1. Ground Water Modeling Concepts
CE EN 547 – BRIGHAM YOUNG UNIVERSITY
Ground Water Modeling Concepts

2. Types of Models Predictive Interpretive Generic
Used to make predictions based on hypothetical future scenarios
Interpretive
Used to gain a better understanding of an aquifer
Generic
Based on a simple classical case, rather than on a real system. Used for academic purposes

3. Model Development Protocol
Define purpose of model
Conceptual model development
Code selection
Data collection
Model design
Calibration
Verification
Prediction
Postaudit
Adapted from Anderson & Woessner, 1992

4. 1. Define Purpose of Model
Types of models
Interpretive
Predictive
How much accuracy is needed?
How much $$ is available for study?
How will answers be used?

5. 2. Develop Conceptual Model
A high level description of the principal features of the system to be modeled
Includes aquifer units, boundary conditions, sources, and sinks
Parsimony - Keep the model simple enough to be manageable, yet complex enough to be useful

6. Characterize Aquifer Units
Try to determine principal hydrogeologic units
Hydrogeologic unit = zone that exhibits common hydraulic properties
May include multiple geologic units

7. 3. Code Selection
Does the code model all of the processes in the conceptual model?
Use the simplest code possible in light of conceptual model and the purpose of the modeling study
Has the code been thoroughly tested and verified?

8. 4. Data Collection Well/borehole logs River/lake stages Pumping data
Maps, aerial photos
Elevations
Observation well data

9. 5. Model Design Construct numerical model
Find reasonable set of parameters and initial conditions

10. 6. Calibration
Modify input parameters until model output matches field-observed values
Observations
Heads at wells
Flows at sources/sinks (rivers, lakes)
Can be automated in some cases with MF2K PES, PEST, UCODE
Should include sensitivity analysis

11. 7. Verification
Calibrate to multiple sets of observation data if possible
In each case, adjust boundary conditions and model stresses to be consistent with conditions present when observations were made

12. 8. Prediction Alter model, run simulation, and make predictions
Can include stochastic analyses

13. 9. Postaudit
When possible, review the model in later years to determine accuracy of predictions
Can be used to improve model

14. One of the most insidious and nefarious properties of scientific models is their tendency to take over, and sometimes supplant, reality.
— Erwin Chargaff
Quoted in J. J. Zuckerman, 'The Coming Renaissance of Descriptive Chemistry', Journal of Chemical Education, 1986, 63, 830.
…all models are approximations. Essentially, all models are wrong, but some are useful.
— George E.P. Box
In George E. P. Box and Norman R. Draper, Empirical Model-Building and Response Surfaces (2007), 414.

15. Case Study
Global Climate Change Modeling

16. Legitimate critique of science?
Classic case of misunderstanding science

17. System being modeled is highly variable. Need an ensemble
System being modeled is highly variable. Need an ensemble. Even then it is highly dependent on assumptions.

18. Skeptic plot. Are the models useless?
Skeptic graphic:

19. Data/facts are easily manipulated
Skeptic skeptic plot.
Data/facts are easily manipulated

20. The IPCC FAR 'Best' BAU projected rate of warming from 1990 to 2012 was 0.25°C per decade. However, that was based on a scenario with higher emissions than actually occurred. When accounting for actual GHG emissions, the IPCC average 'Best' model projection of 0.2°C per decade is within the uncertainty range of the observed rate of warming (0.15 ± 0.08°C) per decade since 1990, though a bit higher than the central estimate.

21. Do your own research!