1. Introduction
Wet summers and dry winters are the characteristic features of the Australian tropical circulation. The transition from dry to wet is termed the monsoon onset, and is often sudden and always associated with enhanced rainfall and strengthening low-level westerly winds. Some previous studies describing the onset of the Australian monsoon are Troup (1961), Davidson et al. (1983), and Drosdowsky (1996). Historically, definitions of the monsoon onset have been based on either the strength of the tropical westerlies or the observed changes in satellite cloudiness and rainfall. A comprehensive study by Drosdowsky (1996, hereafter referred to as D96) used the strength and depth of the monsoon westerlies and upper-tropospheric easterlies, as measured by the radiosonde flights from Darwin, Australia (∼12°S, 131°E) to provide objective onset dates from 1957 to 1992. Although very valuable, this definition is based upon a point observation and need not be indicative of the large-scale changes that precede the strengthening of the westerly winds, nor of the establishment of conditions that support sustained widespread convection (e.g., large-scale ascent, conditional instability, sporadic convection, and tropospheric moistening). To better understand the processes occurring during onset, it is necessary to know when the thermodynamic and kinematic structures begin to change. This does not necessarily correspond with the timing of the wind change. To clarify when the structures begin to change, we use reanalysis data to document the local structure changes during several monsoon onsets. These include the onsets of 1978/79, 1986/87, and 1992/93, which coincide with the Winter Monsoon Experiment (WMONEX), the Australian Monsoon Experiment (AMEX), and the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE). These were seasons when enhanced observational networks were operating over the monsoon region, and thus when the analyses are likely to be most reliable. Onsets during these years have been studied previously by Davidson et al. (1983, hereafter referred to as D83), Hendon et al. (1989, hereafter H89), and McBride et al. (1995, hereafter M95).
Based on analyzed large-scale kinematic and thermodynamic structures, it is shown that (a) sometimes onset may have occurred a little earlier than that reported in D96 and (b) well prior to the strengthening of the monsoon westerlies and upper-level easterlies there existed enhanced ascent and moistening over the monsoon, suggesting that preconditioning for onset had commenced much earlier than that indicated by the zonal wind changes alone. The first aim of this study is to document these structural changes.
Various physical mechanisms have the potential to influence the onset of the Australian monsoon and convective outbreaks over the Tropics generally. They include extratropical–tropical interaction (Lim and Chang 1981; D83; Keenan and Brody 1988), intratropical interactions (Davidson and Hendon 1989), and the Madden–Julian oscillation (MJO; Hendon and Liebmann 1990). There is compelling evidence that all these processes may at times be important (e.g., Hung and Yanai 2004), but operate on different time scales. For example, the annual change in solar insolation and land–sea contrast provide a mechanism for longer-term preconditioning and forcing of the monsoon. The important role of the MJO on onset at intraseasonal time scales during individual years is described in a number of studies (Hendon and Liebmann 1990; Hung and Yanai 2004) and will not be specifically addressed here. On much shorter time scales, the influence of evolving midlatitude weather systems appears to play at least a modulating role. In the current study we focus on the influence of high-latitude flow changes on the tropical circulation. We first provide observational evidence of an association between high-latitude cyclogenesis over the southwest Indian Ocean (SWIO) and short-term modulation of tropical rain events, leading up to and including monsoon onset. Then based on these observations, the second aim of this study is to use idealized simulations to investigate the possible influence of high-latitude cyclogenesis on the monsoon circulation. We show that extratropical cyclogenesis can eventually lead to an enhancement of the low-level cyclonic vorticity and changes in the upper-level horizontal divergence over the Tropics, thereby providing a possible mechanism to trigger large-scale tropical convection, with associated patterns of westerly wind and pressure similar to that predicted theoretically by Gill (1981). Theoretical aspects of high-latitude forcing of tropical motions are described in Webster and Holton (1982), Hoskins and Yang (2000), and the references therein. These studies, briefly summarized in section 3, provide the theoretical basis for the hypothesis tested here. Previous observational evidence of the influence of changes in the Southern Hemisphere winter subtropical jet (STJ) on the flow of the northern (summer) Tropics is described by Straub and Kiladis (2003). They present observations that link enhanced Kelvin wave activity in the central Pacific intertropical convergence zone (ITCZ) with enhanced eastward-propagating Rossby wave activity in the STJ.
The background climatology of the region is illustrated in Fig. 1. Using National Centers for Environmental Prediction (NCEP) reanalysis data, it shows the long-term-mean January mean sea level pressure (MSLP) and 200-hPa wind over the Indian Ocean and Australian region. The MSLP chart shows the heat low over Australia, and the monsoon trough extending from Indonesia, southward through Australia, and then northward into the South Pacific. Of particular significance is the preferred region of low pressure over the southwest Indian Ocean, south of about 50°S. We will show that this area is characterized by major cyclogenesis events, which in the presence of an STJ, can influence the tropical circulation. Between the tropical and high-latitude low-pressure regions lies a significant anticyclone over the Indian Ocean, with trade wind easterlies on its equatorward flank. At upper levels (Fig. 1, lower panel), the tropical circulation is characterized by a major anticyclone over northern Australia, with a ridge axis extending both east and west along approximately 12°S. On the north side of the ridge axis lie the tropical easterlies. To the south lie westerly winds that extend continuously from high latitudes into the Tropics. Over the southern Indian Ocean, the polar front jet (PFJ) associated with the low-pressure region farther south is located near 45°S. The region of frequent cyclogenesis is located beneath the jet exit. In the climatology, the STJ over the eastern Indian Ocean can only be seen as a region of stronger winds (isotachs) extending from the central Indian Ocean toward southwest Australia. Also evident over this region at this time of year is a broad, northwest–southeast-oriented mean trough (indicated by the solid line in Fig. 1), characterized by cyclonically curved, west-southwest–west-northwest flow. We note that the mean flow over the eastern Indian Ocean is suggestive of a twin jet structure, which predominates over this area in all seasons except during southern summer. Examination of individual-day analyses indicates that at times the twin jets are well-defined, separate entities. These upper- and lower-level structures seem consistent with environments conducive to extratropical forcing of tropical circulations (Webster and Holton 1982) and will be used as the basis of our idealized simulations. Based on the climatology, we define regions over which we monitor the structure of the monsoon (region A in Fig. 1), high-latitude cyclogenesis over the southwest Indian Ocean (region B), and the strength of the STJ over the eastern Indian Ocean (region C).
The paper is divided into the following sections. Section 2 describes the data and numerical systems used. Section 3 discusses a simple version of the theory that underpins the observational diagnostics and idealized simulations. Section 4 analyses the local and large-scale changes to the flow prior to the onset of the monsoon in selected years and in a 15-yr onset composite. Section 5 describes results from various idealized numerical simulations. Section 6 is the summary and conclusions.
