• Agee, E. M., 1984: Observations from space and thermal convection: A historical perspective. Bull. Amer. Meteor. Soc., 65 , 938949.

  • Agee, E. M., 1987: Mesoscale cellular convection over the oceans. Dyn. Atmos. Oceans, 10 , 317341.

  • Agee, E. M., and D. E. Lund, 1987: A classification system for atmospheric thermal convection. Proc. Third Conf. on Mesoscale Processes, Vancouver, BC, Canada, Amer. Meteor. Soc., 204–205.

  • Balaji, V., J-L. Redelsperger, and G. P. Klaassen, 1993: Mechanisms for the mesoscale organization of tropical cloud clusters in GATE Phase III. Part I: Shallow cumulus clouds. J. Atmos. Sci., 50 , 35713589.

    • Search Google Scholar
    • Export Citation
  • Berendes, T. A., D. A. Berendes, R. M. Welch, E. G. Dutton, T. Uttal, and E. E. Clothiaux, 2004: Cloud cover comparisons of the MODIS daytime cloud mask with surface instruments at the North Slope of Alaska ARM site. IEEE Trans. Geosci. Remote Sens., 42 , 25842593.

    • Search Google Scholar
    • Export Citation
  • Bourassa, M. A., D. M. Legler, J. J. O’Brien, and S. R. Smith, 2003: SeaWinds validation with research vessels. J. Geophys. Res., 108 .3039, doi:10.1029/2001JC001081.

    • 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 features. J. Atmos. Sci., 59 , 20332056.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., and M. H. Freilich, 2005: Scatterometer-based assessment of 10-m wind analyses from the operational ECMWF and NCEP numerical weather prediction models. Mon. Wea. Rev., 133 , 409429.

    • Search Google Scholar
    • Export Citation
  • Chopra, K. P., 1973: Atmospheric and oceanic flow problems introduced by islands. Advances in Geophysics, Vol. 16, Academic Press, 297–421.

  • Clark, T. L., T. Hauf, and J. P. Kuettner, 1986: Convectively forced internal gravity waves: Results from two-dimensional numerical experiments. Quart. J. Roy. Meteor. Soc., 112 , 899925.

    • Search Google Scholar
    • Export Citation
  • Garay, M. J., R. Davies, C. Averill, and J. A. Westphal, 2004: Actinoform clouds: Overlooked examples of cloud self-organization at the mesoscale. Bull. Amer. Meteor. Soc., 85 , 15851594.

    • Search Google Scholar
    • Export Citation
  • Hubert, L. F., 1966: Mesoscale cellular convection. U.S. Department of Commerce, Environmental Science Services Administration, Meteorological Satellite Laboratory Tech. Rep. 37, 68 pp.

  • Johnson, R. H., T. M. Rickenbach, S. A. Rutledge, P. E. Ciesielski, and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972418.

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

  • King, M. D., Y. J. Kaufman, W. P. Menzel, and D. Tanre, 1992: Remote-sensing of cloud, aerosol, and water-vapor properties from the Moderate Resolution Imaging Spectrometer (MODIS). IEEE Trans. Geosci. Remote Sens., 30 , 227.

    • Search Google Scholar
    • Export Citation
  • King, M. D., S-C. Tsay, S. E. Platnick, M. Wang, and K. N. Liou, 1997: Cloud retrieval algorithms for MODIS: Optical thickness, effective particle radius, and thermodynamic phase. Algorithm Theoretical Basis Documents (ATBD) ATBD-MOD-05, MODIS Science Team, 83 pp. [Available online at http://modis-atmos.gsfc.nasa.gov/reference_atbd.html.].

  • LeMone, M. A., 1976: Modulation of turbulence energy by longitudinal rolls in an unstable planetary boundary layer. J. Atmos. Sci., 33 , 13081320.

    • Search Google Scholar
    • Export Citation
  • LeMone, M. A., and R. J. Meitin, 1984: Three examples of fair-weather mesoscale boundary-layer convection in the tropics. Mon. Wea. Rev., 112 , 19851997.

    • Search Google Scholar
    • Export Citation
  • Malkus, J. S., and H. Riehl, 1964: Cloud structure and distributions over the Tropical Pacific Ocean. Tellus, 16 , 275287.

  • Mapes, B. E., 1993: Gregarious tropical convection. J. Atmos. Sci., 50 , 20262037.

  • Miura, Y., 1986: Aspect ratios of longitudinal rolls and convection cells observed during cold air outbreaks. J. Atmos. Sci., 43 , 2639.

    • Search Google Scholar
    • Export Citation
  • Walter, B. A., and J. E. Overland, 1984: Observations of longitudinal rolls in a near neutral environment. Mon. Wea. Rev., 112 , 200208.

    • Search Google Scholar
    • Export Citation
  • Weckwerth, T. M., J. W. Wilson, and R. M. Wakimoto, 1997: Horizontal convective rolls: Determining the environmental conditions supporting their existence and characteristics. Mon. Wea. Rev., 125 , 505526.

    • Search Google Scholar
    • Export Citation
  • Weckwerth, T. M., T. W. Horst, and J. W. Wilson, 1999: An observational study of the evolution of horizontal convective rolls. Mon. Wea. Rev., 127 , 21602179.

    • Search Google Scholar
    • Export Citation
  • Woodcock, A. H., 1940: Convection and soaring over the open sea. J. Mar. Res., 3 , 248253.

