• Ahijevych, D. A., , C. A. Davis, , R. E. Carbone, , and J. D. Tuttle, 2004: Initiation of precipitation episodes relative to elevated terrain. J. Atmos. Sci., 61 , 27632769.

    • Search Google Scholar
    • Export Citation
  • Arkin, P. A., , and B. N. Meisner, 1987: The relationship between large-scale convective rainfall and cold cloud over the Western Hemisphere during the 1982-84. Mon. Wea. Rev., 115 , 5174.

    • Search Google Scholar
    • Export Citation
  • Arkin, P. A., , and P. Xie, 1994: The global precipitation climatology project: First Algorithm Intercomparison Project. Bull. Amer. Meteor. Soc., 75 , 401419.

    • Search Google Scholar
    • Export Citation
  • Barnes, G. M., , and K. Sieckman, 1984: The environment of fast- and slow-moving tropical mesoscale convective cloud lines. Mon. Wea. Rev., 112 , 17821794.

    • Search Google Scholar
    • Export Citation
  • Carbone, R. E., , T. D. Keenan, , J. Hacker, , and J. Wilson, 2000: Tropical island convection in the absence of significant topography. Part I: Life cycle of diurnally forced convection. Mon. Wea. Rev., 128 , 34593480.

    • Search Google Scholar
    • Export Citation
  • Carbone, R. E., , J. D. Tuttle, , D. A. Ahijevych, , and S. B. Trier, 2002: Inferences of predictability associated with warm season precipitation episodes. J. Atmos. Sci., 59 , 20332056.

    • Search Google Scholar
    • Export Citation
  • Carbone, R. E., , J. D. Tuttle, , and D. A. Ahijevych, 2003: Semidiurnal rainfall variability. Bull. Amer. Meteor. Soc., 84 , 15111512.

  • Chen, S. S., , R. A. Houze Jr., , and B. E. Mapes, 1996: Multiscale variability of deep convection in relation to large-scale circulation in TOGA COARE. J. Atmos. Sci., 53 , 13801654.

    • Search Google Scholar
    • Export Citation
  • Davis, C. A., , K. W. Manning, , R. E. Carbone, , S. B. Trier, , and J. D. Tuttle, 2003: Coherence of warm-season continental rainfall in numerical weather prediction models. Mon. Wea. Rev., 131 , 26672679.

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

  • Keenan, T. D., , and R. E. Carbone, 1992: A preliminary morphology of precipitation systems in tropical northern Australia. Quart. J. Roy. Meteor. Soc., 118 , 283326.

    • Search Google Scholar
    • Export Citation
  • Laing, A. G., , and J. M. Fritsch, 1997: The global population of mesoscale convective complexes. Quart. J. Roy. Meteor. Soc., 123 , 389405.

    • Search Google Scholar
    • Export Citation
  • Laing, A. G., , R. E. Carbone, , V. Levizzani, , and J. D. Tuttle, 2008: The propagation and diurnal cycles of deep convection in northern tropical Africa. Quart. J. Roy. Meteor. Soc., in press.

    • Search Google Scholar
    • Export Citation
  • Levizzani, V., , R. Ginnetti, , A. G. Laing, , and R. E. Carbone, 2006: Warm season precipitation climatology: First European results. Advances in Geoscience, Vol. 7, Academic Press, 15–18.

    • Search Google Scholar
    • Export Citation
  • Machado, L. A. T., , W. B. Rossow, , R. L. Guedes, , and A. W. Walker, 1998: Life cycle variations of mesoscale convective systems over the Americas. Mon. Wea. Rev., 126 , 16301654.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., , and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28 , 702708.

    • Search Google Scholar
    • Export Citation
  • Mathon, V., , H. Laurent, , and T. Lebel, 2002: Mesoscale convective system rainfall in the Sahel. J. Appl. Meteor., 41 , 10811092.

  • Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44 , 2543.

  • McBride, J. L., , and N. Nicholls, 1983: Seasonal relationships between Australian rainfall and the Southern Oscillation. Mon. Wea. Rev., 111 , 19982004.

    • Search Google Scholar
    • Export Citation
  • Moncrieff, M. W., , and J. S. A. Green, 1972: The propagation and transfer properties of steady convective overturning in shear. Quart. J. Roy. Meteor. Soc., 98 , 336352.

    • Search Google Scholar
    • Export Citation
  • Nakazawa, T., 1986: Mean features of 30-60 day variations as inferred from 8-year OLR data. J. Meteor. Soc. Japan, 64 , 777786.

  • Olson, D. A., , N. W. Junker, , and B. Korty, 1995: An evaluation of three decades of quantitative precipitation forecasting at the Meteorological Operations Division of the National Meteorological Center. Wea. Forecasting, 10 , 498511.

    • Search Google Scholar
    • Export Citation
  • Ramage, C. S., 1968: Role of a tropical “maritime continent” in the atmospheric circulation. Mon. Wea. Rev., 96 , 365370.

  • Reeder, M. J., , and R. K. Smith, 1998: Mesoscale meteorology. Meteorology of the Southern Hemisphere, D. Vincent and D. J. Karoly, Eds., Amer. Meteor. Soc., 201–241.

    • Search Google Scholar
    • Export Citation
  • Rotunno, R., , J. B. Klemp, , and M. Wiseman, 1988: A theory for strong long-lived squall lines. J. Atmos. Sci., 45 , 463485.

  • Tuttle, J. D., , and R. E. Carbone, 2004: Coherent regeneration and the role of water vapor and shear in a long-lived convective episode. Mon. Wea. Rev., 132 , 192208.

    • Search Google Scholar
    • Export Citation
  • Wang, C-C., , G-J. Chen, , and R. E. Carbone, 2004: A climatology of warm season cloud patterns over East Asia based on GMS infrared temperature observations. Mon. Wea. Rev., 132 , 16061629.

    • Search Google Scholar
    • Export Citation
  • Wang, C-C., , G-J. Chen, , and R. E. Carbone, 2005: Variability of warm-season cloud episodes over East Asia based on GMS infrared temperature observations. Mon. Wea. Rev., 133 , 14781500.

    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., , R. E. Carbone, , J. Tuttle, , and T. D. Keenan, 2001: Tropical island convection in the absence of significant topography. Part II: Nowcasting storm evolution. Mon. Wea. Rev., 129 , 16371655.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Study domain showing main topographic features. Elevation scale (m) is shown to the right. Circles denote locations of Adelaide (ADL), Albany (ALB), Alice Springs (ASP), Darwin (DWN), Giles (GLS), and Wagga Wagga (WWG).

  • View in gallery

    Maximum surface elevation in each latitude domain. East–west subdivisions of 2.5-km resolution were employed to define maximum elevation.

  • View in gallery

    Impact of variations in Hovmöller thresholds (TBB < −50°, TBB < −35°, and TBB < −15°C) on streak characteristics in 30°–40°S domain for January 2001. Percentage of high cloud (PHC) or frequency of occurrence of TBB less than the indicated threshold obtained from a 0.2° × 0.2° latitude–longitude grid is shown (scale shown to right of panels). See text for description.

  • View in gallery

    Vertical cross section of zonal winds (m s−1) for the domain 120°–150°E averaged over three latitude bands for spring and summer. Source of data: NCEP–NCAR 40-Year Reanalysis.

  • View in gallery

    Selected sample satellite-derived Hovmöller diagrams for (a) midlatitude band 30°–40°S, (b) subtropical band 20°–30°S, and (c) tropical band 10°–20°S. Percentage of high cloud (PHC) is shown to the right with warmer colors indicating higher frequency. Cross sections of maximum surface elevation within each band are also shown below the top Hovmöller diagrams. Shaded ellipse areas denote major “forcing zones.” See text for discussion relating to shaded areas and features.

  • View in gallery

    (Continued)

  • View in gallery

    (Continued)

  • View in gallery

    Example association of streak characteristics in 30°–40°S band with environmental conditions. (a) Hovmöller diagram of PHC (TBB < −35°C) for the period 1–30 Nov 1998. The PHC scale is shown to the right of the panel. (b) Equivalent Hovmöller diagram of 30-kPa daily meridional wind anomaly taken for NCEP–NCAR 40-Year Reanalysis data. (c) Hovmöller diagram showing correlation field employed to define streaks. Correlation scale (×100) shown to right of panel. Streaks defined by the analysis are overlaid. Boxes define periods of streak events that are discussed in text and summarized in Table 4. The maximum elevation within the 30°–40°S latitude band is included for reference.

  • View in gallery

    Case study showing link between observed streaks, diurnal characteristics, and relation of streaks to satellite images. (a) Hovmöller diagram of PHC (TBB < −35°C) for the period 1–30 Nov 2000 for the 30°–40°S latitude band. Arrow highlights 12 Nov 2000. (b) Maximum elevation within the 30°–40°S latitude band. (c) Diurnal cycle of PHC composited over the period 9–23 Nov 2000 (30°–40°S latitude band). (d) Individual GMS satellite images showing diurnal evolution of cloudiness on 12 Nov 2000 for 30°–40°S latitude band. Box in (a) highlights period composited for diurnal cycle shown in (c). FR indicates diurnally forced convection originating over the Flinders Ranges. The subsequent motion is indicated by arrows. GD indicates convection enhanced over the Great Dividing Range. Note LT = UTC + 10 h.

