1. Introduction
Weather conditions strongly influence the occurrence and spread of wildfires. Weather conducive to such fire occurrence and spread is known as “fire weather.” In this article, the weather conditions that promote rapid fire spread and the occurrence of unusually intense fire behavior is called “extreme” fire weather. Extreme fire weather occurred over several periods during the 2019/20 fire season in Australia, particularly along the eastern seaboard. Colloquially known as the “Black Summer,” the fire season was unusually long, extending from July 2019 through to early March 2020 in the state of New South Wales (NSW) alone (Bureau of Meteorology 2019b; Williamson et al. 2016). A total of 6 people died and 572 homes were reported destroyed in NSW as a result of the fires between 29 December and 5 January 2020, out of a total of 26 deaths and 2476 homes destroyed in that state over the entire season (Owens and O’Kane 2020). During the same period, Australian Navy vessels rescued approximately 1000 people trapped by the fires from the far east Victorian township of Mallacoota on 2 January 2020. Especially intense and destructive fire weather was experienced in southeast NSW and far east Victoria (Fig. 1) around New Year’s Eve (NYE) 2019, between 30 December and 4 January 2020.
Map of southeastern New South Wales showing locations discussed in the text. Identifiers with green backgrounds are states/territories: New South Wales (NSW), Victoria (VIC), and the Australian Capital Territory (ACT). Location of this region within Australia is shown in Fig. 2a. Map courtesy of Google.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
The fires burning during this period in southeastern NSW were collectively known as the South Coast complex, and included the “Badja Forest Rd, Courtegany,” “Currowan,” and “Clyde Mountain” fires. The first two ultimately exceeded 300 000 ha each, while the Clyde Mountain fire burnt more than 98 000 ha. Fire behavior at this time included unusually severe fire damage to large areas of vegetation, extreme pyroconvection including the occurrence of fire-generated tornadic vortices (Lareau et al. 2022), as well as atypically active overnight fire behavior. The Badja fire, for example, ran more than 20 km overnight on 30–31 December 2019 (Owens and O’Kane 2020). In this article, aspects of the meteorology contributing to the severity of the NSW event are examined, focusing on 30–31 December 2019.
The extremity of the 2019/20 fire season in southeastern Australia has attracted considerable research attention to date. Several groups have attempted to identify the combination of factors that resulted in the extreme intensity and extensive coverage of the fires. During the season itself, and prior to the NYE event, Nolan et al. (2020) highlighted low moisture content of dead forest fuels resulting from drought conditions as a strong contributing factor to the widespread fires then burning, including in Gondwanan rain forest areas rarely vulnerable to fire. They suggested the fire extent and severity may have been evidence of the impact of changing climate. Abram et al. (2021) noted that the combination of fire-promoting phases of Indian and Pacific Ocean climate drivers in concert with climate change drove several measures of fire severity outside the envelope of historical experience. Published early in the fire season, Lim et al. (2019) warned that weakening of the austral stratospheric polar vortex enhanced the likelihood of dry, hot conditions leading into the eastern Australian summer. Meanwhile, Levin et al. (2021) identified satellite-derived vegetation predictors highlighting vegetation communities most at risk of burning, having noted the extraordinarily high proportion of certain vegetation types that burnt during the 2019/20 fire season.
More broadly, van Oldenborgh et al. (2021) attempted to establish the extent to which anthropogenic climate change was responsible for the extremity of the fire season. They identified temperature increases as a significant factor but noted the complexity of the interaction of climate change with climate drivers influencing the fire season. Similarly, Squire et al. (2021) analyzed a large climate ensemble to establish the likelihood of the extended fire weather and drought event in southeast Australia’s current climate, given the state of critical climate drivers. They demonstrated that still more extreme events are possible even under current climate conditions. Deb et al. (2020) used statistical analysis to confirm commonly understood factors such as drought, high temperatures and low atmospheric moisture content as contributing to the likelihood of destructive fires.
Some groups have investigated consequences of the fires. Peterson et al. (2021) studied the extreme pyroconvective outbreak over the New Year period in southeastern Australia, highlighting the extent of smoke that entered and persisted in the stratosphere over several months. Khaykin et al. (2020) further focused on examining the stratospheric intrusion of smoke from the fires and the significant impacts on circulation, composition and radiative balance within the stratosphere. Collins et al. (2021) examined the extent of severe fire activity on the regions burnt in southeastern Australia during 2019/20. Still others have investigated the health effects of the smoke exposure from the fires over extended periods (Borchers Arriagada et al. 2020; Nguyen et al. 2021), and the effect of the smoke particle load on Pacific Ocean phytoplankton blooms (Tang et al. 2021). On the mesoscale, Peace et al. (2021) examined coupled fire–atmosphere modeling of several 2019/20 fires, including in New South Wales and Victoria over the New Year period. In particular, they investigated the occurrence of fire-generated vortices that developed from some fires. This brief overview is far from exhaustive and does not include, for example, any of the numerous human science examinations of societal precursors or responses to the exceptional fire events.
To date, there has been limited published documentation of synoptic meteorological influences acting on the fires in southeastern Australia, and in particular NSW, during the NYE period of 2019/20. Mills et al. (2022), however, examined several fire events in Victoria during the 2019/20 fire season, including detailed fire reconstructions. Using a high-resolution gridded reanalysis dataset, they identified situations in which overnight fires were driven by low-level jets and in which the overnight McArthur forest fire danger index (FFDI; McArthur 1967) was highest or near highest in the reanalysis dataset for those times and locations.
Here, we examine features of the synoptic weather prevailing over the NYE period and suggest that the passage of upper tropospheric troughs and associated jet streams influenced the fires. We also examine some mesoscale weather influences on fire activity: the occurrence of overnight downslope winds that caused increased fire activity during a diurnally unfavorable time, the progress of cool changes around the coastline and some way inland in far southeast New South Wales and the development of horizontal boundary rolls over fire-affected areas. Each of these phenomena affected fire activity in some way. As such, they are important to better understand the nature of the fire activity during this period. This will, in turn, reinforce the value to fire managers and supporting meteorologists of close examination of these phenomena during severe fire events. In describing these phenomena, this article supports the informal recommendation of the NSW Bushfire Inquiry to closely examine the detail of the fires during the 2019/20 fire season to derive important lessons ahead of future fires (Owens and O’Kane 2020, p. 22).
This article summarizes and extends a report of the NYE fires over southeast NSW (Bureau of Meteorology 2020a). The report was prepared at the request of the NSW State office of the Australian Bureau of Meteorology for the NSW Rural Fire Service (RFS). The RFS was the lead agency responding to the fires experienced through NSW over an approximately 8-month period, from July 2019 to early March 2020. The report, and this article, employ operationally available satellite, automatic weather station (AWS) and numerical weather prediction (NWP) data. Using these data sources better highlights the utility of the content of the article for operational meteorology and fire behavior analysis in future fire events.
The article first identifies the datasets and techniques used in this analysis, then briefly sets a climate or longer-term context for the NYE fire events, drawing on previously published work. The synoptic setting is detailed, and synoptic and mesoscale features identified above are highlighted individually, noting that in some cases they will have interacted with each other to some extent. In the discussion, the importance of the highlighted features is examined within a broader context of multiscale events, particularly in a changing climate. The conclusions make some suggestions for operational application of the material discussed, and for future work.
