numerical simulation of cut-off lows on the australian east coast: sensitivity to sea-surface...

13
INTERNATIONAL JOURNAL OF CLIMATOLOGY, VOL. 12,783-795 (1992) 551.51 5.1 l(94 + 945):551.526.6 NUMERICAL SIMULATION OF CUT-OFF LOWS ON THE AUSTRALIAN EAST COAST: SENSITIVITY TO SEA-SURFACE TEMPERATURE KATHLEEN L. McINNES CSlRO Division of Atmospheric Research. Private Bag I, Mordialloc, Victoria 3195. Australia AND LANCE M LESLIE AND JOHN L. McBRIDE Bureau of Meteorology Research Cenrre, GPO Box 1289K. Melbourne 3001, Australia Received 15 November 1991 Accepted 24 April I992 ABSTRACT The occurrence and synoptic-scale structure of cut-off lows near the Australian east coast are documented. These systems are associated with severe weather in the south-east Australian states of New South Wales and Victoria. Two types of cut- off low formation are identified, described here as coastal lows and blocking lows. Twenty-four hour numerical simulations are performed for four systems in order to investigate the model-reproduced structure of the large-scale rainfall and gale-force wind. The numerical experiments demonstrated that the regional-scale model is capable of simulating the heavy rainfall and gale-force winds associated with these systems. All systems are characterized by a major rain band curling around the south and east of the surface low. To the south of the system the rain band is paralleled by a curved band of gale-strength winds. The sensitivity of these parameters and of mean sea-level pressure to underlying sea-surface temperature (SST) fields also is investigated through numerical experimentation. When the SSTs were increased by between 2" and 3", the simulated cut-off lows had deeper central pressures, usually of 1-2 hPa. In one case the extra deepening was dramatic, being 7 hPa at 24 h. The impact on rainfall was considerable, especially on the peak values, which were increased by between 45 per cent and 80 per cent, and on the areal mean values in 'catchment size' areas surrounding the peak value. In contrast the peak surface winds were increased only by about 10 per cent, but the areal extent of gale-force winds increased greatly, by between 50 per cent and 70 per cent. The large sensitivity of these weather elements to SST is discussed in terms of possible implications for both short-term operational forecasting and for regional effects of enhanced greenhouse climate change. KEY WORDS Cut-off lows eastern Australia Sea-surface temperature numerical simulation 1. INTRODUCTION This paper presents a study of cut-off lows along the east of Australia (Figure 1). In general terms the cut-off low is a discrete synoptic-scale cyclonic low-pressure system. The systems are mid-latitude in type, in that they are cold-cored, which means that the cyclonic circulation increases with height through the middle troposphere. This implies that they are generated and maintained through baroclinic instability processes (or possibly Rossby wave shear instabilities at the tropopause level as described by Hoskins et al. (1985)), as distinct from the latent-heat-related instabilities believed responsible for tropical type, warm-cored cyclones. The term 'cut-off is traditional among Australian forecasters and is borrowed from classical Northern Hemisphere synoptic meteorology (e.g. Pettersen, 1956, chapter 12). In the Australian context a system is called a cut-off if there is a closed circulation at the 500 hPa level; so the system is separated or 'cut-off from the westerly circumpolar vortex. 0889-841 8/92/080783-13Ul.50 0 1992 by the Royal Meteorological Society

Upload: kathleen-l-mcinnes

Post on 15-Jun-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

INTERNATIONAL JOURNAL OF CLIMATOLOGY, VOL. 12,783-795 (1992) 551.51 5.1 l ( 9 4 + 945):551.526.6

NUMERICAL SIMULATION OF CUT-OFF LOWS ON THE AUSTRALIAN EAST COAST: SENSITIVITY TO SEA-SURFACE

TEMPERATURE

KATHLEEN L. McINNES

CSlRO Division of Atmospheric Research. Private Bag I , Mordialloc, Victoria 3195. Australia

AND

LANCE M LESLIE AND JOHN L. McBRIDE

Bureau of Meteorology Research Cenrre, GPO Box 1289K. Melbourne 3001, Australia

Received 15 November 1991 Accepted 24 April I992

ABSTRACT

The occurrence and synoptic-scale structure of cut-off lows near the Australian east coast are documented. These systems are associated with severe weather in the south-east Australian states of New South Wales and Victoria. Two types of cut- off low formation are identified, described here as coastal lows and blocking lows.

