• Austin, G., R. M. Rauber, H. T. Ochs III, and L. J. Miller, 1996: Trade wind clouds and Hawaiian rainbands. Mon. Wea. Rev.,124, 2126–2151.

  • Beard, K. V., D. B. Johnson, and D. Baumgardner, 1986: Aircraft observations of large raindrops in warm, shallow, convective clouds. Geophys. Res. Lett.,19, 991–994.

  • Chong, M., and J. Testud, 1983: Three-dimensional wind field analysis from dual-Doppler radar data. Part III: The boundary condition: An optimum determination based on a variational concept. J. Climate Appl. Meteor.,22, 1227–1241.

  • Fujiwara, M., 1967: Raindrop-size distribution in warm rain as measured in Hawaii. Tellus,19, 393–402.

  • Lavoie, R. L., 1967: Background data for the warm rain project. Tellus,19, 348–353.

  • List, R. L., and J. R. Gillespie, 1976: Evolution of raindrop spectra with collision-induced breakup. J. Atmos. Sci.,33, 2007–2013.

  • Low, T. B., and R. List, 1982: Collision, coalescence, and breakup of raindrops. J. Atmos. Sci.,39, 1591–1618.

  • Miller, L. J., and R. G. Strauch, 1974: A dual-Doppler radar method for the determination of wind velocities within a precipitating weather systems. Remote Sens. Environ.,3, 219–235.

  • ———, J. E. Dye, and B. E. Martner, 1983: Dynamical–microphysical evolution of a convective storm in a weakly sheared environment. Part II: Airflow and precipitation trajectories from Doppler radar observations. J. Atmos. Sci.,40, 2097–2109.

  • ———, C. G. Mohr, and A. J. Weinheimer, 1986: The simple rectification to Cartesian space of folded radial velocities from Doppler radar sampling. J. Atmos. Oceanic Technol.,3, 162–174.

  • Mohr, C. G., and R. L. Vaughan, 1979: An economical procedure for Cartesian interpolation and display of reflectivity data in three-dimensional space. J. Appl. Meteor.,18, 661–670.

  • ———, L. J. Miller, R. L. Vaughan, and H. W. Frank, 1986: The merger of mesoscale datasets into a common Cartesian format for efficient and systematic analyses. J. Atmos. Oceanic Technol.,3, 143–161.

  • Mordy, W. A., and L. E. Eber, 1954: Observations of rainfall from warm clouds. Quart. J. Roy. Meteor. Soc.,80, 48–57.

  • Rauber, R. M., K. V. Beard, and B. M. Andrews, 1991: A mechanism for giant raindrop formation in warm, shallow, convective clouds. J. Atmos. Sci.,48, 1791–1797.

  • Ray, P. S., K. K. Wagner, K. W. Johnson, J. J. Stephens, W. C. Bumgarner, and E. A. Mueller, 1978: Triple-Doppler observations of a convective storm. J. Appl. Meteor.,17, 1201–1212.

  • ———, C. L. Ziegler, W. C. Bumgarner, and R. J. Serafin, 1980: Single- and multiple-Doppler radar observations of tornadic storms. Mon. Wea. Rev.,108, 1607–1625.

  • Rogers, R. R., 1967: Doppler radar investigation of Hawaiian rain. Tellus,19, 433–455.

  • Semonin, F. G., E. A. Mueller, G. A. Stout, and D. W. Staggs, 1968: Radar analysis of warm rain showers. Tellus,20, 227–238.

  • Squires, P., 1952a: The growth of cloud drops by condensation I. General characteristics. Aust. J. Sci. Res.,5, 59–86.

  • ———, 1952b: The growth of cloud drops by condensation II. The formation of large cloud drops. Aust. J. Sci. Res.,5, 473–499.

  • ———, 1958a: The microstructure and colloidal stability of warm clouds. Part I. The relation between structure and stability. Tellus,10, 256–261.

  • ———, 1958b: The microstructure and colloidal stability of warm clouds. Part II. The causes of the variations in the microstructure. Tellus,10, 262–271.

  • Takahashi, T., 1977: A study of Hawaiian warm showers based on aircraft observation. J. Atmos. Sci.,34, 1773–1790.

  • ———, 1981: Warm rain study in Hawaii—Rain initiation. J. Atmos. Sci.,38, 347–369.

  • ———, K. Yoneyama, and Y. Tsubota, 1989: Rain duration in Hawaiian trade wind rainbands—Aircraft observation. J. Atmos. Sci.,46, 937–955.

  • View in gallery
    Fig. 1.

    Location of the seaward dual-Doppler lobe, the radar analysis area, and a typical HaRP flight profile near the Big Island of Hawaii.

  • View in gallery
    Fig. 2.

    (a) Radar reflectivity comparison between CP4 and CP3 on 22 July 1990. The reflectivities were measured by both radars at approximately 0616 UTC and interpolated to a Cartesian grid. This comparison was done at an altitude of 1.2 km; (b) same as (a) except on 22 August 1990 at 1845 UTC.

  • View in gallery
    Fig. 3.

    Horizontal cross section of the radar reflectivity field at 1.0 km measured by CP3 on 22 July 1990 at 0616 UTC. (a) Overlaid are the 2.5 and 3.5 m s−1 vertical motion contours from 0616 UTC at an elevation of 2.0 km. (b) Overlaid are the 2.5 and 3.5 m s−1 vertical motion contours from 0611 UTC at an elevation of 2.0 km advected forward with the mean cell motion to 0616 UTC.

  • View in gallery
    Fig. 4.

    Four horizontal cross sections of the radar reflectivity field measured by CP3 at 1.2 km on 22 August 1990 between 1801 and 1901 UTC. Panels (a) and (b) show the developing stage of the rainband, (c) the mature stage, and (d) the dissipating stage.

  • View in gallery
    Fig. 5.

    Thermodynamic diagram displaying the NCAR Electra sounding taken on 22 August 1990, 100–160 km upstream of the island at approximately 1630 UTC. The shaded area indicates CAPE. A barb and pennant, respectively, represent wind speeds of 1 and 5 m s−1.

  • View in gallery
    Fig. 6.

    Horizontal cross sections of reflectivity through the elevation of maximum reflectivity (a) for cell 1 at 1.2 km measured by CP3 and for (b) cell 2 at 2.8 km measured by CP4 on 22 August 1990 at 1848 UTC, corresponding to region enclosed in box I in Fig. 4a.

  • View in gallery
    Fig. 7.

    Comparison of the temporal evolution of the maximum updraft (bold) and maximum reflectivity (thin) in cells 1 (dashed) and 2 (solid) between 1838 UTC and 1848 UTC on 22 August 1990.

  • View in gallery
    Fig. 8.

    Vertical cross sections through cell 1 at 1848 UTC on 22 August 1990 at the locations shown in Fig. 6: (a) along the rainband and (b) across the rainband. Overlaid are the 4.5, 5.5, 6.5, and 7.5 m s−1 vertical motion contours advected with the mean cell motion from 1843 UTC. Two estimates of the trade wind inversion, based on the sounding and the weak echo region outside the main reflectivity core, are also shown.

  • View in gallery
    Fig. 9.

    Vertical cross sections through cell 2 at 1848 UTC on 22 August 1990 at the locations shown in Fig. 6: (a) along the rainband and (b) across the rainband. Overlaid are the 4.5, 5.5, and 6.5 m s−1 vertical motion contours advected with the mean cell motion from 1845:30 UTC. Two estimates of the trade wind inversion, based on the sounding and the weak echo region outside the main reflectivity core, are also shown.

  • View in gallery
    Fig. 10.

    Horizontal cross section of the vertical motion field at 2.4 km at 1848 UTC on 22 August 1990. Overlaid are horizontal cell-relative streamlines (thin lines) depicting the direction of the main outflow and 50-dBZi contours (bold) depicting the locations of high reflectivity cores in cells 1 and 2.

  • View in gallery
    Fig. 11.

    Vertical cross sections of vertical velocity through cell 1 on 22 August 1990: (a) across-band at 1843 UTC, (b) along-band at 1843 UTC, (c) across-band at 1848 UTC, and (d) along-band at 1848 UTC. Also depicted are the cell-relative streamlines.

