The scientific findings from Principia's analysis of the Applicant's modelling of the South Park ground water system, conducted through computational techniques which represent up-to-date practices in ground water modelling, are listed below. They are grouped together by subject matter, simply for convenience, and presented as numbered paragraphs with clarifying sub-paragraphs and additional figures, wherever appropriate. The sequence in which they are so presented does not bear any special significance.

1. The conceptual framework chosen for the model remains based upon unsubstantiated presumptions regarding the South Park ground water system which have predetermined the modelling outcome. For instance:
   1. the Reinecker Ridge formation is presumed to be completely impermeable and included as the impermeable bottom surface of the model;
   2. the Laramie-Fox Hills formation has been pre-judged to have no influence upon the chosen domain of the model and has then been excluded from this domain, thereby preventing any evaluation of the impact of the proposed pumping upon it; and,
   3. the granitic formations lying to the east of the Elkhorn Thrust Fault zone are assigned the implausible property of transmitting ground water vertically across thousands of feet but none horizontally, thereby violating known physical and hydrogeological principles.
2. The conceptual framework chosen for modelling relies upon unsubstantiated presumptions regarding ground water recharge of the lower water-bearing units of the South Park ground water system. This choice misrepresents the manner, timing and rate of recharge that can actually occur from shallow units to the deeper water-bearing units in the ground water system, in order to create artesian pressure conditions. The consequence of this choice is to artificially increase the rate of downward recharge to this portion of the system, when pressure within the deeper units is reduced by the simulated ground water pumping.
3. The numerical grid cell arrangement chosen for modelling is unsatisfactory for predictive purposes to which the model was applied. For instance:
   1. the model domain and placement of the uniformly-sized square grid cell arrangement is described only in approximate global coordinates which do not allow these cells to be placed in precise juxtaposition with actual features on the ground, such as streams, existing wells, springs, etc.;
   2. the assigned number of active grid cells in the domain for model layers, differs from one model run to another, resulting in unintended modifications to the size and shape of these domains during the model runs made to obtain predictions; and,
   3. the size and shape of chosen uniform square grid cells are too coarse to allow proper representations of stream-aquifer interactions, variations in ground surface elevation, pockets of vegetation from which water loss by evapotranspiration occurs, springs, seeps and ponds, etc.
4. No grid-dependency tests appear to have been conducted to evaluate the sensitivity of results predicted by the model to systematic refinements in grid-cell sizes. In the absence of such tests, it is not possible to demonstrate that the results predicted with the model, which have been quantitatively relied upon, are indeed independent of the chosen size and, in that context alone, sufficiently accurate.
5. The discretization of time periods chosen for modelling is unsatisfactory for predictive purposes to which the model was applied. For instance:
   1. each model run representing either NOCUP or SPCUP conditions was sub-divided into 4-year time spans with such poor output control that the vast, and unnecessary, volumes of generated tabular output have not received careful scrutiny, either by the Applicant or by the model's purported reviewers, as they should have; and,
   2. the chosen stress period of one month was too coarse to allow recognized and rapid variations in stream flows to be properly accounted for.
6. No dependency tests appear to have been conducted to evaluate the sensitivity of results predicted by the model to systematic refinements in sizes of time steps and stress-periods. In the absence of such tests, it is not possible to demonstrate that results predicted with the model, which have been quantitatively relied upon, indeed are independent of the chosen temporal discretization and, in that context alone, sufficiently accurate.
7. The selection of model layers to represent characteristics of the South Park ground water system displays incorrect choices made for purposes of representing geological formations and aquifers in the model in order to predict the technical feasibility and consequences of implementing the proposed Application. For instance:
   1. the chosen model layers, even as composite representations of different geological formations, do not properly represent geological features of the South Park ground water system in quantitative respects;
   2. the assigned spatial distributions of the top and bottom elevations of model layers differ from one model run to the next, which is clearly implausible;
   3. the assigned elevation of the bottom of one model layer lies below the assigned elevation of the top of the underlying layer, which is clearly impossible; and,
   4. the chosen model layering arrangement and prescribed geometric properties cause inexplicable gaps to occur between model layers, which is clearly implausible.
