• Bluestein, H. B., 1992: Synoptic–Dynamic Meteorology in Midlatitudes. Vol. 1. Oxford University Press, 431 pp.

  • Burt, S. D., , and D. A. Mansfield, 1988: The great storm of 15–16 October 1987. Weather, 43, 90110.

  • Canadell, J. G., and Coauthors, 2007: Contributions to accelerating atmospheric CO2 growth from economic activity, carbon intensity, and efficiency of natural sinks. Proc. Natl. Acad. Sci. USA, 104, 18 86618 870.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2006a: The Community Climate System Model version 3 (CCSM3). J. Climate, 19, 21222143.

  • Collins, W. D., and Coauthors, 2006b: The formulation and atmospheric simulation of the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 21442161.

    • Search Google Scholar
    • Export Citation
  • CSIRO and Bureau of Meteorology, 2007: Climate change in Australia. Australian Greenhouse Office Tech. Rep., 148 pp. [Available online at http://www.climatechangeinaustralia.gov.au.]

  • Dowdy, A. J., , G. A. Mills, , and B. Timbal, 2011: Large-scale indicators of Australian east coast lows and associated extreme weather events. Centre for Australia Weather and Climate Research Tech. Rep. 37, 92 pp. [Available online at http://www.cawcr.gov.au/publications/technicalreports.php.]

  • Drosdowsky, W., 2005: The latitude of the subtropical ridge over eastern Australia: The L index revisited. Int. J. Climatol., 25, 12911299.

    • Search Google Scholar
    • Export Citation
  • Eady, E. T., 1949: Long waves and cyclones waves. Tellus, 1, 3352.

  • Fuenzalida, H. A., , R. Sánchez, , and R. D. Garreaud, 2005: A climatology of cutoff lows in the Southern Hemisphere. J. Geophys. Res., 110, D18101, doi:10.1029/2005JD005934.

    • Search Google Scholar
    • Export Citation
  • Furevik, T., , M. Bentsen, , H. Drange, , I. K. T. Kindem, , N. G. Kvamstø, , and A. Sorteberg, 2003: Description and evaluation of the Bergen climate model: ARPEGE coupled with MICOM. Climate Dyn., 21, 2751.

    • Search Google Scholar
    • Export Citation
  • Gordon, C., , C. Cooper, , C. A. Senior, , H. Banks, , J. M. Gregory, , T. C. Johns, , J. F. B. Mitchell, , and R. A. Wood, 2000: The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Climate Dyn., 16, 147168.

    • Search Google Scholar
    • Export Citation
  • Hemer, M. A., , K. L. McInnes, , and R. Ranasinghe, 2012: Climate and variability bias adjustment of climate model-derived winds for an east Australian dynamical wave model. Ocean Dyn., 62, 87104.

    • Search Google Scholar
    • Export Citation
  • Hendon, H. H., , D. W. J. Thompson, , and M. C. Wheeler, 2007: Australian rainfall and surface temperature variations associated with the Southern Hemisphere annular mode. J. Climate, 20, 24522467.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and P. J. Valdes, 1990: On the existence of storm-tracks. J. Atmos. Sci., 47, 18541864.

  • Johns, T. C., and Coauthors, 2006: The new Hadley Centre climate model (HadGEM1): Evaluation of coupled simulations. J. Climate, 19, 13271353.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471.

  • Keable, M., , I. Simmons, , and K. Keay, 2002: Distribution and temporal variability of 500-hPa cyclone characteristics in the Southern Hemisphere. Int. J. Climatol., 22, 131150.

    • Search Google Scholar
    • Export Citation
  • Kent, D. M., , D. Kirono, , B. Timbal, , and F. H. S. Chiew, 2013: Representation of the Australian sub-tropical ridge in the CMIP3 model. Int. J. Climatol., 33, 4857, doi:10.1002/joc.3406.

    • Search Google Scholar
    • Export Citation
  • Kocin, P. J., , and L. W. Uccellini, 2004: Overview. Vol. I, Northeast Snowstorms, Amer. Meteor. Soc., 296 pp.

  • Kounkou, R., , G. A. Mills, , and B. Timbal, 2009: A reanalysis climatology of cool season tornado environments over southern Australia. Int. J. Climatol., 29, 20792090, doi:10.1002/joc.1856.

    • Search Google Scholar
    • Export Citation
  • Leslie, L. M., , and M. S. Speer, 1998: Short-range ensemble forecasting of explosive Australian east coast cyclogenesis. Wea. Forecasting, 13, 822832.

    • Search Google Scholar
    • Export Citation
  • Lindzen, R. S., , and B. Farrell, 1980: A simple approximate result for the maximum growth rate of baroclinic instabilities. J. Atmos. Sci., 48, 16481654.

    • Search Google Scholar
    • Export Citation
  • Lu, J., , C. Deser, , and T. Reichle, 2009: Cause of the widening of the tropical belt since 1958. Geophys. Res. Lett., 36, L03803, doi:10.1029/2008GL036076.

    • Search Google Scholar
    • Export Citation
  • Marshall, G. J., 2003: Trends in the southern annular mode from observations and reanalyses. J. Climate, 16, 41344143.

  • Martin, G. M., , M. A. Ringer, , V. D. Pope, , A. Jones, , C. Dearden, , and T. J. Hinton, 2006: The physical properties of the atmosphere in the new Hadley Centre Global Environmental Model (HadGEM1). Part I: Model description and global climatology. J. Climate, 19, 12741301.

    • Search Google Scholar
    • Export Citation
  • McInnes, K. L., , L. M. Leslie, , and J. L. McBride, 1992: Numerical simulation of cut-off lows on the Australian east coast: Sensitivity to sea-surface temperature. Int. J. Climatol., 12, 783795, doi:10.1002/joc.3370120803.

    • Search Google Scholar
    • Export Citation
  • Mercer, A. E., , and M. B. Richman, 2007: Statistical differences of quasigeostrophic variables, stability, and moisture profiles in North American storm tracks. Mon. Wea. Rev., 135, 23122338.

    • Search Google Scholar
    • Export Citation
  • Mills, G. A., , R. Webb, , N. E. Davidson, , J. Kepert, , A. Seed, , and D. Abbs, 2010: The Pasha Bulker east coast low of 8 June 2007. Centre for Australia Weather and Climate Research Tech. Rep. 23, 62 pp. [Available online at http://www.cawcr.gov.au/publications/technicalreports.php.]

  • Ndarana, T., , and D. W. Waugh, 2010: The link between cut-off lows and Rossby wave breaking in the Southern Hemisphere. Quart. J. Roy. Meteor. Soc., 136, 869885.

    • Search Google Scholar
    • Export Citation
  • Nieto, R., and Coauthors, 2005: Climatological features of cutoff low systems in the Northern Hemisphere. J. Climate, 18, 30853103.

  • Otkin, J. A., , and J. E. Martin, 2004: A synoptic climatology of the subtropical kona storm. Mon. Wea. Rev., 132, 15021517.

  • Pinto, J. G., , T. Spangehl, , U. Ulbrich, , and P. Speth, 2006: Assessment of winter cyclone activity in a transient ECHAM4-OPYC3 GHG experiment. Meteor. Z., 15, 279291.

    • Search Google Scholar
    • Export Citation
  • Pook, M. J., , P. C. McIntosh, , and G. A. Meyers, 2006: The synoptic decomposition of cool-season rainfall in the southeastern Australian cropping region. J. Appl. Meteor. Climatol., 45, 11561170.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., , and I. N. Smith, 2007: Weakening of the Walker circulation and apparent dominance of El Niño both reach record levels, but has ENSO really changed? Geophys. Res. Lett., 34, L18702, doi:10.1029/2007GL030854.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., , and G. Kociuba, 2011: The impact of global warming on the Southern Oscillation Index. Climate Dyn., 37, 17451754.

  • Rayner, N. A., , D. E. Parker, , E. B. Horton, , C. K. Folland, , L. V. Alexander, , D. P. Rowell, , E. C. Kent, , and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Risbey, J. S., , M. J. Pook, , P. C. McIntosh, , C. C. Ummenhofer, , and G. Meyers, 2009: Characteristics and variability of synoptic features associated with cool season rainfall in southeastern Australia. Int. J. Climatol., 29, 15951613.

    • Search Google Scholar
    • Export Citation
  • Seluchi, M. E., , and A. C. Saulo, 1998: Possible mechanisms yielding an explosive coastal cyclogenesis over South America: Experiments using a limited area model. Aust. Meteor. Mag., 47, 309320.