  • View in gallery
    Fig. 12.

    Same as Fig. 11 except for cell 2.

  • View in gallery
    Fig. 13.

    Horizontal cross section of the radar reflectivity field measured by CP3 at 1.2 km on 10 August 1990 at 1746 UTC.

  • View in gallery
    Fig. 14.

    Panels (a)–(g): Horizontal cross sections of the CP3 reflectivity field (corresponding to box II in Fig. 13) between 1731 and 1751 UTC on 10 August 1990. Panels (a)–(d) [(e)–(g)] are at the altitude of maximum reflectivity in high reflectivity core 1 (2). Panels (a1)–(g1): Vertical cross sections of reflectivity at the location of maximum reflectivity in core 1; panels (a2)–(g2) Same as (a1)–(g1), but for core 2.

  • View in gallery
    Fig. 15.

    Altitude and intensity of the maximum reflectivity observed within core 1 and 2 (Fig. 14) between 1726 and 1751 UTC.

  • View in gallery
    Fig. 16.

    (c) Horizontal and vertical cross sections of the CP3 reflectivity field at 1736 UTC 10 August at locations corresponding to panel c of Fig. 14 except that cell-relative streamlines are also depicted on each cross section; (c) Horizontal and vertical cross sections of the vertical velocity field at 1736 UTC 10 August 1990. The horizontal cross section was taken at the altitude of the peak vertical motion, 2.4 km. The vertical cross sections, (c1′) and (c2′), are in the same plane as in panels c1 and c2. Cell-relative streamlines are depicted on all panels.

  • View in gallery
    Fig. 17.

    Vertical profiles of divergence through (a) 10 August updraft shown in Fig. 16 (c2’) and (b) 22 July updraft associated with the high reflectivity core shown in Fig. 18. Also indicated are the estimated heights of the inversion base. Updraft flank divergence profiles are 0.9 km away from the core.

  • View in gallery
    Fig. 18.

    Vertical cross section of the CP3 radar reflectivity on 22 July 1990 at 0616 UTC as indicated by a–a’ in Fig. 3. (a) Overlaid are the 2.5 and 3.5 m s−1 vertical motion contours advected with mean cell motion from 0608:30 UTC. (b) Same as in (a) except advected from 0611 UTC.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 172 22 0
PDF Downloads 51 16 0

The Microphysical Structure and Evolution of Hawaiian Rainband Clouds. Part I: Radar Observations of Rainbands Containing High Reflectivity Cores

Marcin J. SzumowskiDepartment of Atmospheric Sciences, University of Illinois at Urbana–Champaign, Urbana, Illinois, andIllinois State Water Survey, Champaign, Illinois

Search for other papers by Marcin J. Szumowski in
Current site
Google Scholar
PubMed
Close
,
Robert M. RauberDepartment of Atmospheric Sciences, University of Illinois at Urbana–Champaign, Urbana, Illinois

Search for other papers by Robert M. Rauber in
Current site
Google Scholar
PubMed
Close
,
Harry T. Ochs IIIDepartment of Atmospheric Sciences, University of Illinois at Urbana–Champaign, Urbana, Illinois, andIllinois State Water Survey, Champaign, Illinois

Search for other papers by Harry T. Ochs III in
Current site
Google Scholar
PubMed
Close
, and
L. J. MillerNational Center for Atmospheric Research, Boulder, Colorado

Search for other papers by L. J. Miller in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Radar reflectivity factors exceeding 60 dBZ are documented within shallow (<3 km), warm (>0°C), summertime tropical rainbands offshore of the island of Hawaii. Dual-Doppler radar measurements from the Hawaiian Rainband Project are used to document the formation, evolution, and kinematic structure of the high reflectivity cores. The authors show that extremely high radar reflectivities (50–60 dBZ) can develop from echo free regions (−20 dBZ) within approximately 15 min and are preceded by 5–9 m s−1 peak updrafts. High reflectivities (>50 dBZ) typically first formed in the middle or upper part of the clouds. Over the next 10–15 min, the mature high reflectivity cores extended vertically through the cloud depth and then collapsed to the surface as the updrafts weakened. A near-upright orientation of most updrafts producing these high reflectivity cores is conceptually consistent with the idea that large raindrops grow in the highest liquid water content while falling through the updraft core. Strong outflows near the inversion led to the formation of sloped radar echo overhangs surrounding the cells. The bases of the overhangs descended to the surface with time, leading to an overall increase in the width of the rainbands. Short-lived downdrafts were present in the upper part of the clouds in mature and dissipating stages of cells’ life cycles but were not observed in the lower parts of the cloud, even in intense precipitation shafts.

Corresponding author address: Marcin J. Szumowski, Department of Atmospheric Sciences, University of Illinois at Urbana–Champaign, 105 S. Gregory, Urbana, IL 61801.

Email: marcin@atmos.uiuc.edu

Abstract

Radar reflectivity factors exceeding 60 dBZ are documented within shallow (<3 km), warm (>0°C), summertime tropical rainbands offshore of the island of Hawaii. Dual-Doppler radar measurements from the Hawaiian Rainband Project are used to document the formation, evolution, and kinematic structure of the high reflectivity cores. The authors show that extremely high radar reflectivities (50–60 dBZ) can develop from echo free regions (−20 dBZ) within approximately 15 min and are preceded by 5–9 m s−1 peak updrafts. High reflectivities (>50 dBZ) typically first formed in the middle or upper part of the clouds. Over the next 10–15 min, the mature high reflectivity cores extended vertically through the cloud depth and then collapsed to the surface as the updrafts weakened. A near-upright orientation of most updrafts producing these high reflectivity cores is conceptually consistent with the idea that large raindrops grow in the highest liquid water content while falling through the updraft core. Strong outflows near the inversion led to the formation of sloped radar echo overhangs surrounding the cells. The bases of the overhangs descended to the surface with time, leading to an overall increase in the width of the rainbands. Short-lived downdrafts were present in the upper part of the clouds in mature and dissipating stages of cells’ life cycles but were not observed in the lower parts of the cloud, even in intense precipitation shafts.

Corresponding author address: Marcin J. Szumowski, Department of Atmospheric Sciences, University of Illinois at Urbana–Champaign, 105 S. Gregory, Urbana, IL 61801.

Email: marcin@atmos.uiuc.edu

1. Introduction

Beginning with small-scale observational projects (Mordy and Eber 1954) and the pioneering work of Squires (1952a,b; 1958a,b), there has been an increasing awareness of the significance of warm rain processes as a major contributor to total rainfall, especially in the Tropics. The Big Island of Hawaii is a natural laboratory for studying warm rain processes. The location of the island guarantees both warm temperatures and maritime air characteristics, while the trade wind inversion and island lifting provide an ideal environment for formation of shallow convective clouds that are devoid of ice processes. Trade wind convection forms regularly, often organizes into offshore rainbands (Austin et al. 1996), and is responsible for much of the precipitation at Hilo (the wettest city in the United States) and on the windward slopes of the island. Many field experiments have been carried out on the Big Island. The warm rain project in Hawaii in 1965 (Lavoie 1967) investigated the dynamic and microphysical characteristics of Hawaiian clouds. Semonin et al. (1968) used data from two X- band radars to determine climatological parameters associated with trade wind showers and developed positive correlations between the frequency of echoes and the total rainfall. Rogers (1967) observed reflectivity values up to 50 dBZ in convective Hawaiian orographic clouds with a vertically pointing 3-cm Doppler radar. He estimated the most intense updrafts within these clouds to be 7 m s−1 and related these updrafts to substantially increased rainfall rates (up to 40 mm h−1) compared to the nonconvective cases (10 mm h−1). Rogers found good agreement between the location of the strongest updraft and the highest echo intensity. He observed that a transition from a convective to stratiform character often took place in the dissipating stage of a given shower. Drop size distributions were deduced only for stratiform rain. All of his measurements were made over land. Additional drop size distributions and corresponding ZR relationships were provided by Fujiwara (1967). Fujiwara’s observations were limited to surface measurements in orographic precipitation and did not include large raindrops or high rainfall rates. In a later project, Takahashi (1981) concluded from aircraft observations that major raindrop growth in rainbands over the ocean occurred near cloud top, contrary to Rogers’ (1967) earlier radar measurements over land. Drop spectra reported by Takahashi were limited to sizes <1.5 mm. Measurements of drop size distributions from these projects, along with model and laboratory results (List and Gillespie 1976; Takahashi 1977; Low and List 1982) were consistent with the idea that collisional breakup rapidly destroys larger drops in natural clouds and generally limits raindrop diameters to a maximum of about 2.5–3.0 mm.

