Nonlinear Data Assimilation of Coherent Features

Theories of geophysical turbulence and chaotic dynamics lead to the conclusion that assimilating coherent features should lead to improved forecasts. A modular algorithm to assimilate coherent patterns is developed and tested. The algorithm uses nonlinear pattern recognition techniques to identify coherent features and grid morphing to construct the analysis field. The modularity allows feature assimilation to use any standard data assimilation method and can be used together with data assimilation of traditional quantities. The method is tested in a two-layer quasigeostrophic channel by assimilating the location of coherent jets.