Inferring changes in ENSO amplitude from the variance of proxy records

Abstract

One common approach to investigating past changes in El Niño–Southern Oscillation (ENSO) amplitude is through quantifying the variance of ENSO-influenced proxy records. However, a component of the variance of all such proxies will reflect influences that are unrelated to the instrumental climatic indices from which modern ENSO amplitudes are defined. The unrelated component of proxy variance introduces a fundamental source of uncertainty to all such constraints on past ENSO amplitudes. Based on a simple parametric approach to modeling this uncertainty, we present guidelines for the magnitudes of proxy variance change required to robustly infer the following: (i) any change at all in ENSO amplitude and (ii) a change in ENSO amplitude that exceeds the plausible range of unforced variability. It is noted that more extreme changes in proxy variance are required to robustly infer decreases, as opposed to increases, in past ENSO amplitude from modern levels.