8. The selection and spatial assignment of aquifer material property values to numerical grid cells of the five model layers, in order to predict the technical feasibility and consequences of implementing the proposed Application, are incorrect. For instance:
   1. values of specific yield and storage coefficient have been incorrectly assigned to grid cells within model layers prescribed to be under confining pressure;
   2. the spatial distributions of specific yield and storage coefficient values assigned to model grid cells have not been substantiated as valid for the South Park ground water system;
   3. the spatial distributions of hydraulic conductivity values assigned to model grid cells have not been substantiated as valid for the South Park ground water system; and,
   4. the spatial distributions of vertical hydraulic conductance values assigned to model grid cells have not been substantiated as valid for the South Park ground water system.
9. The selection and assignment of material property values to numerical grid cells representing the Elkhorn Thrust Fault zone are incorrect. For instance:
   1. extrapolating the results of a single test made at one specific location to represent the properties of the fault zone wherever it is represented in the model grid cells has not been established as valid for the South Park ground water system; and,
   2. the location and inclination of this thrust fault zone as depicted in the Applicant's own data exhibit differs significantly from the specific model location where it is represented through the model layers and grid cells, thereby mis-characterizing the response of the fault zone to ground water pumping stresses.
10. The spatial assignments of aquifer material property values to numerical grid cells of the five model layers based upon the Applicant's model-calibration adjustments have subsequent to calibration, and entirely inappropriately, been modified for use in making predictions with the model. The consequence of this particular impropriety is to invalidate both the calibration process and the predicted results.
11. The choices and assignments of general-head boundary condition values to boundary grid cells in model layers 1 and 2 have not been established as valid for the appropriate portion of the South Park ground water system. The effect of these choices is an unverifiable pre-determination in rates of ground water inflows and outflows, especially when the model is used to simulate the proposed pumping.
12. The assignment of constant pumping rates to domestic and irrigation wells in order to represent their historical ground water withdrawals has not been established as valid for the South Park ground water system. The effect of such an assignment is mis-characterization of the stresses placed upon the ground water system by historical well pumping.
13. The spatial assignment of recharge rates to model grid cells, and their time-dependent variations, has not been established as valid for the South Park ground water system. For instance:
    1. the values assigned to certain model grid cells located in the vicinity of streams are implausibly high, and the basis of assignments is characterized by several unverified assumptions and errors; and,
    2. the basis of temporal changes to recharge rates assigned to model grid cells has not been recorded but is itself based upon a number of unverified assumptions and contains a number of calculational errors.
    The consequence of such recharge assignments is to mis-characterize the manner in which ground water recharges and discharges may actually occur in the South Park ground water system.
14. The representation of stream flows in the ground water model is seriously incorrect, and has not been established as valid for the South Park ground water system. For instance:
    1. water availability has been artificially created along surface water structures by the assignment of flow quantities at arbitrarily-selected locations, even in the model's representations of historical conditions;
    2. prescribed stream flow rates at model inflow boundaries depart significantly from either gauged flow records or as represented in the Applicant's own surface water model; and,
    3. the routing of flows within several significant streams within the model domain is characterized by significant calculational errors.
    The consequence of such a stream flow representation is to mis-characterize the manner in which streams may actually behave in the South Park ground water system. Thus, the model is incapable of properly accounting for the availability of water flows in these streams and its consequences to the ground water system in South Park.
