Improving ENSO and Its Related SST Forecasts by Accounting for Annual Cycle Effects

The deterministic and probabilistic hindcast skills, for leads from a month to a year, of the current generation of coupled GCMs (CGCMs) derived from the North American Multimodel Ensemble (NMME) are assessed and benchmarked with skills of an empirical model, a “Cyclostationary Linear Inverse Model (CSLIM)”. Both systems produce sea surface temperature (SST) and sea surface height (SSH) forecasts. The CSLIM, constructed from observed tropical (24^oS-24^oN) monthly SST and SSH anomalies during the period 1961-2010, produces forecasts from 1 month to 12 month lead.

The CSLIM deterministic and probabilistic skills are comparable to or better than the CGCM ensemble mean, as well as stationary LIM, at all leads and all locations. Notably, the CSLIM is significantly more skillful than the CGCM at longer forecast leads. For example, significant skills in the ENSO, in both phase and amplitude, up to a year in CSLIM are well compared to relatively low skills in NMME models.

Analysis of the CSLIM suggests that incorporated annual cycle effects improve ENSO phase locking in LIM and lead to improve deterministic and probabilistic skills of ENSO and its related SST anomalies.