    • Search Google Scholar
    • Export Citation
  • Simpson, R. H., 1952: Evolution of the Kona storm, a subtropical cyclone. J. Atmos. Sci., 9, 2435.

  • Sinclair, M. R., , J. A. Renwick, , and J. W. Kidson, 1997: Low-frequency variability of Southern Hemisphere sea level pressure and weather system activity. Mon. Wea. Rev., 125, 25312543.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., , D. Qin, , M. Manning, , Z. Chen, , M. Marquis, , K. Averyt, , M. Tignor, , and H. L. Miller, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

  • Speer, M. S., , P. Wiles, , and A. Pepler, 2009: Low pressure systems off the New South Wales coast and associated hazardous weather: Establishment of a database. Aust. Meteor. Oceanogr. J, 58, 2939.

    • Search Google Scholar
    • Export Citation
  • Tanaka, H. L., , N. Ishizaki, , and A. Kitoh, 2004: Trend and interannual variability of Walker, monsoon and Hadley circulations defined by velocity potential in the upper troposphere. Tellus, 56A, 250269.

    • Search Google Scholar
    • Export Citation
  • Timbal, B., , and W. Drosdowsky, 2012: The relationship between the decline of southeastern Australia rainfall and the strengthening of the subtropical ridge. Int. J. Climatol., doi:10.1002/joc.3492, in press.

    • Search Google Scholar
    • Export Citation
  • Troup, A. J., 1965: The ‘southern oscillation.’ Quart. J. Roy. Meteor. Soc., 91, 490506, doi:10.1002/qj.49709139009.

  • Ulbrich, U., , A. H. Fink, , M. Klawa, , and J. G. Pinto, 2001: Three extreme storms over Europe in December 1999. Weather, 56, 7080.

  • Uppala, S. P., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 29613012.

  • Uppala, S. P., , D. Dee, , S. Kobayashi, , P. Berrisford, , and A. Simmons, 2008: Towards a climate data assimilation system: Status update of ERA-Interim. ECMWF Newsletter, No. 115, ECMWF, Reading, United Kingdom, 12–18.

  • Vecchi, G. A., , B. J. Soden, , A. T. Wittenberg, , I. A. Held, , A. Leetma, , and M. J. Harrison, 2006: Weakening of the tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature, 441, 7376.

    • Search Google Scholar
    • Export Citation
  • Whetton, P., , I. Macadam, , J. Bathols, , and J. O’Grady, 2007: Assessment of the use of current climate patterns to evaluate regional enhanced greenhouse response patterns of climate models. Geophys. Res. Lett., 34, L14701, doi:10.1029/2007GL030025.

    • Search Google Scholar
    • Export Citation
  • Whittaker, L. M., , and L. M. Horn, 1981: Geographical and seasonal distribution of North American cyclogenesis, 1958–1977. Mon. Wea. Rev., 109, 23122322.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Map of Australia showing the region used for the diagnostic method (rectangle), as well as the region used for the observed ECL database (solid black area).

  • View in gallery

    Average monthly number of diagnostic events from 1979 to 2001 for ERA-40 (dotted blue), as well as from 1979 to 2010 for NNR (dashed green) and ERAI (solid orange).

  • View in gallery

    Total number of diagnostic events per year for ERA-40 (blue), NNR (green), and ERAI (orange). Best linear fits to each dataset are shown.

  • View in gallery

    Number of times that the diagnostic is above threshold for (a) ERAI, (b) NNR, and (c) ERA-40, during (left) November–April and (right) May–October.

  • View in gallery

    Diagnostic threshold (calculated for each dataset as the 90th percentile of cyclonic geostrophic vorticity) plotted against horizontal resolution. This is shown for reanalyses (red: ERAI, squares; ERA-40, plus signs, and NNR, times signs) and GCMs (blue: HADCM3.0, asterisks; CCSM3.0, triangles; and BCM2.0, diamonds). The best linear fit to the reanalyses is shown (solid red line) as well as lines parallel to this incorporating the results from all three reanalyses (dotted red lines).

  • View in gallery

    As for Fig. 2, but for GCM data: (a) HADCM3.0 for twentieth (solid) and twenty-first century (dotted), and (b) BCM2.0 for the twentieth (solid), mid-twenty-first (dashed), and late twenty-first century (dotted).

  • View in gallery

    The number of diagnostic events per year for HADCM3.0 (orange) and BCM2.0 (blue). Horizontal bars represent the average values during the periods shown.

  • View in gallery

    Number of times that the diagnostic is above threshold during (left) November–April and (right) May–October, derived from HadCM3.0, for (a) the twentieth and (b) the twenty-first century, and (c) the difference (twenty-first century − twentieth century).

  • View in gallery

    As for Fig. 8, but for the BCM2.0 dataset.

  • View in gallery

    Monthly correlations between the diagnostic calculated from (a) ERAI, (b) NNR, and (c) ERA-40 reanalyses and indices of large-scale circulation (SAM: blue, SOI: green, STR-I: orange, EAC: red). Dotted lines indicate the 95% confidence intervals using a two-sided t-test statistic.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 63 63 16
PDF Downloads 28 28 7

Changes in the Risk of Extratropical Cyclones in Eastern Australia

View More View Less
  • 1 Centre for Australian Weather and Climate Research, Melbourne, Australia
© Get Permissions
Full access

Abstract

The east coast of Australia is a region of the world where a particular type of extratropical cyclone, known locally as an east coast low, frequently occurs with severe consequences such as extreme rainfall, winds, and waves. The likelihood of formation of these storms is examined using an upper-tropospheric diagnostic applied to three reanalyses and three global climate models (GCMs). Strong similarities exist among the results derived from the individual reanalyses in terms of their seasonal variability (e.g., winter maxima and summer minima) and interannual variability. Results from reanalyses indicate that the threshold value used in the diagnostic method is dependent on the spatial resolution. Results obtained when applying the diagnostic to two of the three GCMs are similar to expectations given their spatial resolutions, and produce seasonal cycles similar to those from the reanalyses. Applying the methodology to simulations from these two GCMs for both current and future climate in response to increases in greenhouse gases indicates a reduction in extratropical cyclone occurrence of about 30% from the late twentieth century to the late twenty-first century for eastern Australia. In addition to the absolute risk of formation of these extratropical cyclones, spatial climatologies of occurrence are examined for the broader region surrounding eastern Australia. The influence of large-scale modes of atmospheric and oceanic variability on the occurrence of these storms in this region is also discussed.

Corresponding author address: Andrew Dowdy, Bureau of Meteorology, 700 Collins St., Docklands, VIC 3008, Australia. E-mail: a.dowdy@bom.gov.au

Abstract

The east coast of Australia is a region of the world where a particular type of extratropical cyclone, known locally as an east coast low, frequently occurs with severe consequences such as extreme rainfall, winds, and waves. The likelihood of formation of these storms is examined using an upper-tropospheric diagnostic applied to three reanalyses and three global climate models (GCMs). Strong similarities exist among the results derived from the individual reanalyses in terms of their seasonal variability (e.g., winter maxima and summer minima) and interannual variability. Results from reanalyses indicate that the threshold value used in the diagnostic method is dependent on the spatial resolution. Results obtained when applying the diagnostic to two of the three GCMs are similar to expectations given their spatial resolutions, and produce seasonal cycles similar to those from the reanalyses. Applying the methodology to simulations from these two GCMs for both current and future climate in response to increases in greenhouse gases indicates a reduction in extratropical cyclone occurrence of about 30% from the late twentieth century to the late twenty-first century for eastern Australia. In addition to the absolute risk of formation of these extratropical cyclones, spatial climatologies of occurrence are examined for the broader region surrounding eastern Australia. The influence of large-scale modes of atmospheric and oceanic variability on the occurrence of these storms in this region is also discussed.

Corresponding author address: Andrew Dowdy, Bureau of Meteorology, 700 Collins St., Docklands, VIC 3008, Australia. E-mail: a.dowdy@bom.gov.au

1. Introduction

Extratropical cyclones that develop near the east coast of Australia are known locally as east coast lows (ECLs). The term ECL is generally used to refer to a low pressure system with a closed cyclonic circulation at sea level that forms and/or intensifies in a maritime environment within the vicinity of the east coast of Australia (e.g., Speer et al. 2009).

The hurricane-force winds, heavy rain, and large seas and swells associated with many intense ECLs can cause major socioeconomic impacts on eastern Australia, while providing a large proportion of the major inflows to urban water storages. One such ECL is that which occurred in early June 2007 near the city of Newcastle in New South Wales (NSW). This event, known as the “Pasha Bulker storm,” caused extensive flooding, the grounding of the bulk coal-carrying ship the Pasha Bulker, 10 deaths, and insurance claims of around $1.4 billion (Australian dollars), making it one of the most costly natural disasters in Australia’s history (Mills et al. 2010).

