Seasonal forecasting of subsurface marine heatwaves

Frequency, intensity and duration of subsurface marine heatwaves can be predicted at seasonal timescales and with greater skill than surface events, suggest analyses with an operational seasonal prediction system. Marine heatwaves damage marine ecosystems and services, with effects identified mostly below the ocean surface. To create a truly user-relevant detection system, it is necessary to provide subsurface forecasts. Here, we demonstrate the feasibility of seasonal forecasting of subsurface marine heatwaves by using upper ocean heat content. We validate surface and subsurface events forecast by an operational dynamical seasonal forecasting system against satellite observations and an ocean reanalysis, respectively. We show that indicators of summer events (number of days, strongest intensity, and number of events) are predicted with greater skill than surface equivalents across much of the global ocean. We identify regions which do not display significant surface skill but could still benefit from accurate subsurface early warning tools (e.g., the mid-latitudes). The dynamical system used here outperforms a persistence model and is not widely influenced by warming trends, demonstrating the ability of the system to capture relevant subseasonal variability.