Parameter Optimisation, RR Parameters

Having made the sensitivity analysis we will turn back to the calibration. Based on the sensitivity analysis results the 10 most sensitive parameters are included in the calibration (parameters ranked 1 to 10 in Table 2.2).

The setup is included as Example2/Test2-Optimisation.auc.

Simulation specification
Model Parameters
Objective Functions
Parameter Optimisation
Results

Simulation specification

The same specifications and .rr11 template file applied for the sensitivity anal­ysis are used, and Parameter Optimisation is chosen as the Simulation option.

Model Parameters

In the model parameters table the parameters that are not included in the optimisation are set to constant (CKIF1, TIF1, CKIF2, TIF2). These parame­ters will then be set to their initial values given in Table 2.1. To further simplify the calibration a fixed relation between the maximum water contents in sur­face and root zone storages is used. It is assumed that Lmax is 10 times larger than Umax. In the model parameters table Lmax1 and Lmax2 is set in the Equation field, respectively, 10*Umax1 for Lmax1 and 10*Umax2 for Lmax2.

Objective Functions

The same output measures and objective function as applied for the sensitiv­ity analysis are used.

Parameter Optimisation

The population size is set to 40, and the maximum number of model evalua­tions is set to 800.

Results

The optimised parameters are shown in Table 2.3 and compared with the ref­erence parameters. Since not all NAM parameters have been included in the optimisation some deviations are seen, especially for the Umax and Lmax parameters.