Small-scale explosively developing extratropical cyclones are responsible for the occurrence of a wide variety of extreme weather events throughout many different regions of the world. In Hawaii, where they are known as kona lows, they result in violent hazards such as severe winds, waterspouts, landslides, hailstorms, flash floods, and severe thunderstorms (Otkin and Martin 2004). In the eastern United States, extratropical cyclones can be responsible for causing a variety of different extreme weather events, including severe snowstorms (Kocin and Uccellini 2004). Mercer and Richman (2007) showed that quasigeostrophic diagnostics could be used to distinguish among the North American cyclone types identified by Whittaker and Horn (1981) (i.e., the east coast cyclone, the Colorado cyclone, and the Alberta Clipper), and that North American east coast cyclones have stronger magnitudes of the quasigeostrophic diagnostics compared to the other two types of storms. In Europe these storms have produced intense wind (Burt and Mansfield 1988; Ulbrich et al. 2001), while in South America they have caused severe flooding in some heavily populated regions of Argentina (Seluchi and Saulo 1998). A common feature of these storms is that they are fundamentally baroclinic in nature rather than barotropically driven (i.e., distinct from tropical cyclones).

Mills et al. (2010) argue that many of the more intense Australian ECLs show strong structural and dynamical similarities to the subtropical lows described by Simpson (1952) and Otkin and Martin (2004). An example of the spatial scale of the intense weather impacts associated with an intense ECL can be seen in the Pasha Bulker storm, where the strong wind band and rainband associated with the storm was on the order of 100 km wide at most [see Figs. 26 and 29 in Mills et al. (2010)]. The strong wind bands in the intense European windstorms such as the Great Storm of October 1987 (Burt and Mansfield 1988) or the Lothar windstorm of 1999 (Ulbrich et al. 2001) had similar lateral scales. Additionally, the intensification of these storms can occur with relatively short temporal scales. For example, Leslie and Speer (1998) reported a drop in central pressure of 12 hPa within 12 h for the ECL of 30–31 August 1996.

The impact of these ECLs means that any potential trends in their frequency has particular importance for both coastal infrastructure planning and for water resources planning along the eastern seaboard of Australia. However, although well represented by numerical weather prediction (NWP) models, the small spatial and temporal scale of many ECLs means that they are poorly represented by current global climate models (GCMs), which are typically run with much wider grid spacing than NWP models. The global nature of GCMs and their requirement to solve complex dynamic interactions affecting many different components of the earth’s climate, as well as the lengthy integration times needed, necessitate the use of relatively coarse spatial and temporal resolution (typically 100–500 km), with GCMs being optimized for producing accurate estimates of quantities such as mean temperature trends over large geographic regions, rather than trends in small-scale extreme meteorological phenomena such as ECLs.

Mills et al. (2010) noted that for all 10 of the highest-impact ECLs post-1970 listed by the New South Wales Regional Forecasting Centre of the Australia Bureau of Meteorology (http://www.bom.gov.au/nsw/sevwx/facts/ecl.shtml), a precursor was the development of a large-scale upper-tropospheric cutoff low or large-amplitude trough. They then hypothesized that if this relationship could be demonstrated to hold for a wide range of ECL events, then trends in the frequency of upper-tropospheric cutoff lows in GCM future climate simulations could become a proxy for ECL frequency in coarse-resolution GCM simulations.

Dowdy et al. (2011) tested the relationship between upper-cutoff lows and ECL development by comparing diagnostic indicators of upper-tropospheric trough or low presence with the database of ECLs described by Speer et al. (2009). The diagnostic quantities were based on three different vorticity measures: geostrophic vorticity, isentropic potential vorticity, and the forcing term of the pseudopotential vorticity form of the quasigeostrophic height tendency equation [following Bluestein (1992); see his Eq. (5.8.15)]. These quantities were calculated at the 300-, 400-, and 500-hPa pressure levels from the interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis [ERA-Interim (ERAI); Uppala et al. 2008] from 1989 to 2006.

Of all the diagnostic quantities examined, the extreme cyclonic values of the 500-hPa geostrophic vorticity (greater than ~2 × 10−4 s−1) provided the best representation of the likely occurrence of an ECL, correctly identifying 202 ECLs for the period 1989–2006, with 164 missed events and 127 false alarms, based on a database of observed ECL events (Speer et al. 2009). Although this observed ECL database is currently the best of its kind available, at least some of the false alarms and missed events are due to the somewhat subjective criteria, based on the presence of a closed surface low, used in its definition. Consequently, the observed ECL database may have limitations such as not providing a favorable representation of storms resulting from open systems (i.e., regions of high vorticity without necessarily being closed low pressure systems) or systems that could be well defined at higher altitudes but poorly defined near the surface, as discussed by Dowdy et al. (2011).

A variety of measures have been used in previous studies to examine extratropical cyclogenesis, including baroclinicity measures such as the Eady growth rate (Eady 1949). To compare the diagnostic method used in this study (i.e., based on geostrophic vorticity) with one based on a baroclinicity measure, the Eady growth rate was calculated following the method of Lindzen and Farrell (1980) and Hoskins and Valdes (1990), applied to ERA-Interim reanalyses at 500- and 700-hPa pressure levels.

The relative skill of various diagnostic indicators can be measured using the critical success index (CSI), calculated as the number of hits divided by the sum of the number of hits, misses, and false alarms, thereby rewarding hits while penalizing misses and false alarms. Applying the diagnostic method (as used for geostrophic vorticity) to the Eady growth rate resulted in a maximum CSI value of 0.20 in relation to the database of observed ECL events (Speer et al. 2009), corresponding to 87 hits, 212 missed events, and 143 false alarms. This CSI value is considerably lower than the case for geostrophic vorticity (CSI = 0.41), highlighting the value of the diagnostic method based on geostrophic vorticity as a tool for indicating the likelihood of ECL occurrence.

The diagnostic method developed by Dowdy et al. (2011), based on geostrophic vorticity at 500 hPa, was intended to be coarse enough in scale (both spatial and temporal) to be applicable to current GCMs. The diagnostic was designed to provide an indication of environments favorable to the likely formation of ECLs, so as to examine potential changes in their climatology, rather than for use as a highly accurate forecasting tool for application to finescale NWP models. A clear diagnostic signal was noted by Dowdy et al. (2011) up to 2 days prior to the occurrence of the ECLs, based on composites of cyclonic geostrophic vorticity maxima calculated for different time lags with respect to the day of the ECL event. Results are interpreted throughout this study being in mind the coarse scale of the diagnostic method.

Results of Dowdy et al. (2011) suggest that there is a useful relationship between the occurrence of upper tropospheric lows over eastern Australia and the occurrence of ECLs. In this paper we extend the analysis in two directions. First, we test the sensitivity of the results gained by use of ERAI to the coarser-resolution National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) reanalyses (NNR; Kalnay et al. 1996) and ERA-40 reanalyses (Uppala et al. 2005). This results in a 30-yr climatology of risk of ECL occurrence, enabling an assessment of the influence of large-scale modes of atmospheric and oceanic variability on the risk of ECL formation, as well as enabling an assessment of sensitivity to reanalysis technique and resolution. Second, the method is applied to GCMs to determine the possible impact of anthropogenic climate change on the risk of formation of ECLs into the late twenty-first century.

This paper is structured as follows. The data and methods used are detailed in section 2. In section 3, we describe how the diagnostic is applied to a number of different reanalyses, resulting in an objectively derived 30-yr climatology of the risk of formation of ECL events. As summarized in section 4, the diagnostic is then applied to GCM data to examine how ECLs could be expected to occur in the future, including examinations of interannual, intraseasonal, and spatial variability. Section 5 examines the relationships between the variability of the ECL diagnostic and a number of larger-scale circulation indices (e.g., Southern Oscillation index and southern annular mode). Conclusions are presented in section 6.

2. Data and methodology

We use the diagnostic developed by Dowdy et al. (2011). It is based on the 500-hPa geostrophic vorticity ξ calculated as the Laplacian of geopotential divided by the Coriolis parameter:
e1
where f is the Coriolis parameter and the Laplacian of geopotential.

