4.0 Analysis of the Applicant's Modelling Framework

The Applicant's model of the South Park ground water system is based upon the public-domain computer program known as MODFLOW. The Applicant has undertaken predictive modelling consisting of 48 individual runs of this MODFLOW program. These runs were organized into two series, called the NOCUP and SPCUP series, respectively. The NOCUP series was apparently intended to simulate conditions in the South Park ground water system in the absence of the Applicant's proposed project, while the SPCUP series was intended to simulate conditions including the Applicant's proposed project.

The NOCUP series consisted of two model-run sequences of twelve individual runs each, named NOCUP1A through NOCUP12A, and NOCUP1B through NOCUP12B. Similarly, the SPCUP series consisted of two model-run sequences of twelve runs each, name SPCUP1H through SPCUP12H and SPCUP1J through SPCUP12J. Within each series, each of the runs numbered 1 through 11 covers a simulation time span of 4 years, while the runs numbered 12 cover a 3-year time span. The NOCUP1A and SPCUP1H denote the start run in each of the respective series. The predicted potentiometric heads from the NOCUP1A run form the starting heads for the NOCUP2A run. Each run thus feeds the next in numerical order, so that runs NOCUP1A through NOCUP12A together encompass a 47-year simulation time period. Predicted potentiometric heads from the NOCUP12A run form the starting heads for NOCUP1B run, such that the A and B sequences together form a simulation time period covering 94 years. Similarly the SPCUP H series followed by the J series together encompass the corresponding 94-year simulation time period representing conditions meant to include the Applicant's proposed project.

Each model run of both series covers either a three-year or a four-year time span of water years. Within that time span, model stresses such as well pumping were supplied by the Applicant on intervals of calendar months. Within each month, the model was prescribed to take either 80, 160 or 240 time steps. The aquifer stresses within each month were prescribed as constant; however, the predicted head and head-dependent stresses vary throughout the month in keeping with the actual time steps taken.

The features of this model which are central to its framework were analyzed by Principia. Figure 4-1 depicts the domain and numerical grid system adopted for each of the five model layers, from top to bottom, of which it is composed. The information depicted here is the MODFLOW IBOUND array which establishes those grid cells within which the model is designed to predict heads. The MODFLOW re-wetting feature was invoked by the Applicant to attempt to re-wet cells which dried up, due to the prescriptions made by the Applicant, during the simulation run. However, it may be observed in Figures 4-1, that model grid cells in which it is possible for water to occur, are different both between the two series of simulation runs as well as the between runs in the same series.

Figures 4-2 presents the prescribed, i.e. assigned as individual values to each model grid cell, top elevation distribution for each model layer. Here, the top of Layer 1 is taken to be the ground surface specified in the ET package. The top of Layer 2 differ between the two model series in the area of the Applicant's proposed reservoir. However, the top of Layer 2 also differs in other locations of the domain, both between the two series of simulations as well as the between runs in the same series. Figures 4-3 present the prescribed bottom elevation distribution for each model layer. In the area of the Applicant's proposed reservoir, the bottom of Layer 1 differs between the NOCUP and SPCUP series.

Using these prescriptions, cross-sectional views of the model domain are presented in Figures 4-4. Figure 4-4a shows the cross sections interpreted by interpolating linearly between the values at the center of each grid cell. Figure 4-4b shows the same cross sections, but keeping the value for each cell constant across the width of the cell. In both these figures, the layers are active when showed as color filled. Where white space occurs between layers, an inactive area exists in the model between model layers.

The spatial distributions of model-layer thickness distributions for each model layer are presented respectively in Figures 4-5. Areas depicted in gray represent a thickness of zero to one foot in thickness, which predominantly has a thickness of zero, that is the top and bottom of the layer has been prescribed to be identical and therefore the layer has pinched out. Areas shown in brown have been prescribed a negative thickness, that is the bottom of the layer is prescribed to be above the top of the layer, clearly a physical impossibility.