  • Woodcock, A. H., 1975: Thermals over the sea and gull flight behavior. Bound.-Layer Meteor., 9 , 6368.

  • Young, G. S., and T. D. Sikora, 2003: Mesoscale stratocumulus bands caused by Gulf Stream meanders. Mon. Wea. Rev., 131 , 21772191.

  • Young, G. S., D. A. R. Kristovich, M. R. Hjelmfelt, and R. C. Foster, 2002: Rolls, streets, waves, and more: A review of quasi-two dimensional structures in the atmospheric boundary layer. Bull. Amer. Meteor. Soc., 83 , 9971001. [Also available as an extended electronic supplement at doi:10.1175/BAMS-83-7-Young.].

    • Search Google Scholar
    • Export Citation
  • View in gallery
    Fig. 1.

    MODIS image of a tropical dendritic cumulus formation embedded in a field of fair-weather cumulus. This 107 km × 43 km image was taken over the western Indian Ocean on 28 Sep 2004.

  • View in gallery
    Fig. 2.

    MODIS image of a tropical dendritic cumulus formation with widely scattered fair-weather cumulus between the cloud lines. This 107 km × 43 km image was taken over the western Indian Ocean on 15 Sep 2001.

  • View in gallery
    Fig. 3.

    MODIS image of a tropical dendritic cumulus formation with patchy fair-weather cumulus over a 107 km × 45 km swath of the southeastern Atlantic Ocean on 16 Apr 2003.

  • View in gallery
    Fig. 4.

    A map depicting the location of each center of the 61 dendritic cumulus formations studied.

  • View in gallery
    Fig. 5.

    Alignment dependence to the mean wind direction of tropical dendritic cumulus formations to the background wind speed.

  • View in gallery
    Fig. 6.

    Tropical dendritic cumulus cloud formations with a background wind speed of 5.14 m s−1. MODIS image spanning 96 km × 124 km of the western Indian Ocean on 28 Sep 2004.

  • View in gallery
    Fig. 7.

    Tropical dendritic cumulus cloud formations with a background wind speed of 6.69 m s−1. MODIS image spanning 94 km × 140 km of the western Timor Sea on 5 Nov 2001.

  • View in gallery
    Fig. 8.

    Tropical dendritic cumulus cloud formations with a background wind speed of 10.29 m s−1. MODIS image spanning 94 km × 126 km taken offshore of Tanzania on 25 Sep 2001.

  • View in gallery
    Fig. 9.

    Dependence of dendritic cloud formation alignment on the surface to the 850-mb wind shear.

  • View in gallery
    Fig. 10.

    Dependence of dendritic cloud formation alignment on the surface to the 850-mb shear: after shear-based correction and folding to 0°–360° range.

  • View in gallery
    Fig. 11.

    Combined mesoscale shear image of the surface and the 850-mb layer.

  • View in gallery
    Fig. 12.

    Synoptic–mesoscale shear relationship at 850 mb creating cloud-line branching seen from above.

  • View in gallery
    Fig. 13.

    Suppression of trade inversion by mesoscale shearing flow at 850 mb and the surface.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 108 44 9
PDF Downloads 88 33 7

Dendritic Patterns in Tropical Cumulus: An Observational Analysis

Stephen D. NichollsDepartment of Meteorology, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by Stephen D. Nicholls in
Current site
Google Scholar
PubMed
Close
and
George S. YoungDepartment of Meteorology, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by George S. Young in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

An observational analysis of the structure and synoptic setting of tropical dendritic cumulus formations was undertaken using 30 months of global data from the Moderate Resolution Imaging Spectroradiometer aboard the National Aeronautics and Space Administration Terra satellite, the Quick Scatterometer aboard the SeaWinds satellite, and the National Centers for Environmental Prediction global reanalysis. This analysis yielded 1216 cases of tropical dendritic cumulus formations of which 61 were randomly selected for quantitative study. From these sample cases, it was found that dendritic patterns in shallow cumulus occurred over warm tropical oceans in response to cold air advection. They typically dissipate downstream in regions of cooler water, neutral to warm advection, or deep convection. Moreover, shallow cumulus formations take on a dendritic pattern only in areas where the background wind velocity is between 1.5 and 13 m s−1 in the surface to the 850-mb layer and a shallow layer of conditional instability is present. Individual cumulus clouds in these dendritic formations are arranged in a compound, hierarchical branching pattern in which each element of the pattern takes the form of a Y-shaped cloud line. Analysis of the cloud pattern observations in conjunction with the scatterometer-derived surface winds and the lower-tropospheric wind profiles from reanalysis data reveals that the individual Y elements are aligned closely with the surface wind direction, as linear cloud streets would be. These Y elements are oriented so that their forked end aligns as closely as possible with the surface-to-850-mb shear vector, even when this conflicts with the surface wind direction. A formation mechanism is hypothesized by which the secondary circulation of a towering cumulus line modifies the shear and stability profiles in the adjacent areas to favor shallower cumulus lines oriented at an angle to itself, thus forming a hierarchical branching structure. This hypothesis is supported by stability profiles from the reanalysis data.