Twenty-four hour numerical simulations are performed for four systems in order to investigate the model-reproduced structure of the large-scale rainfall and gale-force wind. The numerical experiments demonstrated that the regional-scale model is capable of simulating the heavy rainfall and gale-force winds associated with these systems. All systems are characterized by a major rain band curling around the south and east of the surface low. To the south of the system the rain band is paralleled by a curved band of gale-strength winds. The sensitivity of these parameters and of mean sea-level pressure to underlying sea-surface temperature (SST) fields also is investigated through numerical experimentation.

When the SSTs were increased by between 2" and 3", the simulated cut-off lows had deeper central pressures, usually of 1-2 hPa. In one case the extra deepening was dramatic, being 7 hPa at 24 h. The impact on rainfall was considerable, especially on the peak values, which were increased by between 45 per cent and 80 per cent, and on the areal mean values in 'catchment size' areas surrounding the peak value. In contrast the peak surface winds were increased only by about 10 per cent, but the areal extent of gale-force winds increased greatly, by between 50 per cent and 70 per cent.

The large sensitivity of these weather elements to SST is discussed in terms of possible implications for both short-term operational forecasting and for regional effects of enhanced greenhouse climate change.

KEY WORDS Cut-off lows eastern Australia Sea-surface temperature numerical simulation

1. INTRODUCTION

This paper presents a study of cut-off lows along the east of Australia (Figure 1). In general terms the cut-off low is a discrete synoptic-scale cyclonic low-pressure system. The systems are mid-latitude in type, in that they are cold-cored, which means that the cyclonic circulation increases with height through the middle troposphere. This implies that they are generated and maintained through baroclinic instability processes (or possibly Rossby wave shear instabilities at the tropopause level as described by Hoskins et al. (1985)), as distinct from the latent-heat-related instabilities believed responsible for tropical type, warm-cored cyclones. The term 'cut-off is traditional among Australian forecasters and is borrowed from classical Northern Hemisphere synoptic meteorology (e.g. Pettersen, 1956, chapter 12). In the Australian context a system is called a cut-off if there is a closed circulation at the 500 hPa level; so the system is separated or 'cut-off from the westerly circumpolar vortex.

0889-841 8/92/080783-13Ul.50 0 1992 by the Royal Meteorological Society

784 K. L. McINNES, L. M. LESLIE AND J. L. McBRlDE

140 145 150 155 160

Figure 1. Map of the model domain of the fine-mesh simulations in this paper, showing place names referred to in the text

The cut-off low has been long recognized as a major source of severe weather in south-east Australia (Foley, 1956; Holland et al., 1987; Wright, 1989). Holland et al. (1987) presented a synoptic overview of a particular type of cut-off (termed by them as ‘east coast cyclones’). Mills and Russell (1992) documented one case of another type of cut-off, associated with widespread flooding through Queensland, New South Wales, and Victoria.

The purpose of the current paper is to document the existence of these two types of cut-off lows, to present the results of numerical simulations of two examples of each type, and to demonstrate (through numerical simulation) a large sensitivity of the associated severe weather to the underlying sea-surface temperatures. In the following section we describe the two types, including the large-scale flow conditions under which they form and a brief climatology. Section 3 describes the experimental methodology for the numerical simulations, including model description. The results of the simulations are then given, including discussion of the synoptic-scale structure of precipitation and strong low-level winds, and the sensitivity of these features to sea-surface temperature. A discussion of the results and possible implications, including implications for global warming, are presented in Section 4.

2. SYNOPTIC CLIMATOLOGY

As described above, the usual Australian operational definition of a cut-off requires the existence of a closed circulation at the 500 hPa level. When the cut-off has no manifestation at the surface, it is referred to by forecasters as an ‘upper-level cold pool’. Such systems are usually associated with intense thunderstorm activity rather than large-scale stratiform rain and strong winds, and will not be dealt with in this paper.