2. Data and methods
Operational numerical weather prediction guidance underpinned this analysis, validated against Bureau of Meteorology Automatic Weather Station (AWS) observations where available. The Bureau of Meteorology ACCESS-G model (BNOC 2016) provided global-scale NWP. ACCESS-G (since upgraded) was at the time a 25-km horizontal resolution, 70 vertical level global model based on the Met Office Unified Model. More detail on ACCESS-G configuration is available in BNOC (2016). Regional-scale (12-km horizontal resolution, hourly) data were also available from the Bureau of Meteorology ACCESS-R model, but features examined with ACCESS-G were near the edge of or slightly beyond the ACCESS-R southern domain. Subsynoptic-scale and mesoscale model analysis was undertaken using ACCESS-C2, an operational high resolution (1.5-km horizontal resolution) convection-permitting model run on a subdomain of ACCESS-R, 38.0000°–27.9965°S, 147.0000°–155.0055°E (Bureau of Meteorology 2018c; Roff et al. 2022).
Remote sensing observations were also critically important for this work, noting that many fires burnt in remote areas poorly served by in situ observations. The Japan Meteorological Agency (JMA) Himawari-8 geostationary satellite routinely provides imagery at 10-min frequency in 16 bands. For over a month during the peak of the Black Summer extended event, including during the period examined here, JMA generously made available 2.5-min rapid scan images over a southeastern Australian subdomain of the normal full disk.
The mesoscale events investigated here fall within the viewshed of the Wollongong Appin radar, although the southern extent of the fire-affected region examined is approximately 200 km from the radar, limiting the detail available from radar observations. The Wollongong radar is located at 34.264°S, 150.874°E (Fig. 1) and is a 2° beamwidth S-band Doppler radar.
Remotely sensed lightning data were available through the study period from the Weatherzone lightning detection network. Lightning occurrences were displayed as an overlay to satellite imagery, color coded in 10-min increments to denote how recently strikes had occurred. These NWP and remote sensing data were all visualized using the Bureau of Meteorology’s operational visualization platform Visual Weather, supplied by IBL Software Engineering.
Five AWSs in the study region reported data at 1-min temporal resolution: Braidwood, Moruya, Nerriga, Nowra, and Ulladulla (Fig. 1), and were used to monitor fire weather conditions and airmass changes during the study period. A running 10-min mean wind was calculated from 1-min observations to then calculate FFDI.
Several nonroutine upper air soundings were available from the Nowra naval base prior to the peak of the event studied here. The soundings were plotted using MetPy (May et al. 2022). Data from each of these sources were examined and compared to document key aspects of the meteorology during the study period and to help establish the underlying dynamics. Throughout this study, times are referred to in UTC, 11 h behind Australian daylight saving time.
3. Climate background to the NYE fires and fire weather
The period leading into NYE 2019/20 was climatologically extraordinary. Drought had gradually developed over much of eastern Australia following significant and widespread rainfall in winter (June–August) of 2016. The rainfall that continued into September 2016 erased most rainfall deficiencies that had accumulated over the previous several years in eastern Australia (Bureau of Meteorology 2017). It is likely that these wetter conditions would have encouraged vegetation recovery and growth during spring (September–November) 2016. Following this wet period, a series of hot and dry events occurred over eastern Australia during the subsequent two years (Bureau of Meteorology 2018a,b, 2019a,c), preconditioning fuels and the landscape for fire. Indeed, climatologically unusual fires occurred as a result of some of these events, well prior to the 2019/20 southern Australian fire season. The Tathra fire (Wilke et al. 2022), for example, occurred in early autumn 2018, later than most southeast NSW fires had typically occurred in the past. Extremes reported during this extended period included Australia’s driest September (2018) and second driest month on record, hottest month (January 2019, by 0.98°C above the previous record in January 2013), and hottest summer (2018/19). In particular, New South Wales experienced several hot, dry records in addition to those recorded nationally. By spring (September–November) 2019, widespread records for seasonal FFDI occurred across many parts of the country (Bureau of Meteorology 2019c). FFDI averaged over spring was highest on record.
Records continued to be broken for heat and dryness over the next several months (Bureau of Meteorology 2020b): 2019 was the hottest (Fig. 2a) and driest (Fig. 2b) year on record for Australia as a whole, with individual months and days also breaching numerous records for heat, dryness and FFDI. December 2019 was the driest December on record, and 18 December had the highest area-averaged maximum temperature across Australia, at 41.88°C. December accumulated FFDI values were record for that month across much of Australia (Fig. 2c).
Decile maps of Australia based on gridded Bureau of Meteorology data for (a) maximum temperature and (b) rainfall for 2019, and (c) accumulated daily FFDI for December 2019, calculated at 0600 UTC each day. The dark blue box in (a) indicates the location of Fig. 1. Rainfall and temperature anomalies are calculated against a standard reference period of 1961–90. FFDI anomalies are calculated against the period 1950–2019.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
Several climate drivers influenced the weather over Australia, and in particular the southeast of the country, in the months leading to NYE 2019/20. Lim et al. (2019), published during the early part of the southern Australian fire season of 2019/20, highlighted the impact a rare Antarctic sudden stratospheric warming event was likely to have on the eastern Australian fire season. The circulation anomaly influenced the occurrence of a record negative Southern Annular Mode, in turn driving hot, dry westerly winds over eastern Australia for an extended period during spring 2019. Lim et al. (2021) reviewed the importance of this sequence of events in retrospect, highlighting their predictive value for fire management purposes.
A positive Indian Ocean dipole (IOD; Ashok et al. 2003; Risbey et al. 2009) occurred during 2019 (Bureau of Meteorology 2019c), also contributing to the decline in rainfall in southeastern Australia during winter and spring months. Significantly, El Niño–Southern Oscillation (ENSO) was neutral during this time (Squire et al. 2021). As Squire et al. (2021) note, had ENSO been in its positive (El Niño) El Niño phase, conditions contributing to the southeast Australian fires could potentially have been more extreme than were observed. Under the conditions that eventuated, an unprecedented fraction of the forest estate in eastern Australia (14% of native woody vegetation, at high severity) burnt during the 2019/20 fire season (Collins et al. 2021).
4. Key observations
Several remote sensing and surface observational datasets are available to characterize conditions during the fires, as noted above. Here, we document a small fraction of these, to inform discussion of the meteorological influences on fire activity during 30–31 December 2019. Reference is made to synoptic weather features at times, particularly the cold front that focused dangerous fire weather over southeast NSW during the study period. Readers are referred to Fig. 8 for locations and movement of these features.
Time series of AWS near the fire grounds in southeast NSW indicated variable fire weather on 30 and 31 December (Fig. 3). Features of the AWS time series discussed below are referenced in Fig. 3 by numbers overlain on the plots. On 30 December, coastal or near coastal sites generally experienced onshore northeasterly, mild and moist air masses (1). Inland, however, northwesterlies prevailed on both days (2), driven by the approaching cold front discussed below. As such, inland locations such as Nerriga experienced severe fire danger (FFDI 50–74,) for an extended period in the afternoon and early evening (3). Braidwood (not shown) is also inland and at similar elevation to Nerriga, between 600 and 650 m above sea level. It experienced an earlier transition to dangerous fire weather conditions on 30 December but more variable wind speed and therefore FFDI on both 30 and 31 December. Wind eased overnight at Nerriga and temperature decreased from the high 30s (°C) to low to mid-20s (4). There was some increase in dewpoint temperature overnight and FFDI decreased. However, FFDI rose rapidly again during the morning of 31 December as the dewpoint temperature fell. Wind speed also increased ahead of the approaching cold front to above 20 kt (knots are used here, for consistency with other plots; 1 kt ≈ 0.51 m s−1) for much of the day (5).