Aircraft data collected during the Joint Hawaiian Warm Rain Project (JHWRP) in 1985 (Beard et al. 1986; Takahashi et al. 1989; Rauber et al. 1991) revealed unexpected microphysical characteristics of Hawaiian rainbands. Raindrops with equivalent spherical diameters as large as 8 mm were recorded in rainshafts of some of the more vigorous cells. Radar reflectivities in excess of 55 dBZ were calculated from measured drop size distributions (Beard et al. 1986; Rauber et al. 1991).

The Hawaiian Rainband Project (HaRP), conducted in July and August of 1990, used aircraft, radar, surface, and upper-air measurements to investigate the formation and dynamical and microphysical structure of rainbands offshore of the Big Island. One goal of HaRP was to study the warm rain process by relating the microphysical evolution of these rainbands to their dynamical structure. HaRP employed the 5-cm wavelength National Center for Atmospheric Research (NCAR) CP3 and CP4 Doppler radar systems as well as the NCAR Electra aircraft. Figure 1 shows the location of the seaward 30° dual-Doppler lobe on the windward side of the island and a typical Electra flight track, including a sounding and a rainband penetration. Similar flight patterns were used in 1985 during JHWRP except that soundings were taken closer to the island. This paper is the first of a series from our group addressing the microphysical objectives of HaRP. In this paper, we focus on the evolution of the structure and kinematics of rainbands containing high reflectivity cores associated with large raindrops. In a later paper (second of the series), we will relate the radar measurements of high reflectivity cores to their microphysical structure, as determined from the JHWRP aircraft data. Future papers in the series will focus on modeling studies.

2. Radar analyses

a. Methodology

The radar baseline during HaRP was 17.5 km long (Fig. 1) and oriented parallel to the shore line. This placed the seaward dual-Doppler lobe in a position to observe large portions of the rainbands forming over the ocean and moving onshore with the trade winds. The scanning procedure involved a 15-min sequence of coplane, surveillance, and Range-Height Indicator (RHI) scans. The sequence typically started with a low angle (0.5°) 360° surveillance scan followed by five coplane volume scans, each taking approximately 2.5 min. The coplane scans were adjusted to attempt to keep the highest coplane above the echo tops. The coplane volume scans were followed by a three-level surveillance scan (0.5°–4.5°) and RHI scans. The entire sequence was timed to begin at each quarter hour and was repeated as long as significant echoes were present.

The radar equivalent reflectivity factor (hereafter called reflectivity) and radial velocity fields were interpolated to a Cartesian grid using the NCAR/Mesoscale and Microscale Meteorology Division (MMM) Sorted Position Radar Interpolation program (SPRINT; Mohr and Vaughan 1979; Miller et al. 1986). The radar analysis domain (shown in Fig. 1) was normally a 45 km × 45 km area extending 4.4 km in the vertical (the exception was 22 July, when a 39 km × 45 km × 4.6 km domain was used to capture the highest echoes). The horizontal grid spacing was 0.3 km, and the vertical spacing was 0.2 km. Editing, dual-Doppler syntheses of the horizontal winds, calculation of derived fields, and creation of graphical displays of the data were done with the NCAR/MMM Custom Editing and Display of Reduced Information in Cartesian space program (CEDRIC; Mohr et al. 1986). In addition, the NCAR/MMM Planned Position Indicator program was used to examine the radar measurements at the original sampling resolution.

Folding, noise, second trip echoes, ground and sea clutter, sidelobe effects, and other sources of error all affect derived horizontal and vertical motion estimates in dual-Doppler analyses. In this analysis, noise was eliminated by applying a threshold based on the signal-to-noise power ratio (SNR). Ground clutter was virtually nonexistent in the analysis domain over the ocean. Sea clutter from side lobes at low elevation scans was minimized by locating the radars so that the lower sidelobes were intercepted by vegetation. Second trip echoes were removed by comparing reflectivity fields from each radar at each point in the Cartesian grid and accepting only those grid points where both radars measured the SNR above the threshold. No velocity folding occurred because all wind speeds were significantly less than the Nyquist velocity (20.4 m s−1 for CP3 and 20.6 m s−1 for CP4).

Small errors in the recorded antenna azimuth angle were typical in HaRP. These errors, which could degrade derived velocity estimates, were determined and corrected by performing cross correlation analyses of the reflectivity fields from both radars. These correlation analyses allowed us to determine accurately the true azimuth angle of the beam, and comparisons between the radars reinforced the validity of the reflectivity and velocity estimates in the vicinity of high reflectivity cores. Examples of such correlations for interpolated volume scans from 22 July (near the beginning of HaRP) and 22 August 1990 (the last day of HaRP) are shown in Figs. 2a and 2b. The correlation coefficients at the optimum azimuth angle corrections were approximately 95% with a standard error of 2.5 dBZ. The sensitivity of these correlations to changes in azimuth correction angles <0.5° was small. The slope of the correlation line was dominated by low reflectivity values; however, a distinct change in slope is apparent above 40 dBZ. Each radar employs separate low and high power receivers to increase its dynamic range. These comparisons point to slight differences in the operation of these receivers. The slope of the linear interpolation for reflectivities above 40 dBZ in general was closer to the perfect correlation (1:1) line. Generally, the maximum value of reflectivity simultaneously observed by each radar within a high reflectivity core differed by no more than 2–4 dBZ. The CP3 low power receiver was less sensitive on average for reflectivities between 20 and 40 dBZ, leading to a slope of the correlation line of less than one. The sensitivities of kinematic fields to varying azimuth correction angles were also tested and the results indicate that within the dual-Doppler lobe changes of less than 1 degree in the azimuth correction angle resulted in maximum vertical motion variations of less then 0.5 m s−1. This is less than estimates of error from other sources, as discussed below.

Three-dimensional kinematic fields were obtained by synthesizing the radial velocity measurements from the two radars within the analysis domain using standard procedures (Miller and Strauch 1974; Ray et al. 1978; Ray et al. 1980; Miller et al. 1983). The radial velocity estimates were advected to a common time, generally close to the middle time of each coplane volume scan. The advection speed and direction were determined from animations of the reflectivity field. Often high reflectivity cores were difficult to track because of their very short life cycles. In such cases the echo advection speed was determined based on both estimates of echo movement and the mean wind averaged over the radar analysis domain at an elevation near the middle level of the rainband.

Vertical motion was calculated from the divergence field at each level using the anelastic continuity equation, prescribed boundary conditions, and the variational integration method (Chong and Testud 1983). The variational integration method was chosen because the stable inversion layer provides justification of the w = 0 upper boundary condition above echo tops. Uncertainties in radial velocity estimates from the HaRP radars (0.3 m s−1) resulted in maximum errors of approximately 1 m s−1 in the derived vertical motion fields near the center of the analysis domain.

