@article {bnh-8318, title = {Improved predictions of severe weather to reduce community impact {\textendash} final project report}, number = {721}, year = {2022}, month = {02/2022}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

Extreme weather often occurs at relatively small scales.\  Accurate forecasts and understanding of such small-scale processes require high-resolution modelling.\  Forecasts are especially useful in severe weather events, since they play an essential role in allowing communities, industry and emergency services to prepare for and mitigate the impacts. Because forecasts are inherently uncertain in the severity, location and duration of an event, preparation needs to be more widespread than the eventual impact {\textendash} but this over-preparation comes at a cost.\  Detailed prediction of the probabilities of severe impacts would avoid the risk of failing to alert areas with the chance of an impact, while minimising the cost of over-warning.\ 

This project has studied the dynamics, predictability, and processes of severe weather, including fire weather, with the purpose of understanding phenomena with severe impact, improving forecasts of severe weather, and better depicting forecast uncertainty in these events. These goals help facilitate better risk management, improve user preparation, reduce adverse outcomes, and enable more effective mitigation.

The project has featured two main strands.

The first strand comprised case studies of severe weather events. For these, we have combined high-resolution numerical weather prediction (NWP) with a wide range of in situ and remotely sensed observations, to better understand both the high-impact event in question, and other events of that class. All except one of the studies used a version of the Bureau{\textquoteright}s operational NWP system, ACCESS. The events comprised two severe mid-latitude systems (an east coast low, and a severe thunderstorm and tornado outbreak), two severe fires, and a tropical cyclone, thereby covering the gamut of severe weather in Australia. Two of the studies featured ensemble simulation, with the east coast low case being the first time that ensemble ACCESS had been run at this resolution within the Bureau and foreshadowing the Bureau{\textquoteright}s new operational ensemble capability. Each case discovered important fine-scale features that contributed to the severity of the event, advancing our knowledge base and ability to respond.

The second strand studied two important phenomena associated with bushfire plumes: the formation of pyrocumulus clouds and ember transport. In each case, we began with idealised, high-resolution simulations of plumes using a large-eddy model. We used this technology, in which the model is run on a 50-m grid, to capture the most energetic size range of the turbulent eddies that are prominent in fire plumes. We found that the mean travel distance of firebrands depends mainly on wind speed and fire intensity, but the spread in the landing positions shifts from being substantially crosswind at light winds, to dominantly along-wind at high winds. This spread is greatly increased by the turbulence in the plume, and the maximum spotting distance can be more than double the mean for this reason.

We also used our plume modelling to study pyrocumulus clouds and analysed the processes that lead to pyrocumulus, with special attention on the relative importance of moisture from two sources, the atmosphere and combustion, and showed that the latter is negligible except in very dry environments. This somewhat controversial result has been confirmed by a conceptual study of the thermodynamics of pyrocumulus formation.

These initial studies laid a firm theoretical framework for our subsequent development of tools to provide predictions in a form, and at a speed, that is suitable for operational use.

For moist pyroconvection, we developed a paradigm to combine the necessary meteorological and fire information, the pyrocumulonimbus firepower threshold, or PFT. The PFT is defined as the power output from a fire at which pyroCb will begin to form and depends solely on meteorological parameters. Fires hotter than the PFT will initiate pyroCb, according to this paradigm, and cooler fires do not. We also developed a way of computing an approximate PFT from either NWP or observed data and tested this in real time during the extraordinarily severe 2019-20 fire season. The results of that trial were outstandingly successful, with nearly all the 30-odd events being captured, and the forecast guidance for non-events was also reported to have been extremely valuable. The Bushfire and Natural Hazards CRC have provided us with utilisation funding to take this tool another step closer to operational use. The tool has also attracted significant international interest, particularly given the recent spate of events in the USA. We have begun a utilisation project to take the PFT work further, detailed in this report.

For ember transport, we developed a simple model to predict plume-dominated transport, incorporating well-known plume modelling concepts, a model of plume turbulence, and a probabilistic model of ember transport within and beneath the plume. The results agree well with our large eddy-based simulations. This model was coupled to CSIRO{\textquoteright}s Spark fire-spread model and tested on the Kilmore East fire of Black Saturday with excellent results. The combination with Spark will make this tool suitable for future operational use, and the ember transport model could also, of course, be coupled to other fire spread simulators, including the coupled fire-atmosphere model ACCESS-Fire. Working with CSIRO and AFAC, we are putting the necessary things in place to be able to convert the prototype implementation of the ember transport parameterisation in Spark, into an operationally robust and supported system. This will both facilitate further research, and provide the framework to allow users to learn, understand and apply the system.

