Analysis of seasonal climate predictability based on long term global sea surface temperature

In the context of current efforts to improve seasonal climate prediction, there is a need to quantify the sources and limits of predictability. Atmospheric processes have a relatively short memory of initial conditions of about two weeks for detailed daily weather prediction. Nevertheless, skilful seasonal forecast is possible in many regions of the world, under specific circumstances. Examples of such conditions are the presence of slow varying boundary conditions (BC) of the atmosphere, such as: a) sea surface temperature anomalies (SSTA) over large oceanic regions, b) persistence of snow cover at high latitudes, and c) development of significant soil moisture anomalies over vast land regions. These boundaries typically evolve on a much slower time scale than daily weather events and atmospheric predictability can be increased as long as the future evolution of such BC can be predicted.

Among these factors, the persistence of SSTA over vast tropical regions has significant effects influencing climate at time scale of season and beyond. It is well established that SSTA tend to have persistence or long memory, due largely to the thermal inertia of the oceans, caused by their heat storage capacity. The ocean communicates its thermal inertia to the atmosphere largely via the surface turbulent fluxes of sensible and latent energy. These fluxes depend upon the sea surface temperature (SST), and several atmospheric parameters. Given the primary importance of SST in the thermal communication between the ocean and atmosphere, and the potential for SST variations to induce slow climatic fluctuations, it is of interest to investigate the nature of temporal persistence of large-scale SST anomalies in the global ocean.

We use the global SSTA and investigate some sources of predictability at seasonal time scale and its impact in various regions of the ocean. Data used are the NOAA Extended Reconstructed Sea Surface Temperature (SST) from 1854 – 2009. The monthly SSTAs are defined relative to the 1971 - 2000 monthly climatology. We show that: 1) SSTA has a memory or persistence that depends largely on regional location in the global ocean, with the largest values in tropical Pacific; 2) A given SSTA distribution from a particular month, can have corresponding similar configurations in the past, largely due to the recurrence of events such as ENSO which tend to affect SSTA distribution over vast regions of the global ocean and have an oscillatory behavior; 3) Correlation of SSTA from different regions of the ocean provide a valuable mean to explore climatic teleconnections. These findings are employed in a statistical model for SSTA prediction and results are discussed in the context of current efforts to improve seasonal forecast.