B11 - Understanding time-varying Earth System properties using data assimilation

The Earth system is not static but continuously changing to adjust to evolving boundary conditions such as those defined by our climate. Vegetation is a prominent example as both its structure and behaviour depend on climatic boundary conditions. Vegetation leaves and overall plant behaviour respond to changing atmospheric CO2 levels, while roots change when exposed to droughts, among other interactions. These changes occur at different time scales, and while some are (at least partially) visible in existing observations such as remotely sensed leaf area, others remain hidden and can only be inferred through inverse modelling, e.g. changing roots. Hydrologic models are used to simulate the changing terrestrial water cycle, but do not currently consider changing vegetation structure and behaviour despite these processes becoming more pronounced with climate change. Hence, it is currently unclear what influence elevated CO2 will have on future streamflow, a critical variable for water supply and for aquatic ecosystems. The question we ask here is whether we, by combining and advancing state of the art understanding of both hydrologic processes and data assimilation, can improve the modelling of these changing system properties in current models? Within this project, the following issues will be addressed:
(a) Develop a dual state-parameter updating framework based on particle methods such as
the ensemble Kalman filter/smoother with particular focus on the value of (process-based) prior constraints on time evolving model parameters utilizing multiple observed data streams.
(b) Application of the framework of a parsimonious and process-based hydrologic model at
data-rich observational sites across Europe. Cross-site evaluation is necessary to understand
interactions of climatic conditions, weather patterns, vegetation and soils in defining water-energy fluxes and changing system properties.