Figures 4-6 display the differences between the bottom elevation prescribed for an overlying layer and the top elevation prescribed for the underlying layer. Areas shown in gray generally indicate that the top elevation of the underlying layer and the bottom elevation of the overlying layer are identical. Areas shown in brown indicate where the difference is negative, i.e. the top elevation of the underlying layer has been prescribed to be above the bottom of the overlying layer, again clearly a physical impossibility. In areas shown by different colors, there is a portion of space between the model layers that is outside the active model domain. In particular, a significant gap is visible between the bottom of Layer 2 and the top of Layer 3 on the eastern side of the domain.

Extracted information regarding material property values assigned to model layers and to grid cells within each model layer, was also analyzed. The prescribed, i.e. assigned as individual values to each model grid cell, distributions of specific yield values within each model layer are depicted in Figures 4-7. Likewise, the prescribed distributions of storage coefficient within the lower four model layers are depicted in Figures 4-8. The model has been specified with MODFLOW LAYCON 1 to describe Layer 1 and the MODFLOW LAYCON 3 to describe Layers 4 through 5. Therefore specific yield is always used for calculating storage in Layer 1. In Layers 4 through 5, storage coefficient is used whenever the predicted head lies above the prescribed top elevation of the layer, and specific yield is used when the predicted head drops below this top elevation.

The hydraulic conductivity values assigned by the Applicant within each of the five model layers are depicted in Figures 4-9 and the vertical conductance values at the four interfaces between these model layers are depicted in Figure 4-10. The prescribed vertical conductance values differ between the NOCUP and SPCUP series, as shown. Since the prescribed values of vertical conductance are non-zero, water can be transmitted vertically between model layers even when there has been a gap prescribed between the layers. Specifically as represented, ground water can be transmitted vertically between model layers 2 and 3 on the east side of the model where there has been a prescribed gap between these model layers of up to several thousand feet.

The prescribed distributions of general-head boundary conditions to the model were also extracted and analyzed. General-head boundary cells appear only in model layers 1 and 2 in the model. Values of boundary condition parameters that were assigned to model layer 1 are depicted in Figure 4-11a and those assigned to model layer 2 in Figure 4-11b. Color bands are used to represent the values conductance prescribed at grid cells. For the same head difference between the supplied reference head and the predicted head in the aquifer, the larger a value of prescribed conductance in a boundary grid cell is, the greater will be the predicted rate of water flow at that boundary. In the MODFLOW computer program, the boundary prescribed by active model cells is treated as impermeable unless ground water flow is postulated to occur using either constant-head cells or general-head boundaries. Constant head cells were not prescribed in the Applicant's model. Therefore the only means of ground water flow either into or out of this model's domain occurs in areas to the north in Layer 1 as shown in Figure 4-11a and in the areas to the east in Layer 2 as shown in Figure 4-11b.

The model representations of historic well pumping, including prescriptions of well locations and pumping rates, were analyzed. Figures 4-12 depict the grid cell locations where wells are prescribed to pump ground water. The background color of the cell represents the magnitude of pumping rate that has been prescribed. Well pumping has been prescribed to occur in all five model layers and the pumping rate has not been prescribed to vary with time. This sequence of figures identifies wells operated historically that are not associated with the Applicant's proposed project and were used in model runs referred to as the NOCUP series. The model representation of proposed SPCUP well pumping rates is presented in Figures 4-13. The pumping rates assigned to these proposed wells in addition to those presumed to pump ground water historically were used in the model runs referred to as the SPCUP series. The proposed well pumping was prescribed to occur from model layers 4 and 5.

The model representations of ground water recharge, including prescriptions of recharge rates to each relevant model grid cell and as a function of time, were analyzed. Recharge as represented by MODFLOW is the portion of surface recharge that actually reaches the local water table. In other instances such recharge is sometimes referred to as deep-percolation recharge. The findings from such analyses are presented here as four figures. Each figure represents cell-by-cell prescriptions of recharge rates during each of the twelve months of a given water year, i.e. twelve months starting on October 1 of a preceding calendar year and ending on September 31 of the current calendar year. These prescriptions are denoted by color-fill zones, whose magnitudes are defined in the legend to each figure in units of inches per month. The recharge is applied to the topmost active layer at each location. Figures 4-14 depict the recharge values for the first model year. These recharge values are intended to represent historical recharges for water year 1950. Figures 15 and 16 depict prescribed recharge values for model years 5 and 8, respectively. Recharge for model year 5 is intended to correspond to historical recharges estimated by the Applicant to occur during water year 1953. Similarly, model year 8 corresponds to water year 1957. These two water years were chosen for illustration since they respectively represent the driest and wettest years from 1950 through 1996 as determined by the total groundwater recharge for that water year. Finally, Figures 4-17 depict the prescribed recharge rates for model year 47 which is intended to correspond to historical recharges for water year 1996. The prescription of recharge rates for the model runs NOCUP1A, NOCUP1B, SPCUP1H and SPCUP1J were identical. Therefore, the prescribed recharge rate during model year 48 is identical to that prescribed for model year 1, model year 49 is the same as model year 2 and so on.

