Calibration Criteria
Use the Calibration Criteria tab to set up how the calibrations are evaluated.
The options you specify are applied to every calibration trial in the Calibration Study. The Calibration Criteria tab contains the following controls:
- Fitness Type - Select the Fitness Type you want to use from the drop down list. In general, regardless of the fitness type you select, a lower fitness indicates better calibration. Fitness Types include: Minimize Difference Squares, Minimize Difference Absolute Values, and Minimize Maximum Difference. For more information, see Calibration Criteria Formulae.
- Minimize Difference Squares - Uses a calibration designed to minimize the sum of squares of the discrepancy between the observed data and the model simulated values. (Model simulated values include hydraulic grades and pipe discharges.) This calibration favors solutions that minimize the overall sum of the squares of discrepancies between observed and simulated data.
- Min. Diff. Absolute Values - Uses a calibration designed to minimize the sum of absolute discrepancy between the observed data and the model simulated values. This calibration favors solutions that minimize the overall sum of discrepancies between observed and simulated data.
- Minimize Max. Difference - Uses a calibration designed to minimize the maximum of all the discrepancies between the observed data and the model simulated values. This calibration favors solutions that minimize the worst single discrepancy between observed and simulated data. Note that the Minimize Maximum Difference Fitness Type is more sensitive to the accuracy of your data than other Fitness Types.
- Head/Flow per Fitness Point - Head and Flow per Fitness Type provide a way for you to weigh the importance of head and flow in your calibration. Set these values such that the head and flow have unit equivalence. You can give higher importance to Head or Flow by setting a smaller number for its Per Fitness Point Value.
- Flow Weight Type - Select the type of weight used: None, Linear, Square, Square Root, and Log. The weighting type you use can provide a greater or lesser fitness penalty.
In general, measurements with larger flow carry more weight in the optimization calibrations than those with less flow. You can exaggerate or reduce the effect larger measurements have on your calibration by selecting different weight types. For example, using no weighting (None) provides no penalty for measurements with lesser flow versus those with greater flow. Using log and square root reduces the fitness penalty for measurements with lesser flow, and using linear or square increases the fitness penalty for measurements with less flow.