The east-coast cut-off lows studied here are synoptic-scale systems extending from the surface to the lower stratosphere. The surface low is synoptic in scale and is located in the trade easterlies, such that it has a distinct cell of the subtropical ridge to the south, separating it from the mid-latitude westerlies. These systems often are associated with widespread stratiform rainfall and flooding. They also can give rise to gale-force winds at the surface. Inspection of past data reveals the existence of two distinct types of east-coast cut-offs. for convenience we shall describe them as coastal lows and blocking lows, although there is no accepted general terminology.

The coastal lows have a number of important features. (i) In the development stage, the upper level low ‘cuts off from a mobile westerly trough. This occurs simultaneously with (or slightly before) the development of the blocking-type cyclone-anticyclone pair at the surface. This sequence is illustrated for the two examples simulated in this study in Figures 2 and 3. (ii) Once formed, the synoptic-scale surface low actually straddles

SIMULATION OF AUSTRALIAN EAST COAST CUT-OFF LOWS 785

DAY 1 DAY 2 DAY 3

Figure 2. Sequences of three charts at 24-h intervals showing the development of a cut-off low of the coastal low type. The upper row is for the 500 hPa level and the lower row is the simultaneous sequence at mean seal-level. This example is case 3 of our numerical simulations;

and the beginning (furthest left) chart is at 2300 UTC, 25 May 1990

DAY 1 DAY 2 DAY 3

Figure 3. Same as Figure 2 showing the development of another example of a coastal low. This is case 4 numerical simulations. Beginning chart is at 2300 UTC, 29 July 1990

786 K. L. McINNES, L. M. LESLIE AND J. L. McBRIDE

.3

L Ava z Ava E AVO I&

SIMULATION OF AUSTRALIAN EAST COAST CUT-OFF LOWS 787

the east coast and moves southward along the coast; signifying that some role is played by either the coastal orography or by the land-sea surface contrasts in the underlying dynamics of these systems. Examples of the movements of the surface system are shown in Figure 4. (iii) These coastal low systems often provide the synoptic environment for a class of intense meso-scale low-pressure systems described by Holland et al. (1987) and numerically simulated by Leslie et al. (1987). These latter systems are only 100-500 km in scale, and occasionally two or more can occur simultaneously within the coastal low (see Figure 6 of Holland et ul. (1987)). Although embedded in the baroclinic coastal low, they have structural similarities to tropical cyclones in that they are highly convective and are warm-cored in the lower troposphere.

The blocking lows are similar in structure to the coastal lows in that they are cold-cored through the depth of the troposphere, are associated with a high-low pair at the surface, the high being continuous with the planetary scale subtropical ridge, are associated with a 'split jet stream' pattern near the tropopause and occur in the general vicinity of the east coast of Australia. They are different to the coastal lows in the following ways. (i) The surface low forms equatorward of a pre-existing blocking high at the surface. There is also a pre- existing high-low blocking pair at upper levels. An example of this is shown in Figure 5. (ii) The actual location of the coastline is incidental to this type of system. They form either inland or to the east of the continent and subsequently drift either eastward or westward.

Both coastal lows and blocking lows are associated with large-scale flooding and strong winds. Inspection of analysed weather maps for the 5-year period 1986-1990 reveals 13 coastal lows and 34 blocking lows located between 20-40s and 140&160"E, making47 cut-offs in total or approximately nine per year. No clear seasonal distribution was observed in the blocking lows; but of the coastal lows, eight of the 13 occurred in spring (September-November). This distribution may not be significant statistically, however, as in their study of the intense meso-scale subcomponents of these systems, Holland et at. (1987) found a strong preference for autumn and winter.

For the numerical simulations in this paper we have chosen two examples of each type of system as outlined in Table I. For each type we have chosen a major rainfall and flooding event (cases 2 and 4) and one minor event (cases 1 and 3).