The temporal variations in recharge rates assigned to the model are depicted in Figures 4-18 and 4-19. Figure 4-18 present the total rate of ground water recharge for the model domain, both as monthly values in acre-feet per month and as annual values in acre-feet per year. This total recharge rate prescribed in the model is composed of individual grid cell contributions which vary significantly with spatial location. Such recharge variability is depicted for eight grid-cell locations chosen for illustrative purposes in Figures 4-19, using the identical format and units for each location, but using different scales to reflect differences in the magnitudes. It will be observed in these figures that the prescription of recharge rate with time is not spatially proportional. The peak values vary from no more than a fraction of an inch per month in some locations to as high as more than 55 inches per month in another.

Analysis of separate procedures used to calculate the recharge rates attributed to precipitation and prescribed in the Applicant's model grid cells was conducted. The findings from this analysis are presented in the group of graphical images prepared by Principia and included here as Figures 4-20 and 4-21. Recharge rates prescribed by the Applicant to the model was apparently intended to include two sources: recharge from precipitation and recharge from return flows of applied irrigation water. Figures 4-20 represent calculations used to determine the contribution of precipitation recharge. Figure 4-20a presents the orographic function used to calculate a baseline recharge value in inches per year as a function of elevation and the type of geologic material at the surface. Figure 4-20b depicts the ground surface elevation distribution used in this calculation. Color fills and contour lines are used to depict the ground surface elevation. Figures 4-20c and 4-20d respectively depict the percentages of the two types of geologic materials identified as Qa and Qb prescribed to occur in model grid cells to which recharge rates are assigned. Based on deposition testimony provided by Dr. Harvey Eastman, Qa is understood to be alluvial material, while Qb is understood to be weathered bedrock materials. Figure 4-20d depicts the method by which the percentage of Qa and Qb were combined by the Applicant. Values represented in Figure 4-20e are based upon the percent of Qa if the percent of Qa is greater than zero, otherwise the percent Qb is used instead. This combined percentage is then used to proportion the amount of precipitation recharge across the blue zone shown in Figure 4-20a. For a given elevation, a percentage of zero in Figure 4-20e will be represented at the bottom of the blue zone, while a percentage of 100 will be represented at the top of the blue zone. Combination of the percentages shown in Figure 4-20e and the elevations shown in Figure 4-20b with the formula represented in Figure 4-20a yields the recharge values shown in Figure 4-20f.

A geologic factor was invented and used by the Applicant to make further adjustments to prescribed recharge rates is depicted in Figure 4-20g. This factor was used to adjust recharge rate upwards in the Michigan Hills area and downwards in the area to the east where Precambrian granites are specified to occur. A spatial factor illustrated in Figure 4-20h was also invented and used additionally by the Applicant to adjust recharge rates such that they increase from the south-east to the north-west within the model domain. The prescribed recharge rate in inches per year shown in Figure 4-20f was thus multiplied by the factors shown in Figures 4-20g and 4-20h to obtain the spatial distribution of recharge shown in Figure 4-20i. Further adjustments to redistribute these precipitation recharge rates to represent temporal variability were also implemented by the Applicant. Figure 4-20j depicts the estimates of recharge rate due to precipitation made for the Applicant by Leonard Rice Consulting Water Engineers (LRCWE) as a blue line. These estimates were then adjusted, according to deposition testimony provided by Dr. Harvey Eastman, to occur at different times as represented by the red line. The consequence of implementing this temporal adjustment to the total recharge prescription for the 47-year simulation time period is illustrated in Figure 4-20k. The right hand vertical axis shows a temporal adjustment factor calculated for each month based on the annual average recharge for the 47-year time span. To calculate the precipitation recharge rate prescribed by the Applicant during a specific month to a specific model grid cell, this temporal factor was multiplied by the appropriate spatial value depicted in Figure 4-20i.

