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

Earth system properties like vegetation leaves and roots are not static, but evolve in response to varying boundary conditions, e.g. due to climate change. Current land surface models struggle to provide robust estimates of how such non-stationarity impacts the water cycle. We propose to develop a dual state-parameter updating framework based on the ensemble Kalman filter/smoother to learn the interaction between process-based prior constraints, dynamic observational constraints and the time variance of model parameters. We will use an existing parsimonious-process-based model and detailed data from study locations across Europe to advance our understanding of this non-stationarity and its implication for the water cycle.