Time series of 1-min resolution observations from (a) Moruya AWS, (b) Nerriga, and (c) Ulladulla AWS on 29–31 Dec 2019. Note that the left-hand y axis scales differently for (a) compared to (b) and (c). The figure legend is in (a) and is the same for all plots. Times are scaled in UTC, consistent with the text. Temperature (T) and dewpoint temperature (D), wind (W; kt), and FFDI are scaled on the left-hand y axis. Wind direction in degrees from north is scaled on the right-hand y axis.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
The minute-scale resolution of the AWS observations allows for detection of quasi-regular pulsing of wind speed and direction, and therefore FFDI, with duration of approximately 30–60 min during both days. Greater excursions of FFDI occurred on 31 December as a result of this pulsing, given the stronger wind (6). An abrupt decrease in FFDI occurred over a 5-min period at the end of the displayed time series (7). This resulted from the passage of the cold front and associated decrease in temperature and increase in dewpoint temperature. Wind direction shifted at the same time from northwest to southeasterly.
Observations at coastal sites indicated mild conditions during local afternoon hours (from 0100 to about 0800 UTC) on 30 December, as noted. Between approximately 2200 and 2300 UTC 30 December (the following morning), however, FFDI was severe at both Moruya (Fig. 3a), on the NSW south coast (8), and Ulladulla (Fig. 3c), some 67 km farther north (see Fig. 1 for site locations). Around 1920 UTC 30 December, onshore south to southeasterly winds at Moruya veered over several minutes to northwesterly, directing the much drier and warmer inland airstream over the coast (9). Apart from a brief period of northerly winds around 2030 UTC, wind remained northwesterly until around 0030 UTC 31 December. FFDI climbed steadily as wind speed and temperature gradually increased, and dewpoint temperature dropped to around 0°C (10). With the arrival of the cold front at Moruya just after 0000 UTC 31 December, FFDI dropped from close to 60 to 9 in less than 15 min, as temperature fell, dewpoint temperature increased, and wind direction backed onshore southeasterly (11). Notably, wind speed increased with the passage of the change, indicating a postfrontal wind surge.
At Ulladulla, a similar but subtly different sequence of events occurred (Fig. 3c). Wind direction had been north to northeasterly from early in the morning, backing northwesterly between approximately 2200 and 2300 UTC 30 December. Consequently, dewpoint temperature was higher initially, but fell to around −4°C immediately following the wind change (12). Wind speed increased with the shift to northwesterlies but oscillated around 20 km h−1. FFDI peaked at 73 in the late morning (between 0000 and 0100 UTC 31 December) with a slight surge in wind speed (13). Then, as the wind started to back farther to southwesterly, wind speed started to decline. FFDI fell abruptly as the wind backed to southerly, with temperature falling from 40°C at 0204 UTC 31 December to 21°C within 8 min, dewpoint temperature rising from −2° to 10°C and FFDI dropping from 64 to 9 in the same period (14). Again, there was a postchange increase in wind speed, persisting for approximately an hour from the south-southwest (15).
Routine upper air soundings were available from Wagga Wagga, at 2200 and 1000 UTC daily. Wagga Wagga lies some 275 km inland from the south NSW coast and is representative of the broad inland air mass. Several nonroutine soundings were also available from Nowra, near the coast and closer to the region of most pronounced fire activity during this period. At Nowra, the atmosphere had dried on the morning of the 30th in comparison with the 29th, with a near-surface dry layer evident from just above the surface to around 800 hPa (Fig. 4). The lapse rate was dry adiabatic through much of the lower troposphere from around 900 hPa to just below 500 hPa. However, there was a strong (approximately 8°C) surface inversion to 950 hPa. Winds to 500 hPa were generally light, at 15 kt or below, and broadly northerly. Wagga Wagga upper air soundings indicated substantial drying of the atmosphere on the morning of 30 December in comparison with the previous day (not shown), with winds northwesterly below 500 hPa generally 10–20 kt. A layer of drier air was evident between approximately 550 and 750 hPa, and the lapse rate was essentially dry adiabatic from near the surface to the top of the dry layer. Soundings from both locations had a thin cap of moist air near the top of the dry adiabatic layer.
Nowra upper-air sounding at 2300 UTC 28 Dec (blue) and 2300 UTC 29 Dec (red). Winds (kt) are plotted on the staff to the right. In both soundings, dewpoint temperature is plotted on the left and dry-bulb temperature is plotted on the right.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
Himawari-8 imagery was available throughout the period, including rapid-scan imagery at 2.5-min temporal resolution. Here, we examine key images that inform understanding of the meteorology discussed later. In addition, some imagery is overlaid with lightning sensor data from the Weatherzone network and with weather radar data from the Wollongong Appin radar.
Water vapor images at 6.2-μm wavelength from (Fig. 5a) 0500 and (Fig. 5b) 1900 UTC 30 December are displayed in Fig. 5. At 0500 UTC 30 December, the cloud band associated with the approaching front is visible on the left of the image, with regions of upper dry air over much of the rest of the image. By 1900 UTC on the same day, the frontal band had advanced over Tasmania with a trailing dry band extending from western Victoria southeastward across Tasmania. Also evident in both images is fire-enhanced convective activity from eastern Victorian fires. In particular, cloud top brightness temperatures were below −60°C especially on the 1900 UTC 30 December image.
Himawari-8 water vapor images at (a) 0500 and (b) 1900 UTC 30 Dec. The nonlinearly scaled color bar indicates temperatures (°C).
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
Visible imagery at 0400 UTC 31 December is displayed in Fig. 6a, together with lightning detections from the Weatherzone network during the previous hour. The same image has radar reflectivity from the Wollongong radar superposed in Fig. 6b, highlighting extensive smoke plumes and pyrocumulonimbus development. By 0530 UTC 31 December, visible imagery of a broader region of southeastern Australia (Fig. 7, with lightning detections) showed several significant features: strong pyroconvection continuing, together with extensive areas of smoke; progress of the cold frontal cloud across southeast Australia; and horizontal cloud streets inland, which had developed earlier during the day.