  • View in gallery

    Diurnal frequency in Hovmöller diagram of PHC (TBB < −35°C) for all seasons (1996–2001) showing (a) spring (November and December) and (b) summer (January and February) for the midlatitude band 30°–40°S, subtropical band 20°–30°S, and tropical band 10°–20°S. The scale indicates the percentage of days on which the PHC is present at the given longitude–UTC hour coordinate. The maximum elevation within each latitude band is included for reference. See text for discussion of arrows.

  • View in gallery

    Diurnal cycle of PHC (TBB < −35°C) from 40° to 20°S in (a)–(c) the summer months of January and February and (d)–(f) the spring months of November and December. Convection over land near time of maximum in diurnal heating cycle is shown in (a) and (d). Evening mature and dissipating land-based convection with transition toward coastal and oceanic convection is shown in (b) and (e). Early morning phase with mature oceanic convection is shown in (c) and (f).

  • View in gallery

    Diurnal cycle of PHC (TBB < −55°C) from the equator to 20°S. (a) Mature convection over land resulting from diurnal heating, (b) dissipation and early transition toward lowlands and coastal oceanic convection, (c) mature oceanic convection associated with offshore flows, and (d) excitation of convection over Cape York Peninsula amid oceanic convection of nocturnal origin. Southward propagation of oceanic convection from New Guinea is coincident with the excitation of early day convection over Cape York. Westward propagation from Cape York (climatologically) assumes the shape of an organized “bow cloud.” Note the especially high amplitude of diurnal variation associated with convection over the Indonesian region (5°S, 110°–115°E). See text for explanation of suspected forcings.

  • View in gallery

    Meridional Hovmöller diagrams of PHC (TBB < −55°C) in the tropics for the five seasons (November–March 1996–2001 inclusive). (a) Western Austral–Asia; (b) eastern Austral–Asia. Coherent patterns suggestive of propagation and phase-locked convection are apparent and these are identified by dashed or dotted lines. See text for explanation.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 84 84 7
PDF Downloads 28 28 4

Propagation and Diurnal Evolution of Warm Season Cloudiness in the Australian and Maritime Continent Region

View More View Less
  • 1 Bureau of Meteorology Research Centre, Bureau of Meteorology, Melbourne, Australia
  • | 2 National Center for Atmospheric Research, Boulder, Colorado
© Get Permissions
Full access

Abstract

Warm season cold cloud-top climatology in the Austral–Indonesian region is examined for evidence of propagating modes of precipitation that originate from elevated heat sources and the diurnal heating cycle. Using satellite-inferred cloudiness from the period 1996–2001 as a proxy for rainfall, this coherent regeneration process and subsequent event propagation is found to consistently occur from the midlatitudes (30°–40°S) to the tropics (10°–20°S) in the Austral region.

Given favorable environmental shear at midlatitudes, long-lived eastward-propagating events are observed to occur regularly with a span and duration typically larger than observed by Carbone et al. The genesis of these events, while intermittent, is directly related to elevated heat sources and the diurnal cycle, similar to the United States. However, given the relatively flat terrain of Australia, an elevated heat source is often insufficient, thus increasing the relative influence of transient synoptic forcing.

In the tropics, the thermal forcing associated with elevated terrain found over the islands of the Maritime Continent and the land–sea interface is increasingly dominant on daily basis. While eastward- and westward-propagating events are found in the more varied environment of the monsoon regime, evidence for meridionally propagating modes is also found. In this manner, complex interactions occur that modify the location and timing of clouds that develop over neighboring oceanic and continental locations. The impact of convection initially linked to the New Guinea highlands and subsequently impacting the Java Sea region is particularly evident affecting the observed diurnal cycle.

The subtropics show characteristics intermediate between the above extremes. With the seasonal cycle, the spring environment favors eastward-propagating events but in summer there is an increasing frequency of diurnally forced quasi-stationary development over elevated terrain enhanced by favorable synoptic conditions. Overall the subtropical summer events have a shorter duration and span than their spring counterparts. The increased environmental steering winds and shear in spring are thought to be the primary reason.

Corresponding author address: T. Keenan, Bureau of Meteorology, P.O. Box 1289K, Melbourne 540000, Australia. Email: t.keenan@bom.gov.au

Abstract

Warm season cold cloud-top climatology in the Austral–Indonesian region is examined for evidence of propagating modes of precipitation that originate from elevated heat sources and the diurnal heating cycle. Using satellite-inferred cloudiness from the period 1996–2001 as a proxy for rainfall, this coherent regeneration process and subsequent event propagation is found to consistently occur from the midlatitudes (30°–40°S) to the tropics (10°–20°S) in the Austral region.

Given favorable environmental shear at midlatitudes, long-lived eastward-propagating events are observed to occur regularly with a span and duration typically larger than observed by Carbone et al. The genesis of these events, while intermittent, is directly related to elevated heat sources and the diurnal cycle, similar to the United States. However, given the relatively flat terrain of Australia, an elevated heat source is often insufficient, thus increasing the relative influence of transient synoptic forcing.

In the tropics, the thermal forcing associated with elevated terrain found over the islands of the Maritime Continent and the land–sea interface is increasingly dominant on daily basis. While eastward- and westward-propagating events are found in the more varied environment of the monsoon regime, evidence for meridionally propagating modes is also found. In this manner, complex interactions occur that modify the location and timing of clouds that develop over neighboring oceanic and continental locations. The impact of convection initially linked to the New Guinea highlands and subsequently impacting the Java Sea region is particularly evident affecting the observed diurnal cycle.

The subtropics show characteristics intermediate between the above extremes. With the seasonal cycle, the spring environment favors eastward-propagating events but in summer there is an increasing frequency of diurnally forced quasi-stationary development over elevated terrain enhanced by favorable synoptic conditions. Overall the subtropical summer events have a shorter duration and span than their spring counterparts. The increased environmental steering winds and shear in spring are thought to be the primary reason.

Corresponding author address: T. Keenan, Bureau of Meteorology, P.O. Box 1289K, Melbourne 540000, Australia. Email: t.keenan@bom.gov.au

1. Introduction

Coherent structure, in the form of the daily generation of warm season propagating rain events, often originating over elevated terrain, was evident in the U.S. radar-based climatology undertaken by Carbone et al. (2002). This finding was significant given the high frequency and long duration of rainfall events (or “streaks” as observed in Hovmöller diagrams). Over the course of propagation eastward of the mountains, “events” consist of a coherent succession of convective systems. The phase speed of these events routinely exceeds the phase speed of upper-tropospheric anomalies and the zonal speed of low–midtropospheric steering winds, thus evoking the term “propagation.”

Given the poor skill associated with the prediction of warm season precipitation (Olson et al. 1995), these findings provide a sense of increased predictability. This was reinforced by Davis et al. (2003) where it was shown that current weather prediction models employing cumulus parameterization schemes do not reproduce the natural space–time distribution of precipitation (especially the propagation characteristics and the link to diurnal forcing over elevated terrain). Tuttle and Carbone (2004) show that the coherent regeneration process depends upon favorable cold pool and low wind level shear interactions when synoptic forcing provides suitable environmental conditions.

Studies focusing on the characteristics of deep convection in other locations but with a specific regional nature have been undertaken by Chen et al. (1996), over the Indo-Pacific ocean warm pool, the Americas by Machado et al. (1998), and in the Sahel region of Africa by Mathon et al. (2002). In a global sense, Laing and Fritsch (1997) previously showed a link between population centers of mesoscale convective complexes, elevated terrain and environmental flow in East Asia, Africa, Europe, South America, and the Austral–Indonesian Region. Given the obvious connection, recent studies have commenced to examine the global characteristics of warm season precipitation over the continents (see summaries of Carbone et al. 2002; Ahijevych et al. 2004; Wang et al. 2005; Laing et al. 2008; Levizzani et al. 2006). These studies are seeking information on the possible universality of forcing, propagation, and the coherent regeneration of warm season precipitation.

This paper examines for the first time the characteristics of warm season precipitation in the Austral–Indonesian region using the approach undertaken by Carbone et al. (2002). Spanning the midlatitudes to equatorial regions, precipitation is related to such diverse forcing as monsoons, frontal zones, and subtropical influences under widely varying environmental flow. This occurs over a low-lying and generally arid continent with little terrain above 2 km. In the equatorial zones of the Maritime Continent there is significant terrain extending to 4–6-km height. With this in mind, the association of such diverse influences upon the characteristic forcing and propagation of warm season precipitation is examined.

