In the Northern Hemisphere, recurrence of transient synoptic-scale Rossby wave packets in the same phase over periods of days to weeks, termed RRWPs, may repeatedly create similar surface weather conditions. This recurrence can lead to persistent surface anomalies. Here, we first demonstrate the significance of RRWPs for persistent hot spells in the Southern Hemisphere (SH) using the ERA-Interim (ERA-I) reanalysis dataset and then examine the role of RRWPs and blocks for heatwaves over south-eastern Australia (SEA).
A Weibull regression analysis shows that RRWPs are statistically associated
with a significant increase in the duration of hot spells over several
regions in the SH, including SEA. Two case studies of heatwaves in SEA in
the summers of 2004 and 2009 illustrate the role of RRWPs in forming
recurrent ridges (anticyclonic potential vorticity – PV – anomalies), aiding in
the persistence of the heatwaves. Then, using a weather-station-based
dataset to identify SEA heatwaves, we find that SEA heatwaves are more
frequent than climatology during days with extreme RRWPs activity over SEA
(high
Since 1900, extreme heat has been responsible for more fatalities in Australia than all other natural hazards combined (Coates et al., 2014). Heatwaves also exacerbate the risk of wildfires, cause surges in power demand, and increase insurance costs (Hughes et al., 2020; Insurance Council of Australia, 2020). Increasingly frequent and severe heatwaves in the midlatitudes in the recent years (Coumou et al., 2013; Perkins-Kirkpatrick and Lewis, 2020; IPCC, 2021) have spurred fruitful research on the atmospheric drivers of heatwaves. Understanding the dynamical mechanisms is particularly important for improving sub-seasonal prediction (Quandt et al., 2017) and for quantifying future changes in heatwaves (Shepherd, 2014; Wehrli et al., 2019).
Several large-scale atmospheric mechanisms and phenomena have been identified as potential drivers of heatwaves in the Northern Hemisphere (NH) extratropics. They include blocking anticyclones (e.g. Barriopedro et al., 2011; Drouard and Woollings, 2018; Kautz et al., 2022), amplified quasi-stationary waves (Teng et al., 2016; Kornhuber et al., 2017), amplified Rossby wave packets (e.g. Fragkoulidis et al., 2018; Kornhuber et al., 2020), and recurrent Rossby wave packets (Röthlisberger et al., 2019). Fragkoulidis et al. (2018) showed that amplified Rossby waves are correlated with surface temperature extremes over NH and used process-based understanding to establish further association for the 2003 and 2010 NH heatwaves.
RRWPs can be considered a subset of amplified Rossby waves with a condition that the transient eddies recur spatially in the same phase on a short timescale of days to weeks. RRWPs are closely related to blocking. RRWPs forming upstream of a block can sustain the block (e.g. Shutts, 1983; Hoskins et al., 1985; Hoskin and Sardeshmukh, 1987). RRWPs can also form downstream of blocks because of the near-constant phase of the wave breaking (trough) on the downstream flank of the blocks (Barton et al., 2016; Röthlisberger et al., 2018). Here, we focus on recurrent Rossby wave packets to explore their importance for heatwaves in south-eastern Australia (SEA).
Broadly, heatwaves in SEA (Fig. 1), comprising the states of Victoria (VIC), New South Wales (NSW), South Australia (SA), and Tasmania (TAS), are associated with slow-moving transient anticyclonic upper-level potential vorticity (PV) anomalies over the Tasman Sea (e.g. Marshall et al., 2014; Parker et al., 2014b; Quinting and Reeder, 2017; Parker et al., 2020). The anticyclonic PV anomalies and the associated subsidence drive heatwaves over VIC (Parker et al., 2014b; Quinting and Reeder, 2017). These anticyclonic PV anomalies can form as part of a synoptic-scale Rossby wave packet (RWP) (King and Reeder, 2021). These RWPs are often initiated several days before the onset of the heatwaves, but they amplify and eventually break anticyclonically over SEA (Parker et al., 2014b; O'Brien and Reeder, 2017).
Map of Australia showing the states of south-eastern Australia (SEA): South Australia (SA), Tasmania (TAS), Victoria (VIC), and New South Wales (NSW). Other states shown are Queensland (QLD), Northern Territory (NT), and Western Australia (WA). Red dots indicate the Australian Bureau of Meteorology's (BoM) monitoring stations used in this study (see Sect. 2).
