An initialised event attribution system was developed at the Bureau of Meteorology using POAMA2, which was used as the operational seasonal forecast system prior to ACCESS-S (Wang et al. 2016). The method was based on the comparison of two sets of forecasts of a particular extreme sub-seasonal event. One forecast was initialised with observed initial conditions as per the operational system and with CO₂ set to the current value. The second forecast was initialised with altered initial conditions, where an estimate of the influence of increasing greenhouse gases over the last 60 years had been removed from the initial conditions. The aim is to forecast the event in a “low CO₂ world". The comparison of these two forecasts was taken to be an estimate of the influence of anthropogenic global warming on the magnitude of the event.

We are now upgrading the event attribution system to use the ACCESS-S framework. The major advantage of an ACCESS-based attribution system, besides being a major upgrade in the model, is the potential for the seamless prediction of extremes on both weather and climate scales. We produce an estimate of the influence of anthropogenic forcing from a difference between the current climate and the climate in earlier years (1860-1950), which represents a period before major impacts from anthropogenic forcing. This pre-industrial mean state is obtained using long simulations of the UKMO GC2 model from CMIP5.  The "low CO₂ world" initial conditions are generated by subtracting the mean difference between the present and past conditions from the initial conditions. Here we will present the method and early results using a case study from the hindcast period. Once developed, this system could sit as part of the sub-seasonal forecast product suite to inform operational staff on the causes of extreme events in the outlook period.