Comparison of different versions of the CSIRO Mark 2 climate model (Watterson, 2002, J Geophys. Res.-D) suggests that air-sea interaction may have a strong influence on intraseasonal tropical variability, including the eastward propagating Madden-Julian Oscillation (MJO). The effect of interaction on the MJO in the new higher-resolution Mark 3 model is the topic of this talk. In standard coupled atmosphere-ocean mode, Mark 3 produces a stronger u₂₀₀ MJO spectral signal than Mark 2, although the peak period remains too high, around 80 d, compared with that from the ERA40 reanalysis data set (35-55d). As for Mark 2, the signal is much smaller in a series of AGCM runs with sea surface temperatures specified from 1949-2001.

The lower tropospheric structure of the coupled-model MJO is represented using EOFs of daily tropical winds from each season. Time-lag regression shows that these bands of convergence propagate eastward across the convective regions of the Indian Ocean and far-west Pacific, particularly for January. There is little propagation of these EOF patterns in the AGCM case. The difference is largely due to air-sea interaction, with the mean equatorial surface westerly belt being an important factor, through its effect on the sign of surface evaporation anomalies. In the westerly belt of the Australian monsoon, the EOF wind anomalies produce evaporation anomalies that boost SSTs to the east of the convergence, aiding propagation.

The role of the equatorial westerly belt is further explored through a partition of years based on the state of ENSO. In simulated years with enhanced central Pacific SSTs (El Ni–o-like), the belt in Mark 3 is extended further east, while in La Ni–a-like years the belt is shorter, but stronger over the Indian Ocean. The propagation of the convergent anomalies varies considerably. In the AGCM, the mean westerlies are greatly modified by the SSTs, and the EOF anomalies are larger in the El Ni–o years, but there is still little propagation. In contrast, there seem smaller differences based on ENSO in the ERA40 data.
 

Convenor: Les Muir.