The diagnostic is produced by first calculating a time series of the minimum value of the 500-hPa geostrophic vorticity within a geographic region of 15° in longitude and 10.5° in latitude (originally determined by the 1.5° resolution of the ERAI dataset) centered on 29°S and 153°E (Fig. 1). The minimum value is selected as this represents the maximum cyclonic vorticity, given that cyclonic vorticity is negative in sign for the Southern Hemisphere. A 1-day running mean is applied to the 6-hourly time series to reduce small-scale temporal variability, as the intention of this study is to use a diagnostic method large enough in scale (spatial and temporal) to be suitable for potential application to GCMs. Days on which the time series exceeds a threshold value, selected as being lower than the 10th percentile of geostrophic vorticity (equivalent to being higher than the 90th percentile of cyclonic vorticity in the Southern Hemisphere), are defined as being indicative of the likely occurrence of ECL formation. This percentile level was chosen as detailed in Dowdy et al. (2011), based on an examination of time series of the diagnostic in relation to days on which ECLs were listed to have occurred in the Speer et al. (2009) dataset, with approximately 1 in 10 days on average being listed in the dataset as corresponding to an ECL event (noting that many ECLs last for multiple days).

Fig. 1.
Fig. 1.

Map of Australia showing the region used for the diagnostic method (rectangle), as well as the region used for the observed ECL database (solid black area).

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

This diagnostic method is not intended to be a direct indication of the occurrence, or nonoccurrence, of an ECL event. Rather, it is an indication of the likelihood of the risk of formation of an ECL, with results presented here being interpreted in this way. It thus differs from some other climatologies of upper tropospheric lows (e.g., Sinclair et al. 1997; Keable et al. 2002; Nieto et al. 2005; Fuenzalida et al. 2005; Pook et al. 2006; Ndarana and Waugh 2010). Many of these use objective methods to populate the climatologies, often seeking local minima in quantities such as the geopotential height field, with other secondary criteria often being applied such as a requirement that the system is “cold-cored” or that the zonal winds reverse on the poleward side of the height minima. The diagnostic method applied in this study does not require a closed low and thus does not discriminate between shear versus curvature vorticity, which have the same dynamic effect of differential advection of vorticity forcing surface development of the storm.

Three different reanalyses are used in this study, allowing for comparisons of results between independent datasets with different spatial resolutions. ERAI and 40-yr ECMWF Re-Analysis (ERA-40) data are used from the European Centre for Medium-Range Weather Forecasts, as well as the NNR reanalyses from the National Centers for Environmental Prediction and National Center for Atmospheric Research. The latitude–longitude grid spacings of the reanalyses are 1.5° for ERAI, 2.5° for ERA-40, and 1.875° for NNR, although the effective grid spacing of the upper-level fields in NNR is 2.5°. The three reanalysis datasets all have a 6-hourly temporal resolution.

It is well known that reanalyses are less reliable prior to the satellite era, and previous studies relying on upper tropospheric signatures to detect the risk of formation of extremes show that it is preferable not to use reanalyses prior to 1979 in the Southern Hemisphere (Kounkou et al. 2009). Reanalyses are used here from 1979 until 2010, with the exception of ERA-40 for which data are only available up to 2001 inclusive.

Climatologies of the diagnostic are produced from all three reanalyses and examined in terms of their interannual, seasonal, and spatial variability, as well as their long-term trend. Spatial variability is examined by calculating the diagnostic centered on different geographic positions, while using the same threshold for all locations based on the 90th percentile of the diagnostic centered on 29°S and 153°E (as shown in Fig. 1), as well as using the same sized geographic area for the diagnostic region (or as close as possible given the resolution of the individual datasets).

A set of GCM experiments has been produced in conjunction with the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4; Solomon et al. 2007): phase 3 of the World Climate Research Program (WCRP) Coupled Model Intercomparison Project (CMIP3). The application of the diagnostic to GCMs requires spatial fields of the 500-hPa geopotential with daily (or shorter) temporal resolution. Twenty-five GCMs are included in the CMIP3 dataset although, because of the data requirements, only three models were available for this study: the third climate configuration of the Met Office Unified Model (HadCM3.0; Gordon et al. 2000; Martin et al. 2006; Johns et al. 2006), the Bjerknes Centre for Climate Research (BCCR) Bergen Climate Model version 2 (BCM2.0; Furevik et al. 2003) and NCAR’s Community Climate System Model, version 3 (CCSM3.0; Collins et al. 2006a,b).

The longitudinal and latitudinal resolutions, respectively, of the GCM data are approximately 2.8° and 2.8° for BCM2.0, 3.75° and 2.5° for HadCM3.0, and 1.4° and 1.4° for CCSM3.0. Data from all three GCM datasets were output at 24-h intervals (valid at 0000 UTC for HadCM3.0 and CCSM3.0 and 1200 UTC for BCM2.0) for time slices from 1960 to 1988 and 2070 to 2098 for HadCM3.0; from 1961 to 1998, 2046 to 2051, 2054 to 2065, and 2081 to 2100 for BCM2.0; and from 1960 to 1999 and 2000 to 2099 for CCSM3.0.

The GCM data for the twenty-first century used in this study for HadCM3.0 and BCM2.0 use a high emission scenario, A2, whereas the data from CCSM3.0 use a middle-high range emission scenario, A1B. Of the six scenarios available from the CMIP3 dataset, the A2 and A1B scenarios are in the higher part of the spectrum in terms of greenhouse gas emissions, which is desirable since observations of greenhouse gas emissions have consistently exceeded expectations of the emission scenarios noted in the Special Report on Emissions Scenarios (SRES; Canadell et al. 2007). As an example, the global average temperature rise for the A2 scenario is expected to be on the order of 2.0° to 5.4°C by the end of the twenty-first century (Solomon et al. 2007), with the temperature range reflecting the range of sensitivity of the individual climate models to the external forcings.

The influence of large-scale modes of atmospheric and oceanic variability on the diagnostic is examined. Indices used are the southern annular mode (SAM: Marshall 2003) and the Southern Oscillation index (SOI: Troup 1965), calculated by Australia’s National Climate Centre. In addition, we use a regional mean sea level pressure (MSLP)-based indicator of the strength of the subtropical ridge (STR-I) over eastern Australia (Drosdowsky 2005) as it has been proposed as a key controller of rainfall across most of southeastern Australia (Timbal and Drosdowsky 2012). Finally, we develop a crude indicator of the strength of the East Australia Current (EAC) based on the average sea surface temperature (SST) in the region bounded by 150°–170°E and 25°–35°S from version 1 of the Hadley Center SST dataset (HadSST-1; Rayner et al. 2003), as sea surface temperature has been proposed as a factor in the development of ECLs (McInnes et al. 1992).

3. Climatology of the risk of ECL formation

a. Temporal variability

Figure 2 shows the monthly average number of diagnostic ECL events for each of the three different reanalyses: NNR, ERAI, and ERA-40. There is a strong similarity among the results obtained from the three reanalyses. The highest monthly average numbers of events occur in July for all three datasets, and the four highest monthly values occur in June, July, August, and September in all three datasets. The two lowest monthly values occur in January and February in all three datasets.

Fig. 2.
Fig. 2.

Average monthly number of diagnostic events from 1979 to 2001 for ERA-40 (dotted blue), as well as from 1979 to 2010 for NNR (dashed green) and ERAI (solid orange).

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

Figure 3 shows the annual number of days for which the diagnostic exceeded its threshold value for each of the three different reanalyses. Large interannual variability is apparent, ranging from about 50 (1989 and 2001) to about 25 (1994, 1997, and 2009). A high degree of consistency is apparent among the three different datasets in terms of how many diagnostic ECL events occur in each individual year, with Pearson correlation coefficients above 0.85 for all pairing combinations of the three reanalyses. Lines of best fit using least absolute deviation have a slight downward slope: −0.9, −0.4, and −0.7 events per year for NNR, ERAI, and ERA-40, respectively. None of these slopes is significant at any meaningful confidence level, as interannual variability is large and the length of the record we decided to use (i.e., post-satellite era) is fairly short. Assessment of long-term climatological trends requires the use of a suitably long period of homogeneous data, with the period of data available for this study (~30 yr) being close to the minimum length that is generally considered to be adequate for such examinations, so it is therefore noted that a longer period of data potentially may, or may not, show a statistically significant trend.

Fig. 3.
Fig. 3.

Total number of diagnostic events per year for ERA-40 (blue), NNR (green), and ERAI (orange). Best linear fits to each dataset are shown.

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

The high degree of consistency among the results obtained from the three different reanalyses provides confidence in the applicability of the diagnostic method to different spatial resolutions. The resultant 30-yr climatology of the risk of formation of ECL events is the largest produced using an objective methodology, making it unambiguous and allowing reproducibility.

b. Spatial variability

Spatial fields of the average annual number of diagnostic ECL events for each reanalysis are shown in Fig. 4. Maximum values are located between about 30° and 40°S centered over the east coast of Australia during winter (right panels) for all three datasets, but farther poleward during summer (left panels) centered roughly over Tasmania (i.e., at about 150°E between 40° and 45°S).

