Efficient nonlinear data assimilation for oceanic models of intermediate complexity

A fully nonlinear particle filter is used on a simplified ocean model, consisting of the barotropic vorticity equation. While common knowledge is that particle filters are inefficient and need large numbers of model runs to avoid degeneracy, the newly developed particle filters need only of the order of 10-100 particles on large scale problems. Also, we show that the scaling is perfect in that increasing the dimension of the system does not need more particles. This opens the possibility for fully nonlinear filtering/smoothing in very high dimensional state spaces, e.g. for numerical weather forecasting.