Figure 4-20l, illustrates the differences between assignments of ground surface elevation values to model grid cells in two different respects: one used in implementing recharge rate adjustments and another used to represent vegetative consumptive use. Differences in assigned cell by cell values can be observed throughout the model domain.

In a manner similar to that used for precipitation recharge rates, analysis of separate procedures used by the Applicant to calculate the recharge rates attributed to irrigation return flows and supplied to model grid cells, was also conducted. The findings from this analysis are presented in Figures 4-21. Figure 4-21a depicts those model grid cells in which a percent value has been prescribed to represent irrigated land. Different colors depict the varying percentages of each grid cell prescribed by the Applicant as covering irrigated lands. Areas shown in gray have been designated as having no irrigated lands. Estimates the areas and irrigation return flows were made for a discrete set of parcels by LRCWE. Figure 4-21b shows the area of each parcel obtained by adding the cell-by-cell area as blue bars, while the total areas are shown as red bars. For irrigated parcel 48, the sum of the grid cell areas is 27.9 acres, while the actual total irrigated area is only 4.0 acres. Figures 4-21c through 4-21w depict the return flow amounts estimated by LRCWE on a monthly and annual basis. The recharge rate from irrigation return flows was calculated by the Applicant through proportioning the return flows according to the number of acres irrigated in each cell. This area for each grid cell is calculated by multiplying the number of acres per cell, i.e. 22.9568, by the percentage of the cell irrigated from this parcel. The number of acres per cell is then divided by the total acreage to derive the proportioning factor. These factors for all the grid cells in a parcel should sum exactly to one if the parcel lies entirely within the model domain, or a value less than one if a portion of the parcel falls outside the model domain. For irrigated parcel 48, the proportioning factors sum to 6.975, resulting in almost seven times the return flows from that parcel than was estimated by LRCWE to actually occur. The total recharge rate supplied by the Applicant to the model is the result of adding precipitation recharge estimates to irrigation return flow estimates for each appropriate grid cell and month during the simulation time period.

The model representations of stream flows and stream routing were analyzed. Such analysis included the spatial locations of stream cells, the values assigned to variables in the stream package of the ground water model, the resulting routing of stream flows and the values assigned to in stream flow rates at model boundaries. Components of the Applicant's surface water model and estimates of available stream flows were also evaluated in the same context. Figures 22 shows how streams are represented in both the NOCUP and SPCUP series of runs. Model grid cells are shown as two halves, the top half representing stream cells that are prescribed to be in connection with model layer 1, and the bottom half representing stream cells similarly in connection with layer 2. This depiction is chosen for display purposes only. In the model itself, the cell center values are in fact used.

Figures 4-22a and 4-22b show stream cells throughout the domain for the NOCUP and SPCUP series of runs, respectively, while Figures 4-22c and 4-22d shows the same information as Figures 4-22a and 4-22b but over a smaller area. In the MODFLOW computer program, stream flows are calculated in terms of segments. The start of each segment is designated in Figures 4-22 by a gray background color. A segment may consist of one or more reaches, each reach corresponding to one model grid cell. In the Applicant's model, the stream stage was prescribed in each such reach. Within a segment, surface water flows in the stream from one cell to the next are prescribed by reach number. Such reaches are generally colored in blue in Figures 4-22. However, in some grid cells, the stream stage increases along successive grid cells in a segment, forcing the flows in the stream to be directed uphill. Such grid cells are identified with a red background in Figures 4-22. It may also be observed that the North Branch Collection System which has been proposed by the Applicant to be constructed as part of the proposed project is simulated as actually present in both the NOCUP and the SPCUP series of model simulation runs.

