Extended Ensemble Filter for High-dimensional Nonlinear State Space Models

There are several functional forms for non-linear dynamical filters. Extended Kalman filters algorithms that used to estimate more accurate values of unknown quantities internal systems from a sequence noisy observation measured over period time. This filtering process becomes computationally expensive when subjected high dimensional data which consequently has negative impact on the filter performance in real is because integration equation evolution covariances extremely costly, especially dimension problem huge case numerical weather prediction.This study developed new filter, First order Ensemble Filter (FoEEF), with extended innovation improve measurement and be able state value data. We propose empirically, lends amenable large models. The derived stochastic state-space models its tested using Lorenz 63 system ordinary differential equations Matlab software.The newly then compared performances three other filters, is, Bootstrap particle (BPF), Bucy (FoEKBF) Second (SoEKBF).The FoEEF improves increase ensemble size. Even as low number ensembles 40, performs good FoEKBF SoEKBF. shows, proposed can register high-dimensional