The value of case studies has been reinforced by our being asked to undertake five case studies of severe fires from the 2019-20 summer. This work will be done jointly with our colleagues from the CRC Coupled fire-atmosphere modelling project, who further developed the ACCESS-Fire model. We expect that this work will provide an impetus to further develop and streamline that capability.

We are also very excited about the potential of the PFT and ember transport tools. While both are relatively early in their development, with the PFT being the more mature, each is presently producing very encouraging results. We are also very conscious that each is raising further scientific questions as they address others, and that both are at the beginning of a steep development and refinement curve that we expect will further improve their accuracy and utility. We are confident that further scientific investment will yield substantial dividends.

}, keywords = {community, Impact, prediction, severe, weather}, issn = {721}, author = {Jeffrey Kepert and KJ Tory and Ching, Eng and Robert Fawcett and Serena Schroeter and W. Thurston and David Wilke and Dragana Zovko-Rajak} } @article {bnh-3924, title = {The contribution of turbulent plume dynamics to long-range spotting}, journal = {International Journal of Wildland Fire}, volume = {26}, year = {2017}, month = {03/2017}, pages = {317-330}, chapter = {317}, abstract = {

Spotting can start fires up to tens of kilometres ahead of the primary fire front, causing rapid spread and placing immense pressure on suppression resources. Here, we investigate the dynamics of the buoyant plume generated by the fire and its ability to transport firebrands. We couple large-eddy simulations of bushfire plumes with a firebrand transport model to assess the effects of turbulent plume dynamics on firebrand trajectories. We show that plume dynamics have a marked effect on the maximum spotting distance and determine the amount of lateral and longitudinal spread in firebrand landing position. In-plume turbulence causes much of this spread and can increase the maximum spotting distance by a factor of more than 2 over that in a plume without turbulence in our experiments. The substantial impact of plume dynamics on the spotting process implies that fire spread models should include parametrisations of turbulent plume dynamics to improve their accuracy and physical realism.

}, doi = {10.1071/WF16142}, url = {http://www.publish.csiro.au/wf/WF16142}, author = {W. Thurston and Jeffrey Kepert and KJ Tory and Robert Fawcett} } @article {bnh-4232, title = {Improved predictions of severe weather to reduce community impact: midterm report 2014-17}, number = {315}, year = {2017}, month = {09/2017}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

This report marks the end of the initially-approved 3{\textonehalf} years of the project. To\ mark that milestone, this annual report is much longer than previous annual\ reports, for it is also the final report of that initial part of the project, and as such\ contains detailed descriptions of the six main activities in the first 3{\textonehalf} years.

We have studied the dynamics, predictability and processes of severe weather, including fire weather, with the purpose of improving forecasts of severe weather and better depicting forecast uncertainty in these events, thereby facilitating better risk management and more cost-effective mitigation.

Two of the six main chapters of this report relate to our work with large-eddy modelling of turbulent plumes. We have used this technology, in which the model is run on a 50-m grid to capture the most energetic size range of the turbulent eddies, to both simulate ember transport and to model pyrocumulus formation. The mean travel distance of firebrands depends mainly on wind speed and fire intensity, but the spread in the landing positions shifts from being substantially cross-wind at light winds, to dominantly along-wind at high winds. This spread is greatly increased by the turbulence in the plume, and the maximum spotting distance can be more than double the mean for this reason.\ 

We have also used our plume modelling to study pyrocumulus clouds. We have analyzed the processes that lead to pyrocumulus, with special attention on the relative importance of moisture from two sources, the atmosphere and combustion, and shown that the latter is negligible except in very dry environments. This somewhat controversial result has been confirmed by a conceptual study of the thermodynamics of pyrocumulus formation.