Corresponding author address: Dr. George S. Young, Department of Meteorology, The Pennsylvania State University, 503 Walker Building, University Park, PA 16802. Email: young@meteo.psu.edu

Abstract

An observational analysis of the structure and synoptic setting of tropical dendritic cumulus formations was undertaken using 30 months of global data from the Moderate Resolution Imaging Spectroradiometer aboard the National Aeronautics and Space Administration Terra satellite, the Quick Scatterometer aboard the SeaWinds satellite, and the National Centers for Environmental Prediction global reanalysis. This analysis yielded 1216 cases of tropical dendritic cumulus formations of which 61 were randomly selected for quantitative study. From these sample cases, it was found that dendritic patterns in shallow cumulus occurred over warm tropical oceans in response to cold air advection. They typically dissipate downstream in regions of cooler water, neutral to warm advection, or deep convection. Moreover, shallow cumulus formations take on a dendritic pattern only in areas where the background wind velocity is between 1.5 and 13 m s−1 in the surface to the 850-mb layer and a shallow layer of conditional instability is present. Individual cumulus clouds in these dendritic formations are arranged in a compound, hierarchical branching pattern in which each element of the pattern takes the form of a Y-shaped cloud line. Analysis of the cloud pattern observations in conjunction with the scatterometer-derived surface winds and the lower-tropospheric wind profiles from reanalysis data reveals that the individual Y elements are aligned closely with the surface wind direction, as linear cloud streets would be. These Y elements are oriented so that their forked end aligns as closely as possible with the surface-to-850-mb shear vector, even when this conflicts with the surface wind direction. A formation mechanism is hypothesized by which the secondary circulation of a towering cumulus line modifies the shear and stability profiles in the adjacent areas to favor shallower cumulus lines oriented at an angle to itself, thus forming a hierarchical branching structure. This hypothesis is supported by stability profiles from the reanalysis data.

Corresponding author address: Dr. George S. Young, Department of Meteorology, The Pennsylvania State University, 503 Walker Building, University Park, PA 16802. Email: young@meteo.psu.edu

1. Introduction

Analyzing mesoscale cloud formations over the open ocean proved difficult until the 1960s because the available observations, all taken in situ, provided only a patchwork of local views and hence little insight into mesoscale cloud organization. Thus, with the notable exception of some aircraft studies such as Malkus and Riehl (1964), few data were available on these mesoscale structures until the Television Infrared Observational Satellite (TIROS) was launched in 1960 and the age of space-borne remote sensing began. TIROS and subsequent meteorological imaging satellites have made it possible to observe mesoscale patterns in what a surface observer would only see as scattered or broken stratocumulus (Hubert 1966; Chopra 1973; Agee 1984; Miura 1986; Young and Sikora 2003; Garay et al. 2004). These patterns suggest the existence of a higher-order convective circulation acting to modulate the basic convective elements (Agee 1987; Agee and Lund 1987).

One unexpected pattern that has been observed via satellite imagery is the actinae (i.e., a raylike starburst) form of stratocumulus (Hubert 1966; Agee 1984; Garay et al. 2004). These cloud patterns reflect mesoscale circulations within the marine stratocumulus clouds above the cool surface currents of the eastern tropical oceans. An equally striking pattern can, however, be seen in relatively shallow cumulus over the warm water regions of tropical oceans. Under appropriate synoptic conditions (discussed below) the shallow and towering cumulus in these regions can align in compound branching (dendritic) patterns as shown in Fig. 1. These dendritic cumulus formations have not received the attention given to open- and closed-cell stratocumulus by Hubert (1966) and Agee (1984), to actiniae stratocumulus by Garay et al. (2004), and to nonbranching, cumulus lines with mesoscale spacing by LeMone and Meitin (1984) and Balaji et al. (1993). Therefore, these cumulus formations are the focus of the present study.

Tropical dendritic cumulus formations merit study both because of their abundance (1216 cases observed in our 30-month study period) and because they can organize shallow atmospheric convection on scales ranging from one hundred to a few thousand kilometers. Following the approach taken by Garay et al. (2004) with actiniae stratocumulus, a quantitative study of dendritic cumulus formations is carried out using satellite imagery and supporting large-scale observations and analyses. The phenomenon is common enough to permit the collection of a large sample, so statistical analysis of the results was undertaken.

This observational study of dendritic cumulus formations used 30 months of data from 1 July 2001 to 31 December 2003. Cases were collected by examining the 250-m resolution true-color images from the Moderate Resolution Imaging Spectroradiometer (MODIS; King et al. 1992, 1997) instrument aboard the National Aeronautic and Space Administration (NASA) Terra satellite. (Examination was restricted to those images archived online at http://rapidfire.sci.gsfc.nasa.gov/realtime/?calendar.) Because a pilot study had shown that dendritic cumulus formations occur only over low latitudes, the examination of the archives was further limited to those images obtained between 25°N and 25°S. Even with this restriction, the number of images examined was still in the thousands. Of these, 1216 showed dendritic cumulus formations.

Because full analysis of 1216 cases was impractical, 61 were chosen at random for quantitative analysis of both the dendritic cloud formation and the synoptic environment in which it formed. This number was predicated on the requirements for subsequent statistical analysis of the results. As 30 cases are generally considered a minimum for useable averages, it was decided to double that number to assure robustness. Any sample size on this order provides a reasonable compromise between analysis effort and statistical robustness. Supporting data were obtained from the multispectral MODIS imagery, the National Aeronautics and Space Agency (NASA) Quick Scatterometer (QuikSCAT; Bourassa et al. 2003) aboard the SeaWinds satellite, and the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis (Kalnay et al. 1996). The MODIS band 3–6–7 composite imagery was inspected to estimate cloud phase. Clouds that appeared to be primarily ice free in these images were assumed to be shallow or towering cumulus rather than deep cumulonimbus (Johnson et al. 1999). The NCEP–NCAR reanalysis provided profiles of vector wind and stability for the lower troposphere along with the information required to calculate lower-tropospheric temperature advection. The reasonableness of the reanalysis surface winds was checked via a comparison with the corresponding QuikSCAT observations, which are 10-m neutral equivalent winds.

