• Adler, R. F., and R. A. Mack, 1984: Thunderstorm cloud height–rainfall rate relations for use with satellite rainfall estimation techniques. J. Climate Appl. Meteor.,23, 280–296.

  • ——, and A. J. Negri, 1988: A satellite infrared technique to estimate tropical convective and stratiform rainfall. J. Appl. Meteor.,27, 30–51.

  • Atlas, D., C. W. Ulbrich, F. D. Marks Jr., E. Amitai, and C. R. Williams, 1999: Systematic variation of drop size and radar–rainfall relations. J. Geophys. Res.,104, 6155–6169.

  • Austin, P. M., and A. C. Bemis, 1950: A quantitative study of the“bright band” in radar precipitation echoes. J. Meteor.,7, 145–151.

  • ——, and R. A. Houze Jr., 1972: Analysis of the structure of precipitation patterns in New England. J. Appl. Meteor.,11, 926–935.

  • Biggerstaff, M. I., and R. A. Houze Jr., 1991: Kinematic and precipitation structure of the 10–11 June 1985 squall line. Mon. Wea. Rev.,119, 3034–3065.

  • ——, and ——, 1993: Kinematics and microphysics of the transition zone of the 10–11 June 1985 squall line. J. Atmos. Sci.,50, 3091–3110.

  • Brown, J. M., 1979: Mesoscale unsaturated downdrafts driven by rainfall evaporation: A numerical study. J. Atmos. Sci.,36, 313–338.

  • Byers, H. R., and R. R. Braham Jr., 1949: The Thunderstorm. U.S. Govt. Printing Office, 287 pp.

  • Chalon, J. P., G. Jaubert, F. Roux, and J. P. LaFore, 1988: The West African squall line observed on 23 June 1981 during COPT 81:Mesoscale structure and transports. J. Atmos. Sci.,45, 2744–2763.

  • Churchill, D. D., and R. A. Houze Jr., 1984: Development and structure of winter monsoon cloud clusters on 10 December 1978. J. Atmos. Sci.,41, 933–960.

  • DeMaria, M., 1985: Linear response of a stratified tropical atmosphere to convective forcing. J. Atmos. Sci.,42, 1944–1959.

  • DeMott, C. A., R. Cifelli, and S. A. Rutledge, 1995: An improved method for partitioning radar data into convective and stratiform components. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 233–236.

  • Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contribution of mesoscale convective weather systems to the warm-season precipitation in the United States. J. Climate Appl. Meteor.,25, 1333–1345.

  • Fujita, T. T., 1955: Results of detailed synoptic studies of squall lines. Tellus,7, 405–436.

  • Goldenberg, S. B., R. A. Houze Jr., and D. D. Churchill, 1990: Convective and stratiform components of a winter monsoon cloud cluster determined from geosynchronous infrared satellite data. J. Meteor. Soc. Japan,68, 37–63.

  • Hartmann, D. L., H. H. Hendon, and R. A. Houze Jr., 1984: Some implications of the mesoscale circulations in tropical cloud clusters for large-scale dynamics and climate. J. Atmos. Sci.,41, 113–121.

  • Hashem, M. S., and M. I. Biggerstaff, 1997: Organization of convection in mesoscale systems. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 483–484.

  • Houghton, H. G., 1968: On precipitation mechanisms and their artificial modification. J. Appl. Meteor.,7, 851–859.

  • Houze, R. A., Jr., 1973: A climatological study of vertical transports by cumulus-scale convection. J. Atmos. Sci.,30, 1112–1123.

  • ——, 1977: Structure and dynamics of a tropical squall-line system. Mon. Wea. Rev.,105, 1540–1567.

  • ——, 1982: Cloud clusters and large-scale vertical motions in the tropics. J. Meteor. Soc. Japan,60, 396–409.

  • ——, 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc.,115, 425–461.

  • ——, 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc.,78, 2179–2196.

  • ——, and E. N. Rappaport, 1984: Air motions and precipitation structure of an early summer squall line over the eastern tropical Atlantic. J. Atmos. Sci.,41, 553–574.

  • ——, S. A. Rutledge, M. I. Biggerstaff, and B. F. Smull, 1989: Interpretation of Doppler weather radar displays of midlatitude mesoscale convective systems. Bull. Amer. Meteor. Soc.,70, 608–619.

  • ——, B. F. Smull, and P. Dodge, 1990: Mesoscale organization of springtime rainstorms in Oklahoma. Mon. Wea. Rev.,118, 613–654.

  • Johnson, R. H., and G. S. Young, 1983: Heat and moisture budgets of tropical mesoscale anvil clouds. J. Atmos. Sci.,40, 2138–2147.