We have prepared three detailed studies of severe weather events. East coast lows are intense low-pressure systems that form over the sea adjacent to the east coast of Australia, most commonly along the New South Wales coast. We analyzed the 20-23 April 2015 event using, for the first time, an ensemble of 24 simulations rather than just a single forecast. The use of an ensemble allows us to better discern the degree of risk, and to account for the inherent uncertainty in any forecast. It also enables insight into the processes that lead to the rapid intensification of these systems.\ 

The Blue Mountains fires of October 2013 were most damaging on the 17th. This was expected to be a day of high fire risk, but the extreme fire spread was not anticipated and the causes were unknown. Our high resolution simulations showed that the downward extension of high upper-level winds to the vicinity of the fire ground, caused by mountain wave activity, was a factor. A dry slot {\textendash} that is, a long, relatively narrow band of dry air {\textendash} which moved over the fire, further contributed to the conditions.

Lastly, we have analyzed a simulation of a secondary eyewall formation and eyewall replacement cycle in a tropical cyclone, yielding better understanding of the underlying processes and the important factors in predicting these developments. Eyewall replacement cycles are associated with marked expansions of the tropical cyclone wind field, leading to a wider damage swath, earlier onset of damaging winds, and increased storm surge and wave damage.\ 

We are pleased that the project will continue for another three years. Our focus will shift to a greater emphasis on utilization activities during this period. In particular, we aim to develop simple methods of calculating ember transport and pyrocumulus development, so as to transfer the knowledge we have developed in these areas into operations. We will also continue to study severe weather events in detail.

}, issn = {315}, author = {Jeffrey Kepert and KJ Tory and W. Thurston and Dragana Zovko-Rajak and Ching, Eng and Robert Fawcett} } @conference {bnh-3883, title = {Thermodynamic considerations of pyrocumulus formation}, booktitle = {AFAC17}, year = {2017}, month = {09/2017}, publisher = {Bushfire and Natural Hazards CRC}, organization = {Bushfire and Natural Hazards CRC}, address = {Sydney}, abstract = {

In favourable atmospheric conditions, large hot fires can produce pyrocumulonimbus (pyroCb) cloud in the form of deep convective columns resembling conventional thunderstorms, which may be accompanied by strong inflow, dangerous downbursts and lightning strikes.\  These in turn may enhance fire spread rates and fire intensity, cause sudden changes in fire spread direction, and the lightning may ignite additional fires.\  Dangerous pyroCb conditions are not well understood and are very difficult to forecast.\ 

Here, a conceptual study of the thermodynamics of fire plumes is presented to better understand the influence of a range of factors on plume condensation.\  Recognising that plume gases are undilute at the fire source and approach 100\% dilution at the plume top (neutral buoyancy), we consider how the plume condensation height changes for this full range of dilution and for a given set of factors that include: environmental temperature and humidity, fire temperature, and fire moisture to heat ratios.\  The condensation heights are calculated and plotted as saturation point (SP) curves on thermodynamic diagrams for a broad range of each factor.\  The distribution of SP curves on thermodynamic diagrams provides useful insight into pyroCb behaviour. Adding plume temperature traces from Large-Eddy Model simulations to the thermodynamic diagrams provides additional insight into plume buoyancy, how it varies with height, and the potential for dangerous pyroCb development.

}, author = {KJ Tory and W. Thurston and Jeffrey Kepert} } @conference {bnh-2944, title = {The effects of turbulent plume dynamics on long-range spotting}, booktitle = {AFAC16}, year = {2016}, month = {08/2016}, publisher = {Bushfire and Natural Hazards CRC}, organization = {Bushfire and Natural Hazards CRC}, address = {Brisbane}, abstract = {

Spotting is a hazardous phenomenon which leads to unpredictable fire behaviour and accelerated fire spread. Spot fires occur when embers are launched by bushfire plumes into the background wind, which then carries the embers a significant distance from the fire front. If the embers land in a suitable fuel bed and are still burning a spot fire may be ignited. The magnitude of the problem is illustrated by Cruz et al. (2012), who provide evidence of long-range spotting in excess of 30 km during the Black Saturday bushfires of February 2009. Therefore a better understanding of the processes that contribute to long-range spotting is essential for the prediction of fire spread. In this study we aim to assess the contribution of turbulent plume dynamics to the process of long-range spotting.

