The underlying large-scale dynamical processes responsible for the development of heat waves in Victoria, southeastern Australia, in summer are presented here. Heat waves are defined as periods of at least three days and two nights for which daily maximum and minimum temperatures exceed the 90th percentile for a particular location and month, using a station daily temperature dataset. Composites of upper-level potential vorticity anomalies from the Interim ECMWF Re-Analysis (ERA-Interim) reveal that heat waves in southeastern Australia are associated with propagating Rossby waves, which grow in amplitude and eventually overturn. The process of overturning generates an upper-level anticyclone over southern Australia and an upper-level trough to the northeast, with maximum amplitudes near the tropopause. The northerly flow associated with the circulation around the surface anticyclone advects hot air from the continental interior over the southeast of Australia, leading to extreme surface temperatures. Composite rainfall shows that precipitation is enhanced in the vicinity of the upper-level trough over northeastern Australia, consistent with adiabatically forced vertical motion, destabilization of the atmosphere, and modified moisture fluxes. Heat waves in the southeast are frequently accompanied by heavy rainfall over the northeast of the continent and adjacent ocean.
Subtropical anticyclones are an important feature of the Southern Hemisphere general circulation (e.g., Taljaard and van Loon 1963; Taljaard 1967; Sinclair 1996). The weather of southern Australia can be characterized as a series of migratory anticyclones, separated by cold fronts or low pressure systems (e.g., Taljaard and van Loon 1963). Although in the winter the seasonal mean of these transitory high pressure systems results in a broad anticyclone centered on the continent, in summer the mean sea level pressure is dominated by a heat low in the northwest of Australia, with a much weaker anticyclone over the remainder of the continent. A strong anticyclone over the ocean to the west and a weaker, more variable anticyclone to the east of the continent are persistent year-round features (Hurrell et al. 1998).
Numerous studies in the Northern Hemisphere (NH) have demonstrated the connection between persistent blocking anticyclones and heat waves (see, e.g., Galarneau et al. 2012, and references therein). There is no formal definition of a blocking anticyclone, and the term is used here to refer to a persistent or slow-moving transient anticyclone. Heat waves are periods of sustained above-average daily maximum and minimum temperatures; elevated overnight temperatures mean there is no relief from the hot daytime conditions. Heat waves are high-impact weather events and are a common natural hazard across Australia, affecting many sectors of the community, industry, and the natural environment. It has been suggested that heat waves have been responsible for more fatalities in Australia over the past two centuries (over 4000) than natural hazards such as tropical cyclones, floods, bushfires, and lightning strikes (Coates 1996). Heat waves in southeastern Australia are frequently accompanied by heavy rainfall over the northeast of the continent and adjacent waters.
In southeastern Australia, heat waves in summer are associated with the pattern of a strong, slow-moving transient anticyclone followed by a cold front (Reeder and Smith 1987). In the lower troposphere, the circulation around the anticyclone advects warm air from the hot continental interior in a northerly or northwesterly flow over southeastern Australia, which is then followed by strong southerly or southwesterly flow during the passage of a summertime cold front (Reeder and Smith 1987, 1992; Engel et al. 2013). Although this synoptic situation is well known, and is also evident on days of high fire danger in southeastern Australia, the fundamental dynamical atmospheric processes that cause the development of heat waves in this region are less well understood.
Potential vorticity (PV) has long been recognized as an important tool for the diagnosis of dynamical processes in the atmosphere. Hoskins et al. (1985) demonstrated that both cutoff cyclones and blocking anticyclones are formed by advection leading to the cutting off of cyclonic and anticyclonic upper-level PV anomalies respectively, and that upper-level PV anomalies may be associated with surface weather systems. A number of climatological studies have linked atmospheric blocking and synoptic-scale Rossby waves (see, e.g., Altenhoff et al. 2008, and references therein). In essence, a block may be viewed as an anticyclonic PV anomaly at upper levels, which forms either as a result of the differential isentropic advection of anomalously low PV air from lower latitudes and/or the diabatically driven cross-isentropic flow of air from lower altitudes. Enhanced precipitation has also been associated with the movement and location of an upper-level PV trough, through the forcing of vertical motion, moisture transport, and the destabilization of the lower troposphere (Massacand et al. 2001; Funatsu and Waugh 2008; Martius et al. 2013).