2. Procedures

a. Dendritic cumulus formation geometry

Each of the archived MODIS images spans an area of 15° of longitude by 20° of latitude and is tilted counterclockwise by 10° from a north–south alignment because of the orbit inclination in the Tropics. This image information was used to estimate the area covered by the dendritic cumulus formations and to correct the cloud formation alignment angles to correspond to standard meteorological coordinates. Because the archived MODIS images are distorted near their cross-track edges (Berendes et al. 2004), only data from the central 45% of the images were analyzed. The alignment angle of a dendritic cumulus formation is defined here as the direction toward which the stems of the Y-shaped elements point. The alignment of the numerous Y elements in each dendritic formation was subjectively averaged by a human analyst prior to quantitative measurement of the angle for each of the 61 cases in the analysis sample. The geographic location of the estimated center of each dendritic cumulus formation and the area covered by the formation were also recorded. The alignment angles were recorded in degrees from true north, the locations in decimal degrees of latitude and longitude, and the area in square kilometers. These data were analyzed, as described below, to determine the geographic regions in which dendritic cumulus formations form, the synoptic setting required for their development, and the relationship between their orientation and the synoptic-scale winds.

b. Synoptic data

To determine the synoptic setting of a lower-tropospheric overocean convective system, the winds, sea, and air temperatures and atmospheric stability are required (e.g., Woodcock 1940, 1975; Miura 1986; Young et al. 2002). Because of the circumglobal distribution of dendritic cumulus formations, synoptic wind data were acquired from the NOAA QuikSCAT satellite-borne scatterometer and NCEP–NCAR reanalysis dataset. The NCEP–NCAR reanalysis was selected because of its global coverage and because the 2.5° resolution yielded grid cells of area similar to the dendritic cumulus formations. The scatterometer data had a much higher resolution of 25 km, more than was required to ascertain the synoptic setting. The scatterometer data were primarily used as a check for the reanalysis surface winds. The reanalysis dataset provided sea surface temperature and air temperatures, as well as wind vectors at the surface, 850 mb, and 700 mb. Differences between these pressure levels were used to assess layer stability and shear vectors at the location of each sampled dendritic cumulus formation. The combination of wind and temperature patterns was used to assess temperature advection at the three lower-tropospheric levels.

3. Results

a. Dendritic cumulus formations

Tropical dendritic cumulus formations exhibit a compound branching, fractal-like structure of shallow and towering cumulus. Closer inspection of the individual cases reveals, however, that the cumulus coverage varies considerably. In some cases the individual cumulus clouds are widely scattered except where they form part of the dendritic pattern (Fig. 2), while in other cases small cumulus form a more uniform background between branches of the dendritic pattern (Fig. 1). Figure 3 shows an intermediate case with patches of scattered cumulus embedded within the dendritic cumulus pattern. Examples from all parts of this spectrum are common in the 1216 cases observed. A histogram of cloud fraction was not computed however, because quantification of coverage by such small clouds is extremely sensitive to the satellite image resolution. The MODIS imagery at 250-m resolution is nonetheless sufficient to conclude that dendritic cumulus formations are a phenomenon of scattered cumulus rather than broken cumulus or stratocumulus.

b. Geography and extent

Figure 4 maps the location of each of the dendritic cumulus formations in the 61-case random sample. All of these cases occurred over warm tropical ocean currents, with a distinct gap in the distribution apparent over the cold Peru Current. Dendritic cumulus formations are also scarce over the warm waters of the western equatorial Pacific and the North Atlantic Tropics, suggesting the existence of two constraints on their global distribution. First, cumulus are much less common than stratocumulus over cold tropical currents (Garay et al. 2004), explaining the dearth of dendritic cumulus formations in such regions. Second, deep-precipitating convection is frequent over the warm equatorial western Pacific and along the intertropical convergence zone (ITCZ) in the North Atlantic Tropics. Thus, an abundance of deep convection appears to disrupt formation of mesoscale dendritic organization among shallow cumulus.

c. Temperature and instability

Both the sea surface temperature and the lower-tropospheric air temperature profile play pivotal roles in the genesis of tropical dendritic cumulus formations, as might be expected from the association of shallow and towering tropical cumulus with lower-tropospheric conditional instability. A sea surface temperature (SST) threshold is apparent in the 61-case sample, as all of the cases occurred with values of at least 24°C. This SST requirement helps explain why dendritic cumulus formations typically develop only between 25°N and 25°S. A shallow layer of conditional instability was also typically present as shown in Table 1. The mean lapse rate in the surface to the 850-mb layer is 6.1°C km−1, conditionally unstable, while the mean lapse rate in the 850–700-mb layer is 5.3°C km−1, stable. Thus, shallow convection is favored over deep, in keeping with the MODIS observations.

d. Triggering mechanisms

Of particular interest for any convective pattern are the triggering mechanisms (Young and Sikora 2003). Because dendritic cumulus formations are composed of locally enhanced lines of shallow cumulus convection, it was hypothesized that they might be either the precursors or the remnants of a mesoscale area of deep convection. To test this hypothesis the presence or absence of deep convection in the locale in the sequence of MODIS images preceding or succeeding each of the 61 sample images was noted. The result was that deep convection was observed in only 25 of the 61 samples the day before and 22 of 61 samples the day after. These results suggest that the deep convective instability is neither a reliable prerequisite nor a result of dendritic cumulus formation. Thus, we need to look further for a triggering mechanism.