Additional error resulted from sidelobe contamination. Sidelobes caused particular difficulty in estimating cloud top elevations in HaRP. Vertical stretching above locations of intense reflectivity due to the first sidelobe was significant, often causing an increase in apparent cloud top elevation of several hundred meters. Sidelobe return from the very strong echoes in the cloud also created artificial divergence above the true cloud top. Intuitively, sidelobe effects should tend to stretch updraft cores upward and slightly increase the elevation and the magnitude of the peak updrafts, but based on our tests, this was not always the case. Although it remains unclear how the sidelobe contamination affects the shape and magnitude of calculated updrafts and downdrafts, it was desirable to remove that contamination near cloud top regions by prescribing the boundary condition near the trade wind inversion since the inversion limits the vertical development of the shallow clouds observed in HaRP. The upper boundary condition of w = 0 was chosen to be one or two grid points (0.2–0.4 km) above the estimated base of the trade wind inversion. This estimate was compared with vertical cross sections of the reflectivity field to ensure that reflectivities greater than 35 dBZ were not present above this level. Reflectivities >35 dBZ cannot be attributed to sidelobes unless the maximum reflectivity in the cloud significantly exceeds 60 dBZ. The vertical velocity at the lower boundary of the radar data was difficult to estimate because the radar did not sample the layer of air between the ocean surface and the lowest coplane angle. The lower boundary condition was estimated by assuming that the measured convergence in the lowest layer of radar data also applied to the unobserved layer immediately below it and that the vertical velocity at the base of the unobserved layer was zero. The validity of this assumption degrades as the height of the lowest coplane increases with distance away from the radar. We estimate from the above considerations, time continuity of the vertical motion fields inherent in all of our analyses, and limited comparisons with aircraft data that the vertical motion estimates inside the 30° dual-Doppler lobe are accurate to within about ±1.5 m s−1. The vertical motions obtained from the syntheses are average values over the grid volume. Peak vertical velocities are reduced by both pulse volume and grid volume averaging inherent in radar processing.

All cross sections presented in this paper display fields interpolated to a Cartesian grid. Interpolation decreased the maximum value of reflectivity observed in the coplane measurements by up to 6 dBZ. Typically, the reduction due to interpolation was about 2–3 dBZ. In this paper, reflectivity values obtained from interpolations to Cartesian space are denoted as dBZi; however, the peak reflectivity attained by a cell during its lifetime is reported from the original data in coplane space and is denoted by dBZr.

For convenience, we use the following terminology: the term rainband will refer to either a continuous or broken elongated region of precipitation (radar reflectivity); a cell will refer to a locally higher region of radar reflectivity, either isolated or embedded within a rainband; and a high reflectivity core will refer to a local region of reflectivity exceeding 50 dBZi. Typically, the diameter of a high reflectivity core is 1–2 km and that of a cell containing a high reflectivity core is 5–10 km. Universal time (UTC) is used throughout the paper (UTC = Hawaii Standard Time + 10 h).

b. Spatial and temporal relationship between reflectivity and vertical motion fields

The spatial and temporal relationship between the vertical motion field and the reflectivity field is determined by the lag time between the formation of the updraft and the subsequent growth and fallout of precipitation. At any given time, the reflectivity represents a snapshot of the spatial distribution of hydrometeors that formed as a direct result of the cell’s kinematic structure a few minutes earlier. Based on our radar analyses, 5–7.5 min typically passed between a local maximum in vertical motion and a corresponding maximum in the reflectivity field. This time lag is consistent with the concept that raindrops grow most rapidly to large sizes when suspended within a region of strong updraft and then fall through the cloud as their terminal velocities exceed the updraft speed. Complications arise because hydrometeors are transported by the horizontal flow as they remain suspended, rise, or fall within the cloud. Furthermore, the peak value of reflectivity observed during the lifetime of an individual cloud generally (but not always) occurred at elevations between 0.8 and 1.6 km, while updraft maxima were located between 2 and 3 km.

Temporal and spatial offsets between reflectivity and updraft maxima can be illustrated by examining a high reflectivity core that developed within a rainband on 22 July. At 0616 UTC the larger of two high reflectivity cores within the rainband (shown in Fig. 3a at an elevation of 1 km) was not accompanied by a local updraft maximum. However, if we advect the vertical motion field from 0611 UTC forward by 5 min using the mean cell motion of 4.4 m s−1 and compare it to the same horizontal cross section of the reflectivity field at 0616 UTC (Fig. 3b), we see a clear correspondence between the two fields. The 35-dBZi contour aligns nearly exactly with the 2.5 m s−1 contour and the peaks in reflectivity are nearly coincident with the peaks in updraft. The reflectivity maximum at 0616 UTC was located 1000 m below the radar-derived updraft maximum at 0611 UTC. A vertical distance of 1000 m over this time period implies a 4 m s−1 mean differential fall speed of the larger raindrops (the mean updraft speed subtracted from the mean terminal velocity of the larger raindrops). In the case studies that follow, we use this advective technique to relate the spatial structure of high reflectivity cores to their parent updrafts.

3. Observations

In this section, we illustrate typical spatial and temporal relationships between rainbands and the high reflectivity cores embedded within them, discuss typical environmental conditions that support the development of high reflectivity cores, and make inferences concerning the microphysical structure of the cores based on the evolution of the reflectivity and kinematic fields. During HaRP, reflectivities exceeding 60 dBZr occurred on three days (22 July, 10 August, and 22 August 1990). We present a detailed analysis of the 22 August event. Aspects of the 10 August and 22 July case studies that contribute additional insight into the development of high reflectivity cores are also presented.

a. 22 August 1990

A sequence of four horizontal cross sections of the radar reflectivity field between 1801 and 1901 UTC on 22 August (Fig. 4) illustrates the evolution of a rainband that originated approximately 60 km offshore. At 1801 UTC (Fig. 4a), a broken rainband moved into the radar analysis domain from the east. Vigorous cells developed along the band by 1838 UTC (Fig. 4b). At 1848 UTC (Fig. 4c), the high reflectivity core in the southern cell in box I reached a peak reflectivity of 62.7 dBZr at 1.2 km. The cells later merged to form a continuous rainband as they moved toward shore (Fig. 4d). High reflectivity cores were only present during the earlier part of the rainband’s life cycle. This case study is of particular interest because one cell developed the highest reflectivity observed during HaRP (62.7 dBZr), while another cell developed very high reflectivity (60.5 dBZr) at the trade wind inversion level near cloud top.

Figure 5 shows the 22 August aircraft sounding taken 100–160 km upstream of the island at about 1630 UTC, 1.5 h before the data shown in Fig. 4a were collected. The mixed layer on 22 August extended from the surface to 960 mb. The lifting condensation level was 935 mb (670 m) and the level of free convection was 880 mb (1170 m). The equilibrium level for buoyant parcels rising from the mixed layer was 722 mb (2760 m). A very strong inversion (4°C/200 m), located between 725 and 705 mb (2735 and 2955 m), capped the trade wind layer. Maximum radar echo tops occurred near 3400 m, leading to a maximum cloud depth of about 2750 m. The convective available potential energy (CAPE) was 98 J kg−1 with maximum buoyancy realized just below the inversion. The maximum updraft estimated from the CAPE was 14.0 m s−1, while the maximum updraft estimated from radar wind synthesis was 8.8 m s−1. The large difference between the equilibrium level (measured 150 km from the island) and the radar echo tops (measured ∼30 km from the island) is due to both overshooting of the convective updrafts and an increase in the inversion height toward the island, a feature observed in other cases where soundings closer to the island were available. Winds were easterly between 5.5 and 8.0 m s−1 in the moist trade wind layer, decreasing to 2.5 m s−1 just above the inversion.

We focus on two cells, designated cell 1 and cell 2 in Fig. 6 (also see box I in Fig. 4c). Figure 6a shows a horizontal cross section of the reflectivity field at 1.2 km, the altitude of the maximum reflectivity observed in cell 1 at 1848 UTC. Figure 6b shows a similar cross section at 2.8 km, the elevation of the maximum reflectivity observed in cell 2. The horizontal dimension of both high reflectivity cores was about 1 km. The reflectivity and vertical motion fields in cells 1 and 2 evolved differently, as illustrated in Fig. 7. The updraft in cell 1 first intensified to a peak of 7.9 m s−1 and then decayed to 4.4 m s−1 over the next 7.5 min. During this 7.5-min period, the peak reflectivity intensified from 44 dBZi to a maximum value of 57 dBZi (62.7 dBZr). The maximum updraft in cell 2 was 7.8 m s−1 when first observed at 1838 UTC. The maximum updraft decreased to 5.8 m s−1 over the next 2.5 min and then reintensified to 7.6 m s−1. Over the next 5 min, the updraft slowly decayed to 6.4 m s−1. This double-peaked structure in updraft intensity also appeared, with a 5 min time lag, in the reflectivity field (Fig. 7). The reflectivity in cell 2 remained below 50 dBZi until 1843 UTC. At 1843 UTC, 5 min after the first updraft maximum occurred, the peak reflectivity rose to 53 dBZi. Following this time, the maximum reflectivity decreased to 50 dBZi, then increased again to 58 dBZi (60.5 dBZr), the maximum value observed during the cell’s lifetime (Fig. 9).

