A Statistical Analysis of Global Inter-Annual Climate Anomalies in Monthly Sea Surface Temperature Records

Warm and cold deviations from the annual sea surface temperature climate in the eastern tropical Pacific have a wide significance. Numerous reports have examined connections between these so-called El-Nino and La-Nina events and anomalies in local climate in areas well removed from eastern extra-tropical Pacific. The analysis of climate anomalies is statistical in nature. We consider a global dataset of monthly sea surface temperature (SST) values compiled on a 2 × 2 global grid from 1945 to 1997 by the National Center for Atmospheric Research (NCAR). The high dimensionality and variable extent of spatial and temporal sampling in this dataset is representative of many climate datasets currently of interest in climate studies in atmospheric science and oceanography. A novel statistical approach to the analysis of climate anomalies is presented. The approach can handle complete records, does not restrict itself to pre-specified seasonally averaged anomalies and produces objectively smoothed estimates of teleconnection patterns associated with anomalies. The methodology involves a combination of time-series seasonal decomposition and 2-D smoothing on the complex ocean domain. Three inter-annual climate anomalies in the SST record are found. The global structure of these anomalies identifies connections between the Pacific, Atlantic and Indian oceans. The relations between the anomalies and established indices of SST climate variations, such as the El-Nino Southern Oscillation, the Pacific Decadal Oscillation and the Cold Tongue Index, are also described. Key-words: climate data, spatio-temporal data analysis, regularization, seasonal decomposition, sea surface temperature, smoothing. ∗Department of Statistics, University College Cork, Ireland. Work supported in part by the IRCSET and the HEA, Ireland.