ISSUE 54 MARCH 2010 COld-frONTAL BUSHFIRE WINDS AND COMPUTER FORECAST MODELS SUMMARY This photo was taken This research has developed a new model from a Firebird 303 JetRanger helicopter to better forecast cold fronts – and outcome over Lawloit, Victoria, that will help better protect communities and just after a recent firefighters. The work was completed through wildfire was hit by extensive study that has led to a better the downburst from a understanding of the complex phenomenon thunderstorm passing of cold fronts. This research has focused on by to the south. The downburst from the understanding two things. Firstly, the physical thunderstorm turned processes that cause cold-frontal wind the northerly flank changes to be what they are – why sometimes into a headfire - with they are abrupt, why sometimes more gradual, the typical pattern why some have sustained strong winds after from a line of fire of the change (Ash Wednesday 1983) and some about 5/6 headfire and 1/6 backing do not (Black Friday 1939). Secondly, how fire. The change in to verify forecasts of wind change timing at fire behaviour was observation sites and from computer forecast very dramatic, and models. increased risks to ground crews who BACKGROUND were positioned to Wind change matters to fire managers. It can the north of the fire. change bushfire activity in a moment, shifting Photo supplied by the flank of the fire to the fire front, suddenly Steve Grant, DSE Fire Management Officer. putting firefighters and communities at risk. A change in wind can also change the rate of fire spread, increase the quantity, distance and direction of downstream spotting, and change the safety status of residents and townships in a flash. There have been several notable ABOUT THIS PROJECT disasters associated with wind changes, Project A2.1: Fire Weather – Fire Danger is part including: of Bushfire CRC Program A: Safe Prevention, • the Mann Gulch Fire in Montana in 1949 Preparation and Suppression • the Ash Wednesday fires across south- Author: Dr Graham Mills (right) is a Bushfire CRC eastern Australia in 1983 Project Leader and Senior Researcher at the Bureau of Meteorology • the 1994 South Canyon fire in Colorado that killed 14 firefighters Acknowledgments: The research team includes Dr Graham Mills (project leader), Xin-Mei Huang • the 1998 Linton bushfires in Victoria, (Bushfire CRC researcher 2004-06), Yimin Ma where five volunteer firefighters (Bushfire CRC researcher 2007-09), and project perished in their vehicle after being collaborator Kevin Parkyn (Head, Severe Weather caught out by a sudden wind change Forecast Section, Victoria). • the fires in Ku-ring-gai Chase For more information about this research, visit the National Park in 2000 that led to Research page at www.bushfirecrc.com or contact the deaths of four National Parks Dr Graham Mills at [email protected] and Wildlife Service staff and the serious injury of three others when © BUSHFIRE CRC LTD 2010 1 Figure 1: A wind change forecast chart. a routine hazard reduction burn time that, as far as possible, match the Applying the model went wrong changes interpreted by the meteorologist’s Very high resolution computer model • the 2005 fires on the Eyre Peninsula verification. This turned out to be a forecasts (mesoscale numerical weather in South Australia that killed nine challenge, as wind changes differ widely in prediction (NWP) models) have been members of the public caught out their intensity, even on the same day. The developed at the Bureau of Meteorology and by sudden changes in the bushfire research developed a Wind Change Rate provide vital guidance to forecasters. behaviour Index (WCRI), which represents the rate of These models predict the wind speed and change of wind direction with time. • the recent Black Saturday fires in direction, temperature, humidity, and Victoria in February 2009 However, because a wind change is more pressure at every gridpoint (a regular lattice • the deaths of several Spanish firefighters significant as the wind speed becomes approximately every five kilometres) every in July 2009. stronger, the wind speed is included in the hour. If these forecasts are considered as a BUSHFIRE CRC RESEARCH calculation of the WCRI. As wind speeds can time series of observations, then a WCRI can be strong before and after the change, and as be calculated at every gridpoint every hour. Verifying wind change forecast charts there is often a lull at the change time, these This has two applications: Due to the importance of wind change on fire decisions were highly complex. By fine tuning • The time of maximum wind change can behaviour, the Bureau of Meteorology offices the algorithm it was able to estimate times in several Australian states have for many be estimated at a gridpoint closest to that compared favourably with the times an observation site and the numerical seasons issued Wind Change Forecast charts estimated by the experienced meteorologists. (Figure 1). Since 2000 the Bureau’s Victorian forecasts of wind change time verified. Regional Forecast Centre has manually One of the by-products of this algorithm