14.07.2013 Views

User's guide of Proceessing Modflow 5.0

User's guide of Proceessing Modflow 5.0

User's guide of Proceessing Modflow 5.0

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Processing <strong>Modflow</strong> 265<br />

Modeling Approach and Simulation Results<br />

The aquifer is simulated using a grid <strong>of</strong> one layer, 15 columns and 15 rows. A regular grid space<br />

<strong>of</strong> 500 ft is used for each column and row. The layer type is 0:confined and the Transmissivity<br />

flag in the Layer Options dialog box is user-specified. Transmissivity and recharge are defined<br />

as estimated parameters with the parameter numbers 1 and 2 (Note that the names <strong>of</strong> these two<br />

parameters are p1 and p2).<br />

Table 6.4 shows the optimized parameter values and the correlation coefficient matrix<br />

calculated by PEST. A similar result obtained by UCODE is shown in Table 6.5. The diagonal<br />

elements <strong>of</strong> the correlation coefficient matrix are always unity. The <strong>of</strong>f-diagonal elements are<br />

always between 1 and -1. The closer an <strong>of</strong>f-diagonal element is to 1 or -1, the more highly<br />

correlated are the parameters corresponding to the row and column numbers <strong>of</strong> that element. For<br />

this example, transmissivity (parameter p1) and recharge (parameter p2) are highly correlated,<br />

as is indicated by the value 0.9572 <strong>of</strong> the correlation coefficient matrix. This means that these<br />

paramters are determined with a high degree <strong>of</strong> uncertainty in the parameter estimation process.<br />

A sensitivity analysis could be used to quantify the uncertainty in the calibrated model caused<br />

by uncertainty in the estimates <strong>of</strong> the aquifer parameters.<br />

For our example, the only discharge is to the river and the only source is recharge. To be in<br />

steady state, these two must balance. Recharge must therefore be equal to 1.125 cfs (the river<br />

gain equals 11.125 cfs - 10 cfs). Spreading over the modeled area:<br />

1.125ft<br />

recharge'<br />

3 /s<br />

(15 × 15)@(500ft × 500ft) ' 2×10&8 ft/s<br />

The estimated parameter values are acceptable.<br />

(6.1)<br />

Table 6.4 Optimized parameter values and the correlation coefficient matrix. Calibration result<br />

from PEST.<br />

Parameter Estimated 95% percent confidence limits<br />

value lower limit upper limit<br />

p1 1.000282E-02 9.724991E-03 1.028859E-02<br />

p2 1.996080E-08 1.985581E-08 2.006578E-08<br />

Note: confidence limits provide only an indication <strong>of</strong> parameter uncertainty.<br />

They rely on a linearity assumption which may not extend as far in<br />

parameter space as the confidence limits themselves - see PEST manual.<br />

Correlation Coefficient Matrix -----><br />

1.000 0.9572<br />

0.9572 1.000<br />

6.4.1 Basic Model Calibration Skill with PEST/UCODE

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!