The next hypothesis to be examined was that low-level temperature advection on the day of the event might play a role in creating the shallow layer of conditional instability necessary for their formation and dominance over other convective modes. Analysis of the NCEP–NCAR reanalysis showed that surface cold advection was present in varying degrees in 97% (i.e., all but two) of the 61 sample cases. This finding provides one reason why dendritic cumulus formations do not occur everywhere the SST exceeds 24°C. Moreover, this low-level cold advection requirement sets them apart from the narrow mode linear cloud streets commonly reported over warm tropical oceans (LeMone 1976; Young et al. 2002). The other common boundary layer convective structure associated with cold advection, wide mode linear rolls (Miura 1986; Young et al. 2002), is composed of broad, unbranching stratocumulus bands, a distinctly different structure than that of the tropical dendritic cumulus patterns discussed here. Wide mode rolls are found primarily poleward of 25° (Miura 1986) while tropical dendritic cumulus patterns are found mainly equatorward of that latitude. This geographic separation suggests that some additional factor, perhaps stability aloft as discussed below, determines which of the two modes occurs.

e. Terminating mechanisms

Also of interest is the mechanism that limits the downwind extent of tropical dendritic cumulus formations. The environmental factors leading to termination of each of the 61 sample cases were examined using a combination of MODIS imagery and NCEP–NCAR reanalysis data. For these cases the downwind edge of the dendritic cumulus formation was usually defined by one of five phenomena: a land-breeze front, a coast, the onset warm or neutral temperature advection at the surface, the edge of a cold current, or the boundary of an area of deep convection. The most common termination mechanism was an area of deep convective such as a mesoscale convective complex or the ITCZ. The second most common is advection of the cloud formation over an area of cooler water. Conversely, land-breeze fronts were the least common cause of termination because most of the sample dendritic cumulus formations occurred in the open sea with only a few adjacent to the coasts of Africa and northeastern South America.

f. Wind speed and direction

As with linear cloud streets (LeMone 1976; Miura 1986; Young et al. 2002), the surface wind and lower-tropospheric shear vectors have a great influence on the alignment of tropical dendritic cumulus formations. The dependence of alignment on surface wind direction can be seen in Fig. 5, which, somewhat surprisingly, shows two modes of alignment, one with the surface (i.e., 10 m) wind and the other into it. Such a bimodal distribution is expected with linear cloud streets because their symmetry yields a 180° ambiguity in alignment. But dendritic cumulus formations have an unambiguous orientation, with the Y stem pointing one way along their alignment axis and the Y prongs pointing the other. From Fig. 5 it can be seen that the Y stem may point either with or into the surface wind direction, with deviations rarely exceeding 30°. Indeed the standard deviation of the sample cases around these two directions is only 14.3° about the 0° alignment and 15.6° about the 180° alignment, about what Young et al. (2002) report for the scatter in wind-relative alignment of linear cloud streets. The key question as to why this bimodal distribution of wind-relative alignment exists is addressed in the next section.

Another key result shown in Fig. 5 is the role of wind speed in determining the alignment of a dendritic cumulus formation, which is illustrated in Figs. 6, 7 and 8 by a dramatic decrease in scatter about both modes of alignment as wind speed increases. This relationship could indicate one of two things. First, it is possible that the QuikSCAT wind directions become more accurate at higher wind speeds, which is not unreasonable if the magnitude of the vector error does not increase as rapidly as the wind speed as reported by Chelton and Freilich (2005). Indeed, unless calms are perfectly analyzed, the wind direction error must reach a maximum at zero wind speed. Second, it is possible that whatever process links wind direction to alignment becomes more effective as wind speed increases. The scatterometer wind directions closely match the reanalysis surface wind directions (bias of 4.5°, standard deviation of 13.8°), but neither are reliable at very low wind speeds. Hence the first possibility cannot be tested conclusively. Linear cloud street studies, such as LeMone (1976) and Weckwerth et al. (1997), show that wind alignment of cumulus requires that the wind speed exceed a minimum value (albeit stability dependent; Weckwerth et al. 1999), giving some support to the second possibility. A broad domain three-dimensional large eddy simulation of moist convection might be required to fully resolve this issue.

The relationship between dendritic cumulus formations and linear cloud streets is clarified somewhat by the wind speed limits in which the two phenomena are observed. While linear cloud streets typically require at least 3 m s−1 of surface wind (Weckwerth et al. 1997), dendritic cumulus formations are observed at speeds ranging from slightly more than 1.5 to just below 13 m s−1 (Table 2). Thus, dendritic cumulus formations and linear cloud streets require similar minimum wind speeds. This finding raises the question as to why a branching mode exists for SST values greater than 24°C but has not been observed under other conditions. The next section will address this issue while collapsing the bimodal distribution mentioned above.