The first important structural difference between cells 1 and 2 was the location of the high reflectivity at the time of their peak intensity. The reflectivity maximum in cell 2 was 1.6 km higher than in cell 1. The high reflectivity core in cell 1 (Figs. 8a,b) extended vertically from the surface to 2.2 km, with the maximum at 1.2 km. The reflectivity near the inversion level varied from 20 to 35 dBZi directly above the high reflectivity core. In cell 2 (Figs. 9a, b), the high reflectivity core extended from 2.4 to 3.2 km with a maximum of 60.5 dBZr located at 2.8 km. The maximum reflectivity in cell 2 was located 200–400 m below the inversion level. Note that the inversion was measured at 2.8 km by the aircraft 160 km upstream of the island and estimated at 3.2 km in the vicinity of these cells.

Based on aircraft observations of raindrop concentrations and size spectra in Hawaiian rainbands (Beard et al. 1986; Rauber et al. 1991), raindrops with diameters 4 mm and greater are necessary to produce the high reflectivities observed in these clouds. The terminal velocity of raindrops ranges from 6.5 m s−1 for a 2-mm diameter drop to 8.8 m s−1 for a 4-mm drop. As shown in Fig. 7, the updraft in cell 1 decayed rapidly after achieving its peak updraft intensity. In contrast, the updraft in cell 2 had two maxima and decayed slower. Because of the reduction in updraft intensity, raindrops contributing to the reflectivity in cell 1 would have fallen from higher altitudes as they grew to large sizes. In cell 2, the updraft was apparently strong enough to keep these raindrops suspended in the cloud top region as they grew. Cell 2 had to have an updraft of at least 9 m s−1 just below the inversion to suspend raindrops near cloud top for the time period required to grow to sufficiently large sizes to produce the observed 60.5 dBZr reflectivity. The maximum theoretical (adiabatic) updraft speed within the clouds, as determined from the aircraft sounding, was 14 m s−1. The true updraft speed in the vicinity of the high reflectivity core was probably closer to an average of the radar-derived updraft, which is an underestimate due to volume averaging and integration procedures, and the adiabatic value, which is generally an overestimate because it neglects mixing and precipitation loading. The value of this average, about 10–11 m s−1, is sufficiently large to account for the suspension of large raindrops within the updraft near the inversion and the extremely high reflectivity observed near the cloud top.

The second important structural difference between cells 1 and 2 was the relative position of the high reflectivity core and the corresponding updraft. Since the position of the cell 1 high reflectivity core was in the middle of the cloud, raindrops contributing to the high reflectivity in cell 1 were likely to have been located in the upper part of cell 1 approximately 5 min earlier (assuming an approximate 4 m s−1 differential fall speed for the larger raindrops). We therefore will compare the reflectivity structure in cell 1 with its updraft calculated 5 min earlier and advected forward using the mean cell motion. In contrast, the raindrops contributing to the peak reflectivity in cell 2 must have grown while suspended in the cloud top region. Since the highest reflectivity in cell 2 was closer to the top of the cloud, and larger raindrops that formed near cloud top would not have fallen as far, we will compare its structure with the updraft calculated from 2.5 min earlier and advected forward with the mean cell motion. The contours overlaid on the 1848 UTC reflectivity fields in Figs. 8 and 9, respectively, represent the position of the two updrafts, advected forward 5.0 and 2.5 min with the mean cell motion. The updraft contours are at every 1 m s−1 beginning at 4.5 m s−1. Taking into account the time difference for growth and fallout of the raindrops, the updraft in cell 1 was located directly above the high reflectivity core. In cell 2, however, the reflectivity core is displaced about 1 km northwest of the updraft core. The reason for this displacement can be explained by examining the kinematic fields in the vicinity of the high reflectivity cores (Figs. 10–12).

Figure 10 shows a horizontal cross section of the cell-relative flow overlaid on the vertical motion field at 2.4 km. The 50-dBZi reflectivity contours from the 1.2 and 2.8 km elevations are also shown for cell 1 and cell 2, respectively. The vertical cross sections through cells 1 and 2 at the positions indicated in Fig. 10 are shown in Figs. 11 and 12. The cell-relative streamlines in both the band-normal and band-parallel cross sections through cell 1 at 1843 UTC (Figs. 11a,b) indicate vertical ascent of the air through the main updraft core and approximately symmetric outflow on either side of the updraft near cloud top. This flow pattern evolved over the next 5 min so that the cell-relative outflow remained nearly symmetric while the updraft tilted slightly toward the southwest (Figs. 11c,d). The cell-relative streamlines in both the band-normal and band-parallel cross sections together indicate that the flow through the updraft in cell 2 was tilted slightly toward the northwest and that the outflow near the inversion was more asymmetric than in cell 1. A much broader updraft was present in the band-parallel direction in cell 2 (Figs. 12b,d). The outflow tended to be concentrated toward the northwest. At 1848 UTC, the updraft in cell 2 was noticeably tilted toward the northwest. Figures 10, 12c, and 12d together show that the outflow from cell 2 was concentrated toward the west-northwest.

The temporal evolution of the updraft intensity and the cell-relative flow through the updraft column shown in Figs. 11a–d suggests that the largest drops primarily fell through the updraft region as it weakened fairly rapidly between 1843 and 1848 UTC (see Fig. 7). Thus the reflectivity maximum was located in the midportion of the cloud (Fig. 8). Because the airflow through the center of the updraft was nearly vertical, the high reflectivity core developed approximately underneath the location of the maximum updraft. In cell 2, the updraft was tilted toward the northwest. Small raindrops following streamlines through the updraft (Figs. 12b,d) would be displaced approximately 1 km toward the northwest by the time they reached the inversion level. Since the updraft was broad in the band-parallel direction and remained strong just below the inversion throughout the period, the growing raindrops, transported northwest in the flow, could remain suspended in the upper part of the cloud. As a result, the highest reflectivity developed high in the cloud and was displaced about 1 km northwest of the maximum updraft.

The 22 August case contrasts the development of two high reflectivity cores that had different updraft structure. The first, a more typical case, had a nearly vertical updraft core that weakened rapidly after reaching its peak intensity. This resulted in the initial formation of a high reflectivity core in the middle of the cloud. The second had a tilted updraft that maintained its strength over a longer duration. In this case, the duration of the strong updraft allowed raindrops to be supported just below the inversion for a time sufficient to explain the observation of very high reflectivities near the cloud top.

b. 10 August 1990

Two rainbands developed on the morning of 10 August (Fig. 13). Ten high reflectivity cores were identified between 1731 and 1831 UTC. The most intense cell reached a peak reflectivity of 61.9 dBZr at 1.6 km. The life cycles of two high reflectivity cores embedded in one of the rainbands (see box II in Fig. 13) are detailed in Figs. 14a–g. These are the best cases documenting the time required for a high reflectivity core to develop.

The evolution of core 1 is shown in Figs. 14a1–g1. In its early stage (a1), prior to the appearance of the high reflectivity, the maximum reflectivity was located approximately 800 m below the trade wind inversion level, suggesting that the initial precipitation formed in the upper part of the cloud. At this time there was no indication of precipitation at or just above cloud base. After 2.5 min (b1) the reflectivity region >20 dBZi extended closer to both the inversion and the surface as precipitation formed through the depth of the cloud. After 5 min (c1), the high reflectivity core appeared in the middle part of the cloud as heavier precipitation reached the ocean surface. The peak reflectivity (58.9 dBZr at 1.6 km, d1) was observed 7.5 min after a1, when the high reflectivity core extended vertically from approximately 800 m below cloud top through cloud base (d1). The elevation of the maximum reflectivity decreased from 1.6 km (d1) to 0.8 km (e1) to the surface as larger raindrops fell through the weakening updraft. Panels f1 and g1 show the dissipating stage, when the maximum reflectivity weakened. The highest reflectivity in g1 was located near the surface, while the echo aloft was weak. The evolution of core 2 (Figs. 14a2–g2) generally followed a similar pattern to core 1, except that the high reflectivity core occupied a larger volume and was more intense. Both high reflectivity cores were present for about 10 min.