15. The representation of stream-aquifer interactions, involving calculations of gaining stream reaches by aquifer discharge and losing stream reaches to aquifer recharge, is also seriously incorrect and has not been established as valid for the South Park ground water system. For instance:
    1. the assignment of temporally invariant values of stream stage to the stream nodes of all streams misrepresents the actual stage that would occur commensurate with the stream discharge at those locations;
    2. the prescription of constant stream stage values, coupled with errors in assigning routing parameter values, forces flow in certain stream segments to flow uphill;
    3. the prescription of stream conductance values, involving stream width, subtended length, streambed thickness and hydraulic conductivity has not been established as valid for the South Park ground water system;
    4. the assignments of elevation values to the bottom of streambeds have not been established as valid for the South Park ground water system;
    5. the summation of gain-loss estimations calculated by the ground water model and fed to the surface water model is incorrect and does not represent reliable quantities derived by a proper method of iterative calculations between the two models; and,
    6. the values of stream flow rates predicted by the ground water model do not resemble those measured at available gauges or as represented in the surface water model.
16. The representation in the model of consumptive use by vegetation and the calculations of ground water ET by the model are incorrect. For instance:
    1. the prescribed grid cell locations of vegetation presumed to consume ground water includes features that have not remained static over decades of water-rights transfers and excludes others, such as wetland areas and fens, that have been identified to exist;
    2. the prescribed values of percent cover of vegetation assumed to consume ground water within a number of model grid cells, are incorrect and have not been established as valid for the South Park ground water system;
    3. the prescription of a constant 6-foot extinction depth to all grid cells, and irrespective of vegetation type, in which vegetative consumption of ground water is presumed to occur, is incorrect and misrepresents both shallow-rooted and deep-rooted vegetation;
    4. the presumption in the model that all vegetative consumptive use occurs only from the water table, ignoring the known opportunistic capture either of precipitation or of vadose zone soil-moisture by vegetation, is incorrect;
    5. the prescription of vegetative consumption of ground water by evapotranspiration to grid cells within model layers is incorrect over large areas of the model domain; and,
    6. the basis for temporal variations in prescribed vegetative consumption of ground water has not been established as valid for the South Park ground water system as derived from either basin-specific knowledge or measurements.
    The consequence of such incorrect representations of vegetative consumptive use is mis-characterization of the quantity of ground water consumed by vegetation. The model is consequently incapable of predicting ground water available for salvage by the proposed pumping, and its timing, from the South Park ground water system. It is thus incapable of establishing the feasibility of the proposed Application.
17. The representation of springs and seeps in the model is incorrect and has not been established as valid for the South Park ground water system. For instance:
    1. the use of the model program's drain package prevents the calculated drain flows from being properly accounted for and routed such that ground water is calculated as disappearing from the ground water system but thereafter is not accounted for as either added to streams or diversion ditches;
    2. the assignment of drain conductance values to spring locations in model grid cells has not been established as valid for the South Park ground water system; and,
    3. the assignment of drain elevation values to spring locations in model grid cells is not based upon measurements or estimations based any upon local tests and hence has not been established as valid for the South Park ground water system.
18. The procedure adopted to calibrate the ground water model remains seriously flawed in significant respects. For instance:
    1. the selection of model calibration targets was incorrect with respect to steady or quasi-steady conditions in the South Park ground water system;
    2. it is improper to implement adjustments to values of physical parameters, such as percent cover of vegetative growth which are measurable or saturated thickness of geological formations, in order to seek model calibration;
    3. it is also improper to implement adjustments to model framework parameters during a quasi-steady state calibration and subsequently to modify the values of these same parameters when making predictions with the model; and,
    4. it is improper to implement parameter-adjustment steps that are involved in model calibrations without first undertaking systematic sensitivity analysis of the model parameters involved.
19. The Applicant's efforts to calibrate the model have been demonstrably unsuccessful. For instance:
    1. comparisons between model predictions and one set of calibration targets, i.e. the estimated water levels or potentiometric head elevations at existing wells, indicate neither generally similar nor locally accurate matches leading to the conclusion that the model is not calibrated to conditions within the South Park ground water system over any time period; and,
    2. an appreciable number of identified calibration targets including the ground water fluxes at stream and spring locations, have been ignored while evaluating the success or failure of the chosen calibration process.