DAY 1 DAY 2 DAY 3

Figure 5. Same as Figure 2, but showing the development of a cut-off low of the blocking low type. This is case 2 numerical simulation. Beginning chart is at 2300 UTC, 17 April 1990

788 K. L. McINNES. L. M. LESLIE AND J. L. McBRIDE

Table I. Summary of the cut-off low events simulated in this study. The date and time given refer to the start time of the 24-h fine-mesh model simulation

Case and type Date-time Comments

1 Blocking low 2300 UTC 9 February 1990

2 Blocking low 2300 UTC

3 Coastal low 2300 UTC

4 Coastal low 2300 UTC

19 April 1990

21 May 1990

31 July 1990

Light rainfall < 50 mm along east coast. Localized thunderstorms and associated strong winds and flash floods in New South Wales and Victoria

Rainfall > 150 mm over Sydney region. Widespread floods over eastern Victoria and south-eastern New South Wales

Heavy rainfall out to sea. Generally light rainfall along the east coast

Widespread heavy rain and gale-force winds along the New South Wales coast

3. NUMERICAL EXPERIMENTS

3. I . Design

For each event, two 24-h numerical simulations were performed: a control simulation and an enhanced SST (sea surface temperature) simulation. The purpose of the control simulation is to determine the synoptic-scale structure of wind and rain in the cut-off lows, as produced by a 'state of the science' operational Numerical Weather Prediction (NWP) model. The purpose of the enhanced SST run is to demonstrate the very large sensitivity of these weather elements to this parameter.

Each case study was made using the Bureau of Meteorology Research Centre (BMRC) limited-area NWP model in a telescoping manner; that is, the primary simulation was performed over south-eastern Australia at 50 km horizontal grid resolution over the domain shown in Figure 1. Each of these 'fine-resolution' simulations was nested in a coarse-resolution forecast (1 50 km grid) over the entire Australian region domain, and that in turn was nested in the BMRC global (spectral rhomboidal 31, 9 level) model.

This procedure was carried out for both control and enhanced SST runs. Sea-surface temperatures for the control runs were taken from monthly mean climatologies. For the enhanced SST runs these were increased by the addition of an anomaly field, which varied in magnitude from approximately 2.5"C at 20"s to about 3.5"C at 40"s over the oceans east of Australia. These magnitudes are consistent with observed interannual fluctuations, though actually they were chosen according to what would be expected under a two-times CO, climate change scenario, and were derived from the GCM climate change experiments described by Whetton and Pittock (1991) using the model of Gordon and Hunt (1987).

To minimize 'spin-up' problems all coarse-resolution regional model forecasts were commenced at t = -24 h so that the 24-h fine-mesh forecasts commenced from initial states in which the effects of the enhanced SSTs had time to penetrate into the troposphere. For the first 24 h the coarse-mesh model was integrated forward; but at each time step, 30 per cent of the model predicted variable at each grid point was replaced by the interpolated analysis value at that point. (Experiments carried out with weightings different from 30 per cent showed little sensitivity after 24 h of spin-up.) The aim of this procedure was to allow the model to adjust to the dramatic changes to the lower boundary values of temperature through vertical mixing, whilst at the same time being constrained by the analyses, For the final 24 h a normal 24-h simulation was carried out. For the 50 km simulations, the model was run only for the latter 24 h, since the initial conditions were taken from the coarse resolution simulations after 24 h of 'spin-up' had taken place. This procedure was applied to both control and enhanced SST runs to ensure consistency in the results.

The limited-area model used in these experiments is a modified version of the Bureau of Meteorology's operational regional NWP model. The model is described in detail by Leslie et al. (1985). The analysis scheme

SIMULATION OF AUSTRALIAN EAST COAST CUT-OFF LOWS 789

is a three-dimensional univariate statistical interpolation scheme with variational blending to achieve mass/wind coupling (Mills and Seaman, 1990). The forecast model is a hydrostatic, primitive equations model, with a sigma-coordinate system in the vertical, 13 vertical levels, and semi-implicit time integration on an Arakawa C-grid. The major difference from the operational model is the parameterization of vertical turbulent exchanges, in which a Mellor-Yamrlda level 2.25 scheme is used (McInnes and Hess, 1992). This parameterization scheme is considerably more sophisticated than the operational parameterization scheme and was deemed necessary because of the important role played by surface fluxes in these experiments.