The aims of this paper are to document the mean synoptic and PV-theta (Θ) structure of heat waves in the southeastern Australian state of Victoria in summer, and to investigate the role of breaking Rossby waves in the formation of the slow-moving transient anticyclones that lead to the development of heat waves in this region. The link between heat waves in southeastern Australia and heavy rainfall over northeastern Australia and the adjacent waters is also discussed.
The research presented here is divided into six sections. The data used and the methodology employed are discussed in section 2, and section 3 defines heat waves in Victoria. Section 4 presents the composites of the synoptic and PV-Θ conditions for summer heat waves in Victoria. Section 5 examines these same conditions on 29 January 2009, during the heat wave prior to the catastrophic Black Saturday bushfires in Victoria on 7 February 2009. The main results and conclusions are discussed in section 6.
2. Methodology and data
Data for the December–February (DJF) summer season are extracted from the Interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim, herein ERAI; Dee et al. 2011) for the 21-yr period from January 1989 to December 2009. These data are available at 6-hourly intervals, on a variety of isobaric and isentropic levels, at 1.5-degree horizontal resolution. Seasonal means are calculated for all DJF months from 1989 to 2009, and anomalies are calculated as the deviation from the seasonal mean.
The high-quality daily temperature (HQDT) dataset for Australia (Trewin 2001) provides corrected daily maximum and minimum temperatures for 99 nonurban stations and four stations located at major cities. The 11 stations in the southeastern state of Victoria are reasonably uniformly located throughout the state (see Table 1). The dataset has been utilized by the Australian Bureau of Meteorology (BoM) both in studies of extreme temperature events and in an operational seasonal temperature prediction scheme, as well as for temperature monitoring at monthly and seasonal time scales.
The Global Precipitation Climatology Project (GPCP) provides precipitation data for the period from 1 October 1996 to 31 August 2009 (Huffman et al. 2001). The dataset used is the GPCP One-Degree Daily Data Set (1DD dataset) version 1.1, created at the Laboratory for Atmospheres, National Aeronautics and Space Administration (NASA) Goddard Space Flight Center. Day-of-the-year means are calculated for the daily rainfall data, and anomalies are taken as the departure from this day-of-the-year mean.
3. Heat waves in Victoria
There is no standard definition of a heat wave. Previous studies of heat waves in Australia have used a variety of definitions: for example, Plummer et al. (1999) examined heat wave periods of a single day/night and three consecutive days/nights, using a 90th percentile threshold. When compiling indices to monitor variations in climate extremes in Australia, Nicholls et al. (2000) included calculations of the percentage of the country warmer than the 90th percentile, as well as the average area of the country experiencing very warm days. Tryhorn and Risbey (2006) defined five indices relating to the intensity, duration, diurnal characteristics, and frequency of heat waves, namely the value of the 90th percentile temperature for each year; the number of runs of consecutive days above this value each year, as well as the average and the maximum run lengths; and the mean of the warmest three consecutive minimum temperatures per year. The use of a percentile-based definition over at least three consecutive days, rather than an absolute threshold such as a particular maximum or minimum temperature, results in a definition that is relative to both the area of interest and the time of year (Perkins and Alexander 2013).
The HQDT dataset is used to determine the monthly 90th percentile maximum and minimum daily temperatures for each of the 11 stations in the state of Victoria during DJF 1989–2009. A heat wave is defined as any period of at least three days for which two criteria are satisfied. First, the daily maximum temperature at one or more stations equals or exceeds the 90th percentile maximum at that station for that month. Second, on at least two of those three days the daily minimum temperature equals or exceeds the 90th percentile minimum (the highest 10% minimum temperatures). This definition identifies 32 summer heat waves of varying lengths in Victoria for the 21-yr period of interest.
