Model calibration

If the model is to be used with confidence, ideally it should be calibrated against measured changes in bed levels or outflow sediment concentrations.

In some cases, the longitudinal pattern of deposition shown in the "Long section" plot and included in the general output file, 'filename.zzd', can be compared to the observed pattern. If too much sediment appears to be depositing close to the upstream end of the reach modelled, then either the sediment input is excessive or the input sizes are too large. The distribution of the material entering at the inflow boundary should be checked.

A comparison between the observed and predicted lateral variation of deposition or erosion might also be attempted using the 'filename.sec' output file imported to a spreadsheet.  If the comparison shows obvious discrepancies in the pattern of section shape changes, then the bed update method, or the exponent for method 3, will need adjusting.  If a large discrepancy between the observed and predicted bed levels at one cross-section occurs, this may be because the cross-section is not representative of the reach in which it occurs. Comparison of the cross-section’s geometry and hydraulic properties with those of the cross-sections upstream and downstream should indicate whether or not this is the case.

In carrying out the model calibration the input variables about which least is known should be varied within physically reasonable bounds. In general the channel geometry data, inflows and water level information at downstream boundaries will be subject to less error than the sediment data.  This is especially the case when sediment inflows are predominantly during extreme flow events of short duration, for which collection of sediment data is difficult.  It is possible, for example, that observed bed material during low flow may not be representative of the bed material grading that develops during high flow events. When calibrating, the number of sediment sizes, their distribution and the parameters affecting sediment inflow may be varied if initial values were only estimates.

Other parameters that might be varied during the calibration are:

• Roughness values.
• Sediment transport equation.
• Sediment transport calibration coefficient.
• Settling velocities and threshold shear stresses if cohesive sediment transport is important.