}, author = {W. Thurston and KJ Tory and Robert Fawcett and Jeffrey Kepert} } @article {bnh-2974, title = {Improved predictions of severe weather to reduce community impact: Annual project report 2015-2016}, number = {169}, year = {2016}, month = {05/09/2016}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

We aim to study the dynamics, predictability and processes of severe weather, including fire weather. We seek also to improve forecasts of severe weather, and to better depict forecast uncertainty in these events, thereby facilitating better risk management and more cost-effective mitigation. Our research into ember transport in smoke plumes and the meteorology of the Blue Mountains bushfires of October 2013 has reached maturity, and journal articles have been written. We continue to develop our work on pyrocumulus clouds, and the east coast low event of April 2015, and have commenced a study of an eyewall replacement cycle in a tropical cyclone. These studies span a wide range of time and space scales and require a range of different methods.

Our ember transport work has confirmed that the mean travel distance of firebrands for a given fire intensity depends mainly on wind speed. However, the spread in the landing positions shifts from being substantially cross-wind at light winds, to dominantly along-wind at high winds. This spread is greatly increased by the turbulence in the plume, and the maximum spotting distance can be more than double the mean for this reason. These sophisticated and computationally intensive calculations can be used to inform the development of physically realistic and computationally cheap parameterizations of ember transport for use in fire models.

We have also used our plume modelling to study pyrocumulus clouds. Intense fire plumes in suitably moist environments can lead to cloud development, with the possibility of strong downbursts {\textendash} one of our simulations is shown on the cover of this report. We have analyzed the processes that lead to pyrocumulus, with special attention on the relative importance of moisture from two sources, the atmosphere and combustion, and shown that the latter is close to negligible. We aim to use the knowledge gained to develop a forecast tool for pyrocumulus formation.

Although the Blue Mountains fires of October 2013 persisted for several weeks, much of the spread occurred on the 17th. While this was expected to be a day of high fire risk, the extreme fire spread was not anticipated and the causes were unknown. Our high resolution simulations showed that the downward extension of high upper-level winds to the vicinity of the fire ground, caused by mountain wave activity, was a factor. In addition, the marked wind change on that day was associated with a dry slot, but the underlying cause of that dry slot seems to be different to previously documented cases.

East coast lows are intense low-pressure systems that form over the sea adjacent to the east coast of Australia, most commonly along the New South Wales coast. We continue our analysis of the event of 20-23 April 2015, using, for the first time, an ensemble of 24 simulations rather than just a single forecast. Collectively, these simulations accurately predict the position and intensity of the low, the strong winds and the rainfall. The differences between them give insight as to the forecast uncertainty, the overall envelope of areas at some risk, and those at highest risk. The ensemble also enables insight into the processes that lead to the rapid intensification of these systems. The risk these systems pose was further illustrated by the two events in June of 2016.

}, issn = {169}, author = {Jeffrey Kepert and KJ Tory and W. Thurston and Simon Ching and Dragana Zovko-Rajak and Robert Fawcett} } @conference {bnh-4112, title = {Large-eddy simulations of pyro-convection and its sensitivity to mositure}, booktitle = {5th International Fire Behaviour and Fuels Conference}, year = {2016}, month = {04/2016}, publisher = {International Association of Wildland Fire}, organization = {International Association of Wildland Fire}, address = {Melbourne}, abstract = {

Intense heating of air in the vicinity of a bushfire leads to deep ascent. If this ascent is deep enough to lift air above the lifting condensation level, cumulus or cumulonimbus clouds form in a process known as moist pyro-convection. There is abundant anecdotal evidence to suggest that pyro-convective clouds may have a significant impact on fire behaviour by (i) amplifying burn and spread rates (Fromm et al., 2012); (ii) enhancing spotting through plume intensification (Koo et al., 2010; Thurston et al., 2015); and (iii) igniting new fires via pyrocumulonimbus lightning, noting that pyrocumulonimbus lightning conditions favour hotter and longer-lived lightning strikes (Rudlosky and Fuelberg, 2011). Pyro-convective clouds are also responsible for the transport of smoke and other aerosols into the stratosphere, resulting in hemisphere-scale smoke transport and substantial climate impacts (Fromm et al., 2010). Therefore a knowledge of the processes that lead to the generation of moist pyro-convection is important for understanding and predicting fire behaviour, as well as the potential climatic influences of large fires. In this study we use large-eddy simulations to investigate the potential for the generation of moist pyro-convection by bushfire plumes. Firstly we perform simulations over a range of fire intensities and environmental moisture levels. Secondly we repeat a subset of these simulations with differing amounts of moisture released by the fire in order to assess the relative roles of environmental moisture and fire-derived moisture in the formation of pyro-convective clouds.

