| one publication added to basket [196911] | Design of a sampling strategy to optimally calibrate a reactive transport model: Exploring the potential for Escherichia coli in the Scheldt Estuary
de Brauwere, A.; De Ridder, F.; Gourgue, O.; Lambrechts, J.; Comblen, R.; Pintelon, R.; Passerat, J.; Servais, P.; Elskens, M.; Baeyens, W.; Kärnä, T.; de Brye, B.; Deleersnijder, E. (2009). Design of a sampling strategy to optimally calibrate a reactive transport model: Exploring the potential for Escherichia coli in the Scheldt Estuary. Environ. Model. Softw. 24(8): 969-981. dx.doi.org/10.1016/j.envsoft.2009.02.004
In: Environmental Modelling & Software. Elsevier: Oxford. ISSN 1364-8152; e-ISSN 1873-6726, meer
|  |
| Trefwoorden |
|
| Author keywords |
Optimal experimental design; Parameter estimation; Parameteruncertainty; Reactive tracer model; Scheldt, Fisher information matrix |
| Auteurs | | Top |
- de Brauwere, A., meer
- De Ridder, F., meer
- Gourgue, O., meer
- Lambrechts, J., meer
- Comblen, R., meer
|
|
|
| Abstract |
For the calibration of any model, measurements are necessary. As measurements are expensive, it is of interest to determine beforehand which kind of samples will provide maximal information. Using a criterion related to the Fisher information matrix as a measure for information content, it is possible to design a sampling scheme that will enable the most precise parameter estimates. This approach was applied to a reactive transport model (based on the Second-generation Louvain-la-Neuve Ice-ocean Model, SLIM) of Escherichia coli concentrations in the Scheldt Estuary. As this estuary is highly influenced by the tide, it is expected that careful timing of the samples with respect to the tidal cycle can have an effect on the quality of the data. The timing and also the positioning of samples were optimised according to the proposed criterion. In the investigated case studies the precision of the estimated parameters could be improved by up to a factor of ten, confirming the usefulness of this approach to maximize the amount of information that can be retrieved from a fixed number of samples. Precise parameter values will result in more reliable model simulations, which can be used for interpretation, or can in turn serve to plan subsequent sampling campaigns to further constrain the model parameters. |
|