We first describe the study domain, methodology, and the environmental conditions. We then provide examples of warm season precipitation in the midlatitudes, the subtropics, and the tropical regions. The observed nature of the diurnal cycle is assessed as a function of region, the importance of elevated sensible heating, and the capacity of atmospheric conditions to support organized deep convection. The similarities and differences in the characteristics of the warm season precipitation for this region are subsequently identified and then related to previous studies.

2. Study domain, data, and methodology

The domain extends from 0° to 40°S and 110° to 160°E incorporating mainland Australia and part of the Maritime Continent region (Ramage 1968) to the north of Australia. As shown in Fig. 1, the main topographic features include the following:

  • Arid to semi-arid conditions with a temperate conditions in the south and east and tropical conditions in the north.
  • The Great Dividing Range on the east side of the Australian continent, extending (with breaks) from 10° to 40°S with a maximum elevation of 2.23 km at Mount Kosciuszko near 35°S, 149°E.
  • Mostly low-plateau conditions with deserts but with elevated regions (<1-km elevation) located in central Australia (MacDonnell and Musgrave Ranges) and Western Australia (Hamersley Ranges and southwest Australia).
  • Significant topographical features associated with mountainous terrain on islands within the Maritime Continent (highest elevation is 5 km at Mount Puncak Jaya, Irian Jaya).

The limited vertical extent of the Australian topography is in contrast to the regions previously studied. Over the continental United States (Carbone et al. 2002), significant forcing is associated with the Rockies that extend from 3–4 km above sea level. In the East Asia region (Wang et al. 2004) the Tibetan Plateau (elevation extending to above 6 km) is associated with a clear diurnal signal in convection.

The primary study areas are within the Australia region extending from 110° to 160°E and are separated to include a midlatitude–subtropical domain (30°–40°S), a subtropical–tropical (20°–30°S) domain, and a tropical (10°–20°S) region. Cross sections of the maximum elevation within each domain are provided in Fig. 2. Complex terrain exists within each latitude band; however, the maxima in nearly all regions are below 1.5 km with the average terrain below 1 km.

To infer the characteristic of the warm season precipitation, satellite data are employed as a proxy for rainfall. A radar network does exist in the region, but it has significant gaps and was not suitable for the purposes of this study. Hence the climatology of warm season precipitation is derived using the Japanese Geostationary Meteorological Satellite (GMS) infrared (IR) data for spring–summer (November to March inclusive) from the period 1996–2001. Hourly 4-km resolution GMS blackbody temperature (TBB) data were interpolated to a 0.2° × 0. 2° latitude–longitude grid. Separate files were created denoting the percentage of high cloud (PHC) or frequency of occurrence of TBB < −15°, −25°, −35°, −45°, and −55°C within each latitude–longitude box. The November–December period is used to define spring, and January–February defines summer. A 5-season database has limitations especially given the known influence of the El Niño–Southern Oscillation on the observed interannual variability rainfall as described by McBride and Nicholls (1983). El Niño (La Niña) effects are present in the 5-season sample, but interannual variability is not the focus of attention in this study.

The subsequent method follows that of Carbone et al. (2002). A Hovmöller analysis of each of the above latitude bands is employed to study the space–time (coherency, longevity,1 span, or zonal distance) characteristics of the satellite-inferred cloudiness. Following this process, a 2D rectangular cosine weighting function is rotated to maximize correlation and stepped through all time–longitude positions (hourly intervals) at 1° increments. A fit to a contiguous event is assumed if the correlation >0.4, the span >100 km, the duration >3 h, and the speed >2.5 m s−1.

Various TBB thresholds have been used to infer the association with rainfall; for example, Arkin and Meisner (1987) employed a TBB threshold of <−38°C; Wang et al. (2004) employed an average brightness temperature, limited to temperatures below 0°C to emphasize deep, high clouds; Laing et al. (2008) employed a brightness threshold of <213 K. Arkin and Xie (1994) show the optimum choice of threshold is not always clear, so environmental conditions in the Australian region are considered briefly to examine potential sensitivities to this choice.

In the Australian region, summer monthly mean tropopause temperatures (see representative sounding station locations in Fig. 1) are −68°C at Albany (30°–40°S), −77°C at Giles (20°–30°S), and −83°C at Darwin (10°–20°S). Respective heights of the monthly mean summer isotherms of the various latitude bands range from 12 to 13 km for −55°C, 9 to 10 km for −35°C, and 6 to 8 km for −25°C. Based on the above, with opaque clouds an IR blackbody temperature of −55°C would reflect deep tropospheric convection or partially opaque deep cirrus near the tropopause, −35°C midtropospheric convection and/or partially opaque anvil cloud, and −15°C relatively shallow convection. The characterization of precipitation originating from shallow clouds is not represented adequately by any of these thresholds.

Examples of Hovmöller diagrams for the 30°–40°S band employing these TBB thresholds are shown in Fig. 3. Event numbers, span, and duration exhibit qualitative sensitivity to threshold, especially where TBB < −55°C is employed. Qualitative consistency is evident between the −35° and −15°C thresholds.

Examination of all spring and summer events for the entire satellite dataset showed a sensitivity to the PHC threshold, with the effect increasing poleward, especially for diagnosing the number of events. For example, changing from a TBB threshold of −35° to −55°C in the midlatitude band (30°–40°S), resulted in a 77% reduction in the number of cases, a 25% reduction in median span, and a 29% reduction in the median duration of events. In the tropics (10°–20°S), where deep convection dominates, the number of events decreased by only 14%, the span by 10%, and the duration by only 7% in going from the −35° to −55°C TBB threshold. Events in the subtropical band (20°–30°S) produced intermediate values. The diagnosed median streak phase speed was less sensitive to the threshold employed and was the most robust statistic (with respect to the TBB thresholds) exhibiting changes in the range <6%–8%.

This analysis suggests that the magnitude of the derived span and duration statistics within each band are influenced by the threshold, a limitation governed by the use of satellite data; that is, the analysis is being impacted by cirrus and anvil emanating from convective clouds and advected downstream, especially in the midlatitudes. Hence not all cases necessarily represent systems with rainfall reaching the ground throughout their duration, while still others may precipitate after an evolution to warmer cloud structures.

The original study undertaken by Carbone et al. (2002) employed radar data to infer precipitation characteristics. Given the lack of radar data in many regions, as in this case, increasing attention has been given to employing satellite data for these purposes (e.g., Wang. et al. 2004, 2005; Laing et al. 2008). Hence Tuttle et al. (2007) have examined four months of U.S. data with coincident satellite and radar to compare streak characteristics from both sources. This initial examination shows that radar-observed streak speeds are 4–5 m s−1 less than the satellite-inferred speeds; the satellite streak duration (median) is 2 h less than radar value, and the satellite-inferred span is slightly greater than the radar-inferred value (radar median 371 km; satellite median value 398 km). Nevertheless, the rainfall climatologies inferred by both datasets are very similar.

For the purposes of this analysis a PHC < −35°C in each 0.2° × 0.2° latitude–longitude grid was generally employed to infer precipitation and define streaks. This value is close to that (−38°C) employed by Arkin and Meisner (1987) but is different from the values used in the East Asia warm season study undertaken by Wang et al. (2004) and the African study undertaken by Laing et al. (2008). The overall statistics for number, span, duration, and median phase, based on this threshold, for each latitude band are summarized in Table 1 for eastward-propagating events.2 Given the environmental differences, some caution should be exercised in comparing these Australian region statistics to those obtained in previous studies, especially those based on radar data.

In a relative sense, streak span, duration, and phase speed increase poleward, although the number of eastward-moving events decreases poleward. These particular trends are independent of the threshold chosen. Over the continental United States in the 30°–48°N band, Carbone et al. (2002) observed 5406 radar streaks over 492 days exhibiting a median duration 4.5 h, with a median phase speed of 13.6 m s−1, and a median span of 350 km. Accounting for the mean radar–satellite differences, in the 30°–40°S domain (closest to U.S. band) the Australian streaks have about the same median phase speed but the duration and span are longer by about 5 h and 200 km, respectively.

3. Environmental conditions

There are significant differences in the environment for the three Australian region latitudes bands under consideration. To characterize these differences the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) 40-Year Reanalysis Project data (Kalnay et al. 1996) have been employed to provide east–west vertical cross sections of the zonal wind averaged over each latitude band domain. These cross sections, shown in Fig. 4, are based on the 40-yr dataset and are produced for spring (November–December) and summer (January–February). Comparison of cross section analyses based on the 40-yr dataset with those derived only using the 5-season satellite dataset showed no differences of significance in the context of this paper.

a. Midlatitude (30°–40°S)

Deep westerly flow and shear is evident across the entire domain and during both spring and summer. At 30 kPa, the mean flow has a relatively uniform west to east gradient with spring (summer) winds near 125°E of 24 (22) m s−1 decreasing to 20 (16) m s−1 near 145°E. At 50 kPa the mean zonal flow exhibits a smaller east–west gradient with spring (summer) winds decreasing from 13 (13) m s−1 near 125°E to 12 (8) m s−1. Shear values (Table 1) derived from the representative stations (see Fig. 1 for locations) for this domain show relatively large and consistent deep westerly shear in the range 2–2.6 × 10−3 s−1 in the middle of the domain during spring decreasing to about 1.6–1.9 × 10−3 s−1 in summer. All shear values are reduced further on the east coast (cf. Adelaide with Wagga Wagga where there is a 14%–18% decrease).