Himawari-8 visible image over southeastern NSW at 0400 UTC 31 Dec and Weatherzone lightning for the previous hour. (b) As in (a), but with the addition of radar reflectivity from Wollongong (Appin) radar. Radar reflectivity scale is in dBZ, with the color scale at the bottom of the figure. Lightning generated by the Currowan fire is circled in red. Lightning from thunderstorms over the Badja fire complex is visible (purple dots) to the southwest, color coded by time of occurrence. The orange dot in (a) identifies the location of Wollongong, and the green dot identifies Batemans Bay.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
Himawari-8 visible image of southeast NSW at 0530 UTC 31 Dec, including lightning detections in the previous hour, color coded by time of occurrence. The region of horizontal boundary layer rolls discussed in the text is highlighted by a blue ellipse, and an example of the numerous smoke plumes is annotated.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
5. Synoptic and mesoscale environment
Here, we describe several synoptic and mesoscale factors that contributed to the severity of the fire weather conditions around NYE, within a context of a broader climate extreme event identified by others, noted above. The features identified are not a comprehensive list of weather phenomena that occurred during the period of interest. Rather, they represent several factors that clearly affected the severity of the fire environment and difficulty of fire suppression. These factors are therefore of particular interest to operational meteorologists and fire behavior analysts, as well as in the study of factors contributing to extremes of fire weather. We examine the following:
-
upper atmospheric support for enhanced ascent and descent in the fire region, associated with vertical jet circulations;
-
overnight downslope winds associated with low-level jet development that led to increased fire spread in the early morning hours of 31 December at Batemans Bay;
-
complex progression of cold fronts over southeastern NSW on 31 December 2019; and
-
occurrence of boundary layer rolls evident from satellite imagery and independently from AWS during the days.
a. Synoptic environment
First, the broad surface synoptic features during the NYE period are identified to provide context for discussion of the upper tropospheric features. In the days prior to 29 December 2019, several weak cold fronts crossed waters south of the Australian continent (Fig. 8a), while a broad region of high pressure lay over the southern continental coast. Heat-related troughs extended from the west across the northern inland to the southeast, with their location varying day to day. By 29 December a more significant cold front approached southwestern Australia (Fig. 8b) as the ridge retreated eastward. By 0000 UTC 30 December (Fig. 8c), the front had reached the west coast of the island state of Tasmania. A prefrontal trough was by this time aligned immediately to its east. A high pressure system was located in the north Tasman Sea east of continental Australia, and a northwesterly pressure gradient began to tighten over the southeast of the continent. This was a classic pattern for dangerous fire weather for that region (Mills 2005; Sharples et al. 2016). Over the next 24 h, the front and trough moved into the far southeast of New South Wales with ridging becoming reestablished over much of southern Australia to its west (Fig. 8d). The most severe fire weather conditions during this period in southeast NSW occurred ahead of and with the passage of the cold front on 30–31 December.
Australian region mean sea level pressure charts at 0000 UTC for (a) 25, (b) 29, (c) 30, and (d) 31 Dec 2019.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
At 300 hPa, a sharp trough extended over the south of Western Australia at 1200 UTC 29 December (Fig. 9a), flanked by a jet stream peaking on the east of the trough axis at 167 kt (86 m s−1). By 0000 UTC 30 December (Fig. 9b), the trough had progressed eastward, with its axis lying over the central Great Australian Bight. It had assumed a negative tilt (axis oriented southwest–northeast) and the eastern jet core had moved southward, farther from the peak of the trough. The western flank of the jet weakened during this time. Over the next 12 h, the trough axis continued to become more negatively tilted as the trough became more open (Fig. 9c). By 0000 UTC 31 December the trough axis impinged on southeast NSW with the jet core extending from immediately south of Tasmania (Fig. 9d).
The 300-hPa Australian region charts from ACCESS-G global NWP model for (a) 1200 UTC 29 Dec, (b) 0000 UTC 30 Dec, (c) 1200 UTC 30 Dec, and (d) 0000 UTC 31 Dec. Blue solid lines represent geopotential height in geopotential decameters, red dotted lines indicate temperature, and shading denotes wind speed (kt) according to the color scale at the bottom of figure.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
The passage on 31 December of both surface pressure trough/front and negatively tilted upper tropospheric trough enhanced vertical motions already present as the northwesterly airstream flowed over the southern reaches of the Great Dividing Range (GDR). This range of mountains inland of the eastern coast is relatively low by global standards, at generally 1000–2000-m elevation, but contains several of the highest mountains in continental Australia. Airflow on the afternoon of 30 December resulted in a topographically forced ascent/descent couple. Figure 10a displays the ACCESS-R 17-h forecast of 500-hPa vertical motion, initialized at 1200 UTC 29 December and valid at 0500 UTC 30 December (1600 Australian eastern daylight saving time). Warm colors (yellow–brown–red) indicate descent while cool colors (green–blue–purple) indicate ascent. Elongated regions of ascent aligned in a northwest–southeast direction over and west of Tasmania at this time correspond to the positions of the prefrontal trough and front, respectively, with bands of descent between the front and trough and to the rear of the front. In addition, ascent/descent couples are visible over southeast NSW, even at 500 hPa, associated with the synoptically forced airflow over the Great Dividing Range. With the approach and passage of the surface front across southeast NSW and the upper-level jet to the south, the pattern of ascent and descent changed over the subsequent 24 h (Fig. 10b). Both the upward and downward branches of the vertical circulation became more intense and narrower during the afternoon of 31 December.
ACCESS-R 17-h forecasts of 500-hPa vertical motion (color bar; Pa s−1) from (a) 1200 UTC 29 Dec model run for 0500 UTC 30 Dec and (b) 1200 UTC 30 Dec model run for 0500 UTC 31 Dec. Red ellipses highlight regions of ascent/descent couples in both panels. Background fields are geopotential height (gpdm) in blue and temperature (°C) in red.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
Figure 11 displays vertical cross sections through the left entrance region of the 300-hPa jet maximum immediately east of Tasmania at 0300 UTC 31 December, from the 1200 UTC 30 December ACCESS-G run. The cross section reveals a downward-sloping region of enhanced potential vorticity extending from a depressed tropopause in the upper trough over Tasmania to the near-surface over the firegrounds, close to the southeast NSW coast (Fig. 11b). In this environment, the presence of an enhanced potential vorticity region would have further destabilized the boundary layer near the fires. Additionally, having descended over a period of several days from at least the upper troposphere, the air in the high potential vorticity region was drier than air outside that region (Fig. 11c). Wind speed in the descending air was also higher (not shown) than that of the existing near-surface environment. Both these latter characteristics would have contributed to worsening fire weather conditions in southeast NSW, and in adjoining regions of eastern Victoria.
Vertical cross section from 46°S, 146°E to 35°S, 150°E, through the left entrance region of the upper-level jet immediately east of Tasmania, at 0300 UTC 31 Dec based on the 1200 UTC 30 Dec run of ACCESS-G. (a) The location and orientation of the cross section is indicated by the arrow. Blue contours indicate geopotential height (gpdm), and red contours represent temperature (°C). Wind speed is indicated by shading with blue–green–orange denoting increasing wind speed. The jet core is highlighted in orange. (b) Potential vorticity (PVU; 1 PVU = 10−6 m2 s−2 K kg−1), with the −2-PVU isopleth indicated in bold and increasingly negative values of PV scaled in blue–violet. Topography is indicated in black in (b) and (c), and elevation (m) is indicated on the left axes. (c) Relative humidity (orange–yellow dry, green moist) and vertical motion (cm s−1), with blue isopleths indicating ascent and red indicating descent.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
ACCESS-G back trajectories (Fig. 12) reveal that, consistent with the cross-sections in Fig. 11, air arriving over far southeast NSW overnight on 30–31 December had, in part, descended from high in the troposphere to the west over previous days. For back trajectories ending between 750 and 2500 m over 37.12°S, 149.4°E at 1800 UTC 30 December, descent had occurred from between 3000 and above 6000 m over the previous 120 h. Air parcels that commenced at the highest levels in the back trajectory analysis descended in excess of 2000 m in the final 12 h of the trajectory path. This latter rapid descent was a result of mountain wave activity as the northwesterly air mass moved over the GDR. In Fig. 14a below, this can be seen as potential temperature contours east of the GDR oscillate strongly in the vertical as a result of the flow over the mountains.
