Interannual Sea Surface Temperature Variability and the Predictability of Tropical Intraseasonal Variability

Abstract

The objective of this study is to examine the relationship between interannual SST variability and the activity and predictability of tropical intraseasonal variability (TISV). A 10-yr simulation forced by climatological SSTs and a 10-member 10-yr (1979?88) ensemble of simulations forced by observed SSTs from the NASA Goddard Laboratory for the Atmospheres GCM coupled to a weakly interacting slab ocean mixed layer are analyzed. The climatological simulation provides a measure of the natural variability associated with TISV, while the observed SST simulations provide an indication of how externally imposed SST anomalies modify this variability. Analysis is conducted on both the eastward-propagating (winter mode) and northeastward-propagating (summer mode) forms of TISV. Indices of TISV are constructed from the amplitude time series of the leading EOFs of intraseasonally bandpassed model precipitation and 850-hPa zonal wind. Analysis of the TISV activity indices from the climatological SST simulation shows that considerable natural variability exists for both the winter and summer TISV modes. Based on the TISV indices constructed, the internal variability of TISV is about 50% of the mean signal. Moreover, the interannual standard deviation in TISV activity from the climatological SST simulation ranges from about the same size to about twice the standard deviation associated with the seasonal evolution of TISV activity. A ratio between an estimate of the externally forced TISV variability in the observed SST simulations to an estimate of natural TISV variability from the climatological SST simulations is constructed to give a measure of TISV predictability. While these predictability values exhibit some seasonal dependence, on average they suggest little to no predictability associated with interannual TISV variations for either the winter or summer TISV mode. However, the ensemble does demonstrate significantly enhanced predictability of the summer TISV mode during the 1982/83 winter. This same characteristic was found during the 1997/98 winter in a second 10-member ensemble that was conducted for the period September 1996 to August 1998. In each of these cases, the ensemble means exhibited a decrease in TISV activity, a feature that is also echoed in the observations. Thus, under very specific circumstances, anomalous SST may have some predictable influence over the level of TISV activity, at least for the summer mode. In addition, the simulations show evidence that generalized intraseasonal variance (i.e., no constraint on spatial structure) displays some predictable characteristics over the central and eastern equatorial Pacific in association with ENSO-related SST anomalies. Considerations of the use of the ocean mixed layer coupling in the context of the above study are discussed as well as the agreement between the observed levels of TISV activity for the periods modeled and the levels simulated by the GCM ensemble.