Fig. 4.
Fig. 4.

Number of times that the diagnostic is above threshold for (a) ERAI, (b) NNR, and (c) ERA-40, during (left) November–April and (right) May–October.

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

The winter maxima in occurrence of ECL diagnostic events over the east coast of Australia are consistent with the location of local maxima in cutoff low occurrence reported in other studies (Ndarana and Waugh 2010; Fuenzalida et al. 2005). Following the arguments of Risbey et al. (2009), this could be partly attributable to the location of the poleward exit of the mean wintertime subtropical jet in this region, the strength and position of which is determined by the cold continent with some contribution from atmospheric blocking.

4. Application of the diagnostic to GCM data

In the previous section it was shown that there is a high degree of consistency among the climatologies of the risk of ECL formation obtained from the three different reanalyses. This provides confidence in the robustness of the diagnostic method to models of different spatial resolution. As a further check of the applicability of the method to GCM data, this section first examines the diagnostic threshold values obtained when the diagnostic method is applied to GCM datasets. The interannual, intraseasonal, and spatial variability of the diagnostic results derived from the GCMs are then examined, providing further means of assessing the applicability of the method to GCM data.

a. The influence of model resolution on the diagnostic threshold

Diagnostic threshold values are shown in Fig. 5 as a function of the spatial resolution of all of the reanalyses and GCMs used in this study. The spatial resolution is calculated as the average of their individual latitudinal and longitudinal resolutions, with the latitudinal resolution scaled by cos(35°) to provide a balance between the distances represented by latitude and longitude at the approximate latitude of the center of the east coast of Australia. The diagnostic threshold is calculated based on percentile values (as described in section 2), with the twentieth-century GCM data being used to calculate the diagnostic threshold for consistency with the reanalysis results.

Fig. 5.
Fig. 5.

Diagnostic threshold (calculated for each dataset as the 90th percentile of cyclonic geostrophic vorticity) plotted against horizontal resolution. This is shown for reanalyses (red: ERAI, squares; ERA-40, plus signs, and NNR, times signs) and GCMs (blue: HADCM3.0, asterisks; CCSM3.0, triangles; and BCM2.0, diamonds). The best linear fit to the reanalyses is shown (solid red line) as well as lines parallel to this incorporating the results from all three reanalyses (dotted red lines).

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

The reanalysis results show that finer-resolution grid spacing generally results in higher-magnitude diagnostic threshold values. The threshold value resulting from applying the diagnostic method to the HadCM3.0 dataset is broadly consistent with expectations based on the thresholds obtained from reanalyses. In contrast, the threshold value for the CCSM3.0 dataset is considerably lower than expected given its finer grid spacing. The threshold value obtained for the BCM2.0 dataset is also lower than could be expected based on the reanalyses, although to a somewhat lesser degree than for CCSM3.0, as the CCSM3.0 threshold value is more than twice as far from the range of values incorporating all three reanalyses (dotted lines in Fig. 5) than the BCM2.0 threshold value.

b. Temporal variability from GCM data

The intraseasonal variability for the HadCM3.0 dataset (Fig. 6a) is broadly similar to that of the reanalyses, with the highest monthly average values occurring in July during the twentieth century. The events are somewhat more tightly grouped around the July maxima than was observed for the reanalyses, particularly for the BCM2.0 dataset, suggesting the possibility that the diagnostic method applied to the GCM data might be underestimating the number of summer events that occur.

Fig. 6.
Fig. 6.

As for Fig. 2, but for GCM data: (a) HADCM3.0 for twentieth (solid) and twenty-first century (dotted), and (b) BCM2.0 for the twentieth (solid), mid-twenty-first (dashed), and late twenty-first century (dotted).

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

The largest projected changes from the twentieth to the twenty-first century occur in July for both the HadCM3.0 and BCM2.0 results. The largest magnitude decreases tend to span the winter months (April–November for HadCM3.0 and May–October for BCM2.0), with little change indicated by both models from December to March, while noting that there is a high degree of variability in the results from month to month.

The CCSM3.0 model did not produce a realistic annual cycle, providing further confirmation (in addition to its threshold value discussed previously) that this particular dataset is not suitable for application of the diagnostic. Potential reasons for this were investigated. It was found that spatial climatologies of the 500-hPa geopotential used by the diagnostic method show small-scale features (local maxima and minima) that make it difficult for the diagnostic to be applied, since the diagnostic method is based on identifying local maxima in geostrophic vorticity from spatial fields of geopotential [Eq. (1)]. A clear explanation for the origin of the small-scale spatial features in the geopotential fields was not apparent.

Figure 7 shows the number of diagnostic events per year calculated for the HadCM3.0 and BCM2.0 datasets. Horizontal bars represent the average values during the periods shown. As the diagnostic method uses a threshold defined by a percentile value, there will be no difference in the mean values for the twentieth century among any of the GCMs or reanalyses.

Fig. 7.
Fig. 7.

The number of diagnostic events per year for HADCM3.0 (orange) and BCM2.0 (blue). Horizontal bars represent the average values during the periods shown.

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

The standard deviations of the twentieth-century GCM-derived data shown in Fig. 7 are 8.2 and 8.8 events per year for the HadCM3.0 and BCM2.0 models, respectively. This is reasonably similar in magnitude to that of the reanalysis climatologies (from Fig. 3) of 6.8, 8.4, and 7.6, for ERAI, NNR, and ERA-40, respectively, providing some degree of encouragement that the GCMs are capturing the variability of ECL occurrence based on the application of the diagnostic method. This result also provides some degree of confidence in the results obtained for the twenty-first century, as the ability of a GCM to produce a reliable simulation of the twentieth century is often used when evaluating the ability of a climate model to produce simulations of future climate (Whetton et al. 2007; CSIRO and Bureau of Meteorology 2007). The standard deviations for the late twenty-first century results are similar to those of the twentieth century for HadCM3.0, but not for BCM2.0 in which a standard deviation of 5.2 occurs. This may be partly due to the short time period over which this is calculated, as well as the reduction in the mean value as compared to the results for the twentieth century. A measure of variability that compensates for a change in the mean value is the difference between the 90th percentile and 10th percentile, divided by the 50th percentile. This quantity, calculated for the geostrophic vorticity used by the diagnostic method, does not show much change between the twentieth and late twenty-first centuries, with values of 2.6 and 2.5 for HadCM3.0 and 3.6 and 3.8 for BCM2.0.

Figure 7 shows a reduction in the average annual number of events from the twentieth to the twenty-first century for both HADCM3.0 and BCM2.0 simulations. The HadCM3.0 results indicate a reduction of 29%, from 36.0 to 25.6 events per year. The BCM2.0 results indicate a similar magnitude reduction of 34% by the later part of this century (corresponding to 24.1 events per year on average), with a reduction of 10% (from 36.5 to 32.8 events) already apparent in the mid-twenty-first century.

As was the case for the results from reanalyses (Fig. 3), a significant long-term trend in the annual number of diagnostic events in the twentieth century was not found. However, as was noted previously, the lack of a significant trend could potentially be related to the limited period of data (i.e., relatively short in terms of assessing long-term climatological trends) combined with the amount of interannual variability in the number of diagnostic events. The slope of fitted lines to the twentieth-century data shown in Fig. 7 was −0.02 events per year for HadCM3.0 and −0.21 events per year for BCM2.0, somewhat smaller than the fitted slopes for the three reanalyses ranging from −0.4 to −0.9 (Fig. 3).

c. Spatial variability from GCM data

To examine the possibility of a spatial shift in the favored region for these storms to occur, spatial climatologies of the annual number of diagnostic events (for summer and winter) are shown for the twentieth and twenty-first centuries, along with the difference (twenty-first − twentieth century), derived from HadCM3.0 (Fig. 8) and BCM2.0 (Fig. 9). The latter period of twenty-first century is used for BCM2.0 to allow a more direct comparison with HadCM3.0.

Fig. 8.
Fig. 8.

Number of times that the diagnostic is above threshold during (left) November–April and (right) May–October, derived from HadCM3.0, for (a) the twentieth and (b) the twenty-first century, and (c) the difference (twenty-first century − twentieth century).

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

Fig. 9.
Fig. 9.

As for Fig. 8, but for the BCM2.0 dataset.