Surface temperature anomalies associated with transient RWPs form, amplify, and decay on synoptic timescales, but the recurrence of RWPs in the same phase on a sub-seasonal timescale can result in persistent surface weather conditions by repeatedly re-enforcing the surface temperature anomalies (e.g. Hoskins and Sardeshmukh, 1987; Davies, 2015). Röthlisberger et al. (2019) termed this phenomenon “recurrent Rossby wave packets” (RRWPs) and demonstrated a statistically significant connection between RRWPs and the persistence of surface temperature anomalies in the Northern Hemisphere (NH). Ali et al. (2021) found that RRWPs are also associated with increased persistence of dry and wet spells in several regions across the globe.
For some impacts, however, it is not only the simple occurrence of an extreme
that defines an extreme but also the duration of the extreme event
that is important. This study addresses that aspect for the hot temperature
extremes in the SH. More precisely, we evaluate the hypothesis whether an
increase in the
This study uses ERA-Interim (ERA-I) reanalysis data (Dee et al., 2011) provided by the European Centre for Medium-Range Weather Forecasts on a
The
In most cases, large values of
For the phase–amplitude information used in Sect. 3.3, it is extracted
using the Fourier decomposition along the longitude of meridionally averaged
(35 and 65
Atmospheric blocking data are computed following the methodology of Schwierz et al. (2004) as in Rohrer et al. (2020) and Lenggenhager and Martius (2019). The detection scheme identifies persistent anticyclonic PV anomalies vertically averaged (VAPV) between 500 and 150 hPa vertical levels. First, the VAPV anomaly is computed from the 30 d running mean climatology of the corresponding time step of the year for the years 1979–2018. An additional 2 d running mean filter is applied to smooth out high-frequency transients. Then the algorithm identifies areas with VAPV
A station-based temperature dataset is used to identify extreme and
persistent heatwaves in SEA. Following the methods developed in Parker et
al. (2014a) and refined in Quinting and Reeder (2017), heatwaves in SEA in December–February (DJF) are detected from temperatures observed at the Australian Bureau of Meteorology's (BoM) monitoring stations (Fig. 1). The BoM's Australian Climate Observations Reference Network – Surface Air Temperature (ACORN-SAT, available at
Hot spells are identified for all SH grid points between 20 and
70
To quantify the effect of RRWPs on the persistence of hot surface weather, we extend an analysis from Röthlisberger et al. (2019) to the SH, including SEA, using the same statistical model setup, a Weibull regression model. This model allows us to model the distribution of the duration of hot spells at each grid point. An advantage of the model of Röthlisberger et al. (2019) is that we do not need to subjectively define the duration of a “significant” spell because the model quantifies the changes in all quantiles of the spell duration modelled. The null hypothesis tested here at each grid point is that RRWPs have no effect on the duration of hot spells at the respective grid point. The Weibull model is only briefly introduced here. Please refer to Röthlisberger et al. (2019) for further details and their Supplement for a detailed introduction to the Weibull model.
To fit the Weibull model to the observed spell duration distribution, a
representative value of the
Here,
Furthermore, fitting Eq. (2) to spell durations at all grid points
results in a spatial field of the AF. The statistical significance of the AF values is evaluated in a two-step approach. First, a
The Weibull analysis reveals that RRWPs are significantly correlated with
the duration of hot spells in several regions within the SH and including
over SEA (Fig. 3). Recall that an AF larger than 1 means that an increase in
Statistically significant acceleration factors (AFs) for hot spells in November–April between 20 and 70
The February 2004 heatwave (7–22 February) lasted for 16 d. More than
60 % of continental Australia recorded temperatures above 39
Figure 4 shows the flow conditions prior to and during the heatwave (Fig. 4b) and the corresponding T2m anomalies over SEA (Fig. 4a). The
Hovmöller diagram (Fig. 4b) shows the 35 and 65
RRWPs and blocks during the 2004 SEA heatwave.
