Item

Abstract

A significant amount of variability in the tropics has been associated with non-Rossby waves, i.e. the equatorial Kelvin, mixed Rossby-gravity and inertia-gravity waves. The existing data assimilation schemes account for these waves only in an implicit way such as internal dynamics within the 4D-Var or regressions within EnKF methods. Global forecast errors have significant projections onto non-Rossby equatorial waves. We speculate that multivariate data assimilation system incorporating mass-wind coupling associated with these waves may significantly reduce tropical analysis uncertainties. A reduction of tropical analysis errors should further propagate to extra-tropics, leading to improvement of global forecast skills in medium and extended range. As a proof of concept, we build the sought 3D-var data assimilation system for the TIGAR (Transient Inertia-Gravity and Rossby wave dynamics) model, that has Rossby and non-Rossby waves as prognostic variables. Using observing system simulation experiments (OSSEs) in the barotropic version of the model, we show that data assimilation “aware” of equatorial wave coupling improves tropical analyses and tropical and extra-tropical forecasts. Furthermore, contributions from winds and mass-field observations in the tropics are compared for their efficiency in reducing extra-tropical forecast errors.