2. Data and numerical systems
a. Comparison of reanalysis datasets
The observational component of this study is based upon NCEP and 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) datasets. To test the robustness of the results, the large-scale indices used in the investigation have been cross-validated. An intercomparison is shown in Fig. 2. It shows daily time series over an extensive 42-day period, 21 November 1978 to 1 January 1979, from the NCEP and ERA-40 datasets of (a) 850-hPa zonal wind, (b) vertical motion, and (c) relative humidity, over the monsoon zone, 14°–6°S, 110°–140°E, and (d) MSLP over the high-latitude domain, 60°–45°S, 80°–100°E. The close agreement in the zonal wind and MSLP time series (also relative vorticity, but not shown here) is indicative of the consistency in the datasets for these standard variables, which are less prone to observational errors and data paucity, and less sensitive to the details in assimilation-system components (e.g., resolution and the cumulus parameterization in the forecast model). Given the well-known difficulties with errors in the divergence and moisture from objective analyses, the time series of vertical motion and relative humidity from the two datasets differ in the magnitudes of these quantities. Nonetheless, the reanalyses are suitable for our purposes since the temporal changes are consistent within each dataset. In particular, we focus on sustained, large-scale changes in these quantities to define thermodynamic and kinematic changes over the Tropics. The major, sustained change at day 33 is clearly evident in both time series, even though the magnitudes of the changes are different.
b. Idealized model configuration
The flow configuration used in this study is an extension of the classic three-dimensional channel model often used for idealized baroclinic instability and frontogenesis studies, the details of which are described in detail by Reeder et al. (1991) and Tory (1997). The basic state follows that of Keyser et al. (1989) with minor changes to take into account the Southern Hemisphere, spherical configuration using a latitude–longitude grid, a σ-coordinate system in the vertical, and an extension into the Tropics with the options of trade easterly flow, monsoon westerly flow, and a subtropical jet in the basic state. The details of each of these changes are listed in the appendix, together with the associated equations. (The basic state is illustrated later in Fig. 13 and discussed further in section 6.)
3. Theoretical background
The tropical response to high-latitude forcing has been investigated by Charney (1963, 1969), Webster and Holton (1982), Hoskins and Karoly (1981), Hoskins and Ambrizzi (1993), Yang and Hoskins (1996), and Hoskins and Yang (2000). This section briefly summarizes those aspects of the theory of particular relevance to the later sections of the paper.
Consider now the specific example of a Rossby wave generated in the extratropics, which subsequently propagates equatorward. Assume U decreases toward the equator and that, for simplicity, |d2U/dy2| ≪ β. In this case, l increases along the ray, defined by the direction tan−1(Cgy/Cgx,), while Cgx and Cgy both decrease along the ray. However, as Cgy decreases less rapidly than Cgx, (l−3 versus l−4 dependence) the ray is refracted equatorward. The critical latitude at which U = c is an effective barrier to the further equatorward propagation of the wave.
Observations presented in the following section suggest that the onset of the Australian monsoon is generally preceded by major high-latitude cyclogenesis in the presence of an STJ. The STJ does not appear to play a significant role in the cyclogenesis, but instead locally increases β*, decreases l (thus increasing Cgx and Cgy), and displaces the critical latitude equatorward, thereby allowing Rossby waves emanating from the high latitudes to penetrate farther into the Tropics. It will be shown that similar processes seem to act in idealized dry simulations of the influence of high-latitude cyclogenesis on the Tropics.
4. Analysis of pre-onset and onset characteristics
In a fashion similar to D96, but for the monsoon region, we define onset to be the day on which over the monsoon region A in Fig. 1: (i) lower-tropospheric westerly winds with magnitude greater than 2.5 m s−1 develop, (ii) the westerlies extend to at least 600 hPa, (iii) easterlies in the upper troposphere overlay the westerlies, and (iv) this structure persists for at least 4 days. This section focuses on the documented onsets in 1978–79, 1986–87, and 1992–93. The aim is to investigate the local structural changes in the circulation leading up to the onset of the monsoon, in order to determine within a day or two when preconditioning began. That is, to determine the commencement of changes in, for example, vorticity, divergence, ascent, and moistening that support sustained and widespread convection. This need not correspond with the onset dates normally defined by the strength and depth of westerlies or large-scale tropical cloudiness.
a. 1978–79 WMONEX onset
Figure 3 shows time–height series of the zonal wind, divergence, and relative humidity over the domain, 15°–5°S, 110°–140°E (region A in Fig. 1) for the period 21 November 1978 to 1 January 1979. There is a marked strengthening and deepening of the monsoon westerlies around 26 December, which are maintained beyond this time and are overlain by easterly winds in the upper troposphere (Fig. 3a). This day corresponds with the onset date defined in D83, who used a widespread outbreak of tropical convection to define onset, but is 10 days before that obtained by D96 (6 January). The reason for the difference appears to be the requirement by D96 that the westerlies strengthen and deepen at the Darwin radiosonde site, and are overlain by easterly winds at upper levels. Naturally there will be years when the dates of onset differ, since D96 is based upon a point estimate and that used here is from a large-scale average. For most years, the two methods give consistent results. For example, in 11 of the 15 seasons considered in the onset composite described in section 4d, the dates of onset from the point estimate and from the large-scale average are within 3 days. The 1978–79 onset date is one of three from 15 years that differ by more than 5 days.
Although there are episodes of enhanced low-level westerly flow prior to onset (near 3 and 15 December), these are associated with upper-level westerly flow and the intrusions of midlatitude troughs into the Tropics. This is not illustrated here but is clearly visible in the analyses. For onset, the upper-level flow is characterized by strengthening easterlies, more typical of a widespread convective event (see also D96).
Figure 3b shows that lower-level convergence and upper-level divergence increases from about 22 December. These changes clearly precede the onset. Figure 3c shows the evolution of relative humidity over the monsoon. Consistent with the vertical motion changes and the moisture changes described in H89 for the AMEX onset, sustained and deep moistening commences from at least 22 December, again around 4 days prior to the wind onset. Other shorter-term moistening events occur near 27 November, 7 December, and 16 December. An interesting association between the moist events, including onset, and earlier upper-level westerly wind maxima in Fig. 3a is noted. Further observational evidence of this association is provided below.
We suggest that preconditioning for onset commenced near 22 December, with organized large-scale ascent, the development of sporadic deep convection, and associated tropospheric moistening. We have marked this time in Figs. 3 and 4 with the solid vertical line. Eventually with maintenance of conditions favorable for large-scale ascent, a major convective outbreak developed that further enhanced the large-scale vertical motion, and then the low-level westerlies/upper-level easterlies, producing patterns of flow (and pressure; not shown) similar to that described theoretically by Gill (1981). However, what are these “conditions favorable for large-scale ascent” and how do they develop? Further discussion on this issue will be presented in section 4.
Figure 4 shows time series of 850-hPa zonal wind over the monsoon (top panel), MSLP over the SWIO (middle panel), and 200-hPa zonal wind over the eastern Indian Ocean (bottom panel). It is important to note that the fixed locations of the large regions over the Indian Ocean are sometimes not ideal for monitoring the cyclogenesis and STJ within each year and from year to year. Because weather systems in this region translate and intensify, and display seasonal and interannual variability in the preferred location for cyclogenesis, the indices based on fixed locations can sometimes be misleading. Clearly if the cyclogenesis commences to the west of the monitoring domain and then the system moves into the domain, then an incorrect late timing is inferred. For this reason, for the 1992–93 season, small adjustments to the locations of regions B and C have been made. These are described in section 4c.