is • A spatial field of WCRI can be verified its wind change forecast charts at that not only is a time of maximum wind generated, and animated in time to selected stations at the end of the fire season. change produced, but the algorithm also provide a powerful visual display of the This verification requires an experienced estimates a change start time and a change wind change and the way it develops, as meteorologist who uses the time series of end time, which suggests that the concept of a it is forecast to move across a region. observations from the verification location, wind change period may have some merit. Such a spatial field of WCRI is illustrated in together with a series of detailed spatial When subjective estimates from experienced Figures 2A and 2B, which show a simulation analyses (‘weather maps’) to estimate a single meteorologists and objective times from the of the cool change on Ash Wednesday 1983. wind change time, which is then compared algorithm were compared, it showed that In these panels the barbs show the direction with the forecasts. This process is demanding over four Victorian fire seasons the subjective and speed of the 10m wind, while the shading and time-consuming. and objective times were within half an hour shows the WCRI, with increasingly dark It was the aim of this research project on greater than 60 percent of occasions, and colours indicating a stronger wind change to develop algorithms to automate this within 2.5 hours on greater than 85 percent of in terms of change in direction and higher process – from the sequence of half-hourly occasions. The subjective time was within the speeds. The first panel shows the simulation observations at a site on a ‘wind change objective wind change period on greater than for 4pm EDST (3:30pm CDST), with the day’, to determine a single wind change 95 percent of occasions. frontal wind change approaching Adelaide, 2 Figure 2A & 2B: Ash Wednesday cool change simulation, according to the Wind Change Rate Index. and just beginning to intensify along the RESEARCH OUTCOMES/ADOPTION western Victorian coastline. The second panel END USER STATEMENT The project team has run the software at the shows the situation five hours later with a “Sudden wind changes and associated end of the past two fire seasons (2007-08 and much more intense change approaching increase in bushfire behaviour has 2008-09) and provided objective verification Melbourne. data to the Victorian Regional Forecast contributed to the deaths or serious Centre. These have been used in its end-of- When forecasters use NWP model guidance long-term injury to those living in season forecast verification. they have to decide whether to “go” with the bushfire prone environments as well guidance, or whether to make some allowance as firefighters. The scientific research The spatial WCRI forecasts have been made for possible error. Knowing the biases and available to forecasters since the 2006-07 fire undertaken in this project provides the error characteristics of the model forecasts season, and form important part of the fire scientists and fire managers with an helps this decision. The verifications of this weather guidance. The project has received research show that the issued forecasts are understanding of this form of complex enthusiastic endorsement of this form of largely unbiased with time (neither too slow weather phenomena influencing fire forecast guidance and the product has been nor too fast on average), and that more than behaviour. The research developed a included in the Bureau’s new Graphical 60 percent are within 2.5 hours, for forecast Wind Change Rate Index (WCRI). The Forecast Editor system that was implemented lead times between nine and 36 hours. accurate forecasting of wind changes in the Victorian Regional Forecast Centre in late 2008. Cool changes and bad fire days together with the WCRI will provide A further aspect of the study was to fire managers with critical information A software package has been developed that understand how cool changes in disastrous for the planning and execution of will allow the verification of the fire weather fire event days differentiate from cool the most appropriate strategy and forecaster’s wind change forecast to be run by the Bureau’s regional offices. This package, changes on more regular fire weather days. tactical resourcing for firefighter and which runs on the Bureau’s workstations, The re-examination of the Ash Wednesday community safety.” fire event revealed that a characteristic of the takes a tabulated set of forecast wind change most extreme days, and those on which the – Russell Rees, Chief Officer, Country times at a set of observing stations, extracts majority of bushfire deaths in south-eastern Fire Authority (Victoria) observational data at those stations from the Australia in the past 45 years have occurred, Bureau’s archives, extracts numerical weather prediction model data from another archive, is the strength of the front at around 1.5 km A diagnostic tool that allows the strength and computer verification statistics for both above the land surface.