Figure 15 shows the evolution of the maximum reflectivities associated with cores 1 and 2 and the altitude at which the maximum reflectivities were observed. The reflectivity in core 2 increased from no echo (approximately −20 dBZi) to 50 dBZi in 15 min and 59.7 dBZi (61.9 dBZr) in 20 min. The evolution of core 1 was similar. The altitude of maximum reflectivity generally decreased with time as the largest raindrops fell toward the surface through the decaying updraft core.

The peak calculated updrafts in the life cycles of both core 1 and core 2 were at 1736 UTC, when the maximum reflectivity reached 51.5 and 36.6 dBZi, respectively. Both updraft maxima (5.3 m s−1 and 5.6 m s−1, respectively) were at 2.2-km elevation (Figs. 16c1′, 16c2′); however, the updraft associated with core 2 was broader and retained its intensity through the next 5 min, while the updraft associated with core 1 weakened. The streamlines shown in the cross sections in Fig. 16 represent the instantaneous cell-relative flow in the plane of the cross section. Two outflow regions on either side of the main updraft were present in the upper parts of the cloud. Based on the sounding (not shown), air rising within the main updraft would attain its maximum vertical velocity close to the inversion level. Figure 17a shows vertical profiles of divergence through the center and the flank of the updraft associated with core 2. The inversion base was measured at 2.5 km by the aircraft ∼100 km upstream of the island and estimated at 2.8 km from the reflectivity pattern in the vicinity of the cell. Near the updraft core, the strongest divergence (7 × 10−3 s−1) occurred approximately 200 m above the inversion base, while ∼1 km from the core, the maximum divergence was located about 200 m below the inversion. The strongest outflow occurred close to the inversion and sloped downward away from the core of the updraft (Fig. 16c2), transporting raindrops out of the main updraft. The gradients in the reflectivity pattern, with the highest reflectivity in the core of the updraft and progressively weaker reflectivity extending outward along the inversion, reflect the fact that raindrops continued to grow in the updraft core, while raindrops ejected from the updraft in the outflow remained small. The regions of weak (<35 dBZi) reflectivity extending outward from the main core at approximately 2.5 km elevation represent the small raindrops “fountained” out of the main updraft in the outflow. The “mushroom” shape of the reflectivity field, the spreading just below the stable inversion layer, and the bottom slope of the reflectivity overhang are all characteristic of this process. The sloped base is caused by the larger ejected raindrops falling rapidly out of the strong outflow layer, while smaller raindrops fall more slowly and are carried outward farther from the updraft. In the maturing stage (Figs. 14e–g), the “mushroom” vertical reflectivity structure characteristic of the earlier stages disappeared as precipitation outside the updraft fell toward the surface, leading to an overall broadening of the rainband containing the cells.

c. 22 July 1990

The 22 July case had the lowest CAPE (36.7 J kg−1) and weakest radar-derived maximum updraft speed (4.6 m s−1) of the three cases where the maximum reflectivity exceeded 60 dBZr. In this case, the most intense high reflectivity core developed in a rainband approximately 15 km offshore (Fig. 3). Vertical cross sections of reflectivity along A-A′ in Fig. 3, shown in Figs. 18a and 18b, further illustrate the relationship between the vertical structure and the time evolution of its parent updraft. The overlaid vertical motion contours in Figs. 18a and 18b are from 7.5 and 5 min earlier, respectively. To obtain the vertical motion contours in Fig. 18a, the vertical motion field from 7.5 min earlier was advected forward at the speed and direction of the cell motion. A vertical cross section at A-A′ was then taken through the advected field. The same was done in Fig. 18b, except that the vertical motion field from 5 min earlier was used. A close correspondence between the high reflectivity core and the updraft core is evident in both figures. The lower portion of the high reflectivity core is directly below the location of the strongest updraft advected forward from 7.5 min earlier, while the upper portion of the high reflectivity core is directly below the location of the strongest updraft advected forward from 5 min earlier. This downstream tilt of the high reflectivity core at 1616 UTC is consistent with larger raindrops falling with a relative velocity of approximately 3–4 m s−1 from the evolving updraft core. The increasing width of the high reflectivity core with height is consistent with the increasing width of the updraft core with time. The divergence profile in the updraft core (Fig. 17b) and the mushroom shape of the reflectivity field in the cross-band direction in Fig. 18 are similar to that observed on 10 August.

4. Characteristics of rainbands and their environment

Environmental variables, stability parameters, and quantities derived from the soundings and radar analyses for the three case studies are summarized in Tables 1 and 2. The environment near the windward shore of the Big Island of Hawaii is typically characterized by 1) a 300–400-m-deep mixed surface layer; 2) a lifting condensation level near 600–700 m; 3) cloud depth between 1.5 and 3.5 km, with clouds top temperatures >5°C; 4) an isothermal layer or an inversion, with base between 2 and 4 km and a dry layer above, that limits vertical development of convection; 5) a conditionally unstable environment within the marine layer below the inversion. The CAPE calculated in the three cases ranged from 35 to 100 J kg−1, allowing for maximum (adiabatic) updrafts between 8.6 and 14.0 m s−1. Vertical wind shear in the marine layer below the trade wind inversion was generally less than 2 m s−1 km−1. Updrafts were generally nearly vertical and the band-relative outflow just below the inversion was nearly symmetric. The only exception in the case studies was cell 2 on 22 August, which developed a 30° tilt and a directional outflow. In this case, the updraft just below the inversion was sufficiently vigorous and broad in the direction of the tilt so that large raindrops could still remain suspended and grow large enough to produce the high reflectivity.

Rainbands were the typical organization of precipitation upstream of the island (Austin et al. 1996). Rainbands first appeared on radar as continuous lines of low reflectivity with embedded cells of higher reflectivity or as lines of cells separated by echo free regions. Typically, the strongest updrafts and highest reflectivities within rainbands were found to occur as rainbands organized and intensified. As the rainbands evolved, updrafts spread laterally and weakened, producing more continuous lines of nearly uniform precipitation. The capping effect of the inversion apparently lead to locally strong outflow regions centered above the strongest updrafts. The strong outflow near the inversion led to the formation of a radar echo overhang surrounding the cells. With time, the base of the overhang descended to the surface, leading to an overall increase of the width of the rainband.

Reflectivities exceeding 50 dBZ were always associated with the strongest updrafts. The horizontal dimensions of the updraft and the high reflectivity core were typically 1–2 km, occupying a small volume fraction of the entire precipitating rainband. The time from the appearance of the first radar echo to the development of a high reflectivity core was approximately 15 min. An intensifying updraft core was first detected approximately 10 min prior to the appearance of the corresponding high reflectivity core. Generally, reflectivities between 20 and 35 dBZ were observed between 1 and 2.5 km above the ocean surface when a local updraft maximum first appeared. The peak updraft always preceded the appearance of peak reflectivities by approximately 5 min, and remained close to its peak speed until the high reflectivity core first appeared. The high reflectivity core typically first appeared in the middle part of the cloud. It then expanded both horizontally and vertically in the minutes following its formation, reaching its peak intensity as the updraft weakened. As the locally strong updraft weakened, the high reflectivity core collapsed to the surface. The dissipating stage of a high reflectivity core typically lasted about 5 min. The lifetime of an individual high reflectivity core was about 10–15 min. The magnitude of the peak reflectivity in a high reflectivity core was not necessarily proportional to the magnitude of the peak updraft. Despite the range of Doppler-derived peak updraft speeds (4.5–8.8 m s−1) and associated CAPE (35–100 J kg−1), the peak reflectivities observed in the three cases were very similar, ranging from 60.2 to 62.7 dBZr. On 10 August a short-lived updraft maximum of 5.6 m s−1 lead to the development of 61.9 dBZr peak reflectivity, while on 22 August a cell with an updraft core of nearly 9 m s−1 (not shown in the figures) was not observed to exceed 55 dBZr. Two other cells with peak updrafts of 7.8 and 7.9 m s−1 reached reflectivities of 60.5 and 62.7 dBZr, respectively.