  • Klazura, G. E., and D. A. Imy, 1993: A description of the initial set of analysis products available from the NEXRAD WSR-88D system. Bull. Amer. Meteor. Soc.,74, 1293–1311.

  • Kummerow, C., and L. Giglio, 1994: A passive microwave technique for estimating rainfall and vertical structure information from space. Part I: Algorithm description. J. Appl. Meteor.,33, 3–18.

  • Lau, K.-M., and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory. J. Atmos. Sci.,44, 950–972.

  • Leary, C. A., and R. A. Houze Jr., 1979a: Melting and evaporation of hydrometeors in precipitation from the anvil clouds of deep tropical convection. J. Atmos. Sci.,36, 669–679.

  • ——, and ——, 1979b: The structure and evolution of convection in a tropical cloud cluster. J. Atmos. Sci.,36, 437–457.

  • LeMone, M. A., and E. J. Zipser, 1980: Cumulonimbus vertical velocity events in GATE. Part I: Diameter, intensity and mass flux. J. Atmos. Sci.,37, 2444–2457.

  • Lewis, J. M., 1975: Test of the Ogura-Cho Model on a prefrontal squall line case. Mon. Wea. Rev.,103, 764–778.

  • Ligda, M. G. H., 1956: The radar observation of mature prefrontal squall lines in the midwestern United States. Swiss Aero Review, No. 11–12, 1–3.

  • Lopez, R. E., 1973: Cumulus convection and larger scale circulations II. Cumulus and mesoscale interactions. Mon. Wea. Rev.,101, 856–870.

  • Malkus, J. S., 1962: Large-scale interactions. The Sea: Ideas and Observations in Progress in the Study of the Seas, M. N. Hill, Ed., Vol. 1, Interscience Publishers, 88–294.

  • Matejka, T., and T. J. Schuur, 1991: The relation between vertical air motions and the precipitation band in the stratiform region of a squall line. Preprints, 25th Int. Conf. on Radar Meteorology, Paris, France, Amer. Meteor. Soc., 501–504.

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

  • Newton, C. W., 1950: Structure and mechanism of the prefrontal squall line. J. Meteor.,7, 210–222.

  • Nitta, T., 1975: Observational determination of cloud mass flux distributions. J. Atmos. Sci.,32, 73–91.

  • Ogura, Y., and H.-R. Cho, 1973: Diagnostic determination of cumulus cloud populations from observed large-scale variables. J. Atmos. Sci.,30, 1276–1286.

  • Riehl, H., and J. S. Malkus, 1958: On the heat balance in the equatorial trough zone. Geophysica (Helsinki), 6, 503–538.

  • ——, and J. Simpson, 1979: The heat balance of the equatorial trough zone, revisited. Contrib. Atmos. Phys.,52, 287–304.

  • Rosenfeld, D., D. B. Wolff, and E. Amitai, 1994: The window probability matching method for rainfall measurements with radar. J. Appl. Meteor.,33, 682–693.

  • ——, E. Amitai, and D. B. Wolff, 1995: Classification of rain regimes by the three-dimensional properties of reflectivity fields. J. Appl. Meteor.,34, 198–211.

  • Roux, F., 1988: The West African squall line observed on 23 June 1981 during COPT 81: Kinematics and thermodynamics of the convective region. J. Atmos. Sci.,45, 406–426.

  • Rutledge, S. A., and R. A. Houze Jr., 1987: A diagnostic modeling study of the trailing stratiform region of a midlatitude squall line. J. Atmos. Sci.,44, 2640–2656.

  • ——, ——, M. I. Biggerstaff, and T. Matejka, 1988: The Oklahoma–Kansas mesoscale convective system of 10–11 June 1985: Precipitation structure and single-Doppler radar analysis. Mon. Wea. Rev.,116, 1409–1430.

  • Ryde, J. W., 1946: The attenuation and radar echoes produced at centimetre wavelengths by various meteorological phenomena. Meteorological Factors in Radio Propagation, Physical Society, 169–188.

  • Simmons, A. J., 1982: The forcing of stationary wave motion by tropical diabatic heating. Quart. J. Roy. Meteor. Soc.,108, 503–534.

  • Simpson, J., R. F. Adler, and G. R. North, 1988: A proposed Tropical Rainfall Measuring Mission (TRMM) satellite. Bull. Amer. Meteor. Soc.,69, 278–295.

  • Smull, B. F., and R. A. Houze Jr., 1985: A midlatitude squall line with a trailing region of stratiform rain: Radar and satellite observations. Mon. Wea. Rev.,113, 117–133.

