|
|
|
|
|
|
|
|
Information - HS32 Quantification of structural error, parameter estimation and uncertainty assessment in groundwater and hydrological catchment modelling
|
|
|
|
Event Information |
|
|
|
|
|
|
Uncertainty analysis and uncertainty propagation is becoming an integral part of hydrological modelling. Using either inverse modelling techniques, forward stochastic modelling methods, data-assimilation or a combination of these, uncertainty about model predictions is expressed in terms of prediction (co)variances, ensemble predictions or interval estimates. Regardless of the method used, almost always the assumption is made that the model is a reasonable, albeit not perfect, representation of reality. It is thus assumed that there is no significant model structural error. One way of considering structural errors in the uncertainty analysis is adding an additional noise component, thereby creating a stochastic model that incorporates the effects of all error sources, including model error. This approach has often been followed in data-assimilation, and has recently been implemented in inverse modelling as well. However, these approaches still assume that errors occurring from misrepresentation of processes or wrong schematisation are random. Systematic model structural errors are still not accounted for. This is especially crucial when a model is to be used for predictions outside some calibration conditions in systems with a strong non-linear behaviour. Consequently, there is great need for methods that are able to detect systematic model errors, such that effort is directed towards improving the model concepts and schematisation first, before any further uncertainty analysis is tried.
This session invites contributions related to the development and application of theories/frameworks/methods that help to identify model structural errors and facilitate a meaningful uncertainty analysis. In particular we invite contributions that address (but are not limited to) the following issues:
1. stochastic modeling - how to properly identify, treat and propagate the various sources of uncertainty, in particular model structural inadequacies, to yield reasonable model output prediction uncertainties;
2. model calibration - how to meaningfully assess parameter uncertainty, with a particular emphasis on the development and application of methods for parameter estimation and uncertainty analysis for spatially distributed environmental models;
3. data assimilation - how to optimally combine model predictions and data, with a particular emphasis on how to exploit multiple sources of data/information to reduce uncertainty in model predictions. Papers are solicited that deal with these issues in groundwater and hydrologic catchment modelling.
|
|
|
|
|
|
|
|
|
|
|
Preliminary List of Solicited Speakers |
|
|
|
|
|
|
|
|
|
|
Back to Session Programme
|
|
|
|