70 Weather and Climate (1986) 6: 70-76 THE CROSS-EQUATORIAL PRESSURE GRADIENT AND SUMMER MONSOON RAINFALL IN NORTHERN AND CENTRAL AUSTRALIA IN JANUARY 1974 A. H. Gordon The Flinders Institute for Atmospheric and Marine Sciences The Flinders University of South Australia Bedford Park, S.A. 5042 ABSTRACT The occurrence of the Australian monsoon is examined in relation to the cross-equatorial gradient which stretched across the equator between the winter-hemisphere subtropical anticyclone and the summer-hemisphere monsoon low pressure system situated over the Australian continent in January 1974. The pressure gradient is calculated in two parts, both parts of the same sign directed from north to south. The first part is expressed as the difference in pressure between 'N and the equator, and the second part as the difference in pressure between the equator and 'S. The day by day sequence of these gradients is compared with daily rainfall values recorded at selected stations in north and central Australia in January 1974. The diagnostic results support the conclusions obtained from numerical trajectory calculations, that a pronounced cross-equatorial gradient is favourable to rainfall in the summer hemisphere. In particular, a strong pressure gradient in the northern winter hemisphere directed southwards appears to be associated with substantial rainfall in the northern part of the area studied, between 11 'S and 18 'S, whereas when a strong gradient extends further southwards into the interior of the continent, heavy rainfall occurs between 18 'S and 'S, and it becomes less rainy north of 18 'S. INTRODUCTION The monsoon is commonly defined in terms of a seasonal reversal in direction of the climatically established mean vector wind (Ramage, 1979). In this work the monsoon is interpreted as the occurrence of spells of continuous rain, which although enhanced by convection, is primarily triggered by the dynamics of the meridional flow. Examination of climatic charts of mean sea level barometric pressure (Lockwood, 1974; Riehl, 1979) suggest that there are only two major monsoon regions in the world, the Asian and the Australian. In these regions recognizable cross-equatorial pressure gradients exist. For example, in July the mean difference in pressure along the 60 "E meridian between 'N and 'S latitude is mb, while in January along the 130 "E meridian between 'S and 'N it is 16 mb. It is probably the vast zonal extent of the meridional pressure gradient rather than its slightly greater magnitude which results in the Indian monsoon developing a greater intensity than its southern hemisphere counterpart. In Bombay the mean monthly rainfall increases from 18 mm in May to 483 mm in June
and 610 mm in July, while the cross-equatorial pressure gradient as defined above has reached a mean value of between 2 and 3 mb per 5 ' of latitude. In Darwin the mean monthly rainfall increases from 12 mm in September to 385 mm in January. Gordon and Taylor (1966, 1975), and Gordon (1967), computed trajectories for selected periods during which a cross equatorial pressure gradient occurred, stretching from the winter hemisphere sub-tropical anticyclone to the summer hemisphere monsoon trough or low pressure region. The computations showed that trajectories computed by the method described tended to converge in the summer hemisphere between 10 ' and ' latitude, depending on the values of the initial wind, pressure gradient and the frictional dissipation constant used. A further example, but on a shorter time scale, was quoted by Gordon (1973) covering the area of Luzon in the Philippines. Some 800 mm of rain fell in Manila during two days when a pressure gradient of 2 mb per 5 ' of latitude extended from the winter to the southern hemisphere. In recent years attention has been directed to the summer monsoon in Australia. Davidson, McBride and McAveney (1984) suggest that day to day changes in tropical convective activity are physically linked to day to day changes in the subtropical high cells and that the secret to the understanding and prediction of changes in monsoon activity on a day to day time scale may lie in the understanding of the detailed structure of the subtropical anticyclone. Davidson (1984) presented composite pressure maps for clear and cloudy monsoon periods. From the maps he found that the crossequatorial flow between 100"E and 140 "E longitude is slightly stronger for the cloudy phase. The total pressure differences between 'N and 'S along the 130 'E meridian were 13 mb and 9 mb, respectively. It should be noted that the period studied, December 1978-January 1979, was not a well marked monsoon such as January 1974. ANALYSIS OF JANUARY 1974 PRESSURE GRADIENT AND RAINFALL DATA The author