Prediction and Optimal Seasonal Precursors of Marine Heatwaves off Western Australia

Marine heatwaves (MHWs) off Western Australia (WA) can cause devastating ecological and socioeconomic impacts. Previous research has shown that WA MHWs are more or less likely to occur depending on the phase of interannual climate variabilities, such as El Niño–Southern Oscillation (ENSO) and the Indian Ocean dipole (IOD). However, the seasonal characteristics of WA MHW predictability and prediction skill associated with these climate modes remain unclear. To address this, we develop a cyclostationary linear inverse model (CS-LIM) using vertically averaged temperature (VAT) to ∼300-m depth as input, and VAT indices are computed for the WA region and ENSO and IOD domains. We investigate seasonal variations in WA MHW prediction skill and predictability by identifying their seasonally varying optimal initial conditions, evolutions, and the contributions of ENSO and IOD. The optimal initial conditions and evolutions typically feature ENSO and IOD patterns, though their phases, timing, flavors, and intensities vary seasonally. An empirical normal mode analysis identifies three dominant modes with coupled ENSO and IOD features. The first (∼4-yr period) features eastern Pacific (EP) ENSO with weak IOD, and its La Niña and negative IOD dominate MHW growth in austral autumn (∼12-month lead). The second (∼7.5-yr period) features central Pacific (CP) ENSO with IOD, and its La Niña with positive IOD influence growth in austral autumn and winter at both 5- and 20-month leads. The third (∼2.5-yr period) features ENSO and strong IOD, and its La Niña with negative IOD dominates austral winter–summer growth (∼5-month lead), while its El Niño and positive IOD dominate at ∼20-month leads.
Significance Statement

Marine heatwaves off Western Australia can occur in any season, damaging marine ecosystems. The climate patterns that precede these events, and may provide predictability, vary seasonally. This study uses a cyclostationary linear inverse model, based on Hasselmann’s stochastic climate framework but incorporating seasonality, to examine how two major climate modes, El Niño–Southern Oscillation and the Indian Ocean dipole, act as precursors to marine heatwaves in different seasons and at different lead times. We find that specific combinations of these modes, with different flavors, intensities, and periods, are linked to marine heatwaves in particular seasons, sometimes showing signs months or even more than a year in advance. These insights can help improve seasonal prediction of marine heatwaves off Western Australia.