Large-scale marine heatwaves (MHWs) have been increasingly observed in the southeastern tropical Indian Ocean (SETIO) with adverse impacts on local marine ecosystems. Existing studies indicated that the eastward-propagating Madden–Julian Oscillation (MJO) critically modulates the occurrence and evolution of MHWs in austral summer. Yet, the potential predictability of MHWs provided by this modulation effect has not been evaluated in coupled forecast systems. This issue is addressed here using the Subseasonal-to-Seasonal forecast of the Flexible Global Ocean–Atmosphere-Land System Model, finite-volume version-2 (FGOALS-f2). The FGOALS-f2 shows useful forecast skills (anomaly correlation coefficient > 0.5) of ~ 15 and ~ 20 days for atmospheric convection and sea surface temperature (SST) anomalies of the MJO. For all summertime MHWs of 2000–2018, 61%, 56%, 44%, and 22% can be correctly forecasted at leads of 10, 15, 20, and 30 days, respectively. Initialized from the Real-time Multivariate MJO (RMM) phase 1, the forecasted occurrence probability of MHW reaches 25% after 10–15 days, evidently higher than those initiated from other phases (e.g., 8% from Phase 5), which is consistent in observation-based statistics. Notably, forecasts initialized from RMM phase 7 achieve an average MHW probability of 13% after 20–25 days, also higher than those from others (e.g., 6% for Phase 3). The forecast error in SST and MHWs tends to increase with the lead time. The error primarily arises from weaker variability of atmospheric convection, leading to slower change and delayed phase transition of SST. This work highlights the essential role of intraseasonal variability in the forecast of MHWs in the SETIO.
Cui et al. (Thu,) studied this question.