The role of shear and convective available potential energy has been well recognized in terms of its importance to organize and determine the lifetime of convection (Moncrieff and Green 1972). While deep shear is important, an essential ingredient in determining the longevity of convection is the relation of the environmental low-level shear to the cold pool circulation. Rotunno et al. (1988) show an optimal state is characterized by a balance between the two circulations enabling cells to grow and decay along a line perpendicular to the low-level shear. Low-level shear in the range 1.5– 3 × 10−3 s−1 (Table 2) is certainly in the range considered appropriate to provide balance for the circulation of typical cold pools.

b. Subtropical (20°–30°S)

In the subtropics, conditions change considerably through the spring to summer transition. During spring the environment is similar to that observed in the 30°–40°S band with deep westerlies (20 m s−1 at 30 kPa) and westerly shear, albeit with some shallow easterly flow evident below 85 kPa. Midatmospheric shear values (to 50 kPa) are reduced by about 30%–40% compared with the 30°–40°S band, although the strong spring westerlies mean only a slight reduction in the deep tropospheric shear (9% using 30 kPa to surface shear). In summer, the easterly flow extends to near 50 kPa and the upper-level westerlies (30 kPa) are reduced to 20% of the spring values. Shear values are correspondingly weaker (25%–50% of the spring values) and less than 10−3 s−1. These subtropical summer environmental conditions are less conducive to supporting long-lived propagating convection.

c. Tropical (10°–20°S)

In the tropical regime the spring conditions show a transition from low-level easterly flow to upper-level westerly flow with a jetlike maximum in the easterlies near 70 kPa (maximum easterly flow is about 4 m s−1 near 70 kPa). This strong low-level easterly shear tends to dominate the deep tropospheric shear values even though there is actually a regime of westerly shear above 70 kPa. These particular conditions and their relation to the organization of convection have been discussed in some detail by Keenan and Carbone (1992). Organized convection with motion characteristically dependent on the jet structure is evident. These conditions are similar to those observed with Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) squall lines (Barnes and Sieckman 1984).

The mean tropical summer environment shows a dramatic change from those observed during spring. Low-level westerly flow replaces the spring easterlies and upper-level easterlies replace the spring upper-level westerlies. These conditions reflect the development of the Australian monsoon. As discussed by Keenan and Carbone (1992), this summer period is in fact made up of periods of active monsoonal flow and break periods (periods when a return to springlike conditions occur). The mean conditions in Fig. 4 reflect a residue of a two opposite zonal wind environments and hence are not an accurate depiction of conditions likely encountered by any one event.

4. Characteristics of the warm season convection

Characteristics of both summer and spring Hovmöller diagrams in each latitude band based on three representative cases are presented in Fig. 5. These months have been selected from the five available seasons to show typical features, with at least one month included to provide more atypical behavior.

a. Midlatitude (30°–40°S)

The midlatitudes are dominated by long-lived slowly eastward-moving “envelopes” of convection that can span the entire domain and be of maritime origin. As evident in Fig. 5a, these long-lived envelopes, for example, E1, are most evident during spring, when they occur 4–8 times per month on average. During summer, the envelope frequency is approximately half that of the spring frequency. Embedded streak activity is evident within these overall “forcing zones” lasting days and extending up to 1000–2000 km in length as observed by Carbone et al. (2002). No significant interseasonal variation in the overall streak statistics was evident.

The “forcing zones” within which the streak events are clustered typically have phase speeds near 3–6 m s−1 (based on a diagnosis of the five seasons), which is comparable to the U.S. value (5–6 m s−1) found by Carbone et al. (2002). However, the envelopes themselves can be manifested in clusters, for example, the periods 2–16 November 1997 (indicated by CEA) and 18–26 February 1997 (indicated by CEB) shown in Fig. 5a. The envelopes tend to be associated with transient westerly disturbances, and the clustering of envelopes results is typically associated with a series of more transient upper-level troughs.

Significant diurnal modulation of the embedded streaks occurs as they pass through the domain, especially in regions of elevated topography and near the time of maximum solar heating. Events A and C in Fig. 5a, are initiated over the elevated terrain of southwest Australia. Event B, an event of oceanic origin, is enhanced as it moves over southwest Australia. Events D, E, F, and G all show enhancement as they reach the Flinders Ranges in the central part of the 30°–40°S domain. Over the eastern-most part of the domain, in the region of the Great Dividing Range, propagating events H and I initiated.

Event J, through the period 9–19 November 2000, is somewhat atypical of the cloudiness observed in the 30°–40°S domain. Here the cloudiness exhibits a quasi-stationary pattern made up of independent diurnally forced events generated at the same location on successive days. The source region coincides with the Great Dividing Range.

The characteristics of Australian propagating events are summarized in Table 1. Typically, events at 30°–40°S have a longer span, duration, and larger propagation speed than observed in China and the United States. For example, the average span, duration, and zonal propagation speed observed by Wang et al. (2004) in China (20°–40°N) were 375.5 km, 7.9 h, and 13.2 m s−1, respectively. The Australian 30°–40°S latitude band produced respective mean3 values of 805 km, 11.6 h, and 21.4 m s−1. Differences in the latitude band and processing may account for the span and duration statistics as discussed earlier. However, using the more robust measure of propagation speed, it is clear that Australian region 30°–40°S streaks are observed to propagate faster than those studied over China and slightly faster than those observed over the United States. A typical median phase speed for the United States in the 30°–48°N latitude band using radar streaks studied by Carbone et al. (2002) was 13 m s−1. Based on the results of Tuttle et al. (2007), this equates to approximately 17 m s−1 for satellite data.

Spring events typically have larger span and duration but similar propagation speeds when compared with summer events (see Table 1). Spring (summer) event frequency is almost identical with occurrence rates of 2.5 (2.3) events per day respectively, that is, small interseasonal variability. The largest 10% value seasonal streak span (duration) is 1700 km (25.5 h). This span is equivalent to 37% of the entire domain, and these events occur on average once every 4.3 days. Ten percentile values for spring (summer) span, duration, and propagation speed are 1800 (1675) km, 24.5 (23) h, and 35.8 (35) m s−1, respectively.

As shown in Table 2, wind shear values are typically larger during spring (by a factor of 40%) than summer. The springtime environment is obviously more favorable for sustained long-lived propagating convection assuming everything else is equal.

The period 4–11 November 1997 in Fig. 5a (denoted by MCV) has the characteristic signature (Carbone et al. 2002) of a system dominated by a mesoscale convective vortex. The streak phase speed in this period was only 2 m s−1.

In summary, the 30°–40°S band showed clear evidence for the intermittent and aperiodic occurrence of propagating events. Events exhibited little interseasonal variations. The streaks do not necessarily originate in regions of elevated terrain, suggesting thermal forcing alone is often insufficient and topography is not nearly as influential as found in the United States and China. Favorable synoptic conditions associated with synoptic-scale troughs seem to be important in the Australian region. Nevertheless, topographic and thermal forcing plays a significant role in the generation and modification of convection in this band. As discussed, comparison with other studies is confounded by differences introduced by the use of radar and different thresholds in the satellite data.

b. Subtropical (20°–30°S)

This region is characterized by considerable inter- and intraseasonal variability. Spring has many of the characteristics observed in the 30°–40°S band. For example, CEC and CED, shown in Fig. 5b, are slowly propagating envelopes of convective bands (moving eastward at 3.5 and 5.8 m s−1, respectively) with embedded streaks almost identical in nature to those observed in spring and summer in the 30°–40°S band.

Median span, duration, and phase speed for easterly propagating events during spring, (see Table 1 for the 20°–30°S band) are somewhat less but comparable to the respective figures associated with spring in the 30°–40°S band. The frequency of long events has decreased compared to the 30°–40°S band, but the overall frequency of occurrence of spring events is a 30% increase over that observed in the 30°–40°S band.

Beginning in later spring and especially in summer, there is an increasing frequency of diurnally forced quasi-stationary events (10 days is typical) of the types indicated by CEE, CEG, and CEF. These events occur over the elevated terrain at the time of maximum solar heating. The Hamersley Ranges on the west coast, the MacDonnell/Musgrave Ranges in central Australia, and the Great Dividing Range on the east coast are source regions for such streaks and or are locations where significant enhancement of events occurs associated with the daily heating cycle.