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

For the twentieth century, local maxima occur during winter over the east coast of Australia for both HadCM3.0 and BCM2.0 (top right panels of Figs. 8 and 9). The winter maxima are located in the region around 30°S for HadCM3.0 and between about 25° and 30°S for BCM2.0, which is about 5°–10° equatorward of the latitudes indicated by the reanalyses (Fig. 4). The high number of events indicated in the Southern Ocean to the southwest of Australia is more likely related to midlatitude lows embedded in the westerlies [such as the storms described by Hoskins and Valdes (1990)] rather than storms that are subtropical in nature and cut off from the westerlies.

For the HadCM3.0 dataset during summer (top left panels of Fig. 8), local maxima do not occur over the east coast of Australia, with the higher values occurring at the latitude of Tasmania (at about 42°S). This is broadly similar to the results from reanalyses (Fig. 4), with one exception being that a well-defined local maximum is not apparent over Tasmania. In contrast, for the BCM2.0 dataset during summer (top left panel of Fig. 9), there is a strong indication of a poleward bias compared to reanalyses, which could account for the relative lack of summer events seen in the annual cycle of its climatology (Fig. 6).

For the twenty-first century, the two models are consistent with each other in showing a large reduction in winter events over the east coast of Australia between about 20° and 30°S. The HadCM3.0 results show that the wintertime local maxima of diagnostic events over eastern Australia weakens, with the reduction being somewhat less pronounced over the far southeast corner of the Australian continent. The BCM2.0 results show a large reduction in the magnitude of the local maxima over the east coast of Australia, without a significant change in its location.

A small increase in winter events is indicated for the twenty-first century over Tasmania and eastward toward New Zealand, although it should be noted that the diagnostic method was designed for use over the east coast of Australia and has not been verified against observations for other geographic regions. It should also be noted that the spatial pattern of this increase is not entirely consistent between both GCMs.

5. Relationship to large-scale circulation indices

In addition to the long-term projection of the risk of formation of ECLs, it is of interest to investigate the relationship between the variability of the diagnostic ECL events and a number of indices representing large-scale modes of internally generally natural variability on interannual time scales. The atmospheric indices used are the SOI, SAM, and STR-I (see details in section 2). Although the conceptual model of subtropical cyclone development is essentially baroclinic in nature, simple thermodynamic arguments suggest that variations in ocean temperatures could potentially influence the risk of ECL cyclogenesis by modulating the available energy and low-level moisture. Consequently, this section also examines the influence of oceanic circulation based on a measure of the strength of the East Australian Current (described in section 2).

The cross correlation between the monthly number of diagnostic events (from reanalyses) and the large-scale circulation indices, both with a 3-month running mean applied, was examined (Fig. 10). This was done by first removing the long-term temporal trends from each dataset by subtracting least absolute deviation linear fits to the data. Cross correlation coefficients were then computed between the resultant time series of diagnostic events and each of the indices of atmospheric and oceanic variability. None of the indices consistently shows high correlation coefficients, although this may partly be due to the relative shortness of the record (~30 yr). Some correlation coefficients reach values of 0.4–0.6, which, although weak, are still of interest.

Fig. 10.
Fig. 10.

Monthly correlations between the diagnostic calculated from (a) ERAI, (b) NNR, and (c) ERA-40 reanalyses and indices of large-scale circulation (SAM: blue, SOI: green, STR-I: orange, EAC: red). Dotted lines indicate the 95% confidence intervals using a two-sided t-test statistic.

Citation: Journal of Climate 26, 4; 10.1175/JCLI-D-12-00192.1

The STR-I, indicting the intensity of the subtropical ridge, produces correlation coefficients that are negative during every month for all three reanalyses, with peak values in the range of −0.5 to −0.6 that are significant at the 95% confidence interval occurring in August for all three reanalyses. Significant values at this confidence level do not occur for any other month of the year for any of the reanalyses. Negative correlation coefficients occur for each month for all three reanalyses. Given that recent studies suggest a strengthening of the subtropical ridge intensity (Kent et al. 2013; Timbal and Drosdowsky 2012), this implies a declining trend in ECL events.

The SOI, an atmospheric representation of the El Niño–Southern Oscillation (ENSO), produces correlation coefficients in the range of 0.3–0.5 during the winter months (e.g., May–August) and close to zero during the summer months (December–February). The only value above the 95% confidence interval occurs during July for NNR, but not for the other reanalyses. Recent studies suggest a trend in ENSO, strongly linked to a weakening of the Walker circulation, toward more frequent conditions that resemble El Niño rather than La Niña (Tanaka et al. 2004; Vecchi et al. 2006; Power and Smith 2007; Power and Kociuba 2011). As El Niño is characterized by strongly negative SOI values, the positive correlation coefficients during winter produced by the SOI are consistent with the expected reduction in the risk of formation of ECL events as indicated by the diagnostic.

The southern annular mode provides a measure of the strength of the westerlies at the latitude of the Australian continent. High positive values of the SAM are characterized by a poleward contraction of the midlatitude westerlies, leading to decreased rainfall over southeast and southwest Australia during winter, and increased rainfall over the southern east coast of Australia during summer (Hendon et al. 2007). The correlation coefficients between the diagnostic and the SAM are not very large and are not significant for any of the three reanalyses for any month of the year, although there is some consistency between the three reanalyses during spring (September–November) with weakly positive correlations being indicated (ranging from about 0.1 to 0.4).

The correlation coefficients are also not very large for the strength of the East Australian Current, representing the potential influence of local SSTs. This suggests very little relationship with the diagnostic ECL events (correlation coefficients in the range −0.3 to 0.3), although there is some suggestion that oceanic conditions could potentially play a role in ECL occurrence during winter, with weakly positive correlations between the diagnostic and the EAC from May to July for all three reanalyses. Additionally, the indication of a weak relationship between the diagnostic and the SOI during winter (with correlation coefficients ranging from 0.3 to 0.5 for all three reanalyses from May to August) further suggests that oceanic conditions could potentially have some influence on ECL formation given that ENSO is a coupled ocean–atmosphere phenomenon. However, the oceanic influence on ECL formation currently remains a significant knowledge gap in the literature.

6. Summary and conclusions

A diagnostic of the risk of ECL formation was applied to three reanalyses (ERAI, NNR, and ERA-40). The similarities in the resultant climatologies provide a high degree of confidence that this diagnostic method can be applied to datasets with differing spatial resolutions, including validating the method for the relatively coarse resolutions of most current GCMs. This allowed the diagnostic to be successfully applied to two GCM datasets (HadCM3.0 and BCM2.0) with the confidence that the method was producing reasonable results similar to expectations based on their particular spatial resolutions.

A high degree of consistency was found among the climatologies derived from the three reanalyses and two GCM datasets (for the late twentieth century), including their seasonal and spatial variability. An example of this is the local maxima in the risk of ECL formation that occur over the east coast of Australia during winter for all of these datasets. Other examples of the consistency among these datasets include July being the month with the highest average number of diagnostic ECL events, and the interannual variability in the number of diagnostic events being similar in magnitude for all datasets.

A significant inconsistency between the datasets is that the region of high diagnostic event occurrence during summer (i.e., greater than about 15 events per season) occurs poleward of about 40°S for BCM2.0 (Fig. 9, upper left panel), whereas it occurs poleward of about 35°S for HadCM3.0 (Fig. 8, upper left panel), which is more similar to where this occurs for the reanalyses (about 32°S for NNR and ERA-40 and about 30°S for ERAI, from Fig. 4). A large poleward bias in the BCM2.0 diagnostic results during summer would be expected to have the effect of reducing the value of the 90th percentile of the diagnostic (i.e., the threshold value). The HadCM3.0 spatial fields (Fig. 8) do not show as large a poleward deviation as compared to reanalyses, which is consistent with its threshold value being closer to the regression line of the reanalyses than occurs for the BCM2.0 (Fig. 5). Additional evidence for this interpretation of the results is provided by the seasonal variation of the diagnostic events (Fig. 6), highlighting the relative lack of summer events for the BCM2.0 results, with the HadCM3.0 results having a closer resemblance to the seasonal variability derived from the reanalyses (Fig. 2).

The diagnostic was also applied to a third GCM dataset, which had a diagnostic threshold considerably different than expected given its spatial resolution and did not realistically reproduce the annual variation. It was found that small-scale local maxima in the spatial fields of geopotential resulted in this dataset not being suitable for the application of the diagnostic method.