Figure 4a shows temporary bursts in the monsoon, prior to onset. Figure 4b shows a major high-latitude cyclogenesis event developing from 20 December, approximately 1 to 2 days prior to the structure changes in the Tropics and some 6 days prior to the onset of the monsoon. We note that the event actually commences near 17 December and to the west of the SWIO region B (see Fig. 1), but is only indicated in the time series from 20 December. The cyclogenesis event is the only major high-latitude cyclogenesis event that occurs within the large monitoring box during this 40-day period. Unique, single cyclogenesis events like this rarely occur in other years. Interestingly, weaker cyclogenesis events appear to precede earlier upper-level westerly wind episodes and moistening events over the Tropics.
Figure 4c shows the remarkable evolution and reorganization of the STJ just prior to onset. Minor fluctuations in the STJ occur from 21 November to 17 December. A marked strengthening in the jet then occurs, which is followed by a pronounced, sudden weakening from 23 to 27 December. We note that the strengthening coincides with the high-latitude cyclogenesis event. In agreement with Troup (1961), a similar evolution of the STJ over the Australian region (weakening or southward displacement) occurs in the vast majority of onsets we have investigated. Idealized modeling studies by Son and Lee (2005) also suggest that the presence of one or two jets is dependent upon the relative strengths of tropical heating and high-latitude cooling, with strong tropical heating favoring a single-jet structure. Since the interpretation from these simulations is based upon steady states and forced tropical heating, cause and effect still remains somewhat unclear. However, the conclusion from each of these studies is that a single-jet structure develops as widespread tropical convection increases. Further study is needed to determine the precise relationship between changes in the midlatitude flow, the evolution of the subtropical jet, and the development of widespread tropical convection.
The horizontal flow changes during onset are illustrated in Fig. 5, which shows analyses of the 850 and 200 hPa wind at 0000 UTC 18, 22, 26, and 30 December. At low levels (Fig. 5, left panels) well prior to onset, the flow was featureless, with weak easterly winds over the Tropics, and weak anticyclones and cold frontal troughs located over the subtropics and high latitudes. A deep extratropical cyclone develops near 55°S, 90°E by 22 December. Following this, the subtropical ridge (STR) over the Indian Ocean near 20°S, 90°E develops (or moves eastward), and the monsoon trough (MT) begins to form along 15°S. By 30 December, the tropical westerlies have been established across the longitudinal extent of the monsoon from 80°E to at least 170°E.
At upper levels (Fig. 5, right panels), a major upper trough develops off the coast of Western Australian just prior to onset. This is consistent with the findings of Keenan and Brody (1988) and Hung and Yanai (2004). The strengthening in the STJ over the eastern Indian Ocean between 20° and 30°S and a connection between the PFJ and STJ is clearly evident in Fig. 5f. A major trough establishes off the coast of Western Australia (Fig. 5g) and coincides with the formation of the MT near 15°S, 115°E. At this time an equatorward-extending trough–ridge–trough–ridge wave pattern extends from the SWIO (∼55°S, 60°E) to north of Australia (∼10°S, 140°E). Note also the anticyclonically curved flow that develops over Borneo and Indonesia as the trough–ridge couplet forms in the eastern Indian Ocean and Australian Tropics. Between these systems the monsoon trough begins to form at low levels. Following onset, the STJ over both the Indian Ocean and Australia becomes either displaced southward (see also Troup 1961), or locally disappears (Fig. 5h). Note that prior to onset, the polar front and subtropical jets are identifiable as separate entities. Post-onset the separation becomes much less obvious and a single-jet structure is evident. The dramatic flow changes over all latitudes can be seen by comparing Fig. 5a with Fig. 5d, and Fig. 5e with Fig. 5h. We propose that these circulation changes are part of a large-scale reorganization of the flow, associated with major high-latitude cyclogenesis in the presence of an STJ, and the subsequent outbreak of widespread tropical convection during onset. Further evidence to support this hypothesis is provided in subsequent sections.
b. 1986–87 AMEX onset
Figure 6 is similar to Fig. 3 but for the onset of the monsoon during AMEX. In this year, the wind onset (Fig. 6a) occurs near 15 January, which corresponds well with the date determined by D96. Westerly winds occur through most of the period prior to onset, but they are weak and shallow. Lower convergence and upper divergence (Fig. 6b) increase between 3 and 6 January. Low- and midtropospheric moistening (Fig. 6c) is diagnosed to slowly increase from 1 January and more rapidly from about 10 January. Prior to this, other transient periods of moistening are evident around 10, 18, and 27 December. The solid vertical lines in Fig. 6 indicate the earliest time that sustained moistening and divergence changes commenced. As in 1978–79, a preconditioning period precedes the onset of monsoon westerly winds. An association over the Tropics between moistening events and upper-level westerly wind maxima is again noted.
The time series in Fig. 7 are similar to those in Fig. 4 but for the AMEX onset. The time series of monsoon westerlies indicates a marked change from easterly to westerly winds commencing near 1 January. This is followed by a period of weak westerly winds until a rapid acceleration in the monsoon occurs after 12 January.
Figure 7b shows the evolution of high-latitude surface pressure over the SWIO. Short-term cyclogenesis events are evident near 9, 17, and 24 December, which, similar to the events in 1978–79, correspond to some extent with the moistening over the Tropics on 10, 18, and 27 December. A period of low pressure over the SWIO commences near 27 December and extends to 8 January with major peaks near 30 December to 3 January, with another near 7 January. This period of low pressure over the SWIO precedes the sustained period of tropical moistening and the onset of the monsoon westerlies.
Figure 7c shows the evolution of the Indian Ocean STJ. During this time zonal winds associated with the STJ are maintained at about 25 m s−1, but with two major reorganizations. The second, during the period 8 to 12 January, precedes the monsoon onset. Interestingly, an earlier cyclogenesis event near 17 December occurs at a time when the Indian Ocean STJ was relatively weak. Accordingly we might expect that, based on theoretical considerations, the tropical response to this event would be weak.
Figure 8 shows 850- and 200-hPa wind analyses during the preconditioning phase. Major high-latitude cyclogenesis had occurred near 60°S, 100°E, three days prior to this time (Fig. 7b). Remnants of this event can be seen over the SWIO. The low-level ridge over the eastern Indian Ocean (near 30°S, 90°E, labeled with an R), and the early stages of the formation of the MT just south of Indonesia can be seen in Fig. 8. At upper levels, the trough over the eastern Indian Ocean extends equatorward, and the trough–ridge–trough–ridge structure is established between the SWIO and the Australian Tropics. Again the trough–ridge pattern forms over the eastern Indian Ocean and Australian Tropics, with anticyclonically curved flow over Borneo and Indonesia between the trough and ridge. Following onset the STJ mostly disappears or moves southward (not shown).
c. 1992–93 TOGA COARE onset
Figure 9 corresponds with Fig. 3 for the TOGA COARE onset. In this year, the wind onset (Fig. 9a) appears to occur near 22 January, which corresponds well with the date suggested by D96. However, we note that low-level westerly winds were present over the domain for nearly the whole period but, as in 1986–87, were generally weak and shallow. Upper-level easterlies overlaying low-level westerlies, albeit relatively weak, were also present around 25 December. During this year two clearly defined MJO events occurred (M95; Hung and Yanai 2004). The first event traversed the monsoon in late December 1992 and the second in late January 1993. For this reason the timing of monsoon onset is somewhat ambiguous in this season.
