Methodology - Response of Flood Events to Land Use and Climate Change - page 68

The best roughness parameters

SOBEK / HEC RAS

Hydrodynamic calculation

Object function

Comparison of calculation and
observation

Not accurate

Accurate

Simplex

Adapt roughness
parameters

Calibration
competed

Fig. 3.7 Schematic calibration procedure Wisser et al. [ 54 ]

The algorithm is very simple. The function to be minimized represents the
differences between calculated and measured water levels. The object function is
based on a function of the roughness parameters which are calibrated. When the
object function is minimal, it means that the roughness parameters are in an
optimal case.
To prevent the roughness parameters from taking on unrealistic and inaccurate
values, lower and upper bounds for the parameters are defined ([ 52 - 55 ]).
The calculation is based on a file with roughness parameters. Then, the cal-
culated water depth and discharges are compared with measurement parameters
using an object function. The simplex algorithm is then used to minimize the
object function. In this step, the new roughness parameters are applied and the
procedures of calculation start again. The process continues until the calculated
measurements reach an acceptable agreement with measurement parameters and
the desired accuracy is obtained. The selected object function is very important
and the following parameters should be considered:

• First, it has to be decided which measurements are to be taken into account
(stations, time series, etc.).
• Second, it has to be decided if all the measurements are equally important or if
weight functions will be used.
• Third, it has to be decided in which way differences between measurements are
quantified (e.g., absolute values, quadratic differences).

Wisser et al. [ 54 ] have suggested a penalty function as another option which has
to be included. In this case, they mentioned that the object function consists of the
sum of the differences between calculations and observation parameters and
another penalty part. If a certain condition is fulfilled, this part is zero. On the other
hand, if that condition is not fulfilled this part is positive [ 54 ]. Wisser et al. [ 54 ]
described the penalty function in a simple way: ''If both depth of water and
discharges are included in the calibration procedures, then a penalty function is