HS8.1.3Model Uncertainties, Parameter Estimation, and Data Assimilation in Surface and Subsurface Hydrology
|Convener: Wolfgang Nowak | Co-Conveners: Thomas Wöhling , Matteo Camporese , Thorsten Wagener|
Predictions of chemical, biological and physical processes in soils, aquifers, rivers and across compartments (e.g., stream-aquifer interactions) are strongly affected by errors in model structure, parameters and forcing data. The determination of optimal parameter sets and the quantification of uncertainty are therefore crucial for reliable predictions. This session invites contributions that focus on improved concepts, approaches and methods for measurement, parameter estimation and uncertainty quantification techniques at all scales (lab/field/catchment), especially studies on:
- new measurement technologies that either can aid in the parameterization of models for surface and subsurface flow and transport, chemical reactions and (micro)biological processes, or can help to diagnose and detect structural model deficiencies.
- parameter estimation schemes which provide a synthesis of spatially distributed parameter fields and prior information of the underlying geology or catchment. This may include approaches for inversion of remote sensing and hydrogeophysical data, especially when equipped with estimates of uncertainty.
- optimal experimental design strategies that maximize information retrieval from measurements and minimize prediction uncertainty.
- the contribution of different observation types (from point-scale measurements to remote sensing data and “soft information”) to uncertainty reduction.
Also, in the past years much progress has been made in the area of parameter uncertainty estimation, but much less attention has been paid to representation of model and forcing data errors. We therefore especially encourage submissions on the following topics:
- novel theories and concepts for spatial and temporal analysis of the model - data mismatch. This ranges from pure optimization-based methods to sequential data assimilation.
- formal and informal statistical frameworks that diagnose, detect and resolve all sources of modelling errors and capture conceptual model uncertainty (e.g. multi-model ensemble systems).