Table 2 lists the number of heat waves versus the number of stations that satisfy the heat wave definition for that event. The three heat waves involving five stations occurred during February 1997, December 1998, and December 1999, and the two most extensive heat waves involving eight stations during February 1997 and January–February 2009, which at 13 days is also the longest heat wave. The average maximum temperature during a Victorian heat wave day is 38.3°C, with temperatures ranging from just 24.0°C in December 1997 at Gabo Island (see Table 1 for station locations), where the 90th percentile maximum temperature for December is 23.5°C, to 46.7°C in February 2009 at Mildura, where the 90th percentile maximum for February is 39.0°C. The mean margin by which the 90th percentile maximum temperature is exceeded is 3.5°C, and the greatest margin was recorded at Cape Otway on 20 January 1997, when the daily maximum of 40°C exceeded the 90th percentile maximum by 11.3°C. In the state capital of Melbourne, the average temperature during a DJF heat wave is 39.3°C and the highest temperature was 45.1°C on 30 January 2009, which is 8.9°C above the 90th percentile maximum (see Table 3). The mean fall in the daily maximum temperature on the day after a heat wave, where the end of the heat wave is defined at individual stations, is 11.0°C. The largest temperature decrease of 25.4°C was recorded at Laverton, where the maximum temperature was 41.4°C on 12 December 1998 and just 16.0°C on the following day. The smallest decrease in daily maximum temperature of 0.5°C was seen at Gabo Island from 31 January to 1 February 2009.
4. Composite of heat waves in Victoria in summer
Composites from the ERA-Interim dataset are constructed for all days (a total of 132) in DJF from January 1989 to December 2009 on which the definition of a heat wave is satisfied for at least one HQDT station in Victoria. Rainfall composites are constructed for all heat wave days in DJF during the period covered by the GPCP dataset (a total of 98 days).
a. Synoptic composite
A composite of synoptic conditions for all Victorian heat wave days reflects the familiar pattern seen on synoptic weather charts during heat waves in the region. A large positive 850-hPa temperature anomaly extends over much of southeastern Australia (Fig. 1a). There is a ridge of anomalously high mean sea level pressure (MSLP) over the Tasman Sea to the east of the continent, colocated with the mean position of the subtropical ridge, and flanked by troughs to the east, west, and south (Fig. 1b). Flow around the surface anticyclone (Fig. 1c) directs warmer air from the interior in a northerly or northwesterly flow over the southeast of the continent. The depth of the layer of warm air over southeastern Australia is evident in the 500–1000-hPa geopotential height anomaly (Fig. 1d), with mean anomalies of 80 m or greater over much of the region. Also evident in the composites are dynamically consistent regions of elevated temperatures, higher MSLP, enhanced northerly flow, and increased thickness over the southern Indian Ocean and the coast of Antarctica between 60° and 90°E. It is clear that these events are not simply an Australian phenomenon, but extend across the Southern Hemisphere (SH). The synoptic composites indicate that a slow-moving transient anticyclone at the surface, accompanied by northerly to northwesterly flow over the southeast of Australia, and a deep layer of warm air over the region are robust features of heat waves in Victoria.
b. PV-Θ composite
PV thinking provides a succinct method through which to gain insights into the dynamics of the large scale atmospheric flow during heat waves. PV is conserved for adiabatic frictionless motion and redistributed by internal diabatic heating processes. It has also proven to be a useful Lagrangian tracer for air parcels. In constructing isentropic PV maps for the NH, Hoskins et al. (1985) noted that the formation mechanism for cutoff cyclones and blocking anticyclones appears to be conceptually the same: differential advection leading to the cutting off of upper-level PV anomalies of the appropriate sign. Throughout this study, the PV field is multiplied by negative one so that a low or negative PV anomaly refers to an anticyclonic anomaly, and a high or positive PV anomaly to a cyclonic anomaly.
A composite of the 350-K PV anomalies for all Victorian heat wave days (Fig. 2) shows an upper-level anticyclonic anomaly over the southeast of Australia with a broad band of cyclonic anomalies wrapping around its eastern flank. The 2-PVU contour (1 PVU = 1 × 10−6 K m2 s−1 kg−1) indicates the dynamical tropopause, and thus the boundary between the tropospheric and stratospheric air masses. Over eastern Australia, this contour illustrates the reversal of the meridional PV gradient that is associated with wave breaking (McIntyre and Palmer 1983, 1984; Song et al. 2011). The structure of the 2-PVU contour corresponds to the LC1 pattern of anticyclonic wave breaking postulated by Thorncroft et al. (1993).