Not all regions of elevated terrain repeatedly develop cloudiness. In Fig. 5b, when CEE is active over the Great Dividing Range, there is no significant activity apparent over the central and western elevated regions of the MacDonnell/Musgrave and Hamersley Ranges. In fact, the envelope of enhanced connective activity can slowly migrate or switch rapidly from one region to another. Arrows indicate an example of this rapid transition in Fig. 5b with CEG switching from the west coast to the east coast over several days. The result is the development of suppressed activity over the western region—the zone previously active for at least 10 days and enhanced convection and streak generation over the Great Dividing Range. This case will be examined in more detail later.

During summer, as shown in Table 1, the frequency of events is almost identical (3 per day) to that observed in the spring, although the summer span and duration values represent a 10%–15% decrease on the spring values. Summer tends to have slightly fewer slow events and more fast events compared to spring. This is a reflection of the increased frequency of diurnally forced events, many of which are either nonpropagating or more rapidly propagating for very short periods.

As discussed in section 2, the environment changes considerably from the 30°–40°S to the 20°–30°S band, especially during the transition from spring to summer. As indicated in Table 2, midtropospheric (50 kPa) to surface shear in spring (summer) decreases by about 35% (75%) from the 30°–40°S to the 20°–30°S band. The average summer shear value in the 20°–30°S band is ∼0.4 × 10−3 s−1, a value less likely to support long-lived organized propagating systems. This is consistent with the observed preponderance of in situ thermal forcing found during summer in this latitude band.

Periods of regular streak occurrence are almost always coincident with a deep northerly component of flow (to 30 kPa) associated with the east side of a trough. Conversely, deep southerly flow suppresses almost all activity and periods of regular activity are suppressed when upper-tropospheric easterly flow is strong. Qualitatively consistent with findings in the United States, propagating modes, once initiated, tend to persist through energetically unfavorable meridians, including deep southerly flow associated with troughs.

Again, shifts in the envelope of convection are coincident with abrupt trough movement, for example, from west to east, and when such shifts occur activity locks on to narrow meridional regions that may be tied to topography (e.g., Hamersley, southwest Australia or the Great Dividing Range) as shown in Fig. 5b with the west to east coast transition of CEG.

An alternative mode or organization in the form a slow westward-propagating envelope (CEH in Fig. 5b) is evident in summer in the 20°–30°S latitude band. Another example originating over the Great Dividing Range can be seen starting near the termination of CEH. These westward-propagating envelopes typically move slowly; for example, CEH moves westward at about 2.2 m s−1. Diurnally forced cloud developments are apparent within this westward-moving envelope over a number of successive days. The cloudiness may persist but is typically reinforced daily at essentially the same location coincident with elevated terrain, that is, over MacDonnell and Musgrave Ranges in central Australia in the case of CEH. The convection is either quasi-stationary or manifested as westward-propagating streaks.

Importantly, the transition to the subtropics has resulted in an increased importance of diurnal forcing, especially during the summer. This diurnally forced convection tends to be phase locked to the terrain, and less likely to propagate from the source region. Significant periods of suppressed activity are also evident. However there is considerable spatial variability. Often when one region is suppressed the enhanced convection is seen to occur on the opposite side of the continent. This implies that thermal forcing from an elevated heat source is by itself less likely to generate the convection when compared to other continents. Synoptic forcing is also an important ingredient and certainly contributes significantly to the observed variations in cloudiness.

Thus the subtropics represent a transition region. Early spring has the characteristics of the midlatitudes. However, during late spring and summer diurnal forcing becomes more evident and a tendency develops for geographically fixed diurnal convection associated with the elevated terrain. Events are repeatedly generated from these zones on a diurnal basis as shear and CAPE permit. Large-scale organization tends to reside in fixed areas for long periods of time with little or no envelope propagation. The elevated terrain over the east of Australia can be an active source for long periods with the rest of Australia inactive. Hence the subtropical summer shows the increasing importance of elevated thermal sources but with less likelihood of long-lived streak propagation.

c. Tropical (10°–20°S)

The tropical belt is characterized by a further increase in the importance of thermal induced forcing with a significant increase in total cloudiness in summer. The spring period event denoted by CEI in Fig. 5c is an example of such geographically fixed and diurnally forced convection repeated for at least 15 days. The source regions are the Cape York Peninsula and the Arnham Escarpment. Propagating events are generated but these typically dissipate within 24 h. Other examples of this source region are also evident during summer in the more suppressed periods.

Events are organized within 1) slow eastward-propagating disturbances, some of which have characteristics often associated with the Madden–Julian oscillation (MJO), originally described by Madden and Julian (1971); 2) faster eastward-moving disturbances possibly associated with Kelvin waves (see Nakazawa 1986); and 3) westward-propagating envelopes with traits of internal equatorial Rossby waves (Matsuno 1966). CEK and CEJ are examples of slow westward-propagating modes with speeds of −4 and −5 m s−1, respectively. CEL is an example of an envelope propagating eastward at ∼7 m s−1. CEL and CEK actually overlap, indicating the presence of both modes simultaneously. The transition from active to break periods, which is evident in the cloudiness, may also take up to 20–30 days, implying an MJO association.

Within these overall envelopes, both eastward- and westward-moving events are apparent. Examples of westward-moving events4 with phase speeds of 20 and 14 m s−1 are shown within CEM. Characteristics of the streaks are summarized in Table 3. Over the five seasons examined there were, on average, 3.5 (2.9) westward- (eastward-) moving events per day for the 10°–20°S domain. The frequency of westward- (eastward-) moving events varies from 2.6 (2.8) per day in spring to 4.2 (3.2) per day in summer; that is, the westward-moving events are more common in the summer. This is almost twice the number observed in 20°–30°S. Note the eastward-moving events generally have a much shorter span (400 km for the westward-moving streaks versus 220 km for eastward-moving systems) and move more slowly (2.2 m s−1).

The environmental conditions summarized in Table 2 imply relatively strong easterly shear to 700 mb and weak shear above. These conditions would imply support for westward-moving organized convection. The environmental conditions in this region and the relation to organized convection have been discussed by Keenan and Carbone (1992). It is important to realize that the region is impacted by a summer monsoonal flow and that the mean sounding employed in this region (especially for summer) is the result of averaging two opposite regimes: one dominated by deep easterly flow (as per spring) with westerly shear above, and another dominated by low-level westerly flow with easterlies aloft. Support for both easterly and westerly moving organized convections comes in these two regimes.

The fact that the eastward-moving systems have a slower phase speed and shorter span may be related to cold pool production in the more saturated environment of the deep westerly monsoon as discussed by Keenan and Carbone (1992). Under these near-saturated conditions, cold pool production is generally weaker and as a consequence maybe less able to sustain a long-lived system. In addition, the monsoon is generally a more disturbed environment with less shear, all conditions considered less conducive to strong propagation and long duration.

5. Sample streak events

The characteristics of typical midlatitude (30°–40°S) propagating events (November 1998) and their relation to the environmental flow are shown in Fig. 6. Two clusters of streaks (labeled A1 and A2) are shown with individual events labeled on the correlation analysis (Fig. 6c). The associated 30-kPa meridional wind anomaly from the NCEP reanalysis data in Hovmöller form is shown in Fig. 6b, and individual streak attributes are summarized in Table 4. Generation is often linked to significant topography but not always (e.g., event A is of oceanic origin). Note the variability in lifetime and the fact that some streaks extend over 2000 km and persist in excess of 36 h.

As found by Carbone et al. (2002), there is strong association between the position of upper-level troughs and the envelope of streak activity. In Australia, these events occur within the poleward (northerly) flow region of the 30-kPa troughs. In both cases the speed of the upper-level troughs (measured either by the speed of the zero meridional velocity or the maximum in northerly winds at 30 kPa) is an approximate match to that of the envelope of streaks (envelope of A1 moves 2–3 m s−1 faster than the associated trough; envelope of A2 moves 2–4 m s−1 faster than trough). However, the individual streaks have phase speeds much faster than the troughs (approximately 3–25 m s−1 for streaks within A1 and approximately 5–11 m s−1 for streaks within A2) with the distribution biased toward propagation speeds almost double that of the troughs.

A case of repetitive diurnal development of events over the Great Dividing Range of eastern Australia is shown in Fig. 7. During the period 8–19 November 2000 there is development of cloudiness in phase with the maximum of the diurnal heating cycle over the area 140°–150°E (see box in Fig. 7a) This region includes the elevated areas of the Flinders Ranges and the Great Dividing Range. Events are of variable duration from this source region with about half of the cases showing evidence of eastward propagation across the Tasman Sea. The composite PHC (Fig. 7c) shows a clear diurnal cycle with maximum in cloudiness near 0800 UTC and in situ dissipation occurring by 1400 UTC. There is evidence of some propagation from this source region to provide an early morning maximum (1800 UTC) in cloudiness over the Tasman Sea. A weak diurnal cycle in cloudiness is also evident over the elevated regions of western Australia.