Lower-level convergence and upper-level divergence (Fig. 9b) increase between 14 and 17 January. Low- and midtropospheric moistenings (Fig. 9c) are diagnosed to increase from approximately 14 January, with a large increase near 22 January. Prior to this, other temporary periods of moistening are evident around 24 December and 6 January, which as in other years, are associated with upper-level westerly wind maxima. The solid vertical line in Fig. 9 indicates the earliest time that sustained moistening and divergence changes have commenced. As in 1978–79 and 1986–87, a preconditioning period commences prior to the onset of monsoon westerly winds.
Figure 10 is similar to Fig. 7 but for the TOGA COARE onset. The time series of monsoon westerlies indicates a marked change from easterly to westerly winds commencing near 14 January, with a marked acceleration near 22 January. The “fake” onset near 25 December and another burst near 5 January (overlain by upper-level westerlies) are also evident.
Figure 10b shows the evolution of high-latitude surface pressure over the SWIO. Cyclogenesis events are evident near 24 December, 3, 11, and 23 January, which similar to the events in 1978–79 and 1986–87 correspond to some extent with the moistening over the Tropics on 24 December, and 6, 14, and 25 January.
Figure 10c shows the evolution of the Indian Ocean STJ. The late January onset date causes some issues regarding monitoring of the higher-latitude flow changes, since the STJ and polar front have relocated somewhat poleward by this time. Accordingly in the time series (Fig. 10), we have shifted poleward by 5° the monitoring zone for SWIO pressure, and by 10° for the STJ. Major reorganizations—strengthening then weakening of the STJ—are associated with the tropical westerly wind maxima and precede each of the tropical moistening events. Interestingly, earlier cyclogenesis events near 25 December (corresponding to the “fake” onset) and 4 January occur at times when the Indian Ocean STJ index was relatively weak or weakening. Accordingly we might expect that the tropical response to these particular, nononset events to be weak.
Figure 11 shows 850- and 200-hPa wind analyses prior to the onset of monsoon westerlies. The left panel shows a mature high-latitude cyclone near 60°S, 80°E (labeled with a generalized “T” for trough). The right panels show the equatorward-extending trough developing over the eastern Indian Ocean, and the establishment of the trough–ridge–trough–ridge structure extending from the SWIO to the Australian Tropics. The trough–ridge couplet over the eastern Indian Ocean and Australian Tropics is again present, with the developing monsoon trough located between these systems but at low levels. These patterns are similar to those described for the 1978–79 and 1986–87 onsets.
d. 15-yr pre- and post-onset composites
The association between high-latitude cyclogenesis, the eastern Indian Ocean STJ, and the monsoon onset is explored with a composite of the days prior to and following onset. The composite comprises analyses around onset for 15 seasons when the monsoon onset was well defined. The onset dates, which were determined from the definition given in section 4 and by examining diagrams similar to Figs. 3a, 6a and 9a are as follows: 26 December 1978, 29 December 1979, 6 January 1981, 11 January 1982, 1 January 1983, 31 December 1983, 15 January 1986, 15 January 1987, 17 December 1987, 13 December 1989, 22 December 1990, 6 January 1992, 13 January 1995, 25 December 1997, and 26 December 1998. Eleven of the 15 dates are within 3 days of those defined by D96. Thirteen of the dates are earlier than in D96. Three dates (1978–79, 1982–83, 1983–84) are more than 5 days different from D96.
Because of differences in the location, amplitude, and phase of various weather systems from year to year, there will necessarily be some smearing and smoothing in time and space within the composites. The rationale in forming the composites is of course that the signal to noise ratio will be enhanced. Figures 12a–f show pre- and post-onset composites of 850- and 200-hPa wind, and MSLP difference from long-term-mean December climatology. The top panels clearly show the strengthening in the 850-hPa monsoon westerlies over regions to the north of Australia from the pre- to the post-onset composite. That is, the composites are truly indicative of pre- and post-onset conditions. Also evident in the pre-onset composite is the high-latitude cyclogenesis near 60°S, 90°E. Cyclonic flow is still evident over this area during post-onset, but the pre-onset circulation is much stronger and of much larger scale. Consistent with a strong STR at pre-onset, the easterly trade winds over the Indian Ocean become less extensive and weaker by post-onset.
At 200 hPa (Figs. 12b,e) the structures that were highlighted for individual years are also present in the composite. The trough–ridge–trough–ridge structure extends from near 60°S, 40°E, to 45°S, 80°E, to 30°S, 100°E and into the Tropics near 10°S, 130°E. The trough and STJ over the eastern Indian Ocean are evident at pre-onset, and completely disappear by post-onset. The PFJ over the SWIO is intense during pre-onset and locally weakens and extends somewhat eastward by post-onset. The evolution from a pre-onset twin jet to a post-onset single-jet structure is evident. The flow over midlatitudes generally becomes more zonal and weakens by post-onset.
The anomalies in MSLP (Figs. 12c,f) mark the high-latitude cyclogenesis over the SWIO in the pre-onset composite. The scale of the anomaly is very large with low pressure extending to the coast of Western Australia. Smaller low-pressure anomalies are still present in the post-onset composite, but now high pressure has developed over southern Australia, and a low-pressure anomaly has formed farther eastward near 50°S, 160°E. Over the Tropics, pressure anomalies evolve from relatively high to relatively low pressure as the monsoon develops.
Figure 12g provides evidence of the Rossby wave propagation from high latitudes into the Tropics, as represented in the composite onset period by streamfunction anomalies (deviations from the zonal mean). The figure is a “skewed Hovmüller diagram” in which time is the ordinate and increases upward from 8 days prior to the mean onset day (O-8) to 3 days prior to onset (O-3), and spatial strips for each day cover the southwest to northeast domain from (60°S, 40°E) to (10°N, 170°E). The bottom left of the diagram thus represents anomalies over the SWIO 8 days prior to onset. The top right of the diagram represents anomalies over the equatorial Pacific 3 days prior to onset. The energy dispersion we expect to see should thus extend from the bottom left to the top right of the diagram as a series of developing troughs and ridges. Although the signal over the western and central Indian Ocean does not show a consistent temporal development, there is still clear evidence of the developing trough and ridge. The subsequent development of the trough–ridge–trough structure over the “better observed” Australian sector is clearly depicted in the diagram. The equatorward extension of the trough off the west Australian coast, documented in earlier studies, is evident and is one synoptic-scale component of the propagation into the Tropics. Note as well the enhanced and then weakening gradients in streamfunction anomaly (and by implication a strengthening and weakening in the STJ) on the eastern flank of the trough during the passage of the wave.
5. Idealized simulations
The initial state for the idealized simulations comprises (a) a zonal PFJ at 40°S, (b) a monsoon trough at 15°S, and (c) an STJ at 25°S. A deep low develops at high latitudes as the PFJ is baroclinically unstable. The two experiments described here are as follows:
Monsoon trough (MT experiment): a PFJ at 40°S and an MT at 15°S.