A ubiquitous feature of the atmospheric flow is the meandering jet stream that forms at the break in the tropopause, where the height of the tropopause decreases abruptly from its maximum in the tropics to lower altitudes in the polar region. A band of enhanced PV gradient, located on isentropic surfaces that transect the tropopause break, is aligned with the jet, and can serve as a waveguide for Rossby waves (Hoskins and Ambrizzi 1993; Schwierz et al. 2004). The mean 350-K meridional winds indicate that, during heat waves in Victoria, the upper-level jet is strongest to the southeast of Australia (Fig. 2). In a climatology of Rossby wave breaking in the SH, Song et al. (2011) found that PV troughs at 350 K on the equatorward side of the midlatitude jet in DJF are predominantly of the LC1 type (see also their Fig. 1 for a schematic of LC1 and LC2 types in the SH). The southwestern Pacific is a preferred region for Rossby wave overturning in DJF (Berrisford et al. 2007), although the exact mechanism underlying the overturning remains unclear, and is beyond the scope of the present study.
The evolution of the 350-K PV anomaly from 4 days prior to the onset of a heat wave until the first day of the heat wave (Figs. 3a–c) illustrates the propagation of Rossby wave packets throughout the SH. During heat waves, these Rossby waves grow in amplitude and eventually overturn to the southeast of Australia. The northwest–southeast tilt of the trough axes, and the deformation of the 2-PVU contour, are consistent with LC1-type wave breaking (Thorncroft et al. 1993). The development of a Rossby wave train over the southern Indian Ocean and to the south of Australia can be seen in the evolution of the 250-hPa meridional wind anomalies and PV field for the same period (Figs. 3d–f). In the region of elevated temperatures over the southern Indian Ocean off Antarctica seen in the synoptic composite, the formation of an upper-level anticyclonic PV anomaly indicates local dynamical conditions that are consistent with those responsible for the formation of heat waves over southeastern Australia.
The mean propagation of the wave packets is illustrated in a Hovmöller (time–longitude) plot of the unfiltered 350-K PV anomalies, averaged over latitude band 30°–50°S (Fig. 4a). This panel shows the mean evolution of the upper-level PV anomalies from 12 days prior to onset up to 7 days after termination of all heat waves in Victoria. From about 5–6 days prior to the onset of a heat wave, the mean upper-level anticyclone is the strongest feature in this latitude band in the SH. The trough to the west strengthens between 4 and 5 days, and that to the east around 3 days, prior to onset. From 3 days prior to onset onward, the midlatitudes south of Australia are dominated by an upper-level anticyclone, with flanking troughs to the east and west. For up to a week after the end of a heat wave, in the mean the midlatitudes are dominated by an anticyclonic 350-K PV anomaly over much of the Indian and Pacific Oceans. A time–longitude plot of the mean 250-hPa meridional wind anomalies and PV field (Fig. 4b) similarly shows the evolution of the wave packets, with (graphically estimated) wavelengths of approximately 7000 km, a phase speed of about 10 m s−1, and a group speed of around 80 m s−1. Note that the PV field and the winds are in quadrature, as expected.
A vertical cross section at 150°E longitude (Fig. 5)—that is, approximately through the center of the anticyclonic anomaly as shown by the thin black line in Fig. 2—shows the evolution of the PV anomalies and winds from two days prior to day one of all heat waves in Victoria. The mean cyclonic PV anomalies are of similar magnitude and spatial extent throughout this period in this cross section (Figs. 5a–c). The anticyclonic anomaly, on the other hand, strengthens considerably by the first day of the heat wave. The PV anomalies extend throughout the depth of the troposphere, with maximum amplitudes near the tropopause. The zonal wind anomalies show the developing deep anticyclonic circulation associated with the anticyclonic PV anomaly, with the maximum wind speeds lying between 200 and 250 hPa. A further notable feature in Fig. 5 is the characteristic bowing of the isentropes in association with the upper-level PV anomalies (see, e.g., Hoskins et al. 1985). A similar cross section of vertical motion (Figs. 5d–f) shows a region of mean ascent from the equator to 20°S, in the region ahead of the upper-level trough over northeastern Australia. On the first day of a heat wave, a region of ascent develops around 40°–60°S, ahead of the mean position of the approaching front. A region of mean descent develops between 30° and 40°S, ahead of the upper-level anticyclone and behind the upper-level trough. In addition to warm advection from the interior of the continent, warming by compression as the air descends along the isentropes contributes to the high temperatures over southeastern Australia. Suppression of cloud in this region of mean descent, which coincides with the surface anticyclonic anomaly, adds to the warming at the surface through increased incoming solar radiation.
