Predictability experiments for the Asian summer monsoon: impact of SST anomalies on interannual and intraseasonal variability

The effects of SST anomalies on the interannual and intraseasonal variability of the Asian summer monsoon have been studied by multivariate statistical analyses of 850-hPa wind and rainfall fields simulated in a set of ensemble integrations of the ECMWF atmospheric GCM, referred to as the PRISM experiments. The simulations used observed SSTs (PRISM-O), covering 9 years characterised by large variations of the ENSO phenomenon in the 1980's and the early 1990's. A parallel set of simulations was also performed with climatological SSTs (PRISM-C), thus enabling the influence of SST forcing on the modes of interannual and intraseasonal variability to be investigated. As in observations, the model's interannual variability is dominated by a zonally-oriented mode which describes the north-south movement of the tropical convergence zone (TCZ). This mode appears to be independent of SST forcing and its robustness between the PRISM-O and PRISM-C simulations suggests that it is driven by internal atmospheric dynamics. On the other hand, the second mode of variability, which again has a good correspondence with observed patterns, shows a clear relationship with the ENSO cycle. Since the mode related to ENSO accounts for only a small part of the total variance, the notion of a quasi-linear superposition of forced and unforced modes of variability may not provide an appropriate interpretation of monsoon interannual variability. Consequently, the possibility of a non-linear influence has been investigated by exploring the relationship between interannual and intraseasonal variability. As in other studies, a common mode of interannual and intraseasonal variability has been found, in this case describing the north-south transition of the TCZ associated with monsoon active/break cycles. Although seasonal-mean values of the Principal Component (PC) timeseries associated with the leading intraseasonal mode shows no significant correlation with ENSO, the 2-dimensional probability PC indices of the two leading modes changes from unimodal in the warm phase of ENSO to bimodal in the cold ENSO phase. These changes are suggestive of some sort of bifurcation in the monsoon properties, with multiple-regime behaviour being established only when the zonal asymmetries in equatorial Pacific SST exceed a threshold value. Although an observational verification of this hypothesis is still to be achieved, the detection of regime-like behaviour in simulations by a complex numerical model gives a stronger support to this dynamical framework than simple qualitative arguments based on the analogy with low-order non-linear systems.