HYSPLIT ACCESS-G 120-h back trajectories ending at 1800 UTC 30 Dec at the location of the black star. (top) Trajectory paths and (bottom) heights. Note that height timeline commences on the right of the panel.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
b. Overnight low-level jet
Fire agency and news media reports indicated that fire activity increased in the hinterland to the west of Batemans Bay overnight on 30 December. This was clearly against the typical cycle of diurnal activity where fires are generally most active during the afternoon. Little evidence of increased wind or other conditions likely to lead to an increase in fire activity occurs in the time series of AWS data (Fig. 3). However, smoke began to appear on the Wollongong radar to the west of Batemans Bay prior to 1600 UTC 30 December (Fig. 13) after the gradual disappearance of smoke from radar returns over the area late in the previous evening. Notably, the smoke plume was quite local, with no other areas of smoke visible on radar along the coast within the Wollongong radar viewshed. This was despite the large amount of fire activity the previous day. A sequence of ACCESS-C vertical cross sections extending west–east through Batemans Bay (including Fig. 14b) indicates strong low-level stability, as deduced from plotted potential temperature. The stability remained largely constant around the time that the smoke plume appeared. This suggests that the increased visibility of the plume was due to greater fire activity, rather than, for example, enhanced visibility of the plume due to decreased stability of its environment.
Wollongong radar 0.5° scan, valid at 1600 UTC 30 Dec. The smoke plume near Batemans Bay is circled in red. The radar reflectivity scale is in dBZ.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
Vertical cross sections through the lowest approximately 6000 m of the atmosphere at (a) 1000 and (b) 1900 UTC 30 Dec and (c) 0000 UTC 31 Dec from the 0000 UTC 30 Dec run of ACCESS-C. Cross-section position is indicated by the inset in (a). Yellow lines represent potential temperature isotherms, shading represents wind speed, with green indicating lighter winds and red/pink indicating stronger winds. Wind barbs represent wind speeds (kt).
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
The east–west cross sections of ACCESS-C wind speed through the location of Batemans Bay identify the development of a low-level jet overnight on 30 December (Fig. 14), strengthening and descending in the lee (eastern, coastal) side of the GDR to the west of Batemans Bay. At 1000 UTC 30 December (Fig. 14a), a layer of enhanced north to northwesterly wind was developing immediately above the GDR. A region of strong northerly wind already occurred offshore, impinging slightly onto the coast. However, a narrow area of lighter winds was evident between the two regions of enhanced wind speed. By 1700 UTC 30 December (Fig. 14b), the low-level jet over the GDR had increased markedly in speed and importantly had extended onto the coastal margin. Finally, by 0000 UTC 31 December (Fig. 14c), the LLJ had contracted to above the highest topography and lifted approximately 1000 m higher than it had been overnight, with a lull again visible over the coast. The coastal lull at that time lay adjacent to a narrow band of southerly wind, indicating that the cold front was surging along the southern NSW coast and starting to extend into the region of the cross section. The ACCESS-C model was unable to resolve the leeside descent of stronger winds all the way to the surface west of Batemans Bay. It does, however, suggest that enhanced fire activity overnight west of Batemans Bay which burnt into the town resulted from the overnight development of the LLJ and its downslope propagation.
Further evidence of the spatial variability in descent of stronger winds to the surface is available in Fig. 15, displaying ACCESS-C winds from the 1200 UTC 30 December run. At 1800 UTC 30 December, close to the time of the fire run into Batemans Bay, little evidence of any strong surface winds can be seen (Fig. 15a). However, immediately above the surface, at 975 hPa, ACCESS-C resolved lobes of enhanced wind speed extending from the eastern slopes of the GDR to the coast (Fig. 15b), likely resulting from downslope winds and dynamical forcing through topography in some areas. These lobes are not regularly spaced but are frequent between approximately Nowra and the far south NSW coast. In this region, an especially well-developed and strong lobe of enhanced wind speed can be seen immediately ahead of the cold front impinging into NSW at that time. Notably, one of the widest lobes extends to the coast over the vicinity of Batemans Bay, and lobes are absent from Moruya (one lies immediately to the south), Ulladulla and Nowra where AWS detected no winds of concern to operational meteorological and fire agency staff during the predawn hours on 31 December.
ACCESS-C wind fields from 1200 UTC 30 Dec run, showing (a) 10-m and (b) 950-hPa winds valid at 1800 UTC 30 Dec, and (c) 950-hPa winds valid at 0000 UTC 31 Dec. Wind speed is measured in kt, colored according to the scale on the right. The location of Batemans Bay is highlighted by a blue star in (a), for reference.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-23-0007.1
c. Complexity of frontal progression
AWS observations (Fig. 3) suggest that the cold front surged along the southern NSW coast during the day on 31 December, before gradually infiltrating inland. The observations discussed above, for example, show the change having passed through Moruya in the late morning, then approximately two hours later at Ulladulla, farther north along the NSW coast. However, the front took approximately two more hours before reaching Nerriga, some 45 km northwest and inland. Similarly, the change passed through Nowra, itself some 10 km from the coast, about 30 min prior to its passage through Nerriga, 25 km to the south but farther inland (not shown).
ACCESS-C model data are consistent with the above observations, and with research on cold fronts conducted, particularly in Australia, over the last two decades (Mills 2002, 2010; Wilke et al. 2022). Figure 15c displays the 950-hPa forecast wind at 0000 UTC 31 December from the ACCESS-C 1200 UTC 30 December run. The complexity of the frontal passage through southeast NSW and adjacent waters is clear from this single image. Strong to locally gale-force (blue in Fig. 15) southwest to southerly winds lie over far southeast NSW waters, backing southeasterly inshore as the wind is directed over the warmer landmass. The front has penetrated some distance inland, but not as far as it extends northward along the coast. At the frontal boundary, a col is evident with very light to no wind. Ahead of the cold front, strong northwest winds occur in a broad band to the west of the GDR. As noted above, however, these were substantially modulated over the coast and hinterland in the lee of the GDR, as a result of downslope winds and likely funneling in some areas.
d. Boundary layer rolls
Himawari-8 visible satellite imagery during 30 and 31 December, e.g., Fig. 7, 0530 UTC 31 December, shows the occurrence of horizontal boundary rolls, particularly inland ahead of the passage of the cold front on 31 December. The rolls are evident as cloud streets, consisting of cumulus clouds, oriented generally northwest–southeast along the direction of low-level wind flow. Rolls were evident as roughly aligned and isolated cumulus from approximately 0300 UTC 30 December west of the GDR, persisting but remaining generally isolated through the afternoon. Higher level cumulus cloud was evident across southeast NSW during the morning of 31 December, ahead of the approaching cold front. However, it was not until the afternoon that low-level cloud streets began to form, aligned with the direction of smoke plumes, as can be seen in Fig. 7.
