Sensitivity of tropical subannual variability to model mixed-layer depth

The influence of air-sea interaction on tropical variability on intraseasonal to seasonal timescales is explored through analysis of three long simulations (of 500 or 1000 years) of the CSIRO atmospheric general circulation model (GCM) with differing ocean specifications: a coupled ocean GCM, a simple 50-m mixed-layer model, or climatological sea surface temperatures (SSTs); together with 50 year simulations with mixed-layer depths of 20 m and 10 m. The analysis focuses initially on low-level equatorial convergent patterns contained in the first two EOFs of both monthly and pentad anomalies of tropical 807-hPa winds in January in the coupled model. Time-lag regression using both monthly and daily data is used to demonstrate that these patterns propagate eastward, although at only half the speed of the observed intraseasonal variability (or Madden-Julian Oscillation, MJO). The propagation speed is faster (as seen in the daily data) in the mixed-layer cases, with the 10-m case producing speeds similar to observations. The specified SST model shows no such propagation. In the interactive models, surface energy fluxes force SST anomalies propagating ahead of the EOF patterns. These fluxes are largely consistent with evaporation perturbed by wind anomalies that augment the mean low-level monsoon westerlies. The SST anomalies then perturb the low-level wind convergence pattern, as is confirmed by a separate SST perturbation experiment. From the examination of other seasons, it is seen that air-sea interaction generally enhances the amplitude of these MJO-like patterns. It also enhances their eastward propagation along westerly wind bands. Analysis of zonal wavenumber one winds confirms the strong sensitivity to mixed-layer depth of the amplitude and period of the eastward propagating component, particularly during October through April. The results suggest that air-sea interaction may have an important influence on the MJO, provided that the SSTs are sufficiently sensitive to atmospheric patterns.