}, author = {W. Thurston and KJ Tory and Robert Fawcett and Jeffrey Kepert} } @conference {bnh-4113, title = {Long-range spotting by bushfire plumes: The effects of plume dynamics and turbulence on firebrand trajectory}, booktitle = {5th International Fire Behaviour and Fuels Conference}, year = {2016}, month = {04/2016}, publisher = {International Association of Wildland Fire}, organization = {International Association of Wildland Fire}, address = {Melbourne}, abstract = {

Spotting is a hazardous phenomenon that leads to unpredictable fire behaviour and accelerated fire spread. Spot fires occur when embers are launched by bushfire plumes into the background wind, which then carries the embers a significant distance from the fire front. If the embers land in a suitable fuel bed and are still smouldering a spot fire may be ignited. The magnitude of the problem is illustrated by Cruz et al. (2012), who provide evidence of long-range spotting in excess of 30 km during the Black Saturday bushfires of February 2009. Therefore a better understanding of the processes that contribute to long-range spotting is essential for the prediction of fire spread. In this paper we aim to assess the contribution of turbulent plume dynamics to the process of long-range spotting.

}, author = {W. Thurston and KJ Tory and Robert Fawcett and Jeffrey Kepert} } @conference {bnh-4114, title = {Mesoscale features related to the Blue Mountains fires of 17 October 2013 revealed by high resolution Numerical Weather Prediction (NWP) modelling}, booktitle = {5th International Fire Behaviour and Fuels Conference}, year = {2016}, month = {04/2016}, publisher = {International Association of Wildland Fire}, organization = {International Association of Wildland Fire}, address = {Melbourne}, author = {Simon Ching and Robert Fawcett and W. Thurston and KJ Tory and Jeffrey Kepert} } @article {bnh-3205, title = {Pyrocumulonimbus forecasting: needs and issues}, number = {239}, year = {2016}, month = {11/2016}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

Pyrocumulonimbus events can substantially change the weather characteristics in the vicinity of fires, which may drastically affect fire behaviour.\  This is of very considerable concern for fire managers, and therefore for fire weather forecasters.\  In particular, the wind around a fire can become very erratic in the presence of pyroCb, with downburst winds a greater risk than at other times.\  Lightning from pyroCb can ignite additional fires, and in extreme cases pyroCb may generate tornadoes (such as occurred during the 2003 Canberra fire).

Forecasters and fire managers need to be aware of the possibility of pyroCb development to allow for the chance of such erratic fire behaviour.\  The environments that support pyroCb development, then, are an important topic of study for the Bushfire and Natural Hazards CRC, and the results of this research project will be of great interest to all involved in the planning for and management of dangerous wildfires.

}, issn = {239}, author = {KJ Tory and Mika Peace and W. Thurston} } @conference {bnh-1614, title = {The Effects of Fire-Plume Dynamics on the Lateral and Longitudinal Spread of Long-Range Spotting Conference Paper 2014}, booktitle = {Bushfire and Natural Hazards CRC and AFAC Wellington Conference 2014}, year = {2015}, month = {02/2015}, abstract = {

The lofting of firebrands from bushfires into a background atmospheric flow can lead to spotting downwind\ of the fire front. Spotting is a hazardous phenomenon because it leads to both unpredictable\ and accelerated fire spread, as winds aloft are often in a different direction from and faster than the\ near-surface winds. Here we use a two-stage modelling process to address some of the uncertainty\ associated with spotting, by quantifying the lateral and longitudinal spread in the landing location of\ potential firebrands and how this spread is affected by the dynamics of the fire plume.

}, author = {W. Thurston and KJ Tory and Jeffrey Kepert and Robert Fawcett} } @article {bnh-2386, title = {Improved predictions of severe weather to reduce community impact: Annual project report 2014-2015}, number = {150}, year = {2015}, month = {11/2015}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

We aim to study the dynamics, predictability and processes of severe weather, including fire weather. We seek also to improve forecasts of severe weather, and to better depict forecast uncertainty in these events, thereby facilitating better risk management and more cost-effective mitigation. So far, we have studied ember transport in smoke plumes, pyrocumulus clouds, the meteorology of the Blue Mountains bushfires of October 2013, and the east coast low event of April 2015, four studies which span a wide range of time and space scales and require a range of different methods.