Downdrafts with magnitudes >1.5 m s−1 synthesized from the radar data were short-lived and present only in the upper part of the clouds in mature and dissipating stages of cells’ life cycles. The strongest downdrafts were observed at the inversion level on 22 August when “overshooting” convective turrets entrained very dry air just above the base of the inversion. Downdrafts were not observed in the lower parts of the clouds, even in intense precipitation shafts (an observation consistent in both radar and aircraft data). Apparently, little evaporation occurred within the humid trade wind environment. Precipitation loading may have suppressed the updraft intensity, but there was no indication that it led to the formation of low-level downdrafts.

5. Microphysical implications

The radar analyses allow us to develop hypotheses concerning the microphysical evolution of the rainbands. The close correspondence between the shape and location of the high reflectivity core and its parent updraft implies that large raindrops, which contribute significantly to the magnitude of the reflectivity, began developing within the strongest updrafts several minutes prior to the observation of the highest reflectivities. Based on aircraft measurements in Hawaiian rainbands (Beard et al. 1986; Rauber et al. 1991), raindrops with diameters of 4 mm and greater were present in the high reflectivity cores. Raindrop terminal velocities range from 4.0 to 8.8 m s−1 for 1 to 4 mm diameter drops. Cells must maintain updrafts with speeds greater than these terminal velocities to suspend, or carry upward, raindrops of these sizes. Growing raindrops (1–2-mm diameter) present near the top of rising cloud turrets in the cases described were easily suspended in the strengthening, nearly vertical updrafts, based on the magnitude of the derived vertical motions in the cells. For larger raindrops, the mean differential fall speed (mean terminal velocity less the mean updraft speed) between the time of the peak updraft and the peak reflectivity was about 4 m s−1. Based on the evolution of the reflectivity and updraft fields, raindrops that grew to the largest sizes and contributed most to the high reflectivity were suspended within the strongest updrafts, where one would expect to find the highest liquid water content.

High reflectivity cores typically first appeared in the middle part of the cloud (the exception was cell 2 on 22 August), shortly after the cloud reached its maximum height and vertical velocity. The radar data implied that large raindrop formation continued throughout the middle and upper part of the clouds for a period of several minutes after updrafts reached their peak intensities. The descent of the base of the high reflectivity core during this time implies that the larger raindrops present in the middle of the clouds fell through cloud base to the surface. Since the maximum reflectivity remained in the middle of the cloud during this period, new raindrops, which had formed near cloud top, must have fallen through the updraft to cloud middle levels. These raindrops, presumably exposed to the largest cloud liquid water content, were responsible for the extremely high (>60 dBZ) reflectivities observed in the most intense part of the high reflectivity cores’ life cycles. Mature high reflectivity cores often extended to the uppermost parts of the cloud and strong vertical reflectivity gradients were often present near the inversion (cloud top). Sharp gradients in reflectivity near cloud top implied that this region supported rapid raindrop growth. As the updrafts weakened, larger raindrops fell rapidly to the surface, leaving smaller raindrops and drizzle in the previous location of the high reflectivity core. At the same time, smaller raindrops previously ejected from the primary updraft core in the reflectivity overhang near cloud top also fell to the surface, creating a broader area of weaker reflectivity.

6. Summary

The radar analyses presented in this paper provide the first available four-dimensional view of the structure and kinematics of shallow trade wind cumuli that produce heavy precipitation through warm rain processes. Important findings are that 1) extremely high radar reflectivities (50–60 dBZ) can develop from echo free regions (−20 dBZ) within approximately 15 min in clouds with depths less than 2–3 km, 2) high reflectivity cores are associated with the stronger (5–9 m s−1) radar-derived updrafts, and 3) the upright orientation of the updrafts that produce the high reflectivity cores is conceptually consistent with the idea that large raindrops grow in the highest liquid water content while falling through nearly vertically oriented updrafts. The data presented here confirm the earliest Doppler measurements in trade wind clouds reported by Rogers (1967), which showed that high reflectivity regions are collocated with the maximum updrafts and that strong reflectivity gradients were common in the upper part of trade wind clouds. As noted by Takahashi (1981) and Takahashi et al. (1989), the summits of trade wind cumuli are regions of active raindrop growth, a point supported by our observation of strong reflectivity gradients near cloud top and highlighted by the maximum radar reflectivity (60.5 dBZr) observed just below cloud top at the level of trade wind inversion base on 22 August.

In future papers we will relate the radar analyses discussed in this paper to aircraft microphysical, thermodynamic and kinematic measurements made in high reflectivity regions, and examine raindrop growth processes within Doppler-derived and model-generated kinematic fields.

Acknowledgments

We thank everyone at the National Center for Atmospheric Research who contributed to the Hawaiian Rainband Project data acquisition and processing. We also wish to thank Neil Laird at the Illinois State Water Survey and Jack Su at the University of Illinois at Urbana–Champaign for their contributions, and Bill Anderson of the National Center for Atmospheric Research for his assistance with SPRINT and CEDRIC. Computing time was provided by the National Center for Atmospheric Research, which is sponsored by the National Science Foundation. This research was sponsored by the National Science Foundation under Grants NSF ATM 9020245 and NSF ATM9223165.

REFERENCES

  • Austin, G., R. M. Rauber, H. T. Ochs III, and L. J. Miller, 1996: Trade wind clouds and Hawaiian rainbands. Mon. Wea. Rev.,124, 2126–2151.

  • Beard, K. V., D. B. Johnson, and D. Baumgardner, 1986: Aircraft observations of large raindrops in warm, shallow, convective clouds. Geophys. Res. Lett.,19, 991–994.

  • Chong, M., and J. Testud, 1983: Three-dimensional wind field analysis from dual-Doppler radar data. Part III: The boundary condition: An optimum determination based on a variational concept. J. Climate Appl. Meteor.,22, 1227–1241.

  • Fujiwara, M., 1967: Raindrop-size distribution in warm rain as measured in Hawaii. Tellus,19, 393–402.

  • Lavoie, R. L., 1967: Background data for the warm rain project. Tellus,19, 348–353.

  • List, R. L., and J. R. Gillespie, 1976: Evolution of raindrop spectra with collision-induced breakup. J. Atmos. Sci.,33, 2007–2013.

  • Low, T. B., and R. List, 1982: Collision, coalescence, and breakup of raindrops. J. Atmos. Sci.,39, 1591–1618.

  • Miller, L. J., and R. G. Strauch, 1974: A dual-Doppler radar method for the determination of wind velocities within a precipitating weather systems. Remote Sens. Environ.,3, 219–235.

  • ———, J. E. Dye, and B. E. Martner, 1983: Dynamical–microphysical evolution of a convective storm in a weakly sheared environment. Part II: Airflow and precipitation trajectories from Doppler radar observations. J. Atmos. Sci.,40, 2097–2109.

  • ———, C. G. Mohr, and A. J. Weinheimer, 1986: The simple rectification to Cartesian space of folded radial velocities from Doppler radar sampling. J. Atmos. Oceanic Technol.,3, 162–174.

  • Mohr, C. G., and R. L. Vaughan, 1979: An economical procedure for Cartesian interpolation and display of reflectivity data in three-dimensional space. J. Appl. Meteor.,18, 661–670.

  • ———, L. J. Miller, R. L. Vaughan, and H. W. Frank, 1986: The merger of mesoscale datasets into a common Cartesian format for efficient and systematic analyses. J. Atmos. Oceanic Technol.,3, 143–161.

  • Mordy, W. A., and L. E. Eber, 1954: Observations of rainfall from warm clouds. Quart. J. Roy. Meteor. Soc.,80, 48–57.

