Abstract
<p>In late January to early February 2019, wide-spread flooding, strong winds and relatively cold temperatures over the north-eastern Australian state of Queensland led to the loss of an estimated 625,000 cattle and 48,000 sheep. The system that caused these impacts was a quasi-stationary monsoon depression that lasted close to 10 days, bringing weekly rainfall totals above 1000 mm in some locations, maximum temperatures 8–12°C below average, and sustained wind speeds of 30-40 km/h. The same weather event caused inundation and damage to more than 3000 homes over the eastern Queensland coastal city of Townsville with an insurance cost of over $1.2 billion AUD (https://www.afr.com/companies/financial-services/insurers-reveal-townsville-flood-cost-warn-region-is-unprofitable-20190804-p52do5). Observations and reanalysis confirm that an active Madden-Julian Oscillation pulse stalled over the western Pacific during the period of the flooding. To the south, a blocking anticyclone over the northern Tasman Sea promoted onshore easterly flow, and with it, the relatively low apparent temperatures (Cowan et al. 2019).</p><p>In the days before the event, the Australian Bureau of Meteorology issued the monthly rainfall outlook for February which provided little indication of the upcoming extreme event. At the time of the event, there was a 50% chance of an El Niño developing during the boreal spring, meaning a tendency towards warmer and drier conditions across the northeast. Here we show that forecasts from the Bureau's newly developed dynamical subseasonal-to-seasonal (S2S) prediction system – Australian Community Climate Earth-System Simulator Seasonal version 1 (ACCESS-S1) – of the weekly-averaged conditions were more skilful. The ACCESS-S1 99-member ensemble forecast a more than doubling of the probability of extreme weekly rainfall totals a week prior to the floods, along with increased probabilities of extremely low maximum temperatures and high winds. Ensemble-mean weekly rainfall amounts, however, were considerably underestimated by ACCESS-S1, even in forecasts initialised at the start of the peak flooding week. This is consistent with other state-of-the-art dynamical S2S prediction systems. Yet one individual ensemble member of ACCESS-S1 managed to forecast close to 85% of the rainfall magnitude across the most heavily impacted region of northwest Queensland in a week 2 forecast. This suggests current S2S prediction systems like ACCESS-S1 are capable at getting close to predicting record-breaking events with at least one week's lead-time. It also appears that accurate prediction beyond two weeks (i.e., a week 3 forecast) of an event like the northern Queensland floods is more difficult to achieve.</p><p>Reference:</p><p>Cowan et al. (2019): Forecasting the extreme rainfall, low temperatures, and strong winds associated with the northern Queensland floods of February 2019, Weather and Climate Extremes, <strong>26</strong>, 100232, https://doi.org/10.1016/j.wace.2019.100232.</p>
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have