On 12 November 2000, typical Hovmöller development was observed. Individual satellite images for this day (Fig. 7d) show cloudiness over the Great Dividing Range (GD in Fig. 7d) and coastal regions of eastern Australia with enhanced deep convection at 0430 UTC in phase with the maximum of the diurnal heating cycle. The deepest convection gradually propagates off the east Australian coast to form a nocturnal maximum in convection in the Tasman Sea by 1630 UTC. Similarly, the Flinders Ranges (FR in Fig. 7d) are another source of deep convective activity in phase with the heating cycle. The dotted arrows in Fig. 7d show the subsequent eastward propagation of convection associated with this latter source as it takes on the classic bow shape at 0830 UTC. In fact, this second generation convection (originally from a region of elevated heating) interacts with the remnants of convection that had developed in the center of the domain at 0430 UTC. This latter rainband was associated with synoptic forcing (at the center of an extratropical low-pressure system) and was not linked directly to initiation over elevated terrain. Through this complex interaction enhanced nocturnal convection (at 1630 UTC) developed over northern Victoria.

This case clearly demonstrates that diurnal forcing associated with elevated terrain and the land–sea contrast is important. However, the existence of other factors is clearly important in the Australian domain (in this case synoptic forcing that produces rainbands within an extratropical low-pressure system).

6. Diurnal cycle

The satellite-inferred diurnal cycle in cloudiness derived from all data (1996–2001) for the three latitude bands and for spring and summer is shown in Fig. 8. In this diagram the diurnal cycle is repeated for clarity. A significant impact is evident across all bands resulting from the interaction of elevated topography and the diurnal heating cycle. Features include the following:

  1. 30°–40°S band
    • (i) A maximum in diurnal forcing (following the solar heating cycle) between 0600 and 0800 UTC in spring and summer.
    • (ii) The Flinders Ranges and the Great Dividing Range are source regions for convection and subsequent eastward propagation (∼8.5 m s−1) of convection across eastern Australia and the Tasman Sea. The eastward propagation is more significant during spring. The southwest Australian Plateau is a weaker source region for convection and the associated eastward propagation less frequent.
    • (iii) A relative minimum exists in observable convection activity over the low lying arid zone between southwest Australia and the Great Dividing Range.
    • (iv) Convection is more active during spring.
  2. 20°–30°S band
    • (i) A marked increase in the frequency of convection occurs (compared with the 30°–40°S band).
    • (ii) A clear diurnal maximum in the forcing is evident across the continent between 0600–1200 UTC, with a morning minimum between 0000–0300UTC. The Great Dividing Range and the elevated Hamersley and MacDonnell Ranges are favored source regions providing a dipole of enhanced convection across the continent.
    • (iii) In spring, the Great Dividing Range is the main Austral source region, but in summer the elevated regions in the western Australian region (see later for coastal effects) become the most frequent and persistent source of convection.
    • (iv) Compared with the 30°–40°S band, convection has a lower frequency for coherent propagation out of the source regions.
  3. 10°–20°S band
    • (i) Increased frequency of convective activity compared with the higher latitude bands with a distinct summer increase in the frequency of convection.
    • (ii) A strong diurnal maximum across the continent (following the solar heating cycle) during the period 0600–1200 UTC coupled with a distinct morning minimum between 0000 and 0300 UTC. Considerable overnight persistence of resulting cloudiness is evident.
    • (iii) Maximum in the diurnal forcing coincident with elevated topography (Cape York Peninsula, Arnham Escarpment, and Kimberly Ranges).
    • (iv) Evidence of westward propagation (approximately 6 m s−1) in spring and summer of the frequency maximum from the Cape York Peninsula across the Gulf of Carpentaria (see arrow in Fig. 8b) suggesting a triggering role in the subsequent generation of convection in the next diurnal cycle over Arnham Land. Summer westward propagation from the Arnham Land Escarpment (see arrow in Fig. 8b) implying a tenuous link to the subsequent diurnal maximum over the Kimberly region.

Overall the Austral region exhibits increasing frequency of convective activity and streak frequency equatorward, but with streaks being longer lived, faster, and having larger spans the more poleward they occur. Equatorward there is an increasing trend for nonpropagating, diurnally forced convection linked to topography with increasing numbers of westward-propagating events.

To aid in the geographic interpretation of these signals the diurnal cycle is now examined at representative times over the subtropical and tropical zones (extending to the equator) of the Austral region.

a. Subtropical zone (20°–40°S)

A clear diurnal signal linked to topographic features is evident in the midlatitudes and subtropical regions shown in Fig. 9. The frequency of overall convective activity clearly increases in the summer and equatorward except where topographic features become significant. At 1600 LT, near the time of maximum solar heating, both the spring and summer show clear impact of the Great Dividing Range and the Artherton Tableland on the development of convection. The subsequent propagation of this cloudiness off the east coast of Australia enhances the frequency of downstream convection in the Tasman Sea. A night and early morning minimum of convection is reflected at 0900 LT across the continent except in the vicinity of the Tasman Sea.

The impact of the elevated heating associated with the Great Dividing Range on the frequency of occurrence of convection at the time of maximum solar heating is greater during spring as noted previously. Also evident is the strong influence of the land–sea interface on both the east and west coasts in the northern subtropical regions. Near Brisbane (see arrows in Fig. 9) a local maximum in convection occurs right at 1600 LT on the coastal interface consistent with this forcing mechanism. On the west coast, and especially summer, the land–sea interface and complex terrain is obviously important in generating local convection (see arrows on west coast in Fig. 9) at 1600 and 2000 LT. Spring and summer sea breezes are obviously important in the generation of convection. In fact, this process appears to be a major source of the increasing frequency of summer convection or the dipole east–west coast shift noted previously; that is, both sea breezes and elevated topography are playing a role in developing the summer maximum in west coast convection.

b. Tropical–equatorial zone (0°–20°S)

The so-called Maritime Continent is among the most active regions for deep moist convection on earth. This region contains tropical northern Australia, New Guinea, and a host of Indonesian islands westward of New Guinea, extending to and beyond the equator. The topography is complex, with mountains up to 5 km MSL (e.g., New Guinea), substantial terrain on the Cape York Peninsula, several escarpments, and smaller islands, both mountainous and flat (Fig. 1). Warm seas separate the landmasses with sea surface temperatures (SSTs) often in excess of 32°C in the wet season. (Because the complexities of the region are formidable, we encourage readers to view an animation of Austral–Asian PHC in association with the following discussion available at http://www.bom.gov.au/bmrc/wefor/wfresact.htm.)

The diurnal cycle of cold cloud tops (PHC) in a geographic framework is illustrated in Fig. 10. At 1900 LT (Fig. 10a), mature convection over land is forced by cumulative shortwave radiation. At this time, northern Australia has a high incidence of convectively generated cirrus over the Cape York Peninsula (140°E), the Northern Territory (130°E), and the Kimberly Region (125°E). Cape York convection, becaue of an early excitation time, is more mature than convection elsewhere and has begun its movement westward toward the Gulf of Carpentaria. Over New Guinea (5°S, 140°E) convection develops equally over south and north slopes of a zonally oriented mountain range. The Indonesian region (5°S, 115°E) exhibits strong activity near opposing shores surrounding the Java Sea. Not evident in Fig. 10a is the activity associated with smaller islands (∼100 km) in the Maritime Continent region. These exhibit a peak of convective development circa 1500 LT owing to the rapid organization of convection from interactions among sea breezes and moist convection in close proximity (e.g., Carbone et al. 2000; Wilson et al. 2001).

By local midnight (Fig. 10b), land-based convection enters a dissipation phase. Seaward movement is evident from most major landmasses. In New Guinea, the earliest stage of propagating of convection is evident, moving from the mountainsides to the sea, over the north- and south-facing slopes. Convection of continental origin moves over the Gulf of Carpentaria in a “bow-shaped” cloud mass, west of the Cape York Peninsula. This bow-shaped cloud pattern is somewhat reminiscent of continental “bow echoes” as defined by radar echoes at midlatitudes. However, the scale of this PHC pattern exceeds 1000 km, nearly 10 times the scale of a typical bow echo as seen by radar.

Figure 10c shows the complete dissipation of land-based convection at 0700 LT. Oceanic convection is well developed between adjacent lands masses, with enhanced PHC located mostly near concave coastlines. It is plausible to assume the convergent maritime atmosphere is the result of offshore flow driven jointly by negatively buoyant outflows from antecedent land-based convection, subsequently amplified by radiative cooling and drainage from elevated terrain. Seas that are surrounded by the larger landmasses systematically exhibit widespread convection and/or regions of enhanced PHC.

The diurnal cycle enters the shortwave heating phase (Fig. 10d) with the early excitation of land-based convection over Cape York. Cape York is a favored location for several reasons, including elevated terrain, the strength and effectiveness of sea breezes on a narrow peninsula, and forcing afforded by the timely arrival of convergent outflow from New Guinea. There is a substantial phase delay in the convective life cycle, as observed over the Indonesian–Java Sea region, when compared with Cape York, due in part to the next-day influence of dissipated New Guinean convection on the Australian continent.