At the head of each model stream segment, MODFLOW does allow users to designate the segment as either receiving stream flows from tributary segments or a diversion from another stream segment or be specified with a prescribed headwater flow. Figures 4-23 depict the headwater flows for some of the stream segments represented in the Applicant's model. It is noteworthy that for both the NOCUP simulation depicted in Figure 4-23a and the SPCUP simulation depicted in Figure 4-23b, the specified headwater flow in Jefferson Creek, Michigan Creek and Tarryall Creek, among others, is prescribed to be precisely constant throughout the simulation period. In contrast, the specified headwater flow for Snyder Creek, Wilson Gulch and Guernsey, among others, are prescribed to vary in a neatly periodic fashion. The North Branch of Park Gulch and Park Gulch, among others, are prescribed to vary more randomly.

For purposes of the SPCUP simulation, stream flow in the North Branch Collection System was specified by the Applicant in at least two different places. The first place is near the location where Tarryall Creek crosses the highway US285. At this location, designated as Segment 81 or Harland/SPCUP Ditch the peak flow has been prescribed to be less than 3 cubic feet per second (cfs). The second place is near the north-west corner of the proposed reservoir and is designated as Segment 122 or Phase 1, Ditch 2. At this location however, the peak flow has been prescribed to exceed 50 cfs.

Stream routing between stream segments as prescribed by the Applicant is illustrated in Figures 4-23 and 4-24. In these figures, blue lines are used to denote stream segments. Where the flow at the top of a segment is specified, a blue dot is shown. Arrows are used to depict the end of segments. Where the flow from a segment enters another segment, the arrow will demonstrate a connection to the head of another stream segment.

For purposes of both the NOCUP and SPCUP model simulations, surface flow from the North Branch Collection System was prescribed to be routed into Tarryall Creek near the location where Tarryall Creek crosses the highway US285. A completely new headwater flow was then created by the Applicant to represent flow nearby in the Harlan Ditch. Similarly, a completely new headwater flow was created near the north-west corner of the reservoir proposed by the Applicant. These prescribed headwater flows introduce new water into the system at those locations in addition to those routed in named creeks.

For purposes of the NOCUP simulations, stream flow down Tarryall Creek at its confluence with Park Gulch, has been prescribed by the Applicant as removed from the system. Flow down Packer Gulch combines with the flow in Park Gulch to form the flow down Tarryall Creek as it flows east from that location. Flows from Packer Gulch are then added to both Tarryall Creek at its confluence with Packer Gulch and to Tarryall Creek at its confluence with Park Gulch.

The model representations of stream-aquifer interactions were also analyzed. The findings from this analysis are presented here as a set of graphical images prepared by Principia and included here as Figures 4-25. Figure 4-25a depicts the values of stream stage, assigned by the Applicant to model grid cells, as compared with the local ground surface elevation also as assigned. Colors within grid cells are used to show the elevation of the assigned stage relative to the model ground surface, extracted from the model program's evapotranspiration package. The model uses this assigned stage to calculate the rate of flow from the stream to the aquifer. The assigned stage for each cell does not vary with time in the model, or between runs. Figure 4-25b similarly depicts the values of assigned stream bed top elevations compared with the assigned ground surface elevation. This value is not used in the simulations since a stream stage was prescribed for every stream cell. Finally, Figure 4-25c depicts values of assigned stream bed bottom elevations compared with the assigned ground surface elevation. When the aquifer water levels are predicted to fall below this prescribed value, leakage from the stream is predicted not to increase further but remains at the same rate that would occur if the predicted aquifer head remained at this elevation.

Figure 4-25d illustrates the difference between the assigned stream stage and the assigned stream bed top elevations. This represents the depth of water in the stream. Figure 4-25e shows the difference between the assigned stream stage and the stream bed bottom elevations. This value represents the maximum driving head for flow from the stream to the aquifer. Figure 4-25f depicts the difference between the prescribed stream bed top and stream bed bottom elevations. This represents the prescribed thickness of the stream bed. Finally, figure 4-25g shows the prescribed stream-bed conductance values. These values represent the leakage from the stream to the aquifer for a unit driving head. It is a function of the stream length, width, stream bed conductance and stream bed thickness. A larger value of stream bed conductance will result in more water being predicted to leak into the aquifer for the same stream stage-aquifer head difference.