Monsoon trough and subtropical jet (MT_STJ experiment): a PFJ at 40°S, MT at 15°S, and an STJ at 25°S.
The zonally uniform initial state is based upon cross sections of analyses made just prior to midlatitude cyclogenesis. Figure 13 compares meridional cross sections of the zonal wind and potential temperature over the Indian Ocean from (a) the NCEP December climatology, (b) a single day prior to the monsoon onset of 1978–79, and (c) the initial condition for the MT_STJ experiment. The right panels show a baroclinic zone centered at 40°S with a sloping tropopause. The associated zonal flow is geostrophic and increases from zero at the surface to a maximum of about 85 m s−1 in the stratosphere near the center of the baroclinic zone. The basic-state zonal flow includes simple representations of the easterly trade and westerly monsoon winds, centered at 25° and 11.5°S respectively. The wind speeds decrease nearly linearly with height, and change sign at σ = 0.75 and 0.5 respectively. The choice was somewhat arbitrary, but shows reasonable agreement with the general structure of the December zonal mean zonal flow over the central Indian Ocean (left panel). Although the jet stream speeds are stronger and the low-level, high-latitude westerly winds weaker in the simulation than in either the climatology or for a single day, the tropical and midlatitude structures are very similar. Note however that the introduction of the MT and STJ slightly alters the flow over the Tropics and subtropics and this affects the interpretation of the simulations. To allow simple comparison between the experiments we will thus mostly show the changes that occur between the initial condition and subsequent integration times.
It takes approximately 3 days of simulation time for the midlatitude flow to show clear signs of cyclogenesis. Consequently we confine our attention to simulation times longer than this. The longitudinal location of the domain of the model is arbitrary. In the figures shown below, some coastlines have been included to provide a sense of scale. The full extent of the domain for this application is approximately 69.2°S, 9.0°N, 85.0°E, 133.7°E. The results are not specific to the region shown. The model is run with free-slip lower boundary conditions (no surface interaction) and periodic nesting boundary conditions. For these reasons and because the model is dry, we do not expect it to reproduce observed tropical flows. The simulations are used to study if, how, and why high-latitude cyclogenesis may influence the flow over the Tropics.
Figure 14 shows for the MT_STJ experiment, the 200- and 950-hPa flow at t = 0 and t = 168 h. The heavy dashed line on the 950-hPa chart shows the location of the monsoon trough separating tropical westerlies from subtropical easterlies. The figure illustrates (a) the development of the high-latitude, low-level cyclone near 50°S, (b) the developing anticyclonic flow extending south to north along about 110°E, (c) the strengthening easterly flow to the north of the anticyclone, (d) the development of zonal asymmetry in the MT as the cyclogenesis evolves, (e) the development of the upper cyclone associated with the cyclogenesis near 55°S, and (f) the equatorward extension to at least 15°S of an upper trough (located over western Australia), which appears to emanate from the upper cyclone but which has tilted westward and become separated from the high-latitude trough. Structures similar to that described here can be seen in the analyses in Fig. 11, which depicts the zonal asymmetry of the MT, and the equatorward-extending upper trough.
Since the simulations are dry, there are no internal, local processes to influence changes in the MT. The structure of the MT changes only through processes initiated from the high-latitude cyclogenesis.
Figure 15 shows the changes in the 200- and 950-hPa flow fields over the Tropics at 4 days into the MT and MT_STJ simulations. At this time, the changes in the tropical winds from the MT experiment are very small (less than 1 m s−1). For the MT_STJ experiment the changes are larger but still small, although there is some evidence of an upper trough extending into the Tropics and an acceleration of the trade easterlies.
Figure 16 is similar to Fig. 15 but depicts the changes after 7 days of simulation time. The influence of the high-latitude cyclogenesis on the tropical flow has increased in both experiments, but the changes in the MT_STJ experiment are much larger. The changes in both experiments have a similar large-scale structure on the order of at least 1500 km. That is, the flow changes are systematic and not associated with model noise or small-scale gravity waves. The intrusion or development of a trough–ridge structure at upper levels into the Tropics is clearly evident and indicates that in the simulation there are regions within 10° of the equator where the easterlies have accelerated and decelerated and the divergence and vorticity have changed. At low levels, the zonal asymmetry in the acceleration of the easterlies appears to be the cause of the zonal asymmetry of the MT noted earlier. There is virtually no change in the low-level monsoon westerlies north of 8°S in each experiment, suggesting that convection and latent heating are of critical importance to their development (e.g., Gill 1981).
In the earlier discussion on the observational analysis it was suggested that northeastward group propagation originating during the high-latitude cyclogenesis may be important for modulating the tropical flow some days later. Since this process is not fundamentally dependent on moist processes, we might expect to see it within the idealized simulations. Figure 17 shows the vector difference between the t = 168 h and t = 0 h, 200-hPa wind for the MT_STJ experiment; T’s and R’s mark where the trough is deepening and the ridge building. Since the winds over the Tropics are much lighter than those over the extratropics we have split the diagram into these domains and scaled the winds so that they appear similar in each domain. There is clear evidence of a trough–ridge–trough–ridge structure emanating from the region of high-latitude cyclogenesis and extending into the Tropics. The solid line on the diagram shows this structure and links the large-scale circulation tendencies over the 7-day period. The structure is suggestive of Rossby wave propagation from high latitudes into the Tropics. Because of assumptions necessary to obtain the theoretical solutions (section 3) and the meridionally varying basic state in the simulations, it is difficult to compare in a simple way the group propagation speeds predicted by the theory and those indicated in the simulations. As part of our future work we plan to examine this more thoroughly from objective analyses and numerical simulations and forecasts.
Finally it is useful to examine if the tropical flow changes associated with the dry dynamics may provide an environment favorable for a large-scale outbreak of convection. Figure 18 shows the change in horizontal divergence and relative vorticity at 200 hPa (top panels) and 950 hPa (bottom panels), during 168 h of simulation time for the MT_STJ experiment. The diagrams suggest that (a) at low levels over the Tropics, the vorticity changes are larger than the divergence changes and related to the enhanced subtropical easterlies, and (b) at upper levels over the Tropics, the divergence changes are larger than the vorticity changes, and are located in two regions, to the east of the upper trough and along an area to the northeast of the trough. The latter region is consistent with the findings of Keenan and Brody (1988) who found a similar location for the upper divergence and tropical cloudiness, but which to date has not been satisfactorily explained. Examination of Figs. 14, 16, 17 and 18 suggests an association with the accelerating easterlies and westerlies on each side of the developing ridge, which is part of the group propagation into the Tropics. This will be further investigated in future studies. The simulations suggest that high-latitude cyclogenesis in the presence of an STJ can influence the vorticity and divergence over the Tropics. Although the changes are not large in magnitude, they are of large spatial scale. We suggest that they are sufficient to provide a favorable environment to trigger convection (upper divergence changes), and to allow efficient feedback between the convection and the large-scale flow (increasing low-level cyclonic vorticity), which can then further intensify the monsoon trough.