The way in which neighboring air parcels converge, diverge, or become trapped in an eddy may be examined through the use of Lyapunov exponents (see, e.g., Cohen and Schultz 2005). The positive Lyapunov exponent (or asymptotic dilatation rate) is calculated as , where the divergence is defined as D = ∂u/∂x + ∂υ/∂y, the total deformation as , the stretching deformation as Est = ∂u/∂x − ∂υ/∂y, the shearing deformation as Esh = ∂u/∂y + ∂υ/∂x, and the relative vorticity as ζ = ∂υ/∂x − ∂u/∂y. Where E2 − ζ2 < 0, the vorticity is greater than the deformation, and the imaginary part of the Lyapunov exponent denotes regions in which adjacent parcels will be isolated from the surrounding flow, or trapped. The vertical extent of the vertically coherent region of mean imaginary Lyapunov exponent is indicated by the heavy black contour in Figs. 5a–c. This region of trapping is coincident with the upper-level anticyclone, and gradually deepens to extend to the lower troposphere by day 1 of a heat wave. The upper level anticyclone will remain stationary as there is no flow across the center. The only ventilation is then in the boundary layer, below about 700 hPa, and will likely be as a result of flow from the hot continental interior, reinforcing the elevated temperatures at the surface.
The PV-Θ composites reveal that anticyclonically overturning Rossby waves, and the resulting upper-level PV anomalies, are the underlying mechanism for the formation of heat waves in Victoria.
On the day prior to the onset of a heat wave in Victoria, there is, in the mean, enhanced precipitation across much of northern Australia and over the Coral Sea to the northeast of the continent (Fig. 6a). This band of anomalous rainfall is associated with an upper-level cyclonic PV trough, which extends from south to north over the Tasman Sea and then curves from east to west across the north of the continent. A region of heavy rainfall over northwestern Australia, which is connected to the equatorial regions to the north and also extends southeastwards over the Southern Ocean, is a notable feature in this panel. On the first day of a heat wave, the upper-level trough and enhanced rainfall over northern Australia are still evident (Fig. 6b). The band of rainfall that extends from the tropics over western Australia has become more pronounced, with rainfall to the south of the continent increasing. As the anticyclone over the southeast of Australia intensifies, precipitation is suppressed over much of the central and southern continent and the Tasman Sea as a result of large-scale descent. This suppression of precipitation results in the characteristic pattern at the start of a heat wave of a ring of anomalous rainfall around the mean position of the anticyclone. In general, the termination of a heat wave in Victoria is marked by the passage of a cold front, and the location of the front and the accompanying precipitation can be seen in Fig. 6c. The rainfall over northern Australia has weakened by this stage.
Heat waves in southeastern Australia are frequently accompanied by heavy rainfall over the northeast of Australia. Several studies find that the periphery of a blocking anticyclone provides a favorable environment for heavy rainfall, as a result of modified lift, instability, and vertical wind shear (Weaver and Nigam 2008; Galarneau et al. 2012). An upper-level PV trough may also enhance rainfall through forced vertical motion and destabilization of the atmosphere, as well as modification of the low-level winds and thereby the moisture flux in the vicinity of the trough (Massacand et al. 2001; Funatsu and Waugh 2008; Martius et al. 2013). In a study of heavy precipitation in the European Alpine region, Massacand et al. (2001) found that cyclonic PV troughs may enhance precipitation through induced ascent as the troughs move, or as flow passes through the anomaly. The location and strength of the trough also modulate the low-level winds, and thereby the regional moisture flux. In general, the tropopause slopes downward and has a lower potential temperature with increasing latitude. As the air from higher latitudes is advected equatorward during Rossby wave breaking, forming the upper-level trough, a cold tropopause potential temperature anomaly will also result. Juckes and Smith (2000) show that cooling in the upper troposphere is an important factor in convective destabilization. These authors noted that the local ascent of isentropes ahead of a moving upper-level cyclonic PV anomaly leads to destabilization as a result of cooling of the atmosphere, and convective available potential energy (CAPE) will increase with the strength and broadness of the upper-level trough. The changes in CAPE were found to be even more significant in the midlatitudes than in the tropics. Whatever the precise physical mechanism, upper-level troughs are frequently the focus of heavy rainfall in both the tropics and extratropics.