The model representations of ground water evapotranspiration (ET), i.e. consumptive use by vegetation from ground water only, was also analyzed. The findings from this analysis are presented here using an analogous format to the recharge rates shown earlier. Figures 4-26, 4-27, 4-28 and 4-29 depict the maximum ET rate by month for water years 1950, 1954, 1957 and 1996, respectively. The years 1950 and 1996 represent the first and last years in the simulation time period selected by the Applicant, respectively, while the years 1954 and 1957 represent dry and wet years, respectively. These figures depict the maximum ET rate prescribed by the Applicant in inches during each month and for each appropriate grid cell. This rate would be attained if the predicted water level within the cell is at or above the prescribed ground surface. The rate is represented in the figure using different colors to designate variations. The specific rates depicted are for the SPCUP series of runs. The NOCUP series of runs differ from the SPCUP series in the area of the proposed reservoir.

The temporal variations in vegetative ET rates specified to the model are depicted in Figures 4-30. Such variability is depicted for eight selected grid-cell locations in Figure 4-30a, using the identical format and units for each location, but with different scales to reflect differences in the magnitudes. The group of eight figures which follow, from Figure 4-30b through 4-30i, present the individual ET rate graphs for the eight selected grid-cell locations. Both monthly values in inches per month and annual values in feet per year are included in each figure of this group. The seasonal fluctuation in calculated ET may be clearly observed, with no ET occurring during the winter months and the maximum ET rate being utilized during late spring or early summer months. ET represented in the Applicant's model was extracted from a designated model layer. Figures 4-31 represents the array of values used to select the model layer from which consumptive use by vegetation is taken. The values differ from run to run within a series and between model run series. It is noteworthy that ET is prescribed to occur from all five model layers in different areas of the model domain, but predominantly from model layers 1 and 2.

Other information relevant to the representation of vegetative consumptive use in the model are presented in Figures 4-32. Figure 4-32a presents the prescribed cell-by-cell values of percent cover of vegetative growth. Percent cover has been defined by the Applicant as that fraction of a model grid cell which is presumed to be covered by vegetation and from which ground water consumptive use occurs. No attempt was made by the Applicant to distinguish different vegetative types and the specific locations where each type occurs. The maximum ET rate was adjusted by two spatial functions. For model cells with known ET, i.e. grid cells where the percent cover shown in Figure 4-32a is greater than zero, the rate was adjusted by function represented in Figure 4-32b. This function increases the ET rate from the north-west to the south east across the model domain. For model cells with no known ET, i.e. grid cells where the percent cover shown in Figure 4-32a is zero, the function shown in Figure 4-32c is still used. The values depicted in Figures 4-32a, 4-32b and 4-32c were combined by multiplying the percent cover of Figure 4-32a with the function shown in Figure 4-32b and by replacing the zero values depicted in Figure 4-32a with the values in depicted Figure 4-32c. The result is a unit spatial function for ET as shown in Figure 4-32d.

Monthly consumptive use rates for vegetation estimated by LRCWE are shown in Figure 4-32e. Multiplying these consumptive use values by the spatial function in Figure 4-32d results in a maximum ET rate for each appropriate grid cell during each month. However, the procedure used in the Applicant's model was a multiplication of the spatial function for ET in depicted in Figure 4-32d by an annual average ET rate of 2.1915 feet per year followed by a further multiplication by the monthly consumptive use value normalized to the average annual rate of 2.03079 feet per year. This practice increases the resulting ET rate by 7.9% over that estimated by LRCWE.

Some of the springs within the area of interest were represented in the ground water model by using the so-called drain package of MODFLOW. This is a component of the computer program embodying the model and allows users to account for discharges of ground water from appropriate locations within a simulated ground water system. It does not however accounting for the fate of the ground water so discharged. The parameters used in such a representation are: drain elevation and drain conductance associated with a grid cell within which a spring is deemed to be contained. Figure 4-33 presents the prescribed grid-cell locations of such springs and the values of drain conductances assigned to the cells. Different colors are used to represent the drain conductance values. A higher conductance value will result in a higher flow rate for the same driving head.


Index | Introduction | Information Reviewed | Spatial Data | Framework | Framework Supplement | Calibration | Calibration Supplement | Predictions | Predictions Supplement | Surface Water | Findings | Findings Supplement | Glossary | Figures
Created 10-Mar-2000 by webmaster@prinmath.com
Copyright © 2000, Principia Mathematica, Inc. - All Rights Reserved