6. Summary and conclusions
Traditional definitions of monsoon onset are based on either the strength and depth of tropical westerly winds, or tropical cloudiness. While valuable, the evidence presented here suggests that a preconditioning phase, characterized by sustained large-scale ascent and tropospheric moistening begins about a week prior to the onset of the deep westerlies and widespread cloudiness.
Evidence is provided that preconditioning and onset are associated with major cyclogenesis over high latitudes in the southwest Indian Ocean. All individual onsets studied so far are characterized by the high-latitude cyclogenesis. Indeed a 15-season composite around the time of onset shows that 5 days prior to the mean onset, a major high-latitude cyclone develops near 60°S, 90°E. An additional critical feature seems to be the existence of a subtropical jet. Although the STJ does not play a role in the cyclogenesis, it appears crucial to the interaction between the midlatitude and tropical circulations. The evolution of the large-scale flow is suggestive of Rossby wave propagation as successive ridges and troughs develop downstream and equatorward from the initial high-latitude cyclogenesis region. This results in the amplification of an equatorward-extending midlatitude upper trough and tropical ridge. This couplet, associated with anticyclonically curved flow out of the Tropics, appears to influence the development of the monsoon trough. More work is needed to understand the precise nature of the low-level changes in the monsoon trough.
We cannot yet address the important issue of the relative roles of the MJO and extratropical forcing on onset. Besides its documented role, it is possible that the MJO, as well as the high-latitude cyclogenesis, may influence the strength of the STJ and thus assist in the group propagation into the Tropics. We hope to study these problems in the future.
Previous studies have suggested the importance on onset and active phases of the monsoon of the subtropical ridge over the eastern Indian Ocean (e.g., D83) and equatorward-extending upper troughs (Keenan and Brody 1988; Hung and Yanai 2004). Onset is often associated also with a marked weakening or relocation of the STJ over Australia (Troup 1961). Each of these observational aspects is consistent with the mechanism proposed here, but this new interpretation provides a more general framework for interpreting the extratropical–tropical interactions. Each component is part of a much more general process, which appears to have its roots in high-latitude cyclogenesis events over the SWIO.
During the transition season in the 6 weeks prior to onset, a number of moist westerly events occur. Generally they are only temporary and overlain by upper-level westerly winds, suggesting an association with midlatitude troughs, which sometimes protrude into the deep Tropics. It still remains unclear what distinguishes onset from the earlier moist westerly events, and what distinguishes the onset cyclogenesis event. Possibly other processes (e.g., seasonal forcing, the MJO, the circulation in the Northern Hemisphere) have not been established for the earlier events. The apparent dual dependence on the intensity of the cyclogenesis and the strength of the STJ also makes this difficult to unravel. Both the observational analysis and the simulations suggest that the simultaneous presence of both is important.
To test the hypothesis that the onset of the Australian monsoon is linked to, or modulated by, extratropical–tropical interaction, idealized, dry simulations are used to study the influence of high-latitude cyclogenesis on the tropical circulation. A three-dimensional baroclinic wave channel model provides the framework. The initial state consists of (a) a zonally constant baroclinic region centered on 40°S, from which the high-latitude cyclogenesis develops, (b) a weak monsoon trough at 15°S, and (c) a subtropical jet at 25°S.
The key findings are as follows: 1) There is evidence in the simulations of northeastward Rossby wave propagation from the cyclogenesis region toward low latitudes. 2) Consistent with theoretical studies, the subtropical jet plays a key role by providing a favorable westerly background flow for group propagation into the Tropics. 3) High-latitude cyclogenesis in the presence of a subtropical jet can influence the meridional location, zonal structure, vorticity, and divergence in the monsoon trough. 4) The vorticity and divergence changes are consistent with enhancement of the monsoon trough (low-level increases in cyclonic vorticity) and the potential for triggering of a large-scale convective outbreak (changes in upper-level divergence).
Further observational and numerical prediction studies of onset are planned. We also hope to extend the idealized simulation methodology to include (a) moist processes, (b) possible influences from the Northern Hemisphere (e.g., cold surges; Chang et al. 1983), (c) simultaneous high-latitude cyclogenesis events in both hemispheres, and (d) the triggering of the MJO and genesis of tropical weather systems.
Acknowledgments
This work was supported in part by NICOP-ONR Grant N00014-02-1-0431. Thanks are also extended to the editor, external reviewers, and BMRC colleagues Drs. Peter May, John McBride, and Jeffrey Kepert for helpful comments during the course of the study.
REFERENCES
Chang, C-P., J. E. Millard, and G. T. Chen, 1983: Gravitational character of cold surges during Winter MONEX. Mon. Wea. Rev., 111 , 293–307.
Charney, J. G., 1963: A note on large-scale motions in the tropics. J. Atmos. Sci., 20 , 607–609.
Charney, J. G., 1969: A further note on large-scale motions in the tropics. J. Atmos. Sci., 26 , 182–185.
Davidson, N. E., and H. H. Hendon, 1989: Downstream development in the Southern Hemisphere monsoon during FGGE/WMONEX. Mon. Wea. Rev., 117 , 1458–1470.
Davidson, N. E., J. L. McBride, and B. J. McAvaney, 1983: The onset of the Australian monsoon during winter MONEX: Synoptic aspects. Mon. Wea. Rev., 111 , 496–516.
Drosdowsky, W., 1996: Variability of the Australian summer monsoon at Darwin: 1957–1992. J. Climate, 9 , 85–96.
Gill, A. E., 1981: Some simple solutions for heat-induced circulation of the tropical atmospheres. Quart. J. Roy. Meteor. Soc., 106 , 447–462.
Hendon, H. H., and B. Liebmann, 1990: The intraseasonal (30–50 day) oscillation of the Australian summer monsoon. J. Atmos. Sci., 47 , 2909–2924.
Hendon, H. H., N. E. Davidson, and B. W. Gunn, 1989: Australian summer monsoon onset during AMEX 1987. Mon. Wea. Rev., 117 , 370–390.
Hoskins, B. J., and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38 , 1179–1196.
Hoskins, B. J., and T. Ambrizzi, 1993: Rossby wave propagation on a realistically longitudinally varying flow. J. Atmos. Sci., 50 , 1661–1671.
Hoskins, B. J., and G-Y. Yang, 2000: The equatorial response to higher-latitude forcing. J. Atmos. Sci., 57 , 1197–1213.
Hung, C-W., and M. Yanai, 2004: Factors contributing to the onset of the Australian summer monsoon. Quart. J. Roy. Meteor. Soc., 130 , 739–758.
Keenan, T. D., and L. R. Brody, 1988: Synoptic-scale modulation of convection during the Australian summer monsoon. Mon. Wea. Rev., 116 , 71–85.
Keyser, D., B. D. Schmidt, and D. G. Duffy, 1989: A technique for representing three-dimensional vertical circulations in baroclinic disturbances. Mon. Wea. Rev., 117 , 2463–2494.
Lim, H., and C-P. Chang, 1981: A theory for midlatitude forcing of tropical motions during winter monsoons. J. Atmos. Sci., 38 , 2377–2392.
McBride, J. L., N. Davidson, K. Puri, and G. Tyrell, 1995: The flow during TOGA-COARE as diagnosed by the BMRC tropical analysis and prediction system. Mon. Wea. Rev., 123 , 717–736.
