Information regarding the predictive model runs made by the Applicant's model of the South Park ground water system, and their volumetric budgets, was subjected to supplementary analysis. Analysis of the time-dependent volumetric budgets corresponding to the 94-year NOCUP model simulation is presented in the set of Figures 6-38. The five components of this set present the month-by-month ground water flow rate predictions exactly as made by the model during each of the 94 years, one for each of the five model layers. For each model layer, these flow rate predictions have been dis-aggregated in these figures by: exchanges of water with the model layer lying above; exchanges of water with the model layer lying below; ground water in storage within the pore space of the layer; water pumped through wells; exchanges of water through the boundaries of the layer via the general-head boundary condition; ground water loss by vegetative consumptive use or evapotranspiration (ET); exchanges of water with streams represented in the layer; discharges of water from springs; and, prescribed ground water recharge to the layer. Values of recharge, i.e. water entering the model layer, are treated as positive and values of discharge, i.e. water leaving the model layer, are treated as negative. Thus, whereas ET values are always negative, exchanges of water with streams can be both positive and negative. Such values can be clearly observed to vary with month of the year, since they are affected by the growing season and surface water availability, respectively.

Analysis of the time-dependent volumetric budgets corresponding to the 94-year SPCUP model simulation is presented in the set of Figures 6-39. The layer distinction, format, scale and method of depiction adopted for these figures are identical to those presented above for the NOCUP simulation in Figures 6-38.

In each set of Figures 6-38 and 6-39, when considered for all five model layers together, the volumetric budget analysis identifies both the origins and destinations of water as accounted for in the model. The magnitude of each flow rate component in relationship to all other components maybe clearly observed in these figures. When all positive and negative components of the volumetric budget for each layer are exactly in balance for each month, the calculational technique adopted by the model is said to function correctly. The magnitudes of each budget component as predicted by the model are consequences of conscious choices and assumptions made by the Applicant's experts. In order to improve clarity of the volumetric budget features, the identical information is presented as annual totals, instead of monthly values, in Figures 6-40 and 6-41 respectively corresponding to NOCUP and SPCUP conditions. The inferences drawn from the latter sets of figures are similar to those drawn from the previous sets; accumulated annual total values allow them to be presented with greater clarity.

The fallacy of attempting a transient model calibration using annual average stresses can be observed in Figures 6-38 and 6-39. In these figures, ET shown as yellow shading appears only during about half the year. Recharge is similarly concentrated during a few months of the year. As a result water levels fluctuate throughout the year. In the water budget, this is reflected in changes in storage, shown in red. A so-called transient calibration that does not take seasonal fluctuations into account cannot properly account for non-linear stresses such as ET, stream-aquifer interactions and spring flows. The Applicant's model has not been calibrated to transient conditions within the South Park ground water system.

Supplementary analysis of the flows in stream segments, associated with the Applicant's uses of surface water from the North Branch Collection System as represented in the model's stream package was also conducted. Figures 6-42 present the flows as predicted by the ground water model within twelve segments that make up the North Branch Collection System, called the Aurora Ditch in the ground water model files, and the surface reservoir in the sequence from Figure 6-42a through 6-42l. The North Branch Collection System is present in both the NOCUP and SPCUP models. In the NOCUP simulation, flow rate in this system is specified to be zero. However, as can be observed in Figure 6-42f, the model predicts this system to serve as a drain to the aquifer. A representation such as this is inappropriate since that particular section of the North Branch Collection System does not currently exist, and has not existed historically. The reality embodied in the NOCUP model simulation run is thus flawed.

At a number of locations along the North Branch Collection System, the rate of stream flow is specified within the model. This specification sometimes reduces and at other times increases the flow rate in this collection system. In some instances, increases in flow rate can be attributed to actual operations such as diversions from streams. However, a corresponding reduction of flow in that stream is not represented in the model. Therefore, the surface flows as represented in the ground water model are not realistic and the ground water model is incapable of predicting stream flows and stream leakage. As a consequence, the ground water model is incapable of representing surface water availability for the project, recharge to the project or impacts as a result of the project.