We are building on our existing modelling of bushfire plumes by using it to study ember transport.\  Embers are added near the base of the plumes and their trajectories calculated from the model winds. \ At higher wind speeds the embers travel further downstream than at lower wind speeds, as expected.\  However, the lateral spread is much broader for lower wind speeds. \ Understanding the spread in landing positions will facilitate the development of computationally affordable and physically realistic means of calculating the expected spotfire distribution in fire spread models.

We have further extended our plume modelling to begin a study of pyrocumulus development. Intense fire plumes in suitably moist environments can lead to cloud development, with intense updrafts and if rain develops the possibility of strong downbursts.\  Such pyro-convection may lead to enhanced and unpredictable fire spread, increased ember transport and spotting, and further ignitions from pyrocumulonimbus lightning. \ Our simulations produce realistic clouds, including the formation of rain and strong downdrafts. \ We will now examine some more cases, with the eventual aim of providing a preliminary forecast tool for pyrocumulus formation.

Although the Blue Mountains fires of October 2013 persisted for several weeks, much of the spread occurred on a single day. While this was expected to be a day of high fire risk, the extreme fire spread was unpredicted and the causes unknown. \ Our high resolution simulations helped identify the downward extension of high winds aloft in the vicinity of the fire ground, due to mountain wave activity. \ In addition, the marked wind change on that day was associated with a dry slot, known to worsen fire behaviour due to extremely low humidity.

East coast lows are intense low-pressure systems that form close to the east coast of Australia, most commonly along the New South Wales coast. They can produce severe wind, wave and flood impacts as in the event of 20-23 April 2015, which we are studying. For the first time, we are conducting this study with an ensemble of 24 simulations, rather than just a single forecast.\  Each simulation begins from a slightly different initial state, giving 24 different, but plausible, forecasts that represent a range of possible outcomes. Collectively, these simulations accurately predict the position and intensity of the low, the strong winds and the rainfall. \ The differences between them will be analysed to determine how predictable aspects of the event were.

}, issn = {150}, author = {Jeffrey Kepert and KJ Tory and W. Thurston and Simon Ching and Robert Fawcett} } @article {bnh-2404, title = {Improved predictions of severe weather to reduce community risk}, year = {2015}, author = {Jeffrey Kepert and W. Thurston and Simon Ching and KJ Tory and Robert Fawcett} } @conference {bnh-2081, title = {Large-eddy simulations of pyro-convection and its sensitivity to environmental conditions - peer viewed}, booktitle = {Adelaide Conference 2015}, year = {2015}, address = {Adelaide, Australia}, abstract = {

Research proceedings from the Bushfire and Natural Hazards CRC \& AFAC Conference in Adelaide, 1-3 September 2015.\ 

}, author = {W. Thurston and KJ Tory and Robert Fawcett and Jeffrey Kepert} } @conference {bnh-2203, title = {Modelling the Fire Weather of the Coonabarabran Fire of 13 January 2013}, booktitle = {Bushfire and Natural Hazards CRC and AFAC Wellington Conference 2014}, year = {2015}, abstract = {

We will exhibit state-of-the-art high-resolution numerical weather prediction simulations and radar imagery for Sunday 13 January 2013, with a specific focus on the region of the Coonabarabran fire which started at around 1600 Eastern Daylight Time (EDT) on 12 January in the Warrumbungle National Park. The simulations show a complicated range of meteorology including weather features that would affect fire behaviour critical for fire-fighter safety.

Features such as convection outflow gust fronts are displayed in the simulations in the north-westerly wind ahead of the main wind change, together with boundary-layer rolls, and sea-breeze-like wind changes proceeding inland from the coast. In addition, small-scale vortices are modelled on the main change: these lead to hazardous local spikes in the modelled Forest Fire Danger Index. Exceptionally strong north/south temperature gradients were observed over inland New South Wales on the Sunday and these are also seen in the simulations.

Sunday 13 January brought difficult conditions for fire fighting. When the fire was declared {\textquotedblleft}out{\textquotedblright} on 24 January, it had burnt an area of 55,210 ha west of Coonabarabran, 53 homes, 131 other buildings and 95\% of the Warrumbungle National Park.

