Low-Frequency Predictability of the Dynamical Extended-Range Forecast Experiment

Abstract

The tropical intraseasonal oscillation is examined using projections of 5-day-averaged wind and height data onto the normal modes of a primitive equation model linearized about a basic state at rest and prescribed global temperature. The dataset utilized in this study is from the Dynamical Extended-Range Forecast (DERF) Experiment, conducted from January 1986 through March 1987. Examination of zonal-wind anomalies in the tropics shows that the forecast model is able to predict propagation of intraseasonal variations more accurately for slow propagation rates. The forecasted amplitude is generally weaker than the analyzed amplitude. Analyzed kinetic energy and error fields exhibit similar horizontal scales for external and internal modes. External Rossby-mode components maximize in the extratropics while Rossby internal modes exhibit patterns that extend over the entire globe. The total energy error of zonally asymmetric flow exhibits dominant contributions from the first and third vertical modes. Midlatitude zonally averaged zonal-wind error is mostly due to internal (baroclinic) Rossby mode structures. Baroclinic error components are shown to be related to the tropical convective rainfall precipitation error using results obtained from a linearized shallow-water equation model. There is some evidence of internal Rossby-mode errors propagating from low into high latitudes, while the reverse is true for external modes. Tropical forecasts of the divergent wind are found to be less dependent on the extratropics than those for the rotational wind, with tropical forcing and dynamics playing a dominant role in the specification of the tropical circulation.