3.2. Results

Control and enhanced SST simulations were carried out for each of the four cut-off situations in Table I. The mean sea-level pressure (MSLP) maps at the end of the 24-h fine-mesh forecast and the associated 24-h accumulated precipitation are shown in Figures 6 and 7 respectively. Commenting first on the meteorology of the four events, all four show a well-defined surface low straddling the east coast of the continent. Cases 2, 3, and 4 have a major rain band curling around the south and east of the system with peak values of between 120 mm and 419 mm accumulated 24-h totals. In the simulations the production of this precipitation is shared almost equally between the model’s stratiform and convection parameterization routines; though the spatial organization is such that in the atmosphere it is almost certainly predominantly stratiform. Case 1, by comparison, is one of very light rainfall (peak value equalling 22 mm); though its structure still is that of a weakly organized rain band south of the system.

Comparison of spot values with actual rainfall observations in south-eastern Australia reveal a good level of model performance in all four cases, particularly in the spatial distribution and areal extent of rainfall. However, magnitudes were underestimated slightly by the model for cases 1, 2, and 3 and overestimated in case 4. This is consistent with the findings of Hess (1990) who demonstrated an extreme sensitivity of model rainfall to horizontal resolution in a simulation of a similar east-coast cyclone event, in which the rainfall distribution bore a similar high-resolution structure to that seen in Figure 7 for the four current events. From Hess’ experiments the accurate simulation of such rainfall patterns requires a model resolution of about 15 km. For reference, spot peak observed 24-h rainfall totals for the four events are as follows: event 1 (Sale, 15 mm; Sydney, 57 mm; Liverpool 34 mm); event 2 (Perisher Valley, t26 mm; Bega, 106 mm; more than 40 mm over most of eastern Victoria); event 3 (Coffs Harbour, 103 mm; Taree, 84 mm); event 4 (Maryville, 99 mm; Liverpool, 80 mm; Bathurst, 65 mm). For locations, see Figure 1.

Figure 8 shows maps of 850 hPa level wind sveed at the end of the four control and enhanced simulations. For the three major rain events (2,3, and 4) there is a curved band of gale-strength winds paralleling the major rain band to the south of the centre of the low. The spatial organization of both the rain and gale-force wind bands extends over approximately 1000 km, signifying the importance of these systems in the forecasting of severe weather on both the daily and climate time-scales.

Comparing now the patterns in Figure 6-8 between the control and the enhanced runs, the most striking result is the strong similarly in the pattern (spatial organization) and in the spatial extent of the rainfall. This can be seen for instance in all four cases by inspection of the location of the 10-mm accumulated rainfall contour, which hardly moves between the control and enhanced runs. The peak rainfall totals within that pattern, however, are greatly increased, as shown in Table II. For all four cases the peak rainfall value increased by between 45 per cent and 80 per cent; and the areal average (over a 300 kmx300 km area surrounding the peak) increased by between 14 per cent and 100 percent.

For MSLP and for 850 hPa wind, the patterns also were very similar between the control and enhanced runs; although the similarities are not as strong as for rainfall. In all cases the centrgl pressure of the low is deeper for the enhanced simulation, usually by 1 or 2 hPa, except for case 3 in which a 7 hPa deeper low was observed.

Looking at comparisons of low-level wind (Figure 8), the major difference is in the areal extent of gale-force (greater then 15 ms-’) winds, rather than in the peak values. As shown in the two right-most columns of Table 11, peak wind values increased by only about 10 per cent, whereas the areal extent of gale-force winds (as counted by number of grid-points) increased by 50 per cent to 70 per cent.

Figure 6. Maps of mean sea-level pressure at the end of the 24-h fine-mesh simulations for each of the four cases. The left column is for the control experiment (C), cases 1-4, and the right is for the enhanced (or increased SST) (E), cases 1-4

Figure 7. Same as Figure 6. but for accumulated 24-h precipitation (mm). Contours are for 10 mm, 50 mm, 100 mm, 150 mm, etc. Peak values are given in Table 11.