During this event, several Rossby wave packets were observed, recurrently
amplifying in the same phase, forming a ridge over SEA. The upper-level flow
over SEA was zonal prior to the heatwave (Fig. 5a). An upper-level ridge
forms over SEA around 5 February prior to the heatwave (Fig. 5b). The flow
becomes more amplified in the subsequent days with a circumglobal amplified
wave pattern apparent around 9 February (Fig. 5c). The amplified wave, part
of a transient and nonstationary Rossby wave packet (RWP; P1 in Fig. 4b)
arrived over the southern Indian Ocean, and an upper-level ridge began to
form over Australia, which amplified further around 13 February (Fig. 5d).
Two further ridges formed over SEA on 16 and 18 February (Fig. 5e, f), each
ridge being part of a transient nonstationary RWP initiated upstream of
Australia (P3, P4 in Fig. 4b). These series of upper-level recurrent ridges
were part of the RRWPs and contributed to the persistence of the heatwave.
These recurrent ridges associated with RRWPs were also detected by the
No blocks were identified directly over SEA during the heatwave, but blocks
were present south of SEA and further downstream (Figs. 4, 5). The RWP
labelled as P1 in Fig. 4b formed downstream of block B1 in the Pacific Ocean (roughly 200
The 2009 heatwave (27 January–9 February), although extensively covered in literature (e.g. Engel et al., 2013; Parker et al., 2014b), has been chosen because it is one of the most severe heatwaves in SEA. It lasted for 14 d. Between 28–31 January and 6–8 February, temperatures in SEA were exceptionally high. On Black Saturday, 7 February, the hot, dry, and windy conditions fuelled many catastrophic fires in VIC, which recorded 173 fatalities, and more than 2133 houses were destroyed (Karoly, 2009; Parker et al., 2014b; VBRC, 2010). During this heatwave, an anticyclone over SEA and the associated north-westerly flow at the surface advected hot continental air into SEA leading to extreme surface temperatures (Parker et al., 2014b). As for the 2004 case, we next present the Hovmöller diagram (Fig. 6) and snapshots of upper-level flow (Fig. 7) to demonstrate the role of transient RWPs and blocks during the heatwaves.
Same as in Fig. 4 but for the February 2009 SEA heatwave.
Same as in Fig. 5 except for the February 2009 SEA heatwave.
Prior to the onset of the heatwave, the flow was already amplified with a
wave breaking over SEA (Fig. 7a). Several RWPs were observed prior to and
during this event (P1 and P2 in Fig. 6b). The RWPs prior to the heatwave
were not in the same phase as those during the heatwave (Fig. 6b), which is
why the value of the
No blocks were identified directly over SEA during the heatwave (Figs. 6, 7).
However, blocks were frequent upstream of SEA from 50 to 70
So far, we have investigated the association of RRWPs with a duration of hot
spells. We also presented two cases of extreme and persistent SEA heatwaves
to show how RRWPs can lead to the formation or replenish the anticyclonic PV
anomalies over SEA. Figure B1 shows another case of SEA heatwave associated
with RRWPs. In the next section, we extend the analysis to a climatological
period (1979–2018) and explore high-
First, we note the co-occurrence of high-
High-
Standardized PV anomalies on the 350 K isentrope with respect to the DJF climatology (1979–2018) for
In contrast, on high-
In addition to the ridge over SEA, circum-hemispheric zonal wavenumber 4 and 5 (WN4, WN5) patterns are present in the composite mean PV fields for high-
Bivariate kernel density estimate using Gaussian kernels in the
complex plane of the Fourier decomposed meridional wind at 250 hPa averaged
between 35 and 65
The WN4 and WN5 components (Fig. 8d, e) of the composite mean PV field for
high-
Figure 8d and e suggest that high-
On high
The phase distribution for WN4 and WN5 is shown here because they emerge as
the dominant patterns in the composite mean (Fig. 8a), whereas the density
distributions for other wavenumbers (
During the 2004 and 2009 SEA heatwaves, we find transient and fast-moving
Rossby waves organized in wave packets recurring in the same phase to form
a ridge over SEA, thereby contributing to the persistence of the heatwave
conditions. This persistence arises by recurrence, in contrast to the
persistence arising from stationary weather features such as slow-moving
Rossby waves (e.g. Wolf et al., 2018) or blocking anticyclones (e.g. Kautz
et al., 2022). The Rossby wave packets observed during the two SEA heatwaves
were not always initiated in the same area. In the 2004 case, these waves
were mostly not in phase upstream of Australia, whereas in the 2009 case,
they were also in phase upstream over the Indian Ocean. Blocks were observed
upstream and downstream during the two heatwaves, which suggests that blocks
could play a role in initiating the RWPs and/or in modulating their phase.
