Role of nonlinear scale interactions in limiting dynamical prediction of lower tropospheric boreal summer intraseasonal oscillations

Abstract

Dominant scale of tropical boreal summer intraseasonal oscillations (BSISOs) being in the range of wave numbers 1–4, dynamical extended range prediction of BSISO is limited by rapid buildup of errors in ultra‐long/planetary waves in almost all prediction models. While the initial errors are largely on the small scales, within 3–5 days of forecasts maximum errors appear in the ultra‐long waves such as the tropical convergence zone. Spectral decomposition of errors with forecast lead time indicate that the initial error in the small scales is already close to its saturation value at these scales, whereas that in ultra‐long waves is about two orders of magnitude smaller than their saturation values. Such an increase of errors in ultra‐long waves cannot be explained as growth of initial errors. It is proposed that the fast growth of errors in the planetary waves is due to continuous generation of errors in the small scales (due to inadequacy of the physical parameterizations such as formulation of cumulus clouds) and upscale propagation of these errors through the process of scale interactions. Basic systematic error kinetic energy and the scale interactions in terms of the wave‐wave exchanges among nonlinear triads are formulated and the above hypothesis is tested through a diagnostic analysis of the error energetics in two different model predictions at the lower troposphere. It has been revealed that nonlinear triad interactions lead to advection of errors from short and synoptic waves (wave number > 10) to long waves (wave numbers 5–10) and from long waves to ultra‐long waves (wave numbers 1–4) and are responsible for the rapid growth of errors in the planetary waves. Unraveling the exact mechanism through which upscale transfer of errors take place may help us in devising a method to inhibit the mingling of small‐scale error with the error in prediction of tropical intraseasonal oscillations and improve extended range prediction of the lower tropospheric BSISOs.