To further illustrate several of the previous points, Fig. 11 exhibits several PHC meridional signals in reduced dimension. The western region of Austral–Asia (Fig. 11a) exhibits a strong maximum associated with the Northern Territory (15°S, 130°E), which includes modest poleward propagation inland from convection of sea-breeze origin. The dotted line (see A) denotes a tenuous continuity in PHC extending from a nocturnal oceanic maximum (at 2°S) to the Northern Territory diurnal maximum some 18 h later. This could represent a coherent regeneration of convection; however, it is problematic and would require a nondissipative mechanism capable of propagating ∼1300 km at ∼20 m s−1.

Recall from the discussion associated with Figs. 5 and 6 that Cape York convection triggers early and propagates slowly westward into the Gulf of Carpentaria. A major factor in the early excitation of Cape York convection is apparent from the meridional depiction of PHC (Fig. 11b). The double maximum of convection near 5°S is associated with elevated terrain in New Guinea. The dissipation of this convection leads to the onset of propagation and regeneration circa 0000 LT. Sequential propagation from the New Guinea highlands impacting the surrounding oceans is indicated by the dashed arrows which radiate poleward (B) and equatorward (C), with the former case extending to 15°S by 1600 LT the following day. This signal subtends the northern third of the Cape York Peninsula and thus constitutes a coherent and phase-locked regeneration of convection spanning two major landmasses. The rate of propagation is 13–14 m s−1, consistent with several dynamical mechanisms, including convectively generated gravity currents and some trapped gravity waves.

Solid lines in Fig. 11b are also suggestive of propagating convective regeneration. In both these cases the sense is toward New Guinea, equatorward from 11°S (see D) and poleward (E) from 2°S. Neither of these signals is related to interaction with the Australian continent. In the first instance the propagation is a sequential development of sea-breeze convection along the coast of southeast New Guinea (11°S, 150°E), from a very narrow peninsula to a progressively wider and more elevated one. In the second instance dissipation of land-based convection over northwest New Guinea (2°S, 132°E) leads to poleward propagation and coherent regeneration of modest amplitude over adjacent seas.

7. Conclusions

Examination of the characteristics of warm season precipitation in Australia and the Maritime Continent region has shown that the “long episodes” of convective streaks originally observed by Carbone et al. (2002) occur at all latitudes, albeit with decreasing frequency poleward. The life cycles of these streaks are linked to diurnally forced heating associated with orography and land–sea contrasts coupled with favorable synoptic conditions and a coherent regeneration process. These streaks do provide episodes exhibiting propagation relative to mean flow as has been observed over North America and East Asia. This is evident in midlatitude regimes dominated by deep westerly shear, tropical regimes dominated by easterly flow, and subtropical regions with environmental conditions intermediate between the two.

The diurnal forcing is evident at all latitudes, although at low latitudes the coherent regeneration becomes increasingly related to both elevated topographic and the land–sea interface. Lower latitudes tend to produce an increased frequency of in situ developments with less tendency to propagate. However, the tropical monsoonal environment does produce almost equal numbers of easterly and westerly propagating events. The westward-moving events are typically faster and of longer span than the eastward-moving streaks. The saturated environment present during the summer westerly monsoon phase may provide less favorable conditions for strong cold pool development. In addition, less persistent shear occurs and this is also necessary to reinforce the coherent regeneration process. The importance of diurnal forcing decreases its influence poleward.

In the midlatitude and subtropical regions northerly meridional flow associated with deep tropospheric troughs markedly increases the likelihood of “long” events. Development of lower-tropospheric shear of order 1.5 × 10−3 s−1 appears to be a necessary condition to maintain long-lived propagating events in both the westerly and easterly flow regimes. Summer events, especially in the subtropics, tend to have less span and duration compared to their spring counterparts. The low frequency of adequate environmental shear and steering winds in summer is a principal discriminating condition that emerges from our analysis. Unfavorable synoptic conditions, for example, the absence of poleward flow, can severely limit the potential for long-lived events. Hence, transient synoptic-scale forcing clearly plays a more important role in the Australian region compared to conditions in North America and East Asia.

The triggering of events by elevated heat sources is evident, but events are less frequent than in Asia and North America, presumably because Australia is relatively flat. Nevertheless, the east coast of Australia, with the elevated terrain of the Great Dividing Range, does provide a major source of events that result in an enhanced frequency of convection downstream in the Tasman Sea. Australia, having its largest mountains on the leeward side of the continent, is less affected by this regime of excitation, organization, and subsequent propagation over land.

Extremely complex interactions are evident especially in the tropics resulting from the diurnal heating cycle, land–sea contrasts, orography, and the regeneration process of convection. Local climate is strongly influenced by these factors especially surrounding Indonesian Islands and northern Australia. Among this class of events are the propagating systems across the Gulf of Carpentaria, originating from the Cape York Peninsula [also known as the north Australian cloud line, described by Reeder and Smith (1998)].

Acknowledgments

This research was supported in part by the National Science Foundation. Thanks to John Tuttle of NCAR for the provision of the analysis software and Michael Whimpey of BMRC for his work in the development of the satellite datasets and the running of the various analysis packages.

REFERENCES

  • Ahijevych, D. A., , C. A. Davis, , R. E. Carbone, , and J. D. Tuttle, 2004: Initiation of precipitation episodes relative to elevated terrain. J. Atmos. Sci., 61 , 27632769.

    • Search Google Scholar
    • Export Citation
  • Arkin, P. A., , and B. N. Meisner, 1987: The relationship between large-scale convective rainfall and cold cloud over the Western Hemisphere during the 1982-84. Mon. Wea. Rev., 115 , 5174.

    • Search Google Scholar
    • Export Citation
  • Arkin, P. A., , and P. Xie, 1994: The global precipitation climatology project: First Algorithm Intercomparison Project. Bull. Amer. Meteor. Soc., 75 , 401419.

    • Search Google Scholar
    • Export Citation
  • Barnes, G. M., , and K. Sieckman, 1984: The environment of fast- and slow-moving tropical mesoscale convective cloud lines. Mon. Wea. Rev., 112 , 17821794.

    • Search Google Scholar
    • Export Citation
  • Carbone, R. E., , T. D. Keenan, , J. Hacker, , and J. Wilson, 2000: Tropical island convection in the absence of significant topography. Part I: Life cycle of diurnally forced convection. Mon. Wea. Rev., 128 , 34593480.

    • Search Google Scholar
    • Export Citation
  • Carbone, R. E., , J. D. Tuttle, , D. A. Ahijevych, , and S. B. Trier, 2002: Inferences of predictability associated with warm season precipitation episodes. J. Atmos. Sci., 59 , 20332056.

    • Search Google Scholar
    • Export Citation
  • Carbone, R. E., , J. D. Tuttle, , and D. A. Ahijevych, 2003: Semidiurnal rainfall variability. Bull. Amer. Meteor. Soc., 84 , 15111512.

  • Chen, S. S., , R. A. Houze Jr., , and B. E. Mapes, 1996: Multiscale variability of deep convection in relation to large-scale circulation in TOGA COARE. J. Atmos. Sci., 53 , 13801654.

    • Search Google Scholar
    • Export Citation
  • Davis, C. A., , K. W. Manning, , R. E. Carbone, , S. B. Trier, , and J. D. Tuttle, 2003: Coherence of warm-season continental rainfall in numerical weather prediction models. Mon. Wea. Rev., 131 , 26672679.

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

  • Keenan, T. D., , and R. E. Carbone, 1992: A preliminary morphology of precipitation systems in tropical northern Australia. Quart. J. Roy. Meteor. Soc., 118 , 283326.

    • Search Google Scholar
    • Export Citation
  • Laing, A. G., , and J. M. Fritsch, 1997: The global population of mesoscale convective complexes. Quart. J. Roy. Meteor. Soc., 123 , 389405.

    • Search Google Scholar
    • Export Citation
  • Laing, A. G., , R. E. Carbone, , V. Levizzani, , and J. D. Tuttle, 2008: The propagation and diurnal cycles of deep convection in northern tropical Africa. Quart. J. Roy. Meteor. Soc., in press.

    • Search Google Scholar
    • Export Citation
  • Levizzani, V., , R. Ginnetti, , A. G. Laing, , and R. E. Carbone, 2006: Warm season precipitation climatology: First European results. Advances in Geoscience, Vol. 7, Academic Press, 15–18.

    • Search Google Scholar
    • Export Citation
  • Machado, L. A. T., , W. B. Rossow, , R. L. Guedes, , and A. W. Walker, 1998: Life cycle variations of mesoscale convective systems over the Americas. Mon. Wea. Rev., 126 , 16301654.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., , and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28 , 702708.

