Accounting for centennial‐scale variability when detecting changes in ENSO: A study of the Pliocene

The El Niño–Southern Oscillation (ENSO) is the dominant mode of interannual climate variability. However, climate models are inconsistent in future predictions of ENSO, and long‐term variations in ENSO cannot be quantified from the short instrumental records available. Here we analyze ENSO behavior in millennial‐scale climate simulations of a warm climate of the past, the mid‐Pliocene Warm Period (mPWP; ∼3.3–3.0 Ma). We consider centennial‐scale variability in ENSO for both the mPWP and the preindustrial and consider which changes between the two climates are detectable above this variability. We find that El Niño typically occurred 12% less frequently in the mPWP but with a 20% longer duration and with stronger amplitude in precipitation and temperature. However, low‐frequency variability in ENSO meant that Pliocene‐preindustrial changes in El Niño temperature amplitude in the NINO3.4 region (5°N–5°S, 170°W–120°W) were not always detectable. The Pliocene‐preindustrial El Niño temperature signal in the NINO4 region (5°N–5°S, 160°E–150°W) and the El Niño precipitation signal are usually larger than centennial‐scale variations of El Niño amplitude and provide consistent indications of ENSO amplitude change. The enhanced mPWP temperature signal in the NINO4 region is associated with an increase in central Pacific El Niño events similar to those observed in recent decades and predicted for the future. This study highlights the importance of considering centennial‐scale variability when comparing ENSO changes between two climate states. If centennial‐scale variability in ENSO has not been first established, results suggesting changes in ENSO behavior may not be robust.