The existence of these numerous forcings raises the possibility of skilful predictions of global temperature for one year ahead. Here indices of the known important climate forcings and influencing phenomena are used to make a mixed empirical and dynamical prediction of the global temperature anomaly from a 1961-90 average. Based on multiple regression, the state of ENSO is the most important predictor on interannual time scales; on greater than decadal time scales the net radiative forcing of the atmosphere is most important. We use a regression model incorporating the above factors. The ENSO predictor is a coupled model prediction of the SST anomaly in the Nino 3.4 region of the Tropical Pacific in the first half of the year ahead using the Met Office GLOSEA model.
Global temperature anomaly forecasts are issued both in best estimate form and as probability forecasts. Intrinsic skill is high (based on cross validation) and equivalent to a correlation of 0.82 for interannual variability and 0.95 when all time scales of variability are considered over 1960-2003. This gives a 2 sigma hindcast annual temperature anomaly uncertainty of +_0.12oC which can be compared to the observed 2 sigma temperature anomaly uncertainty of +-0.06oC. Forecasts are now issued routinely in the previous December as part of an annual press release on global climate made by the UK Met Office. The skill of real-time forecasts issued since 2000 are consistent with cross validation tests, though some details of the forecast procedure have evolved.