  • Rauber, R. M., K. V. Beard, and B. M. Andrews, 1991: A mechanism for giant raindrop formation in warm, shallow, convective clouds. J. Atmos. Sci.,48, 1791–1797.

  • Ray, P. S., K. K. Wagner, K. W. Johnson, J. J. Stephens, W. C. Bumgarner, and E. A. Mueller, 1978: Triple-Doppler observations of a convective storm. J. Appl. Meteor.,17, 1201–1212.

  • ———, C. L. Ziegler, W. C. Bumgarner, and R. J. Serafin, 1980: Single- and multiple-Doppler radar observations of tornadic storms. Mon. Wea. Rev.,108, 1607–1625.

  • Rogers, R. R., 1967: Doppler radar investigation of Hawaiian rain. Tellus,19, 433–455.

  • Semonin, F. G., E. A. Mueller, G. A. Stout, and D. W. Staggs, 1968: Radar analysis of warm rain showers. Tellus,20, 227–238.

  • Squires, P., 1952a: The growth of cloud drops by condensation I. General characteristics. Aust. J. Sci. Res.,5, 59–86.

  • ———, 1952b: The growth of cloud drops by condensation II. The formation of large cloud drops. Aust. J. Sci. Res.,5, 473–499.

  • ———, 1958a: The microstructure and colloidal stability of warm clouds. Part I. The relation between structure and stability. Tellus,10, 256–261.

  • ———, 1958b: The microstructure and colloidal stability of warm clouds. Part II. The causes of the variations in the microstructure. Tellus,10, 262–271.

  • Takahashi, T., 1977: A study of Hawaiian warm showers based on aircraft observation. J. Atmos. Sci.,34, 1773–1790.

  • ———, 1981: Warm rain study in Hawaii—Rain initiation. J. Atmos. Sci.,38, 347–369.

  • ———, K. Yoneyama, and Y. Tsubota, 1989: Rain duration in Hawaiian trade wind rainbands—Aircraft observation. J. Atmos. Sci.,46, 937–955.

Fig. 1.
Fig. 1.

Location of the seaward dual-Doppler lobe, the radar analysis area, and a typical HaRP flight profile near the Big Island of Hawaii.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 2.
Fig. 2.

(a) Radar reflectivity comparison between CP4 and CP3 on 22 July 1990. The reflectivities were measured by both radars at approximately 0616 UTC and interpolated to a Cartesian grid. This comparison was done at an altitude of 1.2 km; (b) same as (a) except on 22 August 1990 at 1845 UTC.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 3.
Fig. 3.

Horizontal cross section of the radar reflectivity field at 1.0 km measured by CP3 on 22 July 1990 at 0616 UTC. (a) Overlaid are the 2.5 and 3.5 m s−1 vertical motion contours from 0616 UTC at an elevation of 2.0 km. (b) Overlaid are the 2.5 and 3.5 m s−1 vertical motion contours from 0611 UTC at an elevation of 2.0 km advected forward with the mean cell motion to 0616 UTC.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 4.
Fig. 4.

Four horizontal cross sections of the radar reflectivity field measured by CP3 at 1.2 km on 22 August 1990 between 1801 and 1901 UTC. Panels (a) and (b) show the developing stage of the rainband, (c) the mature stage, and (d) the dissipating stage.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 5.
Fig. 5.

Thermodynamic diagram displaying the NCAR Electra sounding taken on 22 August 1990, 100–160 km upstream of the island at approximately 1630 UTC. The shaded area indicates CAPE. A barb and pennant, respectively, represent wind speeds of 1 and 5 m s−1.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 6.
Fig. 6.

Horizontal cross sections of reflectivity through the elevation of maximum reflectivity (a) for cell 1 at 1.2 km measured by CP3 and for (b) cell 2 at 2.8 km measured by CP4 on 22 August 1990 at 1848 UTC, corresponding to region enclosed in box I in Fig. 4a.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 7.
Fig. 7.

Comparison of the temporal evolution of the maximum updraft (bold) and maximum reflectivity (thin) in cells 1 (dashed) and 2 (solid) between 1838 UTC and 1848 UTC on 22 August 1990.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 8.
Fig. 8.

Vertical cross sections through cell 1 at 1848 UTC on 22 August 1990 at the locations shown in Fig. 6: (a) along the rainband and (b) across the rainband. Overlaid are the 4.5, 5.5, 6.5, and 7.5 m s−1 vertical motion contours advected with the mean cell motion from 1843 UTC. Two estimates of the trade wind inversion, based on the sounding and the weak echo region outside the main reflectivity core, are also shown.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 9.
Fig. 9.

Vertical cross sections through cell 2 at 1848 UTC on 22 August 1990 at the locations shown in Fig. 6: (a) along the rainband and (b) across the rainband. Overlaid are the 4.5, 5.5, and 6.5 m s−1 vertical motion contours advected with the mean cell motion from 1845:30 UTC. Two estimates of the trade wind inversion, based on the sounding and the weak echo region outside the main reflectivity core, are also shown.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 10.
Fig. 10.

Horizontal cross section of the vertical motion field at 2.4 km at 1848 UTC on 22 August 1990. Overlaid are horizontal cell-relative streamlines (thin lines) depicting the direction of the main outflow and 50-dBZi contours (bold) depicting the locations of high reflectivity cores in cells 1 and 2.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 11.
Fig. 11.

Vertical cross sections of vertical velocity through cell 1 on 22 August 1990: (a) across-band at 1843 UTC, (b) along-band at 1843 UTC, (c) across-band at 1848 UTC, and (d) along-band at 1848 UTC. Also depicted are the cell-relative streamlines.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 12.
Fig. 12.

Same as Fig. 11 except for cell 2.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 13.
Fig. 13.

Horizontal cross section of the radar reflectivity field measured by CP3 at 1.2 km on 10 August 1990 at 1746 UTC.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 14.
Fig. 14.

Panels (a)–(g): Horizontal cross sections of the CP3 reflectivity field (corresponding to box II in Fig. 13) between 1731 and 1751 UTC on 10 August 1990. Panels (a)–(d) [(e)–(g)] are at the altitude of maximum reflectivity in high reflectivity core 1 (2). Panels (a1)–(g1): Vertical cross sections of reflectivity at the location of maximum reflectivity in core 1; panels (a2)–(g2) Same as (a1)–(g1), but for core 2.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 15.
Fig. 15.

Altitude and intensity of the maximum reflectivity observed within core 1 and 2 (Fig. 14) between 1726 and 1751 UTC.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 16.
Fig. 16.

(c) Horizontal and vertical cross sections of the CP3 reflectivity field at 1736 UTC 10 August at locations corresponding to panel c of Fig. 14 except that cell-relative streamlines are also depicted on each cross section; (c) Horizontal and vertical cross sections of the vertical velocity field at 1736 UTC 10 August 1990. The horizontal cross section was taken at the altitude of the peak vertical motion, 2.4 km. The vertical cross sections, (c1′) and (c2′), are in the same plane as in panels c1 and c2. Cell-relative streamlines are depicted on all panels.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 17.
Fig. 17.

Vertical profiles of divergence through (a) 10 August updraft shown in Fig. 16 (c2’) and (b) 22 July updraft associated with the high reflectivity core shown in Fig. 18. Also indicated are the estimated heights of the inversion base. Updraft flank divergence profiles are 0.9 km away from the core.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Fig. 18.
Fig. 18.

Vertical cross section of the CP3 radar reflectivity on 22 July 1990 at 0616 UTC as indicated by a–a’ in Fig. 3. (a) Overlaid are the 2.5 and 3.5 m s−1 vertical motion contours advected with mean cell motion from 0608:30 UTC. (b) Same as in (a) except advected from 0611 UTC.

Citation: Journal of the Atmospheric Sciences 54, 3; 10.1175/1520-0469(1997)054<0369:TMSAEO>2.0.CO;2

Table 1.

Aircraft sounding data.

Table 1.
Table 2.

CP3–CP4 dual Doppler radar derived data.

Table 2.
Save