# HW#9 - Model Calibration

*Note: You may*** **work in pairs on this assignment.

In this exercise we will be first read in a completed MODFLOW model. It is a modified version of the East Texas model (see below). We will then import a set of observation well data and a pair of observed flows. We will then modify the MODFLOW model in an attempt to improve the model calibration.

Before you begin, download the following zip archive:

start.zip

You will need to unzip the files after you download them. Then do the following:

1) Turn on observed flow as one of the properties for the **SourcesSinks** coverage.

2) Create a single arc group out of all of the specified head arcs along the river at the south end of the model. Assign an observed flow = -28,000 to the arc group. Use an interval corresponding to 5% of the observed flow (enter as a positive value).

3) Assign an observed flow to the specified head arc along the river on the right (east) side of the model. Enter an observed flow of -3400 and an interval corresponding to 5% of this value.

4) Download and open the following spreadsheet of observation well data. Create an observation coverage and import the observation wells to GMS. Associate the wells with the top layer. Be sure to go into the coverage setup dialog for the resulting observation well coverage and change the "3D grid layer ..." option to **By layer number**.

obs_wells.xls

5) Save and run the MODFLOW model. Read in the solution and view your results.

6) Adjust the K values in layers 1 & 2 and the recharge until the model is "calibrated". You will not get a perfect match, but you should be able to get total error norms quite low. Be sure to pay attention to both head and flow error. Before you start your calibration, create an Excel spreadsheet named **runlog.xls** and create a table with the following columns:

- Run number
- Run name
- K1 (k in layer 1)
- K2 (k in layer 2)
- Recharge (big polygon only - don't worry about landfill)
- Herr (RMS head error)
- Ferr (RMS flow error)
- TotErr (sum of squared weighted residuals)

For each model run you make, enter a new row in the table showing the parameter values and the model results.

Keep improving your model until you get the total error (sum of squared weighted residuals) **less than 75.0**.

You may not use PEST to calibrate the model. This is a manual calibration exercise.

### Submittal Instructions

Save the GMS project with the completed solution. Zip up all files associated with the project, **including the spreadsheet**.

Upload your zip archive via Learning Suite.