Since rainfall is not, in general, a normally distributed variable, the frequency distribution of GPCP rainfall for heat wave days and that for non-heat-wave days in DJF for the period covered by the GPCP data are shown in Fig. 7. The mean rainfall in a box over northern Australia, 10°–20°S and 110°–150°E, is computed for both heat wave and non-heat-wave days. The distribution for heat wave days is shifted toward higher rainfall events, indicating that the composite rainfall presented in Fig. 6 is not simply dominated by a few very strong precipitation events. Furthermore, 10 000 random samples of 98 days were selected from the DJF months covered by the GPCP dataset, both with and without replacement. In none of these 10 000 samples did the mean rainfall in this box exceed the mean for the 98 heat wave days. This result indicates that the signal evident in the rainfall composites for heat wave days is very unlikely to arise simply by chance.
Rainfall over northeastern Australia during a heat wave in Victoria may therefore be enhanced by favorable conditions on the periphery of the anticyclone, or through forced vertical motion, increased atmospheric instability, upper tropospheric cooling, and modified moisture fluxes associated with the upper-level cyclonic PV trough.
5. Case study: January–February 2009
One of the most severe heat waves in southeastern Australia, the “pre–Black Saturday heat wave,” occurred in Victoria between 27 January and 8 February 2009, with heat wave conditions as previously defined existing for one or more Victorian stations throughout this time. This heat wave was characterized by two periods of exceptionally high temperatures, during 28–31 January and 6–8 February, with very high temperatures persisting over the days in between (National Climate Centre 2009). The prolonged period of warm and dry conditions contributed to the drying of vegetation, increasing the fire risk in this region. Between 28 and 30 January, Melbourne recorded three days with temperatures between 43° and 45°C (see Table 3). On Black Saturday, 7 February, temperatures in Melbourne reached a record of 46.4°C. A large number of catastrophic fires in Victoria on this day, worsened by the hot, dry, and windy conditions, caused 173 fatalities and destroyed more than 2133 houses, decimating several townships (VBRC 2010). To illustrate the conditions which prevail during heat waves in Victoria, the synoptic and PV-Θ conditions are examined for 29 January 2009, the second day on which heat wave conditions existed in the city of Melbourne (see Table 3).
a. Synoptic conditions
The synoptic conditions on this day are consistent with the composites in section 4a and reflect the familiar pattern seen on weather charts during heat waves in Victoria. A large positive 850-hPa temperature anomaly extends over much of southeastern Australia at 0000 UTC [1000 Australian Eastern Standard Time (EST)] 29 January 2009 (Fig. 8a). A region of anomalously high mean sea level pressure over the Tasman Sea to the east of the continent, flanked by troughs to the east and west (Fig. 8b), advects warmer air from the interior of the continent in a northerly or northwesterly flow over the southeast (Fig. 8c). There is a deep layer of warm air over this region (Fig. 8d), with 500–1000-hPa geopotential height anomalies in excess of 160 m over much of Victoria at this time. As seen in the composites above, dynamically consistent conditions exist elsewhere in the SH on this day, with a large positive temperature anomaly, almost equal in magnitude to that over southeastern Australia, over the southern Indian Ocean and extending to the coast of Antarctica between about 60° and 90°E.
b. PV-Θ view
The 350-K PV anomaly at 0000 UTC 29 January 2009 (Fig. 9) shows a region of anomalously anticyclonic PV over the southern continent, with a band of anomalously cyclonic PV wrapping around it. The position of the 2-PVU contour or dynamical tropopause illustrates the reversal of the meridional PV gradient associated with wave breaking, and is consistent with the LC1 mode of anticyclonic wave breaking referred to in the composites above. The PV trough over northeastern Australia is thinning, with wind vectors normal to the 2-PVU contour indicating the region of PV advection, and an upper-level cutoff low forms in this region on the following day (not shown). The 350-K meridional winds (Fig. 9) show the strengthening of the jet stream to the south of Australia, as well as the flow around the upper-level anticyclone and trough. The evolution of the 350-K PV anomalies from four days and two days prior to the onset of the heat wave, to day 1 of the heat wave, illustrates the propagation of Rossby wave packets across the SH (Figs. 10a–c). The poleward excursion of anticyclonic anomalies and equatorward extent of cyclonic anomalies is evident in these panels, as well as the reversal of the meridional PV gradient associated with LC1-type Rossby wave breaking over southeastern Australia and the formation of a cutoff low over the central Pacific Ocean. The 250-hPa meridional wind anomalies and PV field (Figs. 10d–f) show the evolution of the Rossby wave train to the south of Australia.
