Four-Dimensional Ensemble Variational Data Assimilation for parameter estimation with the JULES land surface model
Many problems in land surface data assimilation are best tackled as parameter estimation problems due to the typically non-chaotic and convergent nature of the models. For this reason it can be preferential to use variational data assimilation techniques instead of sequential techniques in order to avoid time-varying parameters. Variational techniques require knowledge of the derivative of the model, this can prove costly to compute and maintain with the release of newer model versions. In this talk we outline the hybrid technique of four-dimensional ensemble variational data assimilation as a possible solution to these problems.
The Joint UK Land Environment Simulator (JULES) is a community land surface model used by the UK Met Office in the production of forecasts and is the land surface scheme of the UK Earth System Model (UKESM). In this talk we will discuss the implementation of a data assimilation scheme for this model and show a number of applications. These include the improvement of parameterisations of soil properties over Ghana through the assimilation of remotely sensed soil moisture and the optimisation of crop model parameters using flux tower data.
Dr. Ewan Pinnington University of Reading