Subseasonal to Interannual Prediction Sensitivities in the GFDL/FMS GCM

W. Stern, A. Rosati, R. Gudgel, M. Harrison, and A. Wittenberg
NOAA Geophysical Fluid Dynamics Laboratory

15th Symposium on Global Change and Climate Variations, Seattle, WA, 12-15 January 2004

Abstract: There is considerable evidence that tropical convection is tied to a variety of phenomena from subseasonal to interannual time scales. For example, intraseasonal oscillations (ISO) or Madden-Julian oscillation (MJO) is known to directly impact the onset/break activity of the Asian monsoon. While on interannual timescales, the El Niño-Southern Oscillation (ENSO) relates to shifts in convection associated with the Walker Circulation. In addition there is evidence that westerly wind bursts associated with ISO may influence the El Niñ-Southern Oscillation (ENSO). Hence, the representation of convection in a GCM and the GCMs ability to represent ISO may impact predictive skill over a range of time and space scales.

In this study the representation/sensitivity of ISO and ENSO in the GFDL FMS GCM, is investigated. Sensitivity of the ISO to modifications in the atmosphere only GCM's (AGCM) convective parameterization has been seen. In particular, this study focuses on the sensitivity of ISO to a simple diffusive cumulus momentum transport (CMT) and setting a minimum bound on lateral entrainment rates for deep convection (Tokioka modification). It is found that adding the Tokioka modification enhances the ISO while the CMT tends to weaken it. In addition to amplitude, the representation of the ISO in the GCM is assessed by validating its propagation characteristics and the geographic location of centers of activity. The representation of ENSO in a fully coupled GCM version (CGCM) also appears to be sensitive to changes to the convective parameterization. To investigate the predictive impact, ensembles of 1 year predictions are performed with the CGCM, for a control and each modification to the GCM convective parameterization. The ocean state is initialized with an ocean data assimilation system (ODA) and the atmospheric states are taken from an ensemble of AGCM integrations. Periods of strong MJO activity in both the GCM and observations, are identified using extended EOF (EEOF) analyses of band passed (i.e., 30 - 90 days) precipitation pentads (5 day means). GCM predictive skill is assessed on subseasonal to ENSO time scales, for a variety of space domains and for periods of active versus inactive ISO.