Puri, K., G. Dietachmayer, G. Mills, N. Davidson, R. Bowen, L. Logan, and L. Leslie, 1998: The new BMRC Limited Area Prediction System, LAPS. Aust. Meteor. Mag., 47 , 203–223.
Reeder, M. J., D. Keyser, and B. D. Schmidt, 1991: Three-dimensional baroclinic instability and summertime frontogenesis in the Australian region. Quart. J. Roy. Meteor. Soc., 118 , 1–28.
Son, S-W., and S. Lee, 2005: The response of westerly jets to thermal driving in a primitive equations model. J. Atmos. Sci., 62 , 3741–3757.
Straub, K. H., and G. N. Kiladis, 2003: Extratropical forcing of convectively coupled Kelvin waves during austral winter. J. Atmos. Sci., 60 , 526–543.
Tory, K. J., 1997: The effect of the continental boundary layer on the evolution of fronts. Ph.D. dissertation, Monash University, 294 pp.
Troup, S., 1961: Variations in upper tropospheric flow associated with the onset of the Australian monsoon. Indian J. Meteor. Geophys., 12 , 217–230.
Webster, P. J., and J. R. Holton, 1982: Cross-equatorial response to middle-latitude forcing in a zonally varying basic state. J. Atmos. Sci., 39 , 722–733.
Yang, G-Y., and B. J. Hoskins, 1996: Propagation of Rossby waves of nonzero frequency. J. Atmos. Sci., 53 , 2365–2378.
APPENDIX
Idealized Model Configuration
The idealized model configuration was designed to investigate the influence of midlatitude baroclinic waves on the Tropics. Thus a three-dimensional channel model was constructed with an extension into the Tropics. Following Keyser et al. (1989) a baroclinic zone was introduced, centered in the midlatitudes, with a 4000-km wavelength surface pressure perturbation added to initiate the growth of the baroclinic wave. A meridional potential temperature specification defined the basic-state thermal structure, and the wind was specified to be in geostrophic balance with this thermal structure. The trade easterlies and monsoon westerlies were introduced in the same manner. The mathematical details of the initial baroclinic zone specification are described in the appendix of Keyser et al. (1989). Changes to their method are as follows.
Southern Hemisphere
The Southern Hemisphere modification required the sign reversal of two parameters ∇θ and μθ (see Table A2 for parameter list).
Spherical coordinates
Keyser et al. (1989) formulated their initial state on an f plane using Cartesian coordinates (x, y in the horizontal, and a pressure-dependent pseudoheight, z, in the vertical). In the Australian Limited Area Prediction System (LAPS; Puri et al. 1998) we have maintained the same formulation, and specified the x–y grid to be square at the center of the midlatitude baroclinic zone (y = y0, ϕ = ϕ0). Thus, a grid spacing, ds, of 50 km at y = y0 yields a latitude and longitude grid spacing of 0.450° and 0.587°, respectively. This results in a constant meridional grid spacing (dy = ds) and a latitude varying zonal grid spacing (dx = ds cosϕ/cosϕ0). (The only zonal dependence in the initial state appears in the surface pressure perturbation.) The choice to make the grid square at the center of the midlatitude baroclinic zone (rather than at the equator) had the additional positive effect of increasing dx everywhere, but more importantly near the poles, where the convergence of longitude meridians requires a reduced time step. This restriction was relaxed further by shifting the center of the midlatitude baroclinic zone 5° closer to the equator (a more typical summertime configuration), and reducing the channel width (excluding tropical extension) from 8000 to 6500 km. [Tory (1997) used the same basic model as Keyser et al. (1989), with a reduced channel width of 6500 km with no noticeable changes in result.]
Pseudoheight in σ coordinates
Keyser et al. (1989) derived the initial flow configuration using a pressure-dependent pseudoheight. Expressed in σ coordinates the pseudoheight becomes dependent on surface pressure requiring the surface pressure to be defined before the other mean flow quantities. Twenty-nine sigma levels are used in all experiments.
Trade easterlies and monsoon westerlies
To incorporate simple representations of trade easterlies and monsoon westerlies in the basic state, a 2000-km extension of the channel into the Tropics was required. Meridional potential temperature gradients (constant in height) were added to set up geostrophic flows (linearly varying in height) of the desired direction and magnitude. Unlike the midlatitude baroclinic region the desired tropical basic-state winds were nonzero at the surface; thus further modifications to the initial surface pressure specification were required. Parameters specifying the maximum meridional gradients in surface pressure and temperature were required to define the baroclinicity and hence vertical gradients in wind speed. To easily choose the subsequent wind profiles these parameters were calculated as functions of the desired surface wind speed and the vertical level at which the winds were to change sign.
Initial-state procedure
The initial state was constructed in the following order. All variables are listed in Table A1 and all parameters are described and specified in Table A2.
- Surface pressure:Otherwise,where psymax1 and psymax2 have been determined from the geostrophic wind equation (see point 4 below) when the geopotential gradient is zero. This allows the maximum surface pressure gradients to be determined from a chosen surface wind speed (Usfc). Thus,
- Pseudoheight:
- Potential temperature:Here the tropopause height, h, is determined in the same manner as Keyser et al. (1989) with the inclusion of additional terms to take into account the tropical winds. Thus the expressionis solved implicitly using a Runge–Kutta method. The solution begins with h, at the poleward edge of the domain, set to the equivalent value for an h profile in an upright baroclinic zone,The terms Tymax1 and Tymax2 are the maximum meridional temperature gradients (constant in height) associated with the trade easterly and monsoon westerly winds respectively. The choice of values for these parameters is somewhat arbitrary. We chose to specify the wind profiles by the surface wind speed and height at which the wind changes sign. Thus, the thermal wind equation was used to formulate the Tymaxn terms as a function of the σ level where u = 0 (σu=0), and the surface wind speed (Usfcn):where the terms ∂u/∂lnσ, ∂lnps/∂y, and ∂T/∂lnσ have been approximated by −Usfcn/ln(σu=0,n), psymaxn/ p0, and (Rθ0/cp) − RN 21(θ0/g)2, respectively. The latter approximation is most valid near the surface.
- Geostrophic wind:providedThe geopotential, φg, is determined hydrostatically. To avoid, divide by zero errors near the equator a minimum value for ϕ has been imposed where sin(2ϕ) appears as a denominator in the equations above, such that ϕl = ϕ when |ϕ| > ϕmin and ϕl = ϕmin when |ϕ| ≤ ϕmin.
January climatology from NCEP reanalysis data for (top) mean sea level pressure in hPa and (bottom) 200-hPa wind in m s−1. Shading indicates winds greater than 15 and 30 m s−1. The highlighted rectangles show regions used to monitor the structure of the monsoon (A: 15°–5°S, 110°–140°E), high-latitude cyclogenesis over the southwest Indian Ocean (B: 60°–45°S, 80°–100°E), and the strength of the Indian Ocean subtropical jet (C: 30°–20°S, 90°–120°E).
