%A Drosdowsky,Wasyl %A Wheeler,Matthew C. %D 2017 %J Frontiers in Earth Science %C %F %G English %K Australian monsoon,Monsoon prediction,tropical prediction,Intraseasonal,Extended-range,Subseasonal,POAMA,dynamical prediction system %Q %R 10.3389/feart.2017.00028 %W %L %M %P %7 %8 2017-April-04 %9 Original Research %+ Dr Matthew C. Wheeler,Research and Development, Bureau of Meteorology,Melbourne, VIC, Australia,matthew.wheeler@bom.gov.au %# %! Extended Prediction of Australian Monsoon %* %< %T Extended-Range Ensemble Predictions of Convection in the North Australian Monsoon Region %U https://www.frontiersin.org/articles/10.3389/feart.2017.00028 %V 5 %0 JOURNAL ARTICLE %@ 2296-6463 %X Extended-range (<35 day) predictions of area-averaged convection over northern Australia are investigated with the Bureau of Meteorology's Predictive Ocean-Atmosphere Model for Australia (POAMA). Hindcasts from 1980-2011 are used, initialized on the 1st, 11th, and 21st of each month, with a 33-member ensemble. The measure of convection is outgoing longwave radiation (OLR) averaged over the box 120°E-150°E, 5°S-17.5°S. This averaging serves to focus on the intraseasonal and longer time scales, and is an area of interest to users. The raw hindcasts of daily OLR show a strong systematic adjustment away from their initial values during the first week, and then converge to a mean seasonal cycle of similar amplitude and phase to observations. Hence, forecast OLR anomalies are formed by removing the model's own seasonal cycle of OLR, which is a function of start time and lead time, a usual practice for dynamical seasonal prediction. Over all hindcasts, the model forecast root-mean-square (RMS) error is smaller than the RMS error of persistence and climatological reference forecasts for leads 3–35 days. Ensemble spread is less than the forecast RMS error (i.e., under-spread) for days 1–12, but slightly greater than the RMS error for longer leads. Binning the individual forecasts based on ensemble spread shows a generally positive relationship between spread and error. Therefore, greater certainty can be given for forecasts with smaller spread.