximum concurrent difference between plans is about an order of magnitude less than the maximums for the L488/497 and Tri-County LD 2 sites. The maximums mainly occurred during 1973. This year had the highest average spring river stage of the 10 years simulated.

A GIS analysis of the areal distribution of average depth to groundwater was conducted. This analysis averages the How Close time series data for the spring season when the Alternative Plan stages would be typically higher than the CWCP stages. Contour plots of the average depth to groundwater during the spring of 1973 and 1976 for both plans are shown in Figures 5.4 through 5.15. The areas of average depth to water less than or equal to 1, 2, and 4 feet are summarized in Table 5.3. The years of 1973 and 1976 were selected to show the impact during a spring with high river stages (1973), and a spring with a lower stage, but with a large difference between plan river stages (1976). The range of depth to groundwater is from 0 feet to 15 feet.

Areas with various average depths to groundwater and number of days with depth to groundwater between 1, 2, and 4 feet were computed using the GIS. These areas quantify high groundwater impacts to farmed areas in the study areas. This analysis shows that the pertinent hydrologic processes are well represented by the model. Tables 5.1 and 5.2 show that the various interactions do not greatly change the maximum area impacted by the river; however, Table 5.3 shows that the interactions are important during the spring months. The average depth to groundwater during the spring varied significantly between sites, water control plans and water years.

The difference between water years 1973 (high stages and precipitation) and 1976 (lower stages and precipitation) is great at the two downstream sites, but not at RM691. The stages included in the comparison of the two water years were observed at the gauge at St. Joseph, Missouri. This gauge is close to the L488/497 site. L488/497 has a greater area with high groundwater in the spring of 1973 than in 1976. This difference between these years is more pronounced downstream at Tri-County LD 2; however, this difference is not significant upstream at the RM691 site. This variation between sites and water years indicates that the river stage influence on the depth to groundwater increases downstream.

The change in the influence of the water control plan on depth to groundwater in the downstream direction is the opposite of the influence of river stage described above. The difference between water control plans is greatest at RM691 and least at Tri-County LD 2. In the spring of 1976, when the average difference between plan river stages is the greatest of the 10-year simulation period, the area with high groundwater of the Alternative Plan is greater than that of the CWCP. This was true for all sites, and for all three depths to groundwater listed in Table 5.3. The critical point made by Table 5.3 is that the river stage has the greatest impact on groundwater downstream where the influence of the water control plan is the least.

Table 5.1. Maximum Area with DTW less than 2 feet, and Maximum Difference Between Plans

Table 5.2. Maximum Area with DTW less than 4 feet, and Maximum Difference Between Plans

Table 5.3. Area with Various Depths to Groundwater for Spring 1973 and Spring 1976