ENSO-conditioned evolution of global mean surface temperature

Michael K. Tippett, Emily Becker

Here we examined how the June--May trajectory of global mean surface temperature (GMST) can be anticipated from recent GMST evolution and upcoming boreal winter Ni\~no-3.4 values. Principal component analysis and dimension reduction of the data led to a simple, interpretable model in which the June--May monthly GMST trajectory is conditioned on two quantities: the average GMST of the prior 12 months and the upcoming December value of Ni\~no-3.4. Including ENSO improves performance relative to persistence baselines that do not include ENSO, especially during the time of the year when atmospheric bridge mechanisms are active. The model associates a December Ni\~no-3.4 anomaly of 1\textdegree C with a GMST increase of about 0.03\textdegree C during June--August followed by a peak increase of about 0.11\textdegree C in the following February. Initialized climate forecasts show broadly similar ENSO--GMST relationships and provide additional skill prior to the ENSO peak.