was interested in examining if there was any day to day connection between the meridional cross-equatorial gradient and rainfall which would support the trajectory calculations already referred to in the previous section. He paid a visit to Darwin and extracted pressure values from the original working charts for January 1974 for 5 ' intervals of latitude and longitude between 'N and 'S and 1 "E and 140 'E. A mean value was then calculated for the belts 'N to the equator, and from the equator to 'S for an average longitude of 130 'E. It is emphasised that both the northern and the southern hemisphere gradients were of the same sign, that is, directed from north to south. January 1974 was a month during which the Australian monsoon was strongly pronounced. It was an extraordinarily wet month in the centre of the continent where, for example, 500 mm fell at Vaughan Springs, some 350 km northwest of Alice Springs. Lake Eyre was transformed into an inland sea in one of its most spectacular fillings. Fig. I shows the day to day variation in the pressure gradient as measured by the difference in pressure in mb between 'N and the equator (solid line), and between the equator and 'S (broken line), together with daily rainfall totals, for three selected stations in northern Australia. The vertical lines represent daily rainfall amounts in mm. The stations chosen are Snake Bay (11 ''S, 130 "O'E) on the northern fringe of Australian territory, Ayers Rock ( ''S, 131 '4'E) in the centre of the continent, and Banka Banka (18 '48'S, 134 '1'E), about half way between the former two stations. The percentages of rainfall in the two periods 1-18 January and 19-31 January have been weighted according to the number of days in each period. It is clearly seen that the pressure gradient during the first part of the month (1-18 January inclusive) was greater in the winter hemisphere while that during the latter part of the month (19-31 January) was greater in the summer hemisphere. It can also be seen that rainfall was more frequent and heavier at Snake Bay during the first part of the month than during the latter part of the month (with the exception of 31 January) and again more frequent at Ayers Rock during the latter part of the month than during the first part. The mid-way station, Banka Banka, shows a fairly even distribution of rain throughout the whole month. The overall relation between the winter hemisphere and summer hemisphere pressure gradients and the rainfall is shown more specifically in Fig. 2. The ordinate shows 71
72 T h e Australian Monsoon JAN 1-18 78% JAN 19-31 22% WEIGHTED (a) M Ap Alb JAN 1-18 8% JAN 19-31 92% WEIGHTED (b) rnm - - 100-100 15 - - 75 15-75 - 50 10-50 5-15 2 0 2 5 3 0 JANUARY 1974 (c) 5 1 0 1 5 2 0 2 5 3 0 JANUARY 1974 Ap mb JAN 1-18 5 3 % JAN 19-31 47% WEIGHTED M 100 15 10 75 50 Fig. 1: Daily variation of the meridional pressure gradient expressed as the difference in mb between 'N and the equator (solid line) and between the equator and 'S (pecked line). Daily rainfall in mm shown by the vertical lines. Data for January 1974. (a) Snake Bay, 11''S (b) Ayers Rock, ''S and (c) Banka Banka, 18'48'S. The percentages of rainfall in the two periods have been weighted according to the number of days in each period. 1 I 1 0 10 1 5 2 JANUARY 1974 30
7 3 10 TABLE 1: LIST OF STATIONS FROM WHICH DAILY RAINFALL AMOUNTS HAVE BEEN OBTAINED. 2 : 15 50 PERCENT Fig. 2: Variation with latitude in northern Australia of the percentage of total rainfall in January 1974 which occurred when the meridional pressure gradient north of the equator exceeded the meridional pressure gradient south of the equator. The pressure gradient was directed from north to south in both hemispheres (high pressure to the north and low pressure to the south). degrees of latitude representing the meridional locations of 19 stations (Table 1) at which daily rainfall has been reported during January 1974. The abscissa represents the percent of the total monthly rainfall amount which occurred during the first part of the month (1-18 January), weighted so that the distribution is referred to equal periods. There appear to be two narrow belts of latitude at which rainfall was greatest during the first part of the month, at about 11 'S and 16 'S. Rainfall was a maximum during the latter part of the month south of 'S. TRAJECTORY COMPUTATIONS It was thought that it would be useful to check if there were any dynamic reasons which would explain the profile exhibited in Fig. 2 and also the broad results shown in Fig. 1. A means of doing this was to compute trajectories of air parcels starting from different latitudes and see where they went. If there was significant convergence of the trajectories it would be expected that