Figure 8. Same as Figure 6 but for wind speed (m s- I ) at 850 hPa. Peak values are given in Table I1

SIMULATION OF AUSTRALIAN EAST COAST CUT-OFF LOWS 793

Table 11. Comparison of 24-h accumulated rainfall and 850 hPa wind for the control (C) and enhanced SST (E) simulations; and the ratios (E/C) of rain and wind between the two simulations

24-h rainfall 850 hPa wind

Number of Case Peak Areal grid points Peak number Simulation type rainfall rainfall >15 ms-' wind

1 Control (C) 22 mm 7 mm 0 1 2 m s - ' Enhanced SST (E) 32 mm 8 mm 0 13 m s - ' Ratio (E/C) 1.45 1.14 - 1.08

2 Control 170 mm 64 mm 63 2 4 m s - ' Enhanced SST 286 mm 134 mm 93 25 m s - I

Ratio (E/C) 1.68 2.09 1.48 1.04 39 18111s-l

Enhanced SST 754 mm 171 mm 66 21 ms- ' Ratio (E/C) 1.80 1-41 1.69 1.17

4 Control 120 mm 31 mm 27 20ms-I Enhanced SST 181 mm 55 mm 42 23 m s - ' Ratio (E/C) 1.51 1.67 1.56 1.15

3 Control 419 mm 121 mm

4. DISCUSSION AND CONCLUSIONS

This paper has presented an investigation into cut-off lows over south-eastern Australia. These lows can produce flood rains, gale- to hurricane-force winds, and raised sea-levels, causing loss of life and property. In Section 2 the synoptic structure of these systems was described and a brief climatology given, based on Bureau of Meteorology archived analyses from 1986 to 1990. In Section 3 a numerical simulation study was performed of four examples, including a study of the effect of a change in SST on the structure of these weather systems.

Two different types of cut-off low associated with severe weather along Australia's east coast have been documented: these are coastal lows which correspond to the synoptic environment of the meso-scale east- coast lows studied by Holland et al. (1987) and blocking lows, which form equatorward of a pre-existing upper level high-low blocking pair. The major large-scale flooding situation studied by Mills and Russell (1992) is an example of a blocking low.

The numerical experiments demonstrated that the regional-scale model is capable of simulating the heavy rainfall and gale-force winds associated with these systems. All systems are characterized by a major rain band curling around the south and east of the surface low. To the south of the system the rain band is paralleled by a curved band of gale-strength winds.

The impact of the enhanced SST fields was to deepen the central (MSL) pressures of the cut-off lows by amounts ranging from 1-2 hPa through to 7 hPa in case 3. Turning to the 24-h accumulated rainfall totals, there are three noteworthy features: firstly, the total area covered by the 2 10-mm isohyet is strikingly similar between the control and enhanced SST forecasts; second, the peak rainfall amounts within the rainfall areas are greatly increased in all four cases in the enhanced SST forecasts, ranging from between 45 per cent and 80 per cent. For major flooding events, increases of this magnitude would have potentially serious consequences. Finally, if an areal average is taken in a catchment-size (300 km x'300 km) region around the peak value, the average increases in all cases from 14 per cent to 100 per cent. In the case of the low-level winds, the percentage increase in peak values was small in all cases, ranging from a few per cent to 25 per cent in case 3. However, the areal extent of gale-force, and stronger, winds was greatly enhanced, by around 60 per cent in all four cases.

794 K. L. McINNES, L. M. LESLIE AND J. L. McBRIDE

MSLP

350 hPa WIND

Figure 9. An example of an east coast cut-off low as produced by the CSIRO-4 current climate simulation. The upper chart is for M S L P lower for 350 hPa wind

The similarity in spatial structure in precipitation and strong wind between control and enhanced runs shows that the cut-off lows are still dominated by the dynamics, as the area and location of upward motion essentially is unchanged. In this regard it is of significance that the enhanced surface temperatures did not have the effect of changing the low precipitation and wind situation (case 1) into a major rain/wind event. Despite that qualification, there are a number of potentially important implications of the results of the enhanced SST experiments.