The climatologies derived from the three reanalyses represent the longest of their type, being based on a systematic method that is readily reproducible. All three climatologies indicate a weak decline in ECL frequency during the late twentieth century, although this trend is not statistically significant based on the available data. The two GCM-derived climatologies (from HadCM3.0 and BCM2.0) both indicate a reduction in the risk of extratropical cyclone occurrence of about 30% from the late twentieth century to the late twenty-first century for eastern Australia. The results obtained from these two GCMs were similar to each other in indicating a large reduction in winter events over eastern Australia from the twentieth to the twenty-first century. The HadCM3.0 results show that the wintertime local maxima of diagnostic events over eastern Australia weakens, with the reduction being somewhat less pronounced over the far southeast corner of the Australian continent. The BCM2.0 results show a large reduction in the magnitude of the local maxima over the east coast of Australia, without a significant change in its location.

This reduction of diagnostic events in the twenty-first century compared to the twentieth century is indicative of a lower frequency of occurrence of upper cutoff lows over eastern Australia. With a view to understanding the potential reasons for this reduction, the variability of the number of diagnostic events was examined in relation to a variety of large-scale indices of atmospheric and oceanic circulation. Few significant correlations were found to occur, indicating that these proxies are generally not suitable for use as diagnostics of ECL occurrence and that the physical reasons for the reduction in the risk of ECL formation may not simply be related to changes in these broad circulation indices.

The most significant correlations were found to be with the intensity of the subtropical ridge during spring, exhibiting correlation coefficients with magnitudes close to the 95% confidence interval for all three reanalysis datasets. The correlation was negative in sign for all months on the year, suggesting that the trend toward stronger subtropical ridge intensity noted in some recent studies (Kent et al. 2013; Timbal and Drosdowsky 2012) could potentially explain part of the reduction in risk of ECL formation. The SOI also showed some reasonably high correlation coefficients during winter. The sign of the correlation is consistent with the expected reduction in the risk of formation of ECL events given that recent studies suggest a trend in ENSO and the Walker circulation toward more frequent conditions resembling El Niño rather than La Niña (Tanaka et al. 2004; Vecchi et al. 2006; Power and Smith 2007; Power and Kociuba 2011). This is the subject of future investigations, as well as examining other potential influencing factors such as the influence of a changing climate on Rossby wave activity, the subtropical jet, or atmospheric blocking.

Although the diagnostic method used is unique to this study, the results presented here are broadly consistent with other studies showing a poleward shift in a variety of different phenomena, including the boundary between the temperate and tropical regions of the world (Lu et al. 2009), the position of the subtropical ridge (Kent et al. 2013), and midlatitude storm tracks (Pinto et al. 2006). Although it is often not possible to unambiguously attribute such changes to specific physical mechanisms, it is interesting that there appears to be a relatively consistent picture of poleward movement emerging from a variety of contrasting methods and phenomena.

An important conclusion of this study is the demonstration of the applicability of this diagnostic method to GCM data. The results presented here are encouraging, particularly in relation to the clear demonstration of the applicability of the diagnostic method to coarse resolution datasets. The diagnostic method produced similar climatologies for ERAI (with a relatively fine spatial resolution) as it did for NNR and ERA-40 (with coarser resolutions similar to that of GCMs). A sufficiently high degree of consistency was found among results obtained from different reanalyses, as well as among different GCM datasets.

The assessment of the influence of climate change on extreme events is important, as this is where many of the consequences of a changing climate are expected to be felt. The Fourth Assessment Report from the Intergovernmental Panel on Climate Change (Solomon et al. 2007) put a critical emphasis on the influence of anthropogenic climate change on extreme events. However, assessing the climatology of extreme events such as subtropical cyclones and their associated severe weather impacts is currently a field that is not well represented in the literature.

The database of observed ECLs (Speer et al. 2009) used by Dowdy et al. (2011) for developing the diagnostic method does not differentiate between the scale and intensity of the storms. The spatial and temporal scale of ECLs is currently a knowledge gap in the literature providing considerable scope for future research. It is intended that research will be undertaken to apply an objective methodology to examine extreme impacts associated with ECLs (based on rainfall observations, wave observations, and wind reanalyses), allowing a systematic examination of the spatial scale of ECL events that have high impact effects.

The diagnostic method used in this study is expected to be a useful and novel approach for examining projected changes in extreme events associated with ECLs, as this is currently a challenge using more traditional methods such as downscaled climate models with bias correction. For example, a recent study based on three different GCMs, downscaled to 60-km horizontal resolution for the Australian region, found that direct forcing of a wave model with the downscaled winds resulted in a suboptimal representation of the wave climate, and although this could be corrected to some degree with bias adjustment techniques, the bias-adjusted wind fields did not improve the ability of the model to reproduce the storm wave climate (Hemer et al. 2012).

As ECLs are poorly represented in GCMs, there is currently a high degree of uncertainty in projections of these extreme weather events. The reduction in the risk of ECLs identified in this paper could potentially lead to a reduction in, or change in distribution of, the rainfall of this region, as well as potential reductions in the damage resulting from extreme wind and wave events associated with ECLs. However, these conclusions are somewhat tentative given that they are derived from only two different GCMs and that they are based on the application of a diagnostic developed only from a dataset of observed ECL events. Consequently, the next stage of this project will examine potential changes in high impact effects (e.g., extreme ocean waves, heavy rainfall, and strong wind events) often associated with ECLs. There is also the potential to extend the diagnostic technique to include more sophisticated thresholding techniques, as well as to enhance the diagnostic method with the addition of some form of lower tropospheric response component such as static stability or Eady growth rate (Hoskins and Valdes 1990). It is also intended to apply the diagnostic method to a wide variety of different GCMs as data become available from the next iteration of the WCRP Coupled Model Intercomparison Project (phase 5; CMIP5).

Acknowledgments

This research was undertaken as part of the Australian Climate Change Science Program. We would like to acknowledge advice used for this paper provided by Richard Dare and James Risbey from the Centre for Australian Weather and Climate Research in their comments on earlier drafts of this study.

REFERENCES

  • Bluestein, H. B., 1992: Synoptic–Dynamic Meteorology in Midlatitudes. Vol. 1. Oxford University Press, 431 pp.

  • Burt, S. D., , and D. A. Mansfield, 1988: The great storm of 15–16 October 1987. Weather, 43, 90110.

  • Canadell, J. G., and Coauthors, 2007: Contributions to accelerating atmospheric CO2 growth from economic activity, carbon intensity, and efficiency of natural sinks. Proc. Natl. Acad. Sci. USA, 104, 18 86618 870.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2006a: The Community Climate System Model version 3 (CCSM3). J. Climate, 19, 21222143.

  • Collins, W. D., and Coauthors, 2006b: The formulation and atmospheric simulation of the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 21442161.

    • Search Google Scholar
    • Export Citation
  • CSIRO and Bureau of Meteorology, 2007: Climate change in Australia. Australian Greenhouse Office Tech. Rep., 148 pp. [Available online at http://www.climatechangeinaustralia.gov.au.]

  • Dowdy, A. J., , G. A. Mills, , and B. Timbal, 2011: Large-scale indicators of Australian east coast lows and associated extreme weather events. Centre for Australia Weather and Climate Research Tech. Rep. 37, 92 pp. [Available online at http://www.cawcr.gov.au/publications/technicalreports.php.]

  • Drosdowsky, W., 2005: The latitude of the subtropical ridge over eastern Australia: The L index revisited. Int. J. Climatol., 25, 12911299.

    • Search Google Scholar
    • Export Citation
  • Eady, E. T., 1949: Long waves and cyclones waves. Tellus, 1, 3352.

  • Fuenzalida, H. A., , R. Sánchez, , and R. D. Garreaud, 2005: A climatology of cutoff lows in the Southern Hemisphere. J. Geophys. Res., 110, D18101, doi:10.1029/2005JD005934.

    • Search Google Scholar
    • Export Citation
  • Furevik, T., , M. Bentsen, , H. Drange, , I. K. T. Kindem, , N. G. Kvamstø, , and A. Sorteberg, 2003: Description and evaluation of the Bergen climate model: ARPEGE coupled with MICOM. Climate Dyn., 21, 2751.

    • Search Google Scholar
    • Export Citation
  • Gordon, C., , C. Cooper, , C. A. Senior, , H. Banks, , J. M. Gregory, , T. C. Johns, , J. F. B. Mitchell, , and R. A. Wood, 2000: The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Climate Dyn., 16, 147168.

    • Search Google Scholar
    • Export Citation
  • Hemer, M. A., , K. L. McInnes, , and R. Ranasinghe, 2012: Climate and variability bias adjustment of climate model-derived winds for an east Australian dynamical wave model. Ocean Dyn., 62, 87104.