    The effect of this lack of model calibration success is that all quantitative support for the Application that is derived from predictions made with this model are entirely unreliable.
20. No proper sensitivity analysis has been conducted on the ground water model used to obtain predictions and to support the application. For instance:
    1. the only sensitivity tests to be conducted were performed on the so-called quasi-steady form of model runs, at the start of the model development procedure when numerous changes to its framework and to prescribed parameter values were being implemented;
    2. no sensitivity tests were performed on the time-dependent model used to make predictions of both historical behavior of the South Park ground water system and the consequences of implementing the proposed project; and,
    3. variations in model predictions to systematic variations in prescribed values of parameters that are either unknown or not measured at all, have not been evaluated.
21. After the incorrect model calibration procedure was implemented with respect to one set of interpreted conditions, the model has not been verified or validated at all with respect to other conditions in South Park ground water system. Thus, the model is incapable of predicting the impacts caused by the proposed Application with any degree of reliability.
22. Due to several ad hoc computational devices embedded in the it, the Applicant's model is incapable of predicting stream depletions reliably.
23. The errors in the Applicant's modelling which have been identified by Principia thus far are too numerous to be individually listed in this summary of findings. However, each type of such errors has been identified in the large number of graphical images created by Principia, based upon analyzing the model, and included in this report.
24. It is extraordinary and unprecedented for a mathematical model introduced in an Application of such gravity, and especially after it has been claimed to be peer reviewed, to retain errors of such significance.

Several other predictive features of the model are worth noting here. Figure 8-1 shows a north-south cross section of the model domain through column 23. The background fill colors designate model layers 1 to 5. The predicted potentiometric heads for layers 1 to 5 are indicated by lines and symbols. Figure 8-1a depicts the predicted heads at model year 47 for the NOCUP simulation. Figure 8-1b depicts the predicted heads at model year 47 for the SPCUP simulation. The Applicant's proposed project pumping was prescribed at rows 56, 60 and 63 along column 23, as well as other locations not along column 23.

Along rows 77, 78 and 88, among other rows, the predicted head in model layers 2, 3 and 4 are almost identical in the Applicant's NOCUP simulation run. However, in the SPCUP simulation run the predicted head in layers 3 and 4 remain almost equal, but the predicted head in layer 2 is artificially forced to remain just slightly above the bottom of layer 2. The predicted head in layer 2 is such that a small saturated thickness is maintained within it, rather than the cell being allowed to dry out as would occur in reality. The consequence of this forcing on the predicted head in layer 2 at cell (88,23) can be clearly observed in Figure 6-22k. It is noteworthy that at this location in the model domain, layer 2 does not represent any known physical feature. Rather, model layers 2 through 5, in combination, are represented as making up the Lower South Park Aquifer. However, the predicted head in model layer 2 is completely determined by the arbitrary choice of the bottom elevation for layer 2.

Figure 8-2 depicts the saturated thickness predicted in model layer 2 during model year 47 for both the NOCUP and SPCUP simulations. Grid cells shown in red are those forced to attain a saturated thickness of less than one foot. In such areas, it is fair to say that the predicted head is entirely fictitious.

Figure 8-3 shows a north-south cross section of the model along model column 25. The background fill colors designate model layers 1 to 5. The predicted head in Layers 2 for two simulations are shown with both lines and symbols. The red line shows the predicted head produced by the Applicant's SPCUP simulation run at the end of model year 47. The dark blue line shows the predicted head produced by a new simulation undertaken by Principia with the identical Applicant's model. In this run, an additional nine acre-feet per year of pumping was postulated as occurring approximately 1.5 miles south of the southern boundary of the Sportsmen's Ranch. The position of the so-called hydrologic divide that was predicted to occur the south of the Applicant's project for both these simulations are indicated using vertical lines. It is noteworthy that the hydrologic divide predicted by the model indeed shifts about 1,000 feet north between the two simulations.