The relationship between the wind speed and the branching angle of the Y prongs shown in Figs. 6, 7 and 8 suggests that a gradual transition may take place from dendritic cumulus formations to linear cloud streets as wind speed increases. In an effort to quantify this hypothetical relationship the typical branching angle each of the sample images was calculated by first thresholding the image to obtain a cumulus mask and then applying the linear feature alignment algorithm of Carbone et al. (2002) to the mask. This objective method, adapted from the Hovmöller diagram feature-tracking application, works imperfectly on a discontinuous cumulus field, but is superior to subjective estimates of the typical spread angle in a complex cloud field. A successful analysis was accomplished on 40 of the 61 sample cases, providing mean cloud-line alignment angles that agreed well with the subjective analysis. The standard deviation of these cloud-line alignment angles was used as a proxy for the average branching angle of each dendritic cumulus formation. However, any correlation between this proxy and the surface wind speed was negligible suggesting that factors other than surface wind speed determine whether linear cloud streets or dendritic cumulus formations occur. A possible explanation for this scatter, in terms of a dependence on both convective intensity and wind shear magnitude, is provided as a consequence of the formation mechanism hypothesized and discussed in the next section.

g. Shear influence and branch structure

To explain the bimodal relationship between the surface wind direction and alignment of tropical dendritic cumulus formations it is necessary to turn to the lower-tropospheric shear. In contrast to the situation for linear cloud streets, where the boundary layer shear and the surface wind are usually in roughly the same direction (e.g., Weckwerth et al. 1997, 1999; Young et al. 2002), these two vectors frequently oppose each other in dendritic cumulus formations. In 41 of the 61 cases sampled, the surface-to-850-mb shear vector was opposed to rather than aligned with the surface wind. Thus, shear and alignment both have a bimodal distribution about the surface wind direction for dendritic cumulus formations. This difference between dendritic cumulus formations and linear cloud streets suggests that the orientation of the boundary layer shear may determine the branching alignment of the dendritic formation. Figure 9 confirms this hypothesis, indicating a strong linear relationship between the two parameters (r2 = 0.70). This correlation can be improved further, to r2 = 0.96, by using the surface-to-850-mb shear merely to resolve the bimodal relationship between surface wind direction and the alignment of the dendritic cloud formation. By using the reverse of the wind direction in those cases where it is more closely aligned to the shear, the bimodal relationship is collapsed. Figure 10 displays the relationship of cloud formation alignment to this new direction parameter. Thus, while the individual Y elements of dendritic cumulus formations align more closely with the surface wind than with the surface-to-850-mb shear, they assume an orientation along the surface wind direction such that the Y prongs point downshear. A regression line between the resulting prediction and the observed alignment has a slope of 0.99 and an intercept of 8.1° suggesting that there is very little multiplicative or additive bias.

One explanation for this shear-dependent alignment lies in the superposition of synoptic-scale shear and the secondary circulations generated by the individual cumulus lines themselves. If these circulations are strong enough they may result in significant variations in the boundary layer shear vector from one side of the cloud line to the other as shown in Fig. 11. Weaker cloud lines adjacent to a strong line would thus experience divergent shear vectors, angling outward from that line. If these cloud lines are shear aligned in the manner of linear cloud streets (Young et al. 2002), they would angle out from the original cloud line with the open end of the resulting branching structure oriented down the synoptic shear as shown in Fig. 12.

This hypothetical explanation raises the question of why, in those synoptic situations where tropical dendritic cumulus formations occur, some cloud lines would be stronger, and thus more shear controlling, than others. One possible explanation is the noticeable increase in stability between the surface-to-850 and the 850–700-mb layers. The stronger cloud lines in a dendritic cumulus formation are often composed of towering cumulus, while the weaker cloud lines are composed of shallow cumulus. Towering cumulus lines would use this stability gradient to their own advantage by lifting it locally while lowering it on either side as shown in Fig. 13. Thus, cumulus lines forming adjacent to preexisting towering cumulus lines would develop in a shallower layer of conditionally unstable air as well as within the modified shear described above. A similar pattern of stability modification has been described in the context of deep precipitating tropical convection (Mapes 1993) and shallow boundary layer convection (Clark et al. 1986). A broad domain three-dimensional large eddy simulation of moist convection might be used to test the applicability of this mechanism to tropical dendritic cumulus formations.

These arguments suggest that linear elements of dendritic cumulus formations differ from linear cloud streets in that their secondary circulations can modify both the background wind and stability profiles experienced by neighboring cloud lines. Thus, they would be favored when synoptic shear is not overwhelmingly stronger than the shear produced by the secondary circulations of the deepest cloud lines and when these cloud lines and their secondary circulations penetrated up into a layer of increased stability. This hypothesis suggests a corollary, not explored here, that nonbranching cloud streets occur when inversion strength is sufficient to limit variations in cloud depth and hence preclude this convective self-scaling process, or when the synoptic shear is much stronger than that of the secondary circulations. Walter and Overland (1984) suggest, however, that it is possible for nonbranching cloud streets to form a scale hierarchy as well, so it appears that the thermodynamically induced variations in cloud-line depth can occur without the kinematically induced variations in cloud line orientation. The range of potential interactions between convective lines has yet to be fully explored. Investigation of the full parameter space may eventually reveal that wide mode rolls, hierarchical nonbranching rolls, and hierarchical dendritic cumulus patterns, do in fact represent separate, but dynamically related, domains within the shear, lid strength, and convective intensity subspace.