The simulation has been performed using the Australian Community Climate and Earth-System Simulator (ACCESS), and involves a sequence of nested limited area model runs embedded in the ACCESS global model run, with a finest grid spacing of 550 m. Our analysis will focus on how well the simulations capture the meteorological factors that promote extreme fire behaviour. The ACCESS model is used at the Bureau of Meteorology for operational numerical weather prediction, but is used here in research mode at resolutions much finer than current operational ones.

}, author = {Robert Fawcett and Yeo, Claire and W. Thurston and Jeffrey Kepert and KJ Tory} } @article {bnh-1882, title = {Pyrocumulonimbus: A Literature Review}, year = {2015}, month = {06/2015}, abstract = {

A pyrocumulus cloud is a dense cumuliform cloud associated with fire or volcanic activity (although here we report only on fire pyrocumulus).\  It is produced by intense heating of air, which leads to deep ascent and subsequent condensation when the rising air becomes saturated due to cooling from adiabatic expansion.\  The condensation is evident in cloud formation.\  The process is similar to conventional convective cloud formation, when a lifting mechanism (e.g., orographic lifting, intersection of two air masses) raises air beyond where the cloud forms (the lifting condensation level) to where the additional condensational heating makes the air positively buoyant (the level of free convection).\  Turbulent entrainment of cooler and drier air from outside the rising airmass dilutes the cloud buoyancy, which can limit the size and growth of the cloud (e.g., fair weather cumulus).\  At the opposite extreme, larger and more intense lifted regions can accelerate to the tropopause. As they cross the tropopause into the much more stable air in the stratosphere, they become cooler than the ambient air, and hence negatively buoyant. However, they may possess enough momentum that they overshoot the level of neutral buoyancy\  Outflowing air at the tropopause gives these cumulonimbus clouds their classic anvil shape (a nimbus cloud is a cloud that produces precipitation).\  Evaporation of moisture by entrained dry air in these clouds leads to cooling and descent and the release of previously suspended precipitation, which can result in heavy downpours and intense downburst winds.

}, author = {KJ Tory and W. Thurston} } @article {bnh-1963, title = {Simulating boundary-layer rolls with a numerical weather prediction model}, journal = {Quarterly Journal of the Royal Meteorological Society}, year = {2015}, pages = {1-14}, chapter = {1}, abstract = {

Boundary-layer rolls have an impact on bushfires, pollution dispersion, the triggering of extreme convective weather events, the air{\textendash}sea interaction and intensification of tropical cyclones and momentum, particle and gas fluxes. Previous numerical modelling studies that investigate the effects of boundarylayer rolls and the processes that control their dynamics have relied upon high-resolution, idealised simulations. Recently, however, numerical weather prediction (NWP) systems have increased in horizontal resolution to the point at which they are capable of explicitly resolving shallow convective circulations. Therefore, there is a need to assess the ability of NWP to accurately forecast the formation, development and breakup of boundary-layer rolls. Here, we assess the ability of the UK Met Office Unified Model (UM), the NWP component of the Australian Community Climate and Earth-System Simulator (ACCESS), to simulate the life cycle of boundary-layer rolls. A suite of simulations is performed, over a range of horizontal resolutions, of boundary-layer rolls observed across Victoria, Australia, on Black Saturday, 7th February 2009. In this case, it is found that a horizontal grid spacing of less than 0.6 km is required to adequately reproduce the scale and evolution of the observed rolls. We note that the boundary layer was particularly deep on this day and as a consequence the boundary-layer rolls were wider than may be typically observed, meaning the grid spacing specified here lies at the upper end of what would be required with a shallower boundary layer. The model output is used to assess how the boundary-layer rolls contribute to the high fire danger on Black Saturday. It is suggested that boundary-layer rolls may increase fire danger by (i) causing wind-direction variability at the surface, which increases the rate of spread of fires; and (ii) enhancing the process of long-range spotting by augmenting the vertical lofting of embers.