    • Search Google Scholar
    • Export Citation
  • Mathon, V., , H. Laurent, , and T. Lebel, 2002: Mesoscale convective system rainfall in the Sahel. J. Appl. Meteor., 41 , 10811092.

  • Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44 , 2543.

  • McBride, J. L., , and N. Nicholls, 1983: Seasonal relationships between Australian rainfall and the Southern Oscillation. Mon. Wea. Rev., 111 , 19982004.

    • Search Google Scholar
    • Export Citation
  • Moncrieff, M. W., , and J. S. A. Green, 1972: The propagation and transfer properties of steady convective overturning in shear. Quart. J. Roy. Meteor. Soc., 98 , 336352.

    • Search Google Scholar
    • Export Citation
  • Nakazawa, T., 1986: Mean features of 30-60 day variations as inferred from 8-year OLR data. J. Meteor. Soc. Japan, 64 , 777786.

  • Olson, D. A., , N. W. Junker, , and B. Korty, 1995: An evaluation of three decades of quantitative precipitation forecasting at the Meteorological Operations Division of the National Meteorological Center. Wea. Forecasting, 10 , 498511.

    • Search Google Scholar
    • Export Citation
  • Ramage, C. S., 1968: Role of a tropical “maritime continent” in the atmospheric circulation. Mon. Wea. Rev., 96 , 365370.

  • Reeder, M. J., , and R. K. Smith, 1998: Mesoscale meteorology. Meteorology of the Southern Hemisphere, D. Vincent and D. J. Karoly, Eds., Amer. Meteor. Soc., 201–241.

    • Search Google Scholar
    • Export Citation
  • Rotunno, R., , J. B. Klemp, , and M. Wiseman, 1988: A theory for strong long-lived squall lines. J. Atmos. Sci., 45 , 463485.

  • Tuttle, J. D., , and R. E. Carbone, 2004: Coherent regeneration and the role of water vapor and shear in a long-lived convective episode. Mon. Wea. Rev., 132 , 192208.

    • Search Google Scholar
    • Export Citation
  • Wang, C-C., , G-J. Chen, , and R. E. Carbone, 2004: A climatology of warm season cloud patterns over East Asia based on GMS infrared temperature observations. Mon. Wea. Rev., 132 , 16061629.

    • Search Google Scholar
    • Export Citation
  • Wang, C-C., , G-J. Chen, , and R. E. Carbone, 2005: Variability of warm-season cloud episodes over East Asia based on GMS infrared temperature observations. Mon. Wea. Rev., 133 , 14781500.

    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., , R. E. Carbone, , J. Tuttle, , and T. D. Keenan, 2001: Tropical island convection in the absence of significant topography. Part II: Nowcasting storm evolution. Mon. Wea. Rev., 129 , 16371655.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Study domain showing main topographic features. Elevation scale (m) is shown to the right. Circles denote locations of Adelaide (ADL), Albany (ALB), Alice Springs (ASP), Darwin (DWN), Giles (GLS), and Wagga Wagga (WWG).

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 2.
Fig. 2.

Maximum surface elevation in each latitude domain. East–west subdivisions of 2.5-km resolution were employed to define maximum elevation.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 3.
Fig. 3.

Impact of variations in Hovmöller thresholds (TBB < −50°, TBB < −35°, and TBB < −15°C) on streak characteristics in 30°–40°S domain for January 2001. Percentage of high cloud (PHC) or frequency of occurrence of TBB less than the indicated threshold obtained from a 0.2° × 0.2° latitude–longitude grid is shown (scale shown to right of panels). See text for description.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 4.
Fig. 4.

Vertical cross section of zonal winds (m s−1) for the domain 120°–150°E averaged over three latitude bands for spring and summer. Source of data: NCEP–NCAR 40-Year Reanalysis.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 5.
Fig. 5.

Selected sample satellite-derived Hovmöller diagrams for (a) midlatitude band 30°–40°S, (b) subtropical band 20°–30°S, and (c) tropical band 10°–20°S. Percentage of high cloud (PHC) is shown to the right with warmer colors indicating higher frequency. Cross sections of maximum surface elevation within each band are also shown below the top Hovmöller diagrams. Shaded ellipse areas denote major “forcing zones.” See text for discussion relating to shaded areas and features.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 5.
Fig. 5.

(Continued)

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 5.
Fig. 5.

(Continued)

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 6.
Fig. 6.

Example association of streak characteristics in 30°–40°S band with environmental conditions. (a) Hovmöller diagram of PHC (TBB < −35°C) for the period 1–30 Nov 1998. The PHC scale is shown to the right of the panel. (b) Equivalent Hovmöller diagram of 30-kPa daily meridional wind anomaly taken for NCEP–NCAR 40-Year Reanalysis data. (c) Hovmöller diagram showing correlation field employed to define streaks. Correlation scale (×100) shown to right of panel. Streaks defined by the analysis are overlaid. Boxes define periods of streak events that are discussed in text and summarized in Table 4. The maximum elevation within the 30°–40°S latitude band is included for reference.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 7.
Fig. 7.

Case study showing link between observed streaks, diurnal characteristics, and relation of streaks to satellite images. (a) Hovmöller diagram of PHC (TBB < −35°C) for the period 1–30 Nov 2000 for the 30°–40°S latitude band. Arrow highlights 12 Nov 2000. (b) Maximum elevation within the 30°–40°S latitude band. (c) Diurnal cycle of PHC composited over the period 9–23 Nov 2000 (30°–40°S latitude band). (d) Individual GMS satellite images showing diurnal evolution of cloudiness on 12 Nov 2000 for 30°–40°S latitude band. Box in (a) highlights period composited for diurnal cycle shown in (c). FR indicates diurnally forced convection originating over the Flinders Ranges. The subsequent motion is indicated by arrows. GD indicates convection enhanced over the Great Dividing Range. Note LT = UTC + 10 h.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 8.
Fig. 8.

Diurnal frequency in Hovmöller diagram of PHC (TBB < −35°C) for all seasons (1996–2001) showing (a) spring (November and December) and (b) summer (January and February) for the midlatitude band 30°–40°S, subtropical band 20°–30°S, and tropical band 10°–20°S. The scale indicates the percentage of days on which the PHC is present at the given longitude–UTC hour coordinate. The maximum elevation within each latitude band is included for reference. See text for discussion of arrows.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 9.
Fig. 9.

Diurnal cycle of PHC (TBB < −35°C) from 40° to 20°S in (a)–(c) the summer months of January and February and (d)–(f) the spring months of November and December. Convection over land near time of maximum in diurnal heating cycle is shown in (a) and (d). Evening mature and dissipating land-based convection with transition toward coastal and oceanic convection is shown in (b) and (e). Early morning phase with mature oceanic convection is shown in (c) and (f).

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 10.
Fig. 10.

Diurnal cycle of PHC (TBB < −55°C) from the equator to 20°S. (a) Mature convection over land resulting from diurnal heating, (b) dissipation and early transition toward lowlands and coastal oceanic convection, (c) mature oceanic convection associated with offshore flows, and (d) excitation of convection over Cape York Peninsula amid oceanic convection of nocturnal origin. Southward propagation of oceanic convection from New Guinea is coincident with the excitation of early day convection over Cape York. Westward propagation from Cape York (climatologically) assumes the shape of an organized “bow cloud.” Note the especially high amplitude of diurnal variation associated with convection over the Indonesian region (5°S, 110°–115°E). See text for explanation of suspected forcings.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Fig. 11.
Fig. 11.

Meridional Hovmöller diagrams of PHC (TBB < −55°C) in the tropics for the five seasons (November–March 1996–2001 inclusive). (a) Western Austral–Asia; (b) eastern Austral–Asia. Coherent patterns suggestive of propagation and phase-locked convection are apparent and these are identified by dashed or dotted lines. See text for explanation.

Citation: Monthly Weather Review 136, 3; 10.1175/2007MWR2152.1

Table 1.

Median duration, speed, span, and number of events for all eastward-moving cases 1996–2001 in the various latitude bands; PHC with TBB < −35°C. Results are formatted for all season (November–March)/spring (November–December)/summer (January–February).

Table 1.
Table 2.

Monthly mean shear values at representative stations for spring and summer in the three latitude bands. See Fig. 1 for station locations.

Table 2.
Table 3.

Summary of all streak characteristics in the 10°–20°S band (PHC with TBB < −55°C).

Table 3.
Table 4.

Summary of characteristics of streaks observed during November 1998.

Table 4.

1

See Carbone et al. (2003) for definitions of these terms.

2

Westward-propagating events increased in frequency with decreasing latitude band. The ratio of eastward- to westward-propagating streaks varied was approximately 3:1 in the 30°–40°S, 2: 1 in the 20°–30°S, and 1:1 in the 10°–20°S band.

3

Note that Table 1 refers to median values.

4

Events are defined in this latitude band employing PHC for TBB < −55°C. Table 1 shows some comparable results employing PHC for TBB < −35°C.

Save