Improving seasonal prediction of global mean surface temperature by incorporating dynamic ENSO realistic forecasts

Abstract Inaccurate autumn-initialized prediction of global mean surface temperature (GMST) limits its practical value for climate-sensitive sectors, largely due to an incomplete understanding of key physical drivers. To improve this, we identified a previously underrepresented El Niño–Southern Oscillation (ENSO)-driven pantropical coupling mechanism that links ENSO evolution to tropical ocean–atmosphere interactions and a coherent autumn–winter global temperature response, and show that lack of this process contributes to increasing errors. Based on this mechanism, we developed a dynamic-statistical framework that incorporates skillful ENSO realistic forecasts into GMST prediction. The new framework extends reliable GMST prediction lead-time from two to four months and reduces hindcast errors by an average of 41% in 64% of years during 1980–2024, with particularly large improvements during ENSO-active periods, especially El Niño years (85%). These gains strengthen seasonal climate early-warning and have broad applications as tropical ocean variability and impacts may intensify under climate change.