4. Conclusions

This paper documents the frequent occurrence of branching Y-element patterns within fields of shallow and towering cumulus. These dendritic cumulus formations occur over warm water tropical oceans (SST > 24°C) with light to moderate winds (1.5–13 m s−1). An additional condition for their existence is that some degree of cold air advection be present at the surface. The combination of cold air advection and warm sea surface favors boundary layer convection as, for example, in cold air outbreaks at higher latitudes. The resulting thermodynamic profile is conditionally unstable in the surface-to-850-mb layer but stable in the 850–700-mb layer. Thus, dendritic cumulus formations occur over tropical waters that are favorable for deep-precipitating convection but in synoptic situations that are unfavorable to it. Temporal analysis supports this result as dendritic cumulus formations are neither reliable precursors for areas of deep precipitating convection nor the reverse. As a result, the geographic distribution of tropical dendritic cumulus formations avoids those regions where synoptic forcing favors deep convection (e.g., the warm pool of the western equatorial Pacific and the Atlantic ITCZ).

Dendritic cumulus formations align with both the surface wind and the surface to the 850-mb shear vector, much as do linear cloud streets. Yet the cloud lines in dendritic formations branch frequently while those in conventional cloud street patterns do not. This difference is hypothesized to result from the formation mechanism of the branching structure itself, the interaction of the secondary circulation of towering cumulus lines with the shear and stability profiles of their surroundings. By locally modifying the shear vector and decreasing the depth of the conditionally unstable layer, this secondary circulation would create conditions favoring shallower cumulus lines branching at a small angle from the preexisting line. This is indeed the pattern observed in dendritic cumulus formations. Moreover the alignment and orientation suggested by this mechanism is a good predictor for the observed dendritic structures.

Future work should test this secondary circulation hypothesis using a broad domain three-dimensional large eddy simulation of moist convection. The results of such a model could also be used to explore the role of gravity wave interaction with boundary layer rolls in producing the mesoscale spacing of cloud lines. Likewise, they could be used to test the conjecture that stronger capping inversions result in spreading (i.e., wide mode) boundary layer rolls while weaker capping inversions lead to dendritic cumulus formations. These studies could also be undertaken using intensive in situ measurements from aircraft and dropsondes.

Acknowledgments

This work was supported by the Office of Naval Research under Grant N00014-04-10539. We wish to thank Dr. Margaret LeMone and two anonymous reviewers for contributing their varied insights on the mechanisms responsible for inducing mesoscale structure in shallow convection.

REFERENCES

  • Agee, E. M., 1984: Observations from space and thermal convection: A historical perspective. Bull. Amer. Meteor. Soc., 65 , 938949.

  • Agee, E. M., 1987: Mesoscale cellular convection over the oceans. Dyn. Atmos. Oceans, 10 , 317341.

  • Agee, E. M., and D. E. Lund, 1987: A classification system for atmospheric thermal convection. Proc. Third Conf. on Mesoscale Processes, Vancouver, BC, Canada, Amer. Meteor. Soc., 204–205.

  • Balaji, V., J-L. Redelsperger, and G. P. Klaassen, 1993: Mechanisms for the mesoscale organization of tropical cloud clusters in GATE Phase III. Part I: Shallow cumulus clouds. J. Atmos. Sci., 50 , 35713589.

    • Search Google Scholar
    • Export Citation
  • Berendes, T. A., D. A. Berendes, R. M. Welch, E. G. Dutton, T. Uttal, and E. E. Clothiaux, 2004: Cloud cover comparisons of the MODIS daytime cloud mask with surface instruments at the North Slope of Alaska ARM site. IEEE Trans. Geosci. Remote Sens., 42 , 25842593.

    • Search Google Scholar
    • Export Citation
  • Bourassa, M. A., D. M. Legler, J. J. O’Brien, and S. R. Smith, 2003: SeaWinds validation with research vessels. J. Geophys. Res., 108 .3039, doi:10.1029/2001JC001081.

    • 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 features. J. Atmos. Sci., 59 , 20332056.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., and M. H. Freilich, 2005: Scatterometer-based assessment of 10-m wind analyses from the operational ECMWF and NCEP numerical weather prediction models. Mon. Wea. Rev., 133 , 409429.

    • Search Google Scholar
    • Export Citation
  • Chopra, K. P., 1973: Atmospheric and oceanic flow problems introduced by islands. Advances in Geophysics, Vol. 16, Academic Press, 297–421.

  • Clark, T. L., T. Hauf, and J. P. Kuettner, 1986: Convectively forced internal gravity waves: Results from two-dimensional numerical experiments. Quart. J. Roy. Meteor. Soc., 112 , 899925.

    • Search Google Scholar
    • Export Citation
  • Garay, M. J., R. Davies, C. Averill, and J. A. Westphal, 2004: Actinoform clouds: Overlooked examples of cloud self-organization at the mesoscale. Bull. Amer. Meteor. Soc., 85 , 15851594.

    • Search Google Scholar
    • Export Citation
  • Hubert, L. F., 1966: Mesoscale cellular convection. U.S. Department of Commerce, Environmental Science Services Administration, Meteorological Satellite Laboratory Tech. Rep. 37, 68 pp.

  • Johnson, R. H., T. M. Rickenbach, S. A. Rutledge, P. E. Ciesielski, and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972418.

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

  • King, M. D., Y. J. Kaufman, W. P. Menzel, and D. Tanre, 1992: Remote-sensing of cloud, aerosol, and water-vapor properties from the Moderate Resolution Imaging Spectrometer (MODIS). IEEE Trans. Geosci. Remote Sens., 30 , 227.