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
(a) Time series for the 42-day period 21 Nov 1978 to 1 Jan 1979 of 850-hPa zonal wind in m s−1 for the zone 15°–5°S, 110°–140°E from NCEP (solid line) and ERA-40 (dashed line). (b) As in (a) but for 400-hPa vertical motion in hPa s−1 (ascent is negative). (c) As in (a) but for relative humidity in %. (d) Mean sea level pressure in hPa for the zone, 60°–45°S, 80°–100°E.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
1978–79 WMONEX onset. Time–height sections of (a) zonal wind, (b) horizontal divergence, and (c) relative humidity, for the zone 15°–5°S, 110°–140°E from the NCEP reanalysis dataset, for the 42-day period 21 Nov 1978 to 1 Jan 1979. Contour intervals are 2.5 m s−1, 1.0 × 10−6 s−1, and 10%, respectively. The solid vertical line marks the approximate start of the monsoon preconditioning period.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
1978–79 WMONEX onset. Time series of (a) zonal wind for the zone 15°–5°S, 110°–140°E, (b) surface pressure for the zone 60°–45°S, 80°–100°E, and (c) zonal wind at 200 hPa for the zone 30°–20°S, 90°–120°E, for the 42-day period 21 Nov 1978 to 1 Jan 1979. Units are m s−1, hPa, and m s−1, respectively.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
1978–79 WMONEX onset. (left) 850- and (right) 200-hPa wind for 0000 UTC (a), (e) 18 Dec 1978; (b), (f) 22 Dec 1978; (c), (g) 26 Dec 1978; and (d), (h) 30 Dec 1978. Contour interval for 850-hPa wind is 10 m s−1 with light and dark shadings for winds greater than 10 and 20 m s−1, respectively. Contour interval for 200-hPa wind is 10 m s−1 with light and dark shadings for winds greater than 20 and 40 m s−1 respectively. The T and R labels indicate where the troughs have deepened and the ridges built.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Same as Fig. 3 but for 1986–87 AMEX onset and the 42-day period 5 Dec 1986 to 16 Jan 1987.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
1986–87 AMEX onset. Time series of (a) zonal wind for the zone 15°–5°S, 110°–140°E, (b) surface pressure for the zone 65°–50°S, 90°–110°E, and (c) zonal wind at 200 hPa for the zone 30°–20°S, 90°–120°E, for the 42-day period 5 Dec 1986 to 16 Jan 1987. Units are m s−1, hPa, and m s−1, respectively.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
1986–87 AMEX onset. (a) 850- and (b) 200-hPa wind fields for 0000 UTC 10 Jan 1987. Contour interval for 850-hPa wind is 10 m s−1 with light and dark shadings for winds greater than 10 and 20 m s−1, respectively. Contour interval for 200-hPa wind is 10 m s−1 with light and dark shadings for winds greater than 20 and 40 m s−1, respectively. The T and R labels indicate where the troughs have deepened and the ridges built.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Same as Fig. 3 but for 1992–93 TOGA COARE onset and the 42-day period 15 Dec 1992–26 Jan 1993.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
1992–93 TOGA COARE onset. Time series of (a) zonal wind for the zone 15°–5°S, 110°–140°E, (b) surface pressure for the zone 65°–50°S, 80°–100°E, and (c) zonal wind at 200 hPa for the zone 40°–30°S, 90°–120°E, for the 42-day period 15 Dec 1992 to 26 Jan 1993. Units are m s−1, hPa, and m s−1, respectively.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Same as Fig. 8 but for 1992–93 TOGA COARE onset for 0000 UTC 23 Jan 1993.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Fifteen-yr composite of 0000 UTC ERA-40 analyses, at 5 days prior to the mean monsoon onset for (a) 850-hPa wind, (b) 200-hPa wind, and (c) MSLP difference (composite − ERA-40 long-term mean). (d)–(f) As in (a)–(c) but for 2 days after monsoon onset. Contour interval for 850-hPa wind is 5 m s−1 with light and dark shadings for winds greater than 5 and 10 m s−1 respectively. Contour interval for 200-hPa wind is 10 m s−1 with light and dark shadings for winds greater than 20 and 40 m s−1 respectively. Contour interval for MSLP is 1 hPa. The T and R labels indicate where the troughs have deepened and the ridges built. (g) Skewed Hovmöller diagram of streamfunction anomalies (deviations from the zonal mean). Time is the ordinate and increases upward from 8 days prior to the mean onset day (O-8), to 3 days prior to onset (O-3). Spatial, southwest to northeast strips for each day cover the domain (60°S, 40°E) to (10°N, 170°E). The bottom left represents anomalies over the SWIO 8 days prior to onset. The top right represents anomalies over the equatorial Pacific 3 days prior to onset. The ellipses and enclosed T and R labels indicate the times and sectors when troughs (positive streamfunction anomalies) and ridges (negative streamfunction anomalies) are developing over specific regions. Units of streamfunction anomalies are m2 s−1 and the contour interval is 1.0 × 106 m2 s−1. Dashed contours indicate negative anomalies. Shading for negative contours is for values less than 12.0 × 106 m2 s−1. Shading for positive contours is for values greater than 5.0 × 106 m2 s−1.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
North–south cross sections (60°S–5°N) of (top) zonal wind (m s−1) and (bottom) potential temperature (K). (left) The long-term mean for December averaged between 90° and 100°E, (middle) a pre-onset period (21 Dec 1978) averaged between 80° and 90°E, and (right) the simulation. Contour intervals are 5 m s−1 for winds and 5 K for potential temperature.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Vector winds at (left) t = 0 h and (right) t = 168 h at (top) 200 and (bottom) 950 hPa for MT_STJ experiment. For left panels, units are m s−1 and the contour interval is 10 m s−1. Levels of shading indicate winds greater than 10 and 20 m s−1. For right panels, units are m s−1 and the contour interval is 20 m s−1. Levels of shading indicate winds greater than 20 and 40 m s−1. The heavy dashed line on the bottom panels indicates the location of the monsoon trough separating monsoon westerlies from trade easterlies.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Vector wind difference [(t = 96 h) − (t = 0 h)] at (top) 200 and (bottom) 950 hPa for (left) MT experiment and (right) MT_STJ experiment. Units are m s−1 and the longest vector is approximately 1 m s−1.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Same as Fig. 15 but for vector wind difference [(t = 168 h) − (t = 0 h)].
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
The [(t = 168 h) − (t = 0 h)] 200-hPa wind for MT_STJ experiment. The T and R labels mark the locations of troughs (or cyclonic circulations) that have deepened and ridges (or anticyclonic circulations) built. The solid line indicates group propagation from high latitudes into the Tropics. The domain has been split into Tropics and extratropics to enable the winds to be scaled to take account of the much weaker winds over the Tropics.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
(left) The MT_STJ experiment, at (top) 200 and (bottom) 950 hPa, showing horizontal divergence (DIVE) changes [(t = 168 h) − (t = 0 h)]. Units are 10−6 s−1 and the contour interval is 0.25 × 10−6 s−1. (right) Same as (left) but for relative vorticity (RVOR) changes. Units are 10−6 s−1 and the contour interval is 2.0 × 10−6 s−1.
Citation: Journal of the Atmospheric Sciences 64, 10; 10.1175/JAS4034.1
Table A1. List of model variables.
Table A2. List of parameters.
