The Bureau of Meteorology (BoM) produces weather forecasts on a 24/7 basis for the Australia region. During the fire season provide direct and timely information for the fire and emergency services community.
What will FIRE-DST do that will add to the weather information already available?
The Bureau’s Numerical Weather Prediction (NWP) system will be configured for extremely high resolution applications over regions of complex topography in severe fire weather conditions. This will entail a considerable tuning and validation exercise as the resolutions envisioned will be an order of magnitude finer than hitherto available operationally. The validation of the expected detailed and complex flow structures, both in space and evolving in time, will be a challenging task.
The Bureau has a new computer-based weather prediction system called the Australian Community Climate and Earth-System Simulator (ACCESS). It is based upon the UK Met Office Unified Model system (www.accessimulator.org.au). It uses improved technology and is much more accurate and reliable.
FIRE-DST involves the Bureau configuring ACCESS for very high resolution forecasts, with application to severe fire weather conditions in rugged terrain. Modelling at this scale really stretch the computers (3000 processors being utilised for recent severe weather simulations), but it is necessary to be able to understand the complex winds, temperatures and humidity that occur during events like Black Saturday.
Once BoM have validated the modelled weather against case-study observations, partners in this project will use the data as an input to their fire behaviour models – the aim is to improve the prediction of the spread, intensity and impact of a fire.
The FIRE-DST project also plans to look at some alternate scenarios – “what if the wind change arrived an hour earlier or later?” It will be some years before computers are powerful enough to run these very high resolution simulations in real time. Within FIRE-DST we plan to also use these simulations to better understand and interpret the resolutions that we can afford to run operationally today.
Once the model configuration has been established, then two or three case studies for recent major fire events will be constructed, and the resulting wind fields offered to the fire behaviour models to assess their sensitivity to the more complex (topographically-affected) information. An important part of the assessment will include measures of uncertainty in the forecasts – what happens if a wind change arrives an hour earlier or later, what happens if the inversion breaks an hour earlier or later, what happens if the wind direction is 10, 20 or 30 degrees different relative to the topography? These uncertainties will be investigated using simple methods, in which the model output fields are modified. These modified output fields, when coupled to the fire spread model, will provide some measure of uncertainty in terms of the range of possible fire spread due to the possible range of meteorological conditions.