Abstract/Summary

Lack of constraint on spatial and long-term temporal variability of the El Niño southern Oscillation (ENSO) and its sensitivity to external forcing limit our ability to evaluate climate models and ENSO future projections. Current knowledge of Holocene ENSO variability derived from paleoclimate reconstructions does not separate the role of insolation forcing from internal climate variability. Using an updated synthesis of coral and bivalve monthly resolved records, we build composite records of seasonality and interannual variability in four regions of the tropical Pacific: Eastern Pacific (EP), Central Pacific (CP), Western Pacific (WP) and South West Pacific (SWP). An analysis of the uncertainties due to the sampling of chaotic multidecadal to centennial variability by short records allows for an objective comparison with transient simulations (mid-Holocene to present) performed using four different Earth System models. Sea surface temperature and pseudo-δ18O are used in model-data comparisons to assess the potential influence of hydroclimate change on records. We confirm the significance of the Holocene ENSO minimum (HEM) 3-6ka compared to low frequency unforced modulation of ENSO, with a reduction of ENSO variance of ∼50 % in EP and ∼80 % in CP. The approach suggests that the increasing trend of ENSO since 6ka can be attributed to insolation, while models underestimate ENSO sensitivity to orbital forcing by a factor of 4.7 compared to data, even when accounting for the large multidecadal variability. Precession-induced change in seasonal temperature range is positively linked to ENSO variance in EP and to a lesser extent in other regions, in both models and observations. Our regional approach yields insights into the past spatial expression of ENSO across the tropical Pacific. In the SWP, today under the influence of the South Pacific Convergence Zone (SPCZ), interannual variability was increased by ∼200 % during the HEM, indicating that SPCZ variability is independent from ENSO on millennial time scales.