    • Search Google Scholar
    • Export Citation
  • Hendon, H. H., , D. W. J. Thompson, , and M. C. Wheeler, 2007: Australian rainfall and surface temperature variations associated with the Southern Hemisphere annular mode. J. Climate, 20, 24522467.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and P. J. Valdes, 1990: On the existence of storm-tracks. J. Atmos. Sci., 47, 18541864.

  • Johns, T. C., and Coauthors, 2006: The new Hadley Centre climate model (HadGEM1): Evaluation of coupled simulations. J. Climate, 19, 13271353.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471.

  • Keable, M., , I. Simmons, , and K. Keay, 2002: Distribution and temporal variability of 500-hPa cyclone characteristics in the Southern Hemisphere. Int. J. Climatol., 22, 131150.

    • Search Google Scholar
    • Export Citation
  • Kent, D. M., , D. Kirono, , B. Timbal, , and F. H. S. Chiew, 2013: Representation of the Australian sub-tropical ridge in the CMIP3 model. Int. J. Climatol., 33, 4857, doi:10.1002/joc.3406.

    • Search Google Scholar
    • Export Citation
  • Kocin, P. J., , and L. W. Uccellini, 2004: Overview. Vol. I, Northeast Snowstorms, Amer. Meteor. Soc., 296 pp.

  • Kounkou, R., , G. A. Mills, , and B. Timbal, 2009: A reanalysis climatology of cool season tornado environments over southern Australia. Int. J. Climatol., 29, 20792090, doi:10.1002/joc.1856.

    • Search Google Scholar
    • Export Citation
  • Leslie, L. M., , and M. S. Speer, 1998: Short-range ensemble forecasting of explosive Australian east coast cyclogenesis. Wea. Forecasting, 13, 822832.

    • Search Google Scholar
    • Export Citation
  • Lindzen, R. S., , and B. Farrell, 1980: A simple approximate result for the maximum growth rate of baroclinic instabilities. J. Atmos. Sci., 48, 16481654.

    • Search Google Scholar
    • Export Citation
  • Lu, J., , C. Deser, , and T. Reichle, 2009: Cause of the widening of the tropical belt since 1958. Geophys. Res. Lett., 36, L03803, doi:10.1029/2008GL036076.

    • Search Google Scholar
    • Export Citation
  • Marshall, G. J., 2003: Trends in the southern annular mode from observations and reanalyses. J. Climate, 16, 41344143.

  • Martin, G. M., , M. A. Ringer, , V. D. Pope, , A. Jones, , C. Dearden, , and T. J. Hinton, 2006: The physical properties of the atmosphere in the new Hadley Centre Global Environmental Model (HadGEM1). Part I: Model description and global climatology. J. Climate, 19, 12741301.

    • Search Google Scholar
    • Export Citation
  • McInnes, K. L., , L. M. Leslie, , and J. L. McBride, 1992: Numerical simulation of cut-off lows on the Australian east coast: Sensitivity to sea-surface temperature. Int. J. Climatol., 12, 783795, doi:10.1002/joc.3370120803.

    • Search Google Scholar
    • Export Citation
  • Mercer, A. E., , and M. B. Richman, 2007: Statistical differences of quasigeostrophic variables, stability, and moisture profiles in North American storm tracks. Mon. Wea. Rev., 135, 23122338.

    • Search Google Scholar
    • Export Citation
  • Mills, G. A., , R. Webb, , N. E. Davidson, , J. Kepert, , A. Seed, , and D. Abbs, 2010: The Pasha Bulker east coast low of 8 June 2007. Centre for Australia Weather and Climate Research Tech. Rep. 23, 62 pp. [Available online at http://www.cawcr.gov.au/publications/technicalreports.php.]

  • Ndarana, T., , and D. W. Waugh, 2010: The link between cut-off lows and Rossby wave breaking in the Southern Hemisphere. Quart. J. Roy. Meteor. Soc., 136, 869885.

    • Search Google Scholar
    • Export Citation
  • Nieto, R., and Coauthors, 2005: Climatological features of cutoff low systems in the Northern Hemisphere. J. Climate, 18, 30853103.

  • Otkin, J. A., , and J. E. Martin, 2004: A synoptic climatology of the subtropical kona storm. Mon. Wea. Rev., 132, 15021517.

  • Pinto, J. G., , T. Spangehl, , U. Ulbrich, , and P. Speth, 2006: Assessment of winter cyclone activity in a transient ECHAM4-OPYC3 GHG experiment. Meteor. Z., 15, 279291.

    • Search Google Scholar
    • Export Citation
  • Pook, M. J., , P. C. McIntosh, , and G. A. Meyers, 2006: The synoptic decomposition of cool-season rainfall in the southeastern Australian cropping region. J. Appl. Meteor. Climatol., 45, 11561170.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., , and I. N. Smith, 2007: Weakening of the Walker circulation and apparent dominance of El Niño both reach record levels, but has ENSO really changed? Geophys. Res. Lett., 34, L18702, doi:10.1029/2007GL030854.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., , and G. Kociuba, 2011: The impact of global warming on the Southern Oscillation Index. Climate Dyn., 37, 17451754.

  • Rayner, N. A., , D. E. Parker, , E. B. Horton, , C. K. Folland, , L. V. Alexander, , D. P. Rowell, , E. C. Kent, , and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

    • Search Google Scholar
    • Export Citation
  • Risbey, J. S., , M. J. Pook, , P. C. McIntosh, , C. C. Ummenhofer, , and G. Meyers, 2009: Characteristics and variability of synoptic features associated with cool season rainfall in southeastern Australia. Int. J. Climatol., 29, 15951613.

    • Search Google Scholar
    • Export Citation
  • Seluchi, M. E., , and A. C. Saulo, 1998: Possible mechanisms yielding an explosive coastal cyclogenesis over South America: Experiments using a limited area model. Aust. Meteor. Mag., 47, 309320.

    • Search Google Scholar
    • Export Citation
  • Simpson, R. H., 1952: Evolution of the Kona storm, a subtropical cyclone. J. Atmos. Sci., 9, 2435.

  • Sinclair, M. R., , J. A. Renwick, , and J. W. Kidson, 1997: Low-frequency variability of Southern Hemisphere sea level pressure and weather system activity. Mon. Wea. Rev., 125, 25312543.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., , D. Qin, , M. Manning, , Z. Chen, , M. Marquis, , K. Averyt, , M. Tignor, , and H. L. Miller, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

  • Speer, M. S., , P. Wiles, , and A. Pepler, 2009: Low pressure systems off the New South Wales coast and associated hazardous weather: Establishment of a database. Aust. Meteor. Oceanogr. J, 58, 2939.

    • Search Google Scholar
    • Export Citation
  • Tanaka, H. L., , N. Ishizaki, , and A. Kitoh, 2004: Trend and interannual variability of Walker, monsoon and Hadley circulations defined by velocity potential in the upper troposphere. Tellus, 56A, 250269.

    • Search Google Scholar
    • Export Citation
  • Timbal, B., , and W. Drosdowsky, 2012: The relationship between the decline of southeastern Australia rainfall and the strengthening of the subtropical ridge. Int. J. Climatol., doi:10.1002/joc.3492, in press.

    • Search Google Scholar
    • Export Citation
  • Troup, A. J., 1965: The ‘southern oscillation.’ Quart. J. Roy. Meteor. Soc., 91, 490506, doi:10.1002/qj.49709139009.

  • Ulbrich, U., , A. H. Fink, , M. Klawa, , and J. G. Pinto, 2001: Three extreme storms over Europe in December 1999. Weather, 56, 7080.

  • Uppala, S. P., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 29613012.

  • Uppala, S. P., , D. Dee, , S. Kobayashi, , P. Berrisford, , and A. Simmons, 2008: Towards a climate data assimilation system: Status update of ERA-Interim. ECMWF Newsletter, No. 115, ECMWF, Reading, United Kingdom, 12–18.

  • Vecchi, G. A., , B. J. Soden, , A. T. Wittenberg, , I. A. Held, , A. Leetma, , and M. J. Harrison, 2006: Weakening of the tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature, 441, 7376.

    • Search Google Scholar
    • Export Citation
  • Whetton, P., , I. Macadam, , J. Bathols, , and J. O’Grady, 2007: Assessment of the use of current climate patterns to evaluate regional enhanced greenhouse response patterns of climate models. Geophys. Res. Lett., 34, L14701, doi:10.1029/2007GL030025.

    • Search Google Scholar
    • Export Citation
  • Whittaker, L. M., , and L. M. Horn, 1981: Geographical and seasonal distribution of North American cyclogenesis, 1958–1977. Mon. Wea. Rev., 109, 23122322.

    • Search Google Scholar
    • Export Citation
Save