Ensemble predictions and ensemble data assimilation have played a significant role in improving forecasting skills for predicting the Northern European winter climate. However, the use of ensemble data assimilation with coupled climate models is still in its infancy. Particular issues related to the multiscale nature of coupled climate models pose problems for the data assimilation methods in their standard implementation. During the research visit, I connect to an ongoing research activity at the University of Bologna that focuses on multiscale data assimilation and estimation of extreme events. In collaboration with this project, we will formulate consistent ensemble data-assimilation methods for multiscale model updating. An additional problem with coupled climate model is the presence of systematic errors and biases. Fortunately, ensemble data assimilation methods have proven helpful in estimating high dimensional model parameters, e.g., in the petroleum sciences. Hence, we will also establish the framework for using ensemble data assimilation to estimate parameters in climate models. The merging of data assimilation with machine learning also offers possibilities to reduce systematic model errors and will be integrated into the assimilation framework. Finally, from ensemble predictions, we can use statistical methods to estimate extreme events, and such estimates are essential in a range of areas, including climate predictions.