After constructing a groundwater model the validity of calculated results are usually checked by comparing them with field data. In order to achieve maximum validity (high similarity between the calculated and observed results) the parameters used in the model have to be optimised. This optimisation of parameters is also called calibration of the model.

Easy comparison of calculated with observed data

With triwaco it is easy to compare calculated with observed heads or fluxes at any observation point or area in the model domain. By creating a file containing the collected data triwaco automatically calculates the deviations, saving the results in a separate file. These results can be graphically displayed using TRIPLOT, for the entire model or for a separate layer.

Calibrating a model like this, however, is often time-consuming so two modules have been developed that automatically perform the parameter optimisation. These modules, called TRCALCON and MONTECARLO, use different methods to achieve optimum parameter values.

TRCALCON

The module TRCALCON uses the Marquet-Levenberg method for parameter optimisation whereby parameter values are adjusted according to deviations between calculated and observed data. The module not only optimises the parameter values but it can also determine reliability intervals for the calculated results.

To use the TRCALCON module a new data set is created in the triwaco project. Only parameters selected for optimisation are added in the new data set. After the selection of parameters (their distribution and the number of necessary calculations) a parameter sensitivity analysis is carried out. The result of the sensitivity analysis is the basis for the procedure of parameter optimisation. After the parameters are optimized, the results are saved in separate files and immediately are available for further calculations.

The TRCALCON module can also determine the reliability of the groundwater modelling results. For this TRCALCON calculates the variance and standard deviation of the model results. In the same way TRCALCON can be used to determine the reliability of path line calculations.

MONTECARLO

With MONTECARLO a large amount of simulation runs are carried out. The user can choose which parameters are used during the optimisation process and what the value distribution for each selected parameter is. Based on these value distributions for every simulation run a set of new parameter values is determined. Selection of a parameter value from the given value distribution is carried out according to the Latin-Hypercube method

To use the MONTECARLO module a new data set is created in the triwaco project. Only parameters selected for optimisation are added in the new data set. After the selection of parameters (their distribution and the number of necessary calculations) a parameter sensitivity analysis is carried out. Based on these results, parameters can be omitted from the optimization process thus increasing efficiency.

Subsequently, the process of parameter optimization commences. Before starting the calculations the module chooses the same amount of parameter value combinations as the amount of given calculations. The results of the simulation runs are then tested against the observed heads and the deviations saved as separate result files. After the calculations are completed, the parameter combination can be chosen from the result files that has the smallest measured deviation (the optimum).