[Page 6] |

34. (gms) This exercise is based on the Bicknell Bottoms site that we studied in one of the homework exercises this semester.

Click here to download a zip archive containing a set of files associated with this project. Open the GMS project file named **bbottoms.gpr**. This project file represents a solution to the homework exercise, minus the background images (remove in order to make the zip archive smaller). The objective of this exercise is to parameterize the model, calibrate the model, and then perform a stochastic simulation and analysis.

**Click here to download a copy of the solution.**

I recommend doing all of your work on a local drive so that you have fast disk access.

**1. Parameterize the model**

First we will parameterize the model and perform a forward run.

a. Using the polygons in the conceptual model, locate the polygon zones in the recharge and hydraulic conductivity coverages according to the following maps:

Hydraulic Conductivity:

Recharge:

Enter a set of Key values for the polygons using the codes shown below. Then initialize the parameters in the parameters dialog and assign the parameter values shown in the table.

Type |
Zone |
Key Value |
Param Value |

HK | A | -100 | 2 |

HK | B | -200 | 0.5 |

RECH | A | -300 | 0.0001531 |

RECH | B | -400 | 0.0001837 |

RECH | C | -500 | 0.0004593 |

RECH | D | -600 | 0.00007655 |

b. Save your model as **bicknell_forward.gpr** as run in forward mode. Check to ensure that your results look correct.

**2. Calibrate the model**

Next we will import some observation data and calibrate the model.

a. Save your project as **bicknell_calib.gpr**.

b. Change to parameter estimation mode.

c. Edit your parameter data to solve for all six parameters and do a log transformation on all six parameters.

d. Download the following file to your local drive: obs wells.txt. This contains a set of head values from observation wells in the region. Create an observation wells coverage in the conceptual model and then import the file to the coverage. Leave the calibration target (interval) at the default value.

e. Turn on observed flow as an attribute for the River coverage. Select the river segment shown below and enter an observed flow value of **-75 m^3/d**. Enter an appropriate interval.

f. Save and run your model to calibrate the parameters.

g. After the solution is complete, import your optimal values to the starting values for the parameters.

h. Save your model one more time.

**3. Perform a stochastic analysis.**

Finally, we will perform a stochastic analysis.

a. Open the **bicknell_forward.gpr** project file you saved in part (1). Save a copy of this project as **bicknell_stoch.gpr**.

b. Turn on the stochastic modeling option.

c. Edit the parameters to randomize all six parameters. Do not log transform. Do a simple parameter randomization with 20 model instances. Use the following standard deviations:

Type |
Zone |
Key Value |
Std Dev |

HK | A | -100 | 2 |

HK | B | -200 | 1 |

RECH | A | -300 | 0.0002 |

RECH | B | -400 | 0.0002 |

RECH | C | -500 | 0.0002 |

RECH | D | -600 | 0.0002 |

d. Go into the **Output Control** settings and turn OFF the following output options: **HFF** and **List** files. This will make the file easier to upload.

e. Save and run your stochastic simulation.

f. Read in the solution and save your project again.

g. After reading in the solution, browse through each of the solutions and look at the flow budget for each. Open up a spreadsheet and copy the Drain discharge to the spreadsheet for each of the 20 simulations. You should be able to select and copy/paste using the clipboard.

h. Analyze the data in the spreadsheet and calculate the probability that the discharge to the drains is greater than **5000 m^3/day**. Be careful with the sign (discharge is negative). Show your work clearly.

Zip up all of your files (including the spreadsheet) from each of the three steps into a single zip archive.

**GRADING**

Your solution will be graded as follows

1. Forward run (3 pts)

2. Inverse run

a-c (2 pts)

d (3 pts)

e (1 pt)

f-h (3 pts)

3. Stochastic run

a-f (3 pts)

g-h (3 pts)

(Upload instructions and links went here)