Figure E1 presents the relationship between
The relevance of RRWPs for persistent SEA heatwaves documented in these two
case studies is consistent with the results of the Weibull regression
analysis, which reveals a significant positive statistical link between the
duration of hot spells over SEA and RRWPs. The PV composite for high-
A reverse causal link between surface temperature anomalies during SEA
heatwaves and
We find that RRWPs are associated with a significant increase in the persistence of hot spells in the SH. In several parts of SEA, including the states of South Australia, New South Wales, Victoria, and Tasmania, longer hot spells coincide with high-amplitude RRWPs (Fig. 3). Other regions over land where RRWPs are statistically associated with hot-spell duration include South America: southern Brazil, Bolivia, and parts of Argentina and Chile.
We have demonstrated the role of RRWPs in building persistent ridges during
two cases of SEA heatwaves: the 2004 and 2009 heatwaves. Both heatwaves
featured RRWPs comprised of transient Rossby waves, which were in phase
regionally but not hemisphere wide. Blocks were not directly observed over
SEA, but the case studies suggest that blocks upstream and downstream played
an important role in initiating the Rossby wave packets and modulating their
phase. We further investigated the co-occurrence of RRWPs during the most
persistent and extreme SEA heatwaves using the
We find that days with
The following open questions remain: what is the role of blocks in initiating RRWPs and modulating their phase? The case studies and the PV composites suggest that blocking might play an important role. What is the role of background flow in setting up RRWPs and modulating their phase? The interaction of RRWPs with other well-known climate oscillation patterns such as the ENSO and the Southern Annular Mode also needs to be investigated further. Better understanding of the interplay between these features might offer an opportunity to improve sub-seasonal forecasts during RRWP events.
Same as in Fig. 4 but for January 2014 SEA heatwave.
Occurrence of high
Figure D1 shows PV anomalies for all SEA heatwaves days identified in this
study. The PV anomalies for SEA heatwaves feature anticyclonic PV anomalies
over SEA with cyclonic PV anomalies to the north and south of it, which is
similar to Fig. 2 in Parker et al. (2014b), who show PV anomalies for
Victorian heatwaves. However, the wavenumber pattern seen in Fig. 8a for SEA
HDs and high
To further analyse the spatial distribution of RRWPs relative to blocks in
the SH, we focus on two longitudinal subdomains that show a high blocking
frequency in the DJF climatological mean: the South Pacific (130–50
Time-lagged Hovmöller composites of
In the Pacific Ocean, blocks coincide with positive
Code for calculating the
SMA led the study, performed the analysis, and wrote the first draft. MR provided the code for the Weibull regression model, which SMA suitably modified for this study. TP produced the SEA heatwaves dataset and contributed to the analysis of heatwaves. SMA, MR, KK, and OM contributed in designing the project. All the co-authors contributed to the interpretation and discussion of the results and contributed to the first draft.
The contact author has declared that none of the authors has any competing interests.
Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
S. Mubashshir Ali is grateful to Alexandre Tuel and Pauline Rivoire for discussions and Simon Milligan for editing the text. Olivia Martius and S. Mubashshir Ali acknowledge Marco Rohrer for the blocking algorithm. The authors acknowledge the European Centre for Medium-Range Forecasts (ECMWF) for producing the ERA-I dataset and the Australian Bureau of Meteorology for producing the ACORN-SAT dataset. The authors are also grateful to the two anonymous reviewers and Volkmar Wirth for their constructive comments which helped to improve this work.
S. Mubashshir Ali and Olivia Martius were funded from the Swiss National Science Foundation (grant no. 178751). Matthias Röthlisberger was funded by the European Research Council under the European Union's Horizon 2020 research and innovation programme (INTEXseas; grant no. 787652). Kai Kornhuber was partially supported by the National Science Foundation (NSF; project no. AGS-1934358).
This paper was edited by Michael Riemer and reviewed by Volkmar Wirth and two anonymous referees.