On both 30 and 31 December at the inland site of Nerriga, very clear “pulsing” of the wind speed, and consequently FFDI, is visible (Fig. 3b). Also evident are wind direction changes associated with the wind speed pulses, which can serve to broaden a fire front. In cases described elsewhere (Engel et al. 2012; Thurston et al. 2016; Wilke et al. 2022), the pulsing results from the passage of boundary layer rolls over the AWS site. The 100- and 400-m resolution model simulations of the Tathra fire environment, for example, undertaken by Wilke et al. (2022) explicitly resolve horizontal boundary rolls. The modeled rolls moved into the region of an automatic weather station at around the time that station began recording pulsing of the weather parameters used to calculate fire danger, and near the time that cloud streets became visible on satellite imagery in the general region. These observations provided good evidence of the connection between the rolls, cloud streets and AWS pulsing. While the operational model data described here are unable to resolve such rolls, the observational AWS data point to their presence. Importantly, irrespective of the underlying mechanism causing the pulsing, its occurrence indicates potentially dangerous variability in weather conditions, with rapid changes in fire behavior likely.
Similar pulsing in AWS data can be seen at the coastal sites of Ulladulla and Moruya, but is lower frequency and less well-defined, potentially as a result of disruption of the roll structure by passage of the diurnally heated air mass over the GDR.
6. Discussion
Factors acting at several scales from the planetary to mesoscale contributed to the severity of individual days and hours of weather and fire activity during the NYE period over southeast NSW. On the planetary scale, a rare southern stratospheric warming event during winter–spring of 2019 (Lim et al. 2019) contributed to development of a negative Southern Annular Mode. This, together with a positive Indian Ocean dipole, resulted in warmer, drier weather following an already dry and warm period in eastern Australia. This period was itself at least partly attributable to climate change (van Oldenborgh et al. 2021). The passage of cold fronts, especially those associated with negatively tilted upper tropospheric troughs, directed still warmer air masses over southeastern Australia. The negatively tilted trough events included dry air with high wind speeds that had descended from relatively high in the atmosphere. Favorable topographic alignment then led to further focusing of wind in downslope wind events and in mountain wave activity, as well as funneling through mountain valleys (Sharples 2009). The development of horizontal boundary rolls introduced additional small-scale variability to fire weather conditions. As a result of this cascade of multiscale effects, precipitated by the passage of a cold front across the region, fire was able to propagate across the landscape in uncharacteristic fashion, such as its diurnally unusual run into Batemans Bay in the early hours of 31 December. The cascading of several spatial and temporal scales is increasingly recognized as important in many of the most destructive severe weather events (e.g., Kaplan et al. 2021, in the case of fire and thunderstorms).
The interaction of scales highlights the complexity of this and similar forecast situations. In such cases, operational meteorologists may find difficulty summarizing all important details, and similarly fire behavior analysts may face difficulty accounting for variability in fire behavior modeling. Additional complexity is introduced with the advent (in Australia, following the 2019/20 summer) of high-resolution ensemble NWP. The current Australian convection-allowing high resolution ensemble model, ACCESS-CE, provides a relatively small ensemble; however, time-lagging ensemble output can result in ensembles of several dozen members. The volume of data involved in characterizing likely weather outcomes can prove challenging in any attempt to provide insight into the development of especially severe events. This is particularly the case when characterizing variability for regions of complex topography. Notwithstanding such difficulties, it is important to identify ways to characterize variability in space and time and across ensembles of probabilistic outputs, given the critical importance of accurate weather forecasting for fire management, and for other high impact weather events. The difficulties also highlight the value of ingesting gridded weather data into fire behavior models at the maximum possible spatial and temporal resolution.
Interestingly, despite the descent of high PV air to low levels, there is little or no evidence of significant “jumps” in FFDI associated with rapid increases in wind speed or decreases in relative humidity during the event studied here. Such abrupt changes in fire danger values have been observed in other events where PV descent has been documented to low levels in association with wildfires (Mills 2008; Fox-Hughes 2012, 2015 and references therein, Love et al. 2021 and Georgiev et al. 2022). It is clear that the high PV air mass impinged overnight into the region of interest. In view of the strong evidence from cross sections and back trajectory plots that the air mass had descended into the region from high in the troposphere, it is likely that the injection of dry air contributed to the lack of recovery of air mass (and vegetation) moisture content overnight. Hence, the high PV air mass would have contributed to the level of fire activity both overnight and into 31 December ahead of the cooler, more humid, cold frontal air mass.
Further to the above, Batemans Bay was likely subject to enhanced fire activity due to development of a LLJ which descended close to the surface in the lee of the GDR. Enhanced winds were not, however, evident at AWS to the north or south of Batemans Bay, and other coastal townships were not as badly affected by overnight fire progress (however, inland, the Badja fire progressed rapidly during the overnight hours). The descent of strong winds along the southeast NSW coast was therefore not uniform and was likely dependent on several factors including the leeside topography and ambient stability. The presence of fire to the west of Batemans Bay would have affected stability and may have contributed to the descent of stronger wind to the surface, as documented in coupled fire–atmosphere modeling (Peace et al. 2022). ACCESS-C did detect leeside-enhanced winds near the coast during the early morning of 31 December, but these winds were not modeled to reach the surface. They therefore require careful analysis to observe, given the large amount of other guidance available during this period and the fact that they are not uniformly present across the landscape. In other events, particularly in the absence of landscape-scale fires, such winds may not reach the surface at all. This suggests that specific guidance products alerting to the potential presence of downslope winds would be useful to operational meteorological and fire agency staff (and to emergency managers in general, given the impact severe downslope winds can have, even in the absence of fire, see e.g., Abatzoglou et al. 2021).
Wilke et al. (2022) document the occurrence of boundary layer rolls in high resolution (subkilometer horizontal scale) numerical weather modeling of the Tathra fire in southeast NSW during March 2018. They note that the pulsing of FFDI, due to periodic increases in wind speed with the passage across the landscape of the rolls, likely contributed to the transport of embers across water barriers. Operational ACCESS-C NWP output, at 1.5-km horizontal resolution, is unable to resolve the scale of the rolls. As suggested in Wilke et al. (2022), however, it would be useful to be able to characterize their impact on the variability of fire weather conditions, based on broader-scale atmospheric properties resolvable by current generation numerical weather prediction models.
7. Conclusions
The 2019/20 bushfire season in Australia was record-breaking in several respects. NSW bore the highest cost of any Australian state or territory during the spring and summer, and this study has focused on aspects of the meteorology experienced during one of the most intense periods of fire weather and fire activity in NSW. While there is increasing literature on impacts and antecedent conditions relating to the fires, this paper focuses on several synoptic and mesoscale weather drivers of the fire event:
-
role of upper atmospheric structure and forcings in contributing to extreme fire behavior and pyroCb activity;
-
low-level jet driving diurnally unusual overnight fire activity;
-
complexity of the wind change through a topographically diverse region; and
-
horizontal rolls modulating fire activity through wind speed variability.
Each of these factors interacted with others, making for a temporally and spatially complex event.
