Predictability of Marine Heatwaves off Western Australia Using a Linear Inverse Model

Abstract

Abstract Marine heatwaves (MHWs) off Western Australia (110°–116°E, 22°–32°S; herein, WA MHWs) can cause devastating ecological impacts, as was evidenced by the 2011 extreme event. Previous studies suggest that La Niña is the major large-scale driver of WA MHWs, while the Indian Ocean dipole (IOD) may also play a role. Here, we investigate historical WA MHWs and their connections to these large-scale climate modes in an ocean model (ACCESS-OM2) simulation driven by a prescribed atmosphere from JRA-55-do over 1959–2018. Rather than analyzing sea surface temperature, the WA MHWs and climate mode indices were characterized and investigated in vertically averaged temperature (VAT) to ∼300-m depth to afford the longer ocean dynamic time scales, including remote oceanic connections. We develop a cyclostationary linear inverse model (CS-LIM; from 35°S to 10°N, across the Indo-Pacific Ocean), to investigate the relative contributions of La Niña VAT and positive IOD VAT to the predictability of WA VAT MHWs. Using a large ensemble of CS-LIM simulations, we found that ∼50% of WA MHWs were preceded about 5 months by La Niña, and 30% of the MHWs by positive IOD about 20 months prior. While precursor La Niña or positive IOD, on their own, were found to correspond with increased WA MHW likelihood in the months following (∼2.7 times or ∼1.5 times more likely than by chance, respectively), in combination these climate mode phases were found to produce the largest enhancement in MHW likelihood (∼3.2 times more likely than by chance). Additionally, we found that stronger and longer La Niña and/or positive IOD tend to lead stronger and longer WA MHWs. Significance Statement This study examines seasonal-to-interannual time-scale predictability of marine heatwaves off Western Australia. We developed and applied a linear inverse model, informed by numerical model results, to generate a large number of 60-yr temperature simulations across the broader Indian–Pacific Ocean region to quantify this marine heatwave predictability. We found that La Niña typically increases the likelihood of marine heatwaves off Western Australia about 5 months (3–7 months) later, while positive Indian Ocean dipole events increase their likelihood about 20 months (18–22 months) later. Marine heatwaves can severely impact local marine ecosystems and the economy. Our findings are expected to be valuable for marine heatwave prediction system development on time scales that can be beneficial to marine ecosystem conservation and fishery management.