}, keywords = {Shallow convection; mesoscale modelling; high resolution NWP; terra incognita; Black Saturday bushfires; atmospheric boundary layer; wind direction variability}, url = {http://www.cawcr.gov.au/staff/jdk/Kepert_papers/Thurston_etal_2015_qjrms_acc.pdf}, author = {W. Thurston and Robert Fawcett and KJ Tory and Jeffrey Kepert} } @proceedings {BF-4343, title = {Applications of Very High Resolution Atmospheric Modelling for Bushfires}, year = {2013}, url = {http://www.bushfirecrc.com/resources/research-report/applications-very-high-resolution-atmospheric-modelling-bushfires}, author = {Jeffrey Kepert and Robert Fawcett and KJ Tory and W. Thurston} } @article {BF-4282, title = {A comparison of the fire weather characteristics of the Melbourne dust storm (1983) and Black Saturday (2009): a high-resolution ACCESS case study}, journal = {20th International Congress on Modelling and Simulation}, year = {2013}, month = {12/2013}, pages = {167-172}, chapter = {167}, abstract = {We present the results of high-resolution simulations of the fire weather over Melbourne during its notorious dust storm on 8 February 1983. The simulations were performed using the Australian Community Climate and Earth-System Simulator (ACCESS), and involved a sequence of nested model runs starting with a global model run initialised with an ERA-Interim reanalysis initial condition. The ACCESS model is used at the Bureau of Meteorology for operational numerical weather prediction, but is used here in research mode at resolutions much finer than those currently used operationally. The day of the dust storm saw the passage of a significant cool change across Victoria, with many similarities in the simulations to that of Black Saturday (7 February 2009), against which we compare it. Wind changes such as these two can have a significant (and dangerous) impact on the behaviour of bushfires in southeast Australia, and their prediction forms an important component of fire weather prediction in this part of the country. While the wind change on 8 February 1983 was accompanied by some fire activity, that activity was much less than the major fire activity which accompanied the significant wind change eight days later on Ash Wednesday (16 February 1983). Why the dust storm wind change was not associated with severe fires, while those on Ash Wednesday and Black Saturday were, is a challenging question. Catastrophic fires are quite rare, whereas significant fire weather events that could potentially lead to catastrophic fires (if the non-meteorological prerequisites were to fall into place) are much more common and their study can lead to further understanding of the fire events.}, keywords = {Ash Wednesday, Black Saturday, Fire weather, Melbourne{\textquoteright}s dust storm, numerical weather prediction}, url = {http://www.mssanz.org.au/modsim2013/A3/fawcett.pdf}, author = {Robert Fawcett and Alan Wain and W. Thurston and Jeffrey Kepert and KJ Tory} } @proceedings {BF-4339, title = {The Eyre Peninsula Fire of 11 January 2005: an ACCESS case study}, year = {2013}, url = {http://www.bushfirecrc.com/resources/research-report/eyre-peninsula-fire-11-january-2005-access-case-study}, author = {Robert Fawcett and W. Thurston and Jeffrey Kepert and KJ Tory} } @conference {bnh-4115, title = {Large-eddy simulations of bushfire plumes in the turbulent atmospheric boundary layer}, booktitle = {20th International Congress on Modelling and Simulation}, year = {2013}, month = {12/2013}, address = {Adelaide}, abstract = {

Spot fires are a hazardous phenomenon which can lead to unpredictable fire behaviour and accelerated fire spread. Spot fires occur when firebrands are lofted into strong ambient winds and ignite new fires downwind. Anecdotal evidence suggests that this lofting and transport of firebrands can be responsible for the ignition of spot fires at large distances, up to tens of kilometres ahead of the fire front. A thorough knowledge of the potential for lofting from a fire is therefore desirable in order to accurately predict the fires rate of spread and coverage.

The extent to which firebrands are lofted and transported away from a fire is largely determined by both the intensity of the fire convective column and the strength of the ambient winds. Previous work on the response of fire plumes to background winds has principally relied on theoretical or highly idealised numerical models. Here we use high-resolution three-dimensional numerical simulations, performed with the UK Met Office Large-Eddy Model, to investigate the behaviour of bushfire plumes. We begin by simulating the dry, neutral atmospheric boundary layer for a range of wind speeds. Simulations are run to a quasi-steady state, ensuring that the flow displays realistic turbulence properties. Plumes are then produced by imposing a localised positive surface heat flux anomaly at the model surface. The sensitivity of the size, shape and intensity of the plume{\textquoteright}s updraft to the interaction between the plume and the turbulent atmospheric boundary layer is explored, with reference to the potential for spotting.

}, author = {W. Thurston and KJ Tory and Robert Fawcett and Jeffrey Kepert} }