A vertical cross section at 142.5°E longitude (Fig. 11a)—that is, approximately through the center of the anticyclonic PV anomaly as indicated by the thin black line in Fig. 9—again shows PV anomalies that extend throughout the troposphere, with maximum amplitudes near the tropopause. The characteristic bowing of the isentropes associated with the upper-level PV anomalies is also evident. A cross section of vertical motion (Fig. 11b) shows a broad region of ascent between about 10° and 20°S ahead of the upper-level trough, as well as upward motion around 45°S between the anticyclone and the approaching front to the south. Descent between about 25° and 35°S, between the upper-level anticyclone and cyclone, indicates the region where additional warming by compression occurs. Above the boundary layer in the mid to upper troposphere the warming due to subsidence is of approximately the same magnitude as that due to advection, on the order of 1–2 K day−1 over southeastern Australia. The region in which the vorticity exceeds the deformation, denoted by the imaginary part of the Lyapunov exponent, is again coincident with the upper-level anticyclone and extends to the lower troposphere. This indicates that air parcels in the upper anticyclone are likely to remain trapped there for some time, contributing to the development of the heat wave.
The pre–Black Saturday heat wave was accompanied by heavy rainfall in parts of Queensland, which resulted in the severe flooding of several rivers along the northern coast (Australian Bureau of Meteorology 2009). A tropical low developed on the monsoon trough over the Gulf of Carpentaria (approximately 20°S, 140°E) between 27 and 30 January 2009, bringing heavy rainfall to northern Queensland. A second low on the monsoon trough then developed east of Cairns (17°S, 145°E) to become category 1 tropical cyclone (TC) Ellie. TC Ellie crossed the coast of northeastern Queensland early on 2 February 2009, bringing further heavy rainfall to the region. Another low then developed on the monsoon trough east of Cairns on 4 February 2009, producing additional heavy precipitation for several more days (Australian Bureau of Meteorology 2009).
The features described above are shown in the GPCP rainfall and 350-K PV field for several days during the pre–Black Saturday heat wave (Fig. 12). On 26 January 2009, the day prior to the onset of the heat wave (Fig. 12a), there is enhanced precipitation over northern Australia in the vicinity of the 350-K trough. Precipitation associated with TC Hettie is visible over the Pacific Ocean to the northeast of Australia. A region of heavy rainfall associated with TC Dominic can be seen over western Australia, with a band of rainfall extending southeastward in association with the cyclone outflow region. Mesoscale bands of heavy precipitation in association with TC outflow regions have been referred to as predecessor rain events (PREs) by Galarneau et al. (2010), and are sustained by tropical moisture transported poleward by the TC. On the first day of the heat wave, 27 January (Fig. 12b), there are regions of anomalously heavy rainfall associated with the upper-level trough over the northwest of the continent, and with the tropical low over the Gulf of Carpentaria. The rainfall associated with the outflow region of TC Dominic over southwestern Australia now extends farther toward the southeast. A band of precipitation in the region behind the upper-level anticyclone, and ahead of the following trough, can be seen to the south of the continent. On 31 January, the precipitation associated with TC Ellie can be seen over the northeast of Australia (Fig. 12c). On 8 February, the last day of the heat wave, a band of rainfall extends across the Coral Sea and the north of the continent in the vicinity of the upper-level trough (Fig. 12d). Although a cold front passed through Victoria on the night of 7 February 2009, very little rainfall was associated with this change (Engel et al. 2013), and thus the passage of the front is not marked by precipitation in this panel. TC Freddy is located over the Indian Ocean to the northwest of Australia on this date.
Berry et al. (2012) investigated the connection between summer rainfall in northwestern Australia and the propagation of synoptic PV maxima within the Australian monsoon, finding that rainfall is significantly modulated in the vicinity of such PV maxima, with enhanced rainfall in the region of mean isentropic ascent ahead of the cyclonic anomalies. The authors estimated that on the order of half of the summer rainfall in some regions of northwestern Australia occurs in the vicinity of PV maxima. The isentropic ascent is a result of the characteristic bowing of the isentropes referred to previously, and mean motion of the anomalies with respect to the background flow. Berry et al. (2012) also hypothesized that the PV maxima, which were first detected over the northeast of Australia, comprised the debris of Rossby wave breaking over the east coast of the continent. The anomalously heavy rainfall during the pre–Black Saturday heat wave may therefore be related to the modulation of precipitation that has been associated with synoptic cyclonic PV anomalies (Berry et al. 2012).