It is hoped that the analysis conducted in the study will prove useful to operational meteorologists and fire behavior analysts, both in Australia and in similarly fire-prone regions globally. The complexity of the interactions of meteorological and topographic features (and those of fires themselves when they grow sufficiently large) highlights the value of being able to ingest high resolution numerical weather prediction model data directly into fire behavior and other impact models. It also points to the value of increasingly detailed in situ and remote sensing observations for fire and other severe weather monitoring.
Acknowledgments.
Alan Wain derived ACCESS-G back trajectory plots. Rob Schaap archived considerable quantities of operational guidance material over the 2019/20 Christmas and New Year holiday period, which facilitated this study. Jeff Kepert, David Wilke, Graham Mills, and Barry Hanstrum provided helpful feedback on drafts of this article. Detailed comments by anonymous reviewers were very helpful in clarifying the content of the manuscript. The author acknowledges the tireless efforts of Bureau of Meteorology colleagues and other emergency service personnel over many months during the extended 2019/20 fire season. Notwithstanding the scale of the devastation, their efforts saved many lives and much of human and natural value. The preparation of this manuscript was funded under the Australian Climate Service.
Data availability statement.
All numerical weather prediction model data used in this study are archived on the Australian National Computing Infrastructure https://nci.org.au/our-services/data-collections-management, catalogued as https://dx.doi.org/10.25914/608a99ac28fa1. Remote sensing data are similarly archived at NCI, with radar data catalogued as https://dx.doi.org/10.25914/JJWZ-0F13 and Himawari-8 data as https://dx.doi.org/10.25914/61a609adf0f7e.
REFERENCES
Abatzoglou, J. T., B. J. Hatchett, P. Fox‐Hughes, A. Gershunov, and N. J. Nauslar, 2021: Global climatology of synoptically‐forced downslope winds. Int. J. Climatol., 41, 31–50, https://doi.org/10.1002/joc.6607.
Abram, N. J., and Coauthors, 2021: Connections of climate change and variability to large and extreme forest fires in southeast Australia. Commun. Earth Environ., 2, 8, https://doi.org/10.1038/s43247-020-00065-8.
Ashok, K., Z. Guan, and T. Yamagata, 2003: Influence of the Indian Ocean dipole on the Australian winter rainfall. Geophys. Res. Lett., 30, 1821, https://doi.org/10.1029/2003GL017926.
Borchers Arriagada, N., A. J. Palmer, D. M. Bowman, G. G. Morgan, B. B. Jalaludin, and F. H. Johnston, 2020: Unprecedented smoke‐related health burden associated with the 2019–20 bushfires in eastern Australia. Med. J. Aust., 213, 282–283, https://doi.org/10.5694/mja2.50545.
Bureau National Operations Centre, 2016: APS2 upgrade to the ACCESS-G Numerical Weather Prediction System. BNOC Operations Bull. 105, 32 pp., http://www.bom.gov.au/australia/charts/bulletins/APOB105.pdf.
Bureau of Meteorology, 2017: Special Climate Statement 58—Record September rains continue wet period in much of Australia. Bureau of Meteorology, 42 pp., http://www.bom.gov.au/climate/current/statements/scs58.pdf.
Bureau of Meteorology, 2018a: Special Climate Statement 65—Persistent summer-like heat sets many April records. Bureau of Meteorology, 33 pp., http://www.bom.gov.au/climate/current/statements/scs65.pdf.
Bureau of Meteorology, 2018b: Special Climate Statement 66—An abnormally dry period in eastern Australia. Bureau of Meteorology, 31 pp., http://www.bom.gov.au/climate/current/statements/scs66.pdf.
Bureau of Meteorology, 2018c: APS2 upgrade of the ACCESS-C Numerical Weather Prediction system. NMOC Operations Bull. 114, Bureau of Meteorology, 33 pp., http://www.bom.gov.au/australia/charts/bulletins/BNOC_Operations_Bulletin_114.pdf.
Bureau of Meteorology, 2019a: Special Climate Statement 68—Widespread heatwaves during December 2018 and January 2019. Bureau of Meteorology, 70 pp., http://www.bom.gov.au/climate/current/statements/scs68.pdf.
Bureau of Meteorology, 2019b: Special Climate Statement 71—Severe fire weather conditions in southeast Queensland and northeast New South Wales in September 2019. Bureau of Meteorology, 35 pp., http://www.bom.gov.au/climate/current/statements/scs71.pdf.
Bureau of Meteorology, 2019c: Special Climate Statement 72—Dangerous bushfire weather in spring 2019. Bureau of Meteorology, 28 pp., http://www.bom.gov.au/climate/current/statements/scs72.pdf.
Bureau of Meteorology, 2020a: Meteorological report on disastrous bushfires in southeastern NSW, 30 December 2019–4 January 2020. New South Wales Rural Fire Service Rep., Bureau of Meteorology, 56 pp.
Bureau of Meteorology, 2020b: Special Climate Statement 73—Extreme heat and fire weather in December 2019 and January 2020. Bureau of Meteorology, 17 pp., http://www.bom.gov.au/climate/current/statements/scs73.pdf.
Collins, L., R. A. Bradstock, H. Clarke, M. F. Clarke, R. H. Nolan, and T. D. Penman, 2021: The 2019/2020 mega-fires exposed Australian ecosystems to an unprecedented extent of high-severity fire. Environ. Res. Lett., 16, 044029, https://doi.org/10.1088/1748-9326/abeb9e.
Deb, P., H. Moradkhani, P. Abbaszadeh, A. S. Kiem, J. Engström, D. Keellings, and A. Sharma, 2020: Causes of the widespread 2019–2020 Australian bushfire season. Earth’s Future, 8, e2020EF001671, https://doi.org/10.1029/2020EF001671.
Engel, C. B., T. P. Lane, M. J. Reeder, and M. Rezny, 2012: The meteorology of Black Saturday. Quart. J. Roy. Meteor. Soc., 139, 585–599, https://doi.org/10.1002/qj.1986.
Fox-Hughes, P., 2012: Springtime fire weather in Tasmania, Australia: Two case studies. Wea. Forecasting, 27, 379–395, https://doi.org/10.1175/WAF-D-11-00020.1.
Fox-Hughes, P., 2015: Characteristics of some days involving abrupt increases in fire danger. J. Appl. Meteor. Climatol., 54, 2353–2363, https://doi.org/10.1175/JAMC-D-15-0062.1.
Georgiev, C. G., S. A. Tjemkes, A. Karagiannidis, J. Prieto, and K. Lagouvardos, 2022: Observational analyses of dry intrusions and increased ozone concentrations in the environment of wildfires. Atmosphere, 13, 597, https://doi.org/10.3390/atmos13040597.
Kaplan, M. L., and Coauthors, 2021: The multi-scale dynamics organizing a favorable environment for convective density currents that redirected the Yarnell Hill fire. Climate, 9, 170, https://doi.org/10.3390/cli9120170.
Khaykin, S., and Coauthors, 2020: The 2019/20 Australian wildfires generated a persistent smoke-charged vortex rising up to 35 km altitude. Commun. Earth Environ., 1, 22, https://doi.org/10.1038/s43247-020-00022-5.