Results predicted by the model under both NOCUP and SPCUP conditions were also subjected to supplementary analysis. The set of Figures 6-43 depict results predicted by the model under NOCUP conditions, in NOCUP year 47. This set, consisting of ten figures, presents a cross-sectional view of potentiometric head values predicted by the model in the context of the local model layering assigned by the Applicant. Each figure in the set corresponds to a given east-west cross section identified by a model row number. The identical format for depiction is used in each figure of the set, with vertical exaggeration which allows both layer geometry and predicted potentiometric head elevations to be identified. The set of Figures 6-44 depict results predicted by the model under SPCUP conditions, in SPCUP year 47. This set, also consisting of ten figures, presents a cross-sectional view of potentiometric head values predicted by the model in the context of the chosen local model layering. Each figure in the set also corresponds to the identical east-west cross sections as presented in Figures 6-43 and employs the identical format.

Several characteristics in behavior of this model can be observed in these figures. Figures 6-43a, 6-44a, 6-45c and 6-46c depict cross sections through the Michigan Hills area of South Park. Very steep gradients of predicted potentiometric heads in Layers 2, 3 and 4 can be observed at the edges of Michigan Hill where low conductivity zones have been prescribed. In Layer 5 where these low conductivity zones do not exist, the predicted water levels are similar to those in surrounding areas of the domain. The predicted water levels in Layers 4 and 5 become strongly coupled when Layer 4 is prescribed to be thin. Comparisons between, for example, Figures 6-43d and 6-44d illustrate this characteristic. The predicted heads at the last three grid nodes on the east side in Layer 4, differ markedly from the rest of Layer 4 but match very well with Layer 5. Similarly, predicted heads on the western edge of the model in Layer 4 track those predicted in Layer 5 rather than follow the trend in the remainder of Layer 4. This characteristic is primarily caused by Layer 4 being prescribed as relatively thin in these areas. At the western edge of the domain, Layer 4 in fact represents the same geologic formation as Layer 5 and is its near-surface manifestation. Thus, as a result of the Applicant's choices in defining model layers to contain different formations, it is not possible to interpret just what the predicted head in a model layer means, since its meaning changes from location to location.

The predicted water levels depicted in Figures 6-43g and 6-44g illustrates just how poorly the model represents reality. At the western edge of the domain, Layer 2 denotes the surficial layer, within which ground water levels are predicted to occur over one hundred feet above the local ground surface. Such an implausible prediction occurs throughout the model prediction time spans from the quasi-steady state calibration to the end of the predictive model. It does appears that the Applicant has either failed to detect or chose to ignore such extraordinary errors which occur about a mile from the proposed project location.

Numerous other examples exist in the model domain where predicted water levels in the surficial layer lie well above the local ground surface. Figures 6-43j and 6-44j depict an area to the east of the model domain where a mile-wide lake' which is several tens of feet deep is predicted by the model. This lake does not exist in reality, and thus demonstrates the absence of reality as represented by the Applicant's model.

Figure 6-47a depicts the location of two stream gages called TarComCo and UpPark. The TarComCo gage currently exists and is operated by the City of Aurora. The UpPark gage is proposed by the Applicant in the Proposed Decree. The Applicant has proposed to use these gages to quantify stream depletions caused by the proposed project. In order to do this reliably, the gages must lie outside the zone of influence of the project pumping. Figure 6-47a shows that at the UpPark gage, project impacts as predicted by the Applicant's model are still significant, so that this gage is not appropriate for determining project impacts. The impacts predicted by the Applicant's model at the TarComCo gage appear insignificant. However, Figure 6-47b shows that this result is pre-determined by choices embedded in the model framework. In Figure 6-47b, a cross section of the model is shown along Tarryall Creek. Water enters the western edge of the model through a general head boundary and flows in model layer 1 towards the east. The TarComCo gage appears in this section of model layer 1. Impacts to the TarComCo gage have been minimized by the representation of model layer 1 as rather steeply sloping with a constant source of water to the west and having an impermeable bottom not in connection with the rest of the South Park ground water system.