    • Search Google Scholar
    • Export Citation
  • King, M. D., S-C. Tsay, S. E. Platnick, M. Wang, and K. N. Liou, 1997: Cloud retrieval algorithms for MODIS: Optical thickness, effective particle radius, and thermodynamic phase. Algorithm Theoretical Basis Documents (ATBD) ATBD-MOD-05, MODIS Science Team, 83 pp. [Available online at http://modis-atmos.gsfc.nasa.gov/reference_atbd.html.].

  • LeMone, M. A., 1976: Modulation of turbulence energy by longitudinal rolls in an unstable planetary boundary layer. J. Atmos. Sci., 33 , 13081320.

    • Search Google Scholar
    • Export Citation
  • LeMone, M. A., and R. J. Meitin, 1984: Three examples of fair-weather mesoscale boundary-layer convection in the tropics. Mon. Wea. Rev., 112 , 19851997.

    • Search Google Scholar
    • Export Citation
  • Malkus, J. S., and H. Riehl, 1964: Cloud structure and distributions over the Tropical Pacific Ocean. Tellus, 16 , 275287.

  • Mapes, B. E., 1993: Gregarious tropical convection. J. Atmos. Sci., 50 , 20262037.

  • Miura, Y., 1986: Aspect ratios of longitudinal rolls and convection cells observed during cold air outbreaks. J. Atmos. Sci., 43 , 2639.

    • Search Google Scholar
    • Export Citation
  • Walter, B. A., and J. E. Overland, 1984: Observations of longitudinal rolls in a near neutral environment. Mon. Wea. Rev., 112 , 200208.

    • Search Google Scholar
    • Export Citation
  • Weckwerth, T. M., J. W. Wilson, and R. M. Wakimoto, 1997: Horizontal convective rolls: Determining the environmental conditions supporting their existence and characteristics. Mon. Wea. Rev., 125 , 505526.

    • Search Google Scholar
    • Export Citation
  • Weckwerth, T. M., T. W. Horst, and J. W. Wilson, 1999: An observational study of the evolution of horizontal convective rolls. Mon. Wea. Rev., 127 , 21602179.

    • Search Google Scholar
    • Export Citation
  • Woodcock, A. H., 1940: Convection and soaring over the open sea. J. Mar. Res., 3 , 248253.

  • Woodcock, A. H., 1975: Thermals over the sea and gull flight behavior. Bound.-Layer Meteor., 9 , 6368.

  • Young, G. S., and T. D. Sikora, 2003: Mesoscale stratocumulus bands caused by Gulf Stream meanders. Mon. Wea. Rev., 131 , 21772191.

  • Young, G. S., D. A. R. Kristovich, M. R. Hjelmfelt, and R. C. Foster, 2002: Rolls, streets, waves, and more: A review of quasi-two dimensional structures in the atmospheric boundary layer. Bull. Amer. Meteor. Soc., 83 , 9971001. [Also available as an extended electronic supplement at doi:10.1175/BAMS-83-7-Young.].

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

MODIS image of a tropical dendritic cumulus formation embedded in a field of fair-weather cumulus. This 107 km × 43 km image was taken over the western Indian Ocean on 28 Sep 2004.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 2.
Fig. 2.

MODIS image of a tropical dendritic cumulus formation with widely scattered fair-weather cumulus between the cloud lines. This 107 km × 43 km image was taken over the western Indian Ocean on 15 Sep 2001.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 3.
Fig. 3.

MODIS image of a tropical dendritic cumulus formation with patchy fair-weather cumulus over a 107 km × 45 km swath of the southeastern Atlantic Ocean on 16 Apr 2003.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 4.
Fig. 4.

A map depicting the location of each center of the 61 dendritic cumulus formations studied.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 5.
Fig. 5.

Alignment dependence to the mean wind direction of tropical dendritic cumulus formations to the background wind speed.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 6.
Fig. 6.

Tropical dendritic cumulus cloud formations with a background wind speed of 5.14 m s−1. MODIS image spanning 96 km × 124 km of the western Indian Ocean on 28 Sep 2004.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 7.
Fig. 7.

Tropical dendritic cumulus cloud formations with a background wind speed of 6.69 m s−1. MODIS image spanning 94 km × 140 km of the western Timor Sea on 5 Nov 2001.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 8.
Fig. 8.

Tropical dendritic cumulus cloud formations with a background wind speed of 10.29 m s−1. MODIS image spanning 94 km × 126 km taken offshore of Tanzania on 25 Sep 2001.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 9.
Fig. 9.

Dependence of dendritic cloud formation alignment on the surface to the 850-mb wind shear.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 10.
Fig. 10.

Dependence of dendritic cloud formation alignment on the surface to the 850-mb shear: after shear-based correction and folding to 0°–360° range.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 11.
Fig. 11.

Combined mesoscale shear image of the surface and the 850-mb layer.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 12.
Fig. 12.

Synoptic–mesoscale shear relationship at 850 mb creating cloud-line branching seen from above.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Fig. 13.
Fig. 13.

Suppression of trade inversion by mesoscale shearing flow at 850 mb and the surface.

Citation: Monthly Weather Review 135, 5; 10.1175/MWR3379.1

Table 1.

Lower tropospheric and SSTs associated with the sample.

Table 1.
Table 2.

Minimum, maximum, and average of the fastest, slowest, and average surface winds speed, respectively, in the area of dendritic cumulus formations for the 61 sampled cases.

Table 2.
Save