HS1.8Data Assimilation for integrated hydrological models and Earth System Models
|Convener: Harrie-Jan Hendricks Franssen | Co-Conveners: Clemens Simmer , Insa Neuweiler , Jehan Rihani|
This session focuses on data assimilation across compartments, for example for integrated hydrological models or earth system models. The advantage of such models over models describing a single compartment is that due to the more physically consistent representation of exchange fluxes, observations in different compartments can be used more consistently in order to predict fluxes and states in the whole system. This refers to water and energy fluxes or, for example, carbon and nitrogen in earth system models.
Although this approach could potentially lead to improved predictions, a main limitation is the large number of unknown parameters and initial states that need to be dealt with.
Data assimilation is in principle able to optimally combine model estimates and measurements, and improve model parameterization. In the context of integrated hydrological models or earth system models, it is important to consider the cross-compartmental influence of measurements. For example, leaf area index data not only provide information on vegetation status, but are also linked to soil moisture content of the unsaturated zone. Another example is soil moisture data, which can help to improve the characterization of states in the soil, aquifer beneath, vegetation and the atmospheric boundary layer. In order to fully explore the value of these data for specific predictions, an integrated modelling framework in combination with a data assimilation framework across terrestrial system compartments is needed. On the other hand, efficient data assimilation approaches are needed in order to allow predictions in a large and complex modeling system.
We welcome contributions on the development of data assimilation systems and applications of data assimilation for models which include multiple compartments of the terrestrial system. Contributions could focus on short- or long-term predictions, parameter estimation, data value or design of monitoring networks. Also field studies addressing the topic would be interesting contributions.