Global ocean surface and subsurface temperature forecast skill over subseasonal to seasonal timescales

Abstract

Subseasonal to seasonal forecasts of ocean temperatures, including extreme events such as marine heatwaves, have demonstrated utility in informing operational decision-making by marine end users and managing climate risk. Verification is critical for effective communication and uptake of forecast information, together with understanding ocean temperature predictability. The forecast skill of surface and subsurface ocean temperature forecasts from the Bureau of Meteorology’s new ACCESS-S2 seasonal prediction system are assessed here over an extended 38-year hindcast period, from 2 weeks to 6 months into the future. Forecasts of sea surface temperature (SST), heat content down to 300 m (HC300), bottom temperatures on continental shelves, and mixed layer depth are compared to both satellite observations and ocean reanalyses for the globe and the Australian region, using a variety of skill metrics. ACCESS-S2 demonstrates increased SST skill over its predecessor ACCESS-S1 at subseasonal timescales for all variables assessed. Heat content skill is particularly high in the tropics but reduced in subtropical regions especially when compared to persistence. Forecast skill for ocean temperature is higher in the austral summer months than winter at lead times up to 2 months in the Western Pacific region. Mixed layer depth is poorly predicted at all lead times, with only limited areas of skill around Australia and in the south-west Pacific region. Probability of exceedance forecasts for the 90th percentile as an indicator for marine heatwave conditions, shows adequate skill for SST, HC300 and bottom temperatures, especially near shelf regions at shorter lead times. This work will underpin the future development of an operational marine heatwave forecast service, which will provide early warning of these events and thus valuable preparation windows for marine stakeholders.