vertical motion would occur in the convergent area, causing precipitation. The convergence of trajectories moving from the winter hemisphere and crossing the equator into the summer hemisphere under the influence of a continuous pressure gradient directed from the winter to the summer 75 Station Latitude Longitude mm of rain in January 1974 1 Darwin 12'23'S 130'44'E 189.7 2 Snake Bay 11''S 130 '40'E 618.0 3 Minjilang 11'9'S 132'35'E 432.3 4 Elcho Island 12'2'S 135'34'E 379.7 5 Yirrkala Mission 12'15'S 136 '53'E 247.0 6 Port Keats Mission 14'14'S 129'31'E 671.0 7 El Sharana 13'31'S 132 '31'E 265.4 8 Mountain Valley 14'5'S 133'49'E 154.2 9 Larrimah 15'35'S 133'13'E 229.1 10 Borroloola 16 '41'S 136 '18'E 265.6 11 Timber Creek 15'39'S 130 '29'E 278.0 12 Rabbit Flat '13'S 1301'E 324.0 13 Banka Banka 18 48S 134'1'E 334.7 14 Avon Downs '21'S 137'29'E 635.0 15 Jervois 22 '57'S 136 '9'E 240.7 16 Barrow Creek 21'32S 133'53'E 383.4 17 Vaughan Springs 22 '18'S 130 '51'E 502.0 18 Ayers Rock ''S 131'4'E 242.0 19 Wave Hill 17 '27'S 130'50'E 308.2 hemisphere has already been discussed and documented. The mathematical technique involving the solution of a pair of ordinary differential equations, and the stepwise time iteration of the resulting algebraic equations have been described in detail (Gordon and Taylor, 1966, 1975). The method shows the importance of the variation of the Coriolis parameter with latitude, the beta effect, on typical monsoon rainfall situations. Analytic solution of the problem has also been investigated (Byron-Scott and Gordon, 1985). In Fig. 3 three idealized cases have been studied, simulating typical meridional pressure gradients corresponding to the cases shown in Fig. 1. The vertical lines represent the meridional extent of the trajectories as measured along the ordinate. The abscissa, in effect, represents different starting latitudes from 'N to 'S. Thus 9 trajectories have been considered starting from 9 different latitudes. The pressure gradient is assumed to be zonal as is, of course, the Coriolis parameter. Thus the trajectories are longitudinally independent, a reasonable assumption for the
74 width of longitude studied. The presence of monsoon troughs or low pressure centres would naturally amplify the convergence in the preferred locations. The horizontal lines are drawn normal to the end points of the trajectories so that the closeness of the horizontal lines relative to their initial distance apart of 5 of latitude, will represent a measure of convergence. In Fig. 3(a) a pressure gradient of 4 mb per 5 ' of latitude has been assumed north of the equator and a pressure gradient of I mb per 5 ' of latitude south of the equator. Convergence is displayed by the density of the horizontal lines occurs at about 12 'S and at about 17 'S. This result agrees quite well with the rainfall maxima shown in Fig. 2 for the first part of the month when the pressure gradient is greater north of the equator than south of the equator. Fig. 3(b) shows the case where the pressure gradients for the northern and southern hemispheres have been reversed. In this case strong convergence occurs further south. Fig. 3(c) shows the case where the pressure gradient is 4 mb/5 ' of latitude in both hemispheres. In this case convergence is spread out between about 18 'S and 'S latitude and is not so tightly concentrated as in Fig. 3(b). These particular cases were computed for frictionless motion and for parcels starting from rest. However, the introduction of friction in the boundary layer into the equations and the use of some sub-geostrophic initial velocities greater than zero do not affect the overall result of convergence, but do shift the belts marginally. See Gordon and Taylor (1966) for discussion of relationship between convergence of the trajectories and consequent convergence of airflow in the boundary layer and vertical motion. Fig. 4 shows a typical family of trajectories for the case of frictionless motion where parcels start with an easterly velocity of 5 m sec. The latter velocity is almost geostrophic at 'N for a pressure gradient of 2 mb per 5 ' of latitude, but sub-geostrophic for latitudes nearer the equator. Convergence occurs between 100 and 15 'S. Trajectories for the exact conditions INITIAL LATITUDE 8 15 z 15 1 (a) INITIAL LATITUDE (b) INITIAL LATITUDE 10 10 5 5 a 0 0 0 11:- 4 5 5 10 10 15 5 15 1 Fig. 3: Meridional extent of trajectories start ng every 5' latitude from W to 'S. (a) pressure gradient of 4 mb/5' latitude north of the equator and I mb/5' latitude south of the equator. (b) pressure gradient of I mb/5' latitude north of the equator and 4 mb/5' latitude south of the equator. (c) pressure gradient of 4 mb/5' latitude in both summer and winter hemispheres.
