The increasing demand for climate information that spans seasonal to multi-annual time scales poses a challenge for current prediction systems, which are traditionally designed for specific forecast horizons. This study addresses this gap by proposing a new method to generate seamless climate information from seasonal to decadal time scales. We develop a constraining approach based on ensemble member selection, in which decadal prediction members are selected to match the seasonal forecast ensemble mean of sea surface temperature. The method leverages the higher skill of seasonal predictions in capturing interannual climate variability, particularly El Niño–southern oscillation, to constrain decadal forecasts using the most recent climate information. Results show that the method to constrain decadal predictions improves the forecast skill over the Niño3.4 region up to 12 months and enhances the near-surface temperature predictions over broad parts of the globe, with modest improvements in precipitation. This work highlights the practical potential of combining seasonal and decadal prediction systems and offers a first step toward operational, seamless climate services across monthly to multi-year timescales.

Read original article

Read original article