Local particle filter for geofluid models: a review and beyond
In this paper, my colleague Alban Farchi and I review recent advances in the development of local particle filters meant to track geofluid models. We have attempted to classify all flavours of local particle filters that appeared recently. We also perform extensive and careful numerical comparisons using these data assimilation algorithms on the Lorenz 96 model as well as on a two-dimensional barotropic model. In addition, we propose new schemes; notably one based on optimal transport resampling performed in state space. The best schemes, including the new one, and with proper tuning, are almost as good as a well-tuned local ensemble Kalman filter in weakly nonlinear conditions, where the the particle filter has no particular edge. The review, entitled Review article: Comparison of local particle filters and new implementations, is published in Nonlinear Processes in Geophysics.