Interannual to Decadal Climate Predictions: Where Are We Now and What Have We Learned? (Invited Presentation)

Prediction of interannual to decadal climate variability has been an operational activity for more than five years now, mainly through the WMO Lead Centre for Annual to Decadal Climate Prediction. The Lead Centre publishes a Global Annual to Decadal Climate Update every year; it generates a lot of interest from the press, primarily for the prediction of the probability of near-surface global mean temperature exceeding 1.5 degrees C, which generates many headlines. This is just the tip of the iceberg of a large community of researchers working on understanding, improving and applying these forecasts to create climate information that can be acted upon. It is becoming clear that large ensembles are needed to isolate a spuriously weak climate signal. An ensemble of ocean initial conditions for predictions might help to sample the uncertainty from poorly observed ocean states, but there can be surprises, such as an increase in ensemble spread when there are more observations. There is increased interest in the interannual timescales, which still benefit from skill in predictions of the El Niño Southern Oscillation. New mechanisms on the interannual timescales are being discovered and the protocol for decadal climate predictions in CMIP7 is likely to encourage the running of hindcasts for these timescales. However, there are some worrying problems with decadal predictions, forecasts for recent decades show worse skill than hindcasts from earlier years and this must be investigated and understood. One way to gain confidence in the forecasts is if we can use large ensembles of single-forcing runs to attribute forecast signals.