SIMULATION OF AUSTRALIAN EAST COAST CUT-OFF LOWS 795

Firstly, for regional flood and gale forecasting this large sensitivity to the underlying SST is of major importance; and a possible implication is that real-time (as distinct from climatological) SST fields are necessary for short-term (namely 24 h) operational forecasts. The second obvious implication is in the area of climate change scenarios resulting from enhanced greenhouse warming. The above results imply that if the same types of weather system occur under greenhouse warming, they are likely to produce much higher peak- value rainfalls and much larger areal extent of gale-force winds. Of course, under an enhanced greenhouse climate, both the frequently and the large-scale structure of cut-off lows may be changed substantially, as welt as the position and strength of the upper level jets and various other factors affecting rainfall. To gain some indication of these effects, the one times CO, and two times CO, runs of the CSTRO 4 level model (Gordon and Hunt, 1987; Whetton and Pittock, 1991) have been examined for the existence of cut-off lows near south- east Australia. Cut-off lows with the correct large-scale structure do occur in these simulations, as shown in the example in Figure 9. Over the last 3 years of the 25-year one times CO, simulation a total of 40 cases of cut-off lows were counted over the area encompassing eastern South Australia, southern Queensland, New South Wales, and Victoria. This averages to about 13 cases a year, which compares favourably with present- day annual averages. For the last 3 years of the doubled CO, simulation, 28 cases were counted, averaging to about nine per year. Although the temporal extent of the data set is limited, it appears from this analysis that fewer cut-off developments may occur under enhanced greenhouse conditions.

ACKNOWLEDGEMENTS

The authors are grateful to Drs A. B. Pittock, B. F. Ryan, and J. L. Evans of CSIRO and Dr G. A. Mills of BMRC for useful scientific discussion, to Malcolm Haylock of CSIRO for his help in processing the CSIRO GCM data and to D. Pike of BMRC for help in figure preparation and model output diagnostics. Funding for this work was provided by DASET and the State Governments of Victoria and New South Wales.

REFERENCES

Foley, J . C. 1956. ‘500 mb contour patterns associated with the occurrence of widespread rains in Australia’, Aust. Meteorol. Mag., 13,

Gordon, H. B. and Hunt, B. G . 1987. ‘Interannual variability of the simulated hydrology in a climatic model -implications for drought’,

Hess, G. D. 1990. ‘Numerical simulation of the August 1986 heavy rainfall event in the Sydney area’, J . Geophys. Res. 95(D3), 2073-2082. Holland, G. J., Lynch, A. H. and Leslie, L. M. 1987. ‘Australian east-coast cyclones. Part I: overview and case study’. Mon. Wea. Rev., 115,

Hoskins, B. J., McIntyre, M. E. and Robertson, A. W. 1985. ‘On the use and significance of isentropic potential vorticity maps’, Q.J.R.

Leslie, L. M., Mills, C . A,, Logan, L. W., Gauntlett, D. J., Kelly, G. A,, Manton, M. J., McGregor, J. L. and Sardie, J. M. 1985. ‘A high

Leslie, L. M., Holland, G. L. and Lynch, A. H. 1987. ‘Australian east-coast cyclones. Part 11: numerical modelling study’, Mon. Wea. Reu.,

McInnes, K. L. and Hess, G. D. 1992. ‘The implementation of a high resolution physics package in the Australian Region Limited Area

Mills, G. A. and Seaman R. S. 1990. ‘The BMRC Regional Data Assimilation System’, Mon. Wea. Rev., 118, 1217-1237. Mills, G. A. and Russell, I. 1992. ‘The April 1990 floods over eastern Australia: synoptic description and assessment of the regional NWP

Petterssen, S . 1956. Weather Analysis And Forecasting. Volume I . Motion and Motion Systems, 2nd edition, McGraw-Hill, New York,

Whetton, P. and Pittock, A. B. 1991. Australian region intercomparison oj’the results oj’some general circulation models used in enhanced

Wright, W. J . 1989. ‘A synoptic climatological classification of winter precipitation in Victoria’, Aust. Meteorol. Mag., 37, 217-219.

1 18.

C h a f e Dynamirs, 3, 19-33.

3024-3036.

Meteorol. Soc., 1 11. 877-946.

resolution primitive equations N W P model for operations and research’, Aust. Meteorol. Mag., 33, 11-36,

115, 3037 3053.

model’, Aust. Meteorol. Mag., 40, 21-31.

forecast guidance’, Weather und Forecasting, 7, in press.

428 pp.

yreenhouse experiments, CSIRO Division of Atmospheric Research Technical Paper No. 21, CSIRO, Melbourne Australia.