There is a further connection between strong tropical convection and heat waves. Parker et al. (2013) examined the connection between strong tropical convection, in the form of TCs, and the strength of the upper-level anticyclonic PV anomaly responsible for the formation of heat waves in southeastern Australia. This study found that, if the relative phasing between the TC and the large scale flow in the midlatitudes is favorable, perturbation of the waveguide by the upper-level TC outflow indirectly reinforces the upper-level anticyclone through wave amplification and enhanced downstream ridging. Furthermore, the material advection of low-PV air from regions of convection associated with the TC outflow (PREs) may directly reinforce the anticyclone, and thereby enhance the conditions necessary for the formation of heat waves in Victoria. It was shown that, in the pre–Black Saturday heat wave, both these effects contributed to the reinforcement of the upper-level anticyclone. The band of rainfall over southwestern Australia seen in the case study is a feature of the composite of precipitation during heat waves presented in section 4c and Fig. 6b, indicating that the two-way tropical–extratropical interaction is a common feature of heat waves in southeastern Australia.
6. Summary and conclusions
Motivated by a lack of understanding of the fundamental underlying physical mechanisms responsible for the formation of heat waves over southeastern Australia in summer, this study has explored the dynamics of the large-scale flow during heat waves in this region. This has been accomplished using a combination of global reanalysis products and remote sensing rainfall estimates. Using a high-quality daily temperature dataset, heat waves in the southeastern Australian state of Victoria have been defined based on the exceedance of the 90th percentile maximum and minimum temperatures for a period of at least three consecutive days and two nights, respectively. This definition resulted in the identification of 32 heat waves of varying intensity and extent over the southeastern Australian state of Victoria during DJF from 1989 to 2009.
ERAI composites of synoptic conditions showed that heat waves in southeastern Australia are accompanied by a slow-moving transient surface anticyclone over the Tasman Sea to the east, which directs warm continental air in a northerly or northwesterly flow over the southeast, causing extreme temperatures. The elevated temperatures extend from the surface in a deep layer over the region, as shown by the mean geopotential height anomalies. One of the key findings from the PV-Θ composites is that heat waves in southeastern Australia in summer are associated with upper-level PV anomalies, which form as the result of anticyclonic (LC1 type) Rossby wave breaking on the equatorward side of the mean position of the midlatitude jet. A region of parcel trapping, described by a deep layer of imaginary Lyapunov exponent, indicated that the upper-level anticyclone would remain stationary and hence that the conditions responsible for the formation of heat waves would persist. At the same time, upper-level cyclonic PV troughs were found to enhance rainfall over the northeast of the continent through enhanced vertical motion, increased instability, and modifications to moisture flux. This dipole of extreme heat in the southeast and heavy precipitation in the northeast of the continent and adjacent ocean is a common feature of heat waves in Victoria. These mechanisms were illustrated for the most severe heat wave in the study period, the pre–Black Saturday heat wave of January–February 2009.
Several strong links between tropical convection and heat waves have been made. Not only is convection and thus precipitation enhanced in the region of mean isentropic ascent ahead of the cyclonic anomalies that form as a result of Rossby wave overturning during heat waves, but strong convection in phase with the large-scale flow may reinforce the upper-level anticyclone that is responsible for the formation of heat waves in southeastern Australia. In addition, the vertical gradient of diabatic heating in the band of convection associated with the TC outflow region provides a source of anticyclonic PV in the upper atmosphere, which may be advected into the upper-level anticyclone and further reinforce the dynamical conditions responsible for the formation of heat waves in southeastern Australia.
The authors are grateful to Thomas Spengler, Nicholas Klingaman, and Andrew Marshall for their insightful comments. We would also like to thank Neville Nicholls and Thomas Chubb for their assistance with the statistical analysis used in this work. TJP has been supported in part by the Australian Research Council Centre of Excellence for Climate System Science. GJB has been supported by Australian Research Council Grant FS100100081. Reanalysis data used in this study were provided by the ECMWF, GPCP rainfall data were obtained from NASA Goddard Earth Sciences Data and Information Service Center, and the Australian Bureau of Meteorology provided the HQDT dataset.