75 Fig. 5: Mean topography for the 1000 mb surface for January 1974 for the Attstralasian region. LONGITUDE Fig. 4: Computed trajectories in the Central Pacific for the conditions shown. Initial winds were 5 m sec-1 from the east; 12-hourly positions are indicated by small open circles. The zonal pressure gradient was very small throughout the area of interest. (After Gordon and Taylor. 1966). given in Fig. 3 show very similar patterns to those in Fig. 4. For an in-depth discussion of the dynamical beta effect on trajectories in the tropics, see Gordon (1985). Fig. 5 shows the mean topography of the 1000 mb surface for January 1974 for the Australasian region. A difference of 60 m is equivalent to about 8 mb. It is noted that there is a cross-equatorial gradient of the contours between the winter sub-tropical high and the continental low. Although the presence of a monsoon depression over the continent may have helped to induce rainfall, the theoretical calculations referred to indicate that the meridional pressure gradient can explain heavy rainfall in the summer hemisphere tropics without the need to invoke zonal pressure gradients. CONCLUSIONS The results of the diagnostic and numerical analysis of the relation between monsoon rainfall and a cross-equatorial winter to summer hemisphere pressure gradient during January 1974 in the Australian region support previous work on the subject for other regions. Although it is not useful to relate individual days, periods of two to three weeks during which pressure gradients continuously straddle the equator relate positively to the occurrence of substantial monsoon-type rainfall in the summer hemisphere. The belts of latitude at which rain occurs depend upon the strength of the pressure gradient itself, and also whether the gradient is stronger in the winter hemisphere than in the southern hemisphere and vice-versa. Broadly speaking, a stronger gradient in the winter hemisphere causes a belt of rainfall at about 12 'S, whereas if a strong gradient penetrates deeply into the summer hemisphere the belt of heavy continuous rain shifts to about '- ACKNOWLEDGEMENTS The author is indebted to Dr. Geoffrey Love, Bureau of Meteorology, Darwin, N.T. for his assistance in providing access to the working charts for the period studied. REFERENCES Bryon-Scott, R. A. D. and A. H. Gordon, 1985: The critical trajectory for cross-equatorial flow and onset of the Australian Summer monsoon. Extended abstracts of the Second ABRMS Australian Conference on Tropical Meteorology, Perth, W.A. pp. 9-11. Davidson, N. E., 1984: Short-term fluctuations in the
76 Australian monsoon during winter MONEX, Monthly Weather Review, 112, 1697-1708. Davidson, N. E., J. L. McBride and B. J. McAveney, 1984: Divergent circulations during the onset of the 1978-79 Australian monsoon, Monthly Weather Review, 112, 1684-1696. Gordon, A. H., 1985: Lagrangian interpretations of atmospheric motion. Ph.D. thesis. Flinders University of South Australia. Gordon, A. H., 1967: A Lagrangian approach to problems in tropical meteorology, Weather, 22, 11, 455-468. Gordon, A. H., 1973: The Great Philippine floods of 1972, Weather, 28, 404-415. Gordon, A. H. and R. C. Taylor, 1966: Lagrangian dynamics and low latitude weather, H.I.G. Report, 66-12, Honolulu, Hawaii, 32 pp. Gordon, A. H. and R. C. Taylor, 1975: Computations of surface layer air parcel trajectories and weather in the oceanic tropics. International Indian Ocean Expedition Monographs, 7, East-west press, Honolulu, Hawaii, 112 pp. Lockwood, J., 1974: World Climatology, Arnold, London, 330 pp. Ramage, C. S., 1971: Monsoon Meteorology, Academy Press, London and New York, 296 pp. Riehl, H., 1979: Climate and Weather in the tropics, Academic Press, London and New York, 611 pp.