Lareau, N. P., N. J. Nauslar, E. Bentley, M. Roberts, S. Emmerson, B. Brong, M. Mehle, and J. Wallman, 2022: Fire-generated tornadic vortices. Bull. Amer. Meteor. Soc., 103, E1296–E1320, https://doi.org/10.1175/BAMS-D-21-0199.1.
Levin, N., M. Yebra, and S. Phinn, 2021: Unveiling the factors responsible for Australia’s Black Summer fires of 2019/2020. Fire, 4, 58, https://doi.org/10.3390/fire4030058.
Lim, E.-P., H. H. Hendon, G. Boschat, D. Hudson, D. W. J. Thompson, A. J. Dowdy, and J. M. Arblaster, 2019: Australian hot and dry extremes induced by weakenings of the stratospheric polar vortex. Nat. Geosci., 12, 896–901, https://doi.org/10.1038/s41561-019-0456-x.
Lim, E.-P., and Coauthors, 2021: The 2019 Southern Hemisphere stratospheric polar vortex weakening and its impacts. Bull. Amer. Meteor. Soc., 102, E1150–E1171, https://doi.org/10.1175/BAMS-D-20-0112.1.
Love, P. T., P. Fox-Hughes, T. A. Remenyi, N. Earl, D. Rollins, G. Mocatta, and R. Harris, 2021: A characterisation of synoptic weather features often associated with extreme events in southeast Australia. Stage 1–Common features of recent events. University of Tasmania Tech. Rep. 679.2021, 107 pp., https://www.bnhcrc.com.au/sites/default/files/managed/downloads/a_characterisation_of_synoptic_weather_features_final_report.pdf.
May, R. M., and Coauthors, 2022: MetPy: A Meteorological Python library for data analysis and visualization. Bull. Amer. Meteor. Soc., 103, E2273–E2284, https://doi.org/10.1175/BAMS-D-21-0125.1.
McArthur, A. G., 1967: Fire behaviour in eucalypt forests. Forestry and Timber Bureau Leaflet 107, 36 pp., https://catalogue.nla.gov.au/catalog/2275488.
Mills, G. A., 2002: A case of coastal interaction with a cool change. Aust. Meteor. Mag., 51, 203–221.
Mills, G. A., 2005: A re-examination of the synoptic and mesoscale meteorology of Ash Wednesday 1983. Aust. Meteor. Mag., 54, 35–55.
Mills, G. A., 2008: Abrupt surface drying and fire weather Part 1: Overview and case study of the South Australian fires of 11 January 2005. Aust. Meteor. Mag., 57, 299–309.
Mills, G. A., 2010: A westward propagating roll cloud and cool change on Tasmania’s North Coast. Aust. Meteor. Oceanogr. J., 60, 237–247, https://doi.org/10.22499/2.6004.002.
Mills, G. A., O. Salkin, M. Fearon, S. Harris, T. Brown, and H. Reinbold, 2022: Meteorological drivers of the eastern Victorian Black Summer (2019–2020) fires. J. South. Hemisphere Earth Syst. Sci., 72, 139–163, https://doi.org/10.1071/ES22011.
Nguyen, H. D., and Coauthors, 2021: The summer 2019–2020 wildfires in east coast Australia and their impacts on air quality and health in New South Wales, Australia. Int. J. Environ. Res. Public Health, 18, 3538, https://doi.org/10.3390/ijerph18073538.
Nolan, R. H., M. M. Boer, L. Collins, V. Resco de Dios, H. G. Clarke, M. Jenkins, B. Kenny, and R. A. Bradstock, 2020: Causes and consequences of eastern Australia’s 2019–20 season of mega-fires. Global Change Biol., 26, 1039–1041, https://doi.org/10.1111/gcb.14987.
Owens, D., and M. O’Kane, 2020: Final report of the NSW Bushfire Inquiry. 466 pp., https://www.dpc.nsw.gov.au/assets/dpc-nsw-gov-au/publications/NSW-Bushfire-Inquiry-1630/Final-Report-of-the-NSW-Bushfire-Inquiry.pdf.
Peace, M., and Coauthors, 2021: Coupled fire-atmosphere simulations of five Black Summer fires using the ACCESS-Fire model. Black Summer Final Rep. 105.2021, 74 pp., https://www.bnhcrc.com.au/sites/default/files/managed/downloads/coupled_fire-atmosphere_simulations_black_summer_final_report.pdf.
Peace, M., J. Greenslade, H. Ye, and J. D. Kepert, 2022: Simulations of the Waroona fire using the coupled atmosphere–fire model ACCESS-Fire. J. South. Hemisphere Earth Syst. Sci., 72, 126–138, https://doi.org/10.1071/ES22013.
Peterson, D. A., and Coauthors, 2021: Australia’s Black Summer pyrocumulonimbus super outbreak reveals potential for increasingly extreme stratospheric smoke events. npj Climate Atmos. Sci., 4, 38, https://doi.org/10.1038/s41612-021-00192-9.
Risbey, J. S., M. J. Pook, P. C. McIntosh, M. C. Wheeler, and H. H. Hendon, 2009: On the remote drivers of rainfall variability in Australia. Mon. Wea. Rev., 137, 3233–3253, https://doi.org/10.1175/2009MWR2861.1.
Roff, G., and Coauthors, 2022: APS2-ACCESS-C2: The first Australian operational NWP convection-permitting model. J. South. Hemisphere Earth Syst. Sci., 72, 1–18, https://doi.org/10.1071/ES21013.
Sharples, J. J., 2009: An overview of mountain meteorological effects relevant to fire behaviour and bushfire risk. Int. J. Wildland Fire, 18, 737–754, https://doi.org/10.1071/WF08041.
Sharples, J. J., and Coauthors, 2016: Natural hazards in Australia: Extreme bushfire. Climatic Change, 139, 85–99, https://doi.org/10.1007/s10584-016-1811-1.
Squire, D. T., and Coauthors, 2021: Likelihood of unprecedented drought and fire weather during Australia’s 2019 megafires. npj Climate Atmos. Sci., 4, 64, https://doi.org/10.1038/s41612-021-00220-8.
Tang, W., and Coauthors, 2021: Widespread phytoplankton blooms triggered by 2019–2020 Australian wildfires. Nature, 597, 370–375, https://doi.org/10.1038/s41586-021-03805-8.
Thurston, W., R. J. Fawcett, K. J. Tory, and J. D. Kepert, 2016: Simulating boundary‐layer rolls with a numerical weather prediction model. Quart. J. Roy. Meteor. Soc., 142, 211–223, https://doi.org/10.1002/qj.2646.
van Oldenborgh, G. J., and Coauthors, 2021: Attribution of the Australian bushfire risk to anthropogenic climate change. Nat. Hazards Earth Syst. Sci., 21, 941–960, https://doi.org/10.5194/nhess-21-941-2021.
Wilke, D. J., J. D. Kepert, and K. J. Tory, 2022: The meteorology of the Tathra bushfire. Wea. Forecasting, 37, 581–600, https://doi.org/10.1175/WAF-D-21-0084.1.
Williamson, G. J., L. D. Prior, W. M. Jolly, M. A. Cochrane, B. P. Murphy, and D. M. Bowman, 2016: Measurement of inter- and intra-annual variability of landscape fire activity at a continental scale: The Australian case. Environ. Res. Lett., 11, 035003, https://doi.org/10.1088/1748-9326/11/3/035003.
