Atmospheric water vapour transport in ACCESS‐S2 and the potential for enhancing skill of subseasonal forecasts of precipitation

Extended warning of above‐average and extreme precipitation is valuable to a wide range of stakeholders. However, the sporadic nature of precipitation makes it difficult to forecast skilfully beyond one week. Subseasonal forecasting is a growing area of science that aims to predict average weather conditions multiple weeks in advance using dynamical models. Building on recent work in this area, we test the hypothesis that using large‐scale horizontal moisture transport as a predictor for precipitation may increase the forecast skill of the above‐median and high‐precipitation weeks on subseasonal time‐scales. We analysed retrospective forecast (hindcast) sets from the Australian Bureau of Meteorology's latest operational subseasonal‐to‐seasonal forecasting model, ACCESS‐S2, to compare the forecast skill of precipitation using integrated water vapour transport (IVT) as a proxy, compared to using precipitation forecasts directly. We show that ACCESS‐S2 precipitation generally produces more skilful forecasts, except over some regions where IVT could be a useful additional diagnostic for warning of heavy precipitation events. ACCESS‐S2 simulates integrated water vapour transport (IVT) with more skill than precipitation on subseasonal time‐scales. IVT is highly correlated with precipitation. Therefore, we assess whether IVT can be used as a proxy for precipitation and improve forecast skill at longer lead ‐times (weeks 2–4). We find enhanced skill of precipitation where we have used IVT as a proxy rather than modelled precipitation in some regions. Shown are the Critical Success Index for forecasting weeks with above 50th percentile precipitation in (a) December–January–February (DJF) and (c) June–July–August (JJA), and above 80th percentile precipitation in (b) DJF and (d) JJA using ACCESS‐S2 IVT as a predictor for precipitation. Predictions are verified against the precipitation observed in the Global Precipitation Climatology Project. Week n indicates the forecast lead time.