  • ——, and ——, 1987: Dual-Doppler radar analysis of a midlatitude squall line with a trailing region of stratiform rain. J. Atmos. Sci.,44, 2128–2148.

  • Sommeria, G., and J. Testud, 1984: COPT 81: A field experiment designed for the study of dynamics and electrical activity of deep convection in continental tropical regions. Bull. Amer. Meteor. Soc.,65, 4–10.

  • Srivastava, R. C., T. J. Matejka, and T. J. Lorello, 1986: Doppler radar study of the trailing anvil region associated with a squall line. J. Atmos. Sci.,43, 356–377.

  • Steiner, M., and R. A. Houze Jr., 1997: Sensitivity of the estimated monthly convective rain fraction to the choice of Z–R relation. J. Appl. Meteor.,36, 452–462.

  • ——, ——, and S. E. Yuter, 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor.,34, 1978–2007.

  • ——, J. A. Smith, S. J. Burges, C. V. Alonso, and R. W. Darden, 1999: Effect of bias adjustment and rain gauge data quality control on radar rainfall estimation. Water Resour. Res.,35, 2487–2503.

  • Tao, W.-K., and J. Simpson, 1984: Cloud interactions and merging: Numerical simulations. J. Atmos. Sci.,41, 2901–2917.

  • ——, S. Lang, J. Simpson, and R. Adler, 1993: Retrieval algorithms for estimating the vertical profiles of latent heat release: Their applications for TRMM. J. Meteor. Soc. Japan,71, 685–700.

  • Tokay, A., and D. A. Short, 1996: Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. J. Appl. Meteor.,35, 355–371.

  • Toracinta, E. R., K. I. Mohr, E. J. Zipser, and R. E. Orville, 1996: A comparison of WSR-88D reflectivities, SSM/I brightness temperatures, and lightning for mesoscale convective systems in Texas. Part I: Radar reflectivity and lightning. J. Appl. Meteor.,35, 902–918.

  • Wilheit, T. T., A. T. C. Chang, M. S. V. Rao, E. B. Rodgers, and J. S. Theon, 1977: A satellite technique for quantitatively mapping rainfall rates over the oceans. J. Appl. Meteor.,16, 551–560.

  • Yanai, M., S. Esbensen, and J.-H. Chu, 1973: Determination of bulk properties of tropical cloud clusters from large-scale heat and moisture budgets. J. Atmos. Sci.,30, 611–627.

  • Yuter, S. E., and R. A. Houze Jr., 1995: Three-dimensional kinematic and microphysical evolution of Florida cumulonimbus. Part II: Frequency distributions of vertical velocity, reflectivity, and differential reflectivity. Mon. Wea. Rev.,123, 1941–1963.

  • Zipser, E. J., 1969: The role of organized unsaturated convective downdrafts in the structure and rapid decay of an equatorial disturbance. J. Appl. Meteor.,8, 799–814.

  • ——, 1977: Mesoscale and convective-scale downdrafts as distinct components of squall-line structure. Mon. Wea. Rev.,105, 1568–1589.

  • ——, and K. Lutz, 1994: The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability? Mon. Wea. Rev.,122, 1751–1759.

  • View in gallery

    Location of the Houston WSR-88D. The ellipse indicates the 150-km range limit of the data used in this study

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 1052 UTC on 2 May 1993. Intensity according to the grayscale. (b) Same as in (a) but for 1701 UTC on 22 Jun 1996. (c) SHY95 echo classification for (a). Light (dark) shading indicates convective (stratiform) classification. (d) Same as (c) but for the reflectivity field denoted in (b)

  • View in gallery

    Conceptual model of the precipitation trajectories and mean vertical motions throughout a leading-line trailing-stratiform squall line system (from Biggerstaff and Houze 1993)

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2132 UTC on 23 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a). The heavy, dark line in (a), (c), and (d) denotes the location of the vertical cross section shown in Fig. 5

  • View in gallery

    Vertical cross section of equivalent radar reflectivity (dBZe) for the 23 May 1993 squall line system at 2132 UTC. The cross section was taken from (X, Y) coordinates (−150, −32) to (50, −32) km in Fig. 4. Contours are every 5 dB as labeled, with various degrees of shading for values exceeding 35 dBZe. Horizontal tick marks are every 2 km. The traces at the top show the echo classification by the two algorithms as indicated with heavy (light) shading denoting stratiform (convective) rain. The heavy, dashed line in the lower portion of the diagram defines the altitude of the working level used in the SHY95 algorithm

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 0135 UTC on 6 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a). The heavy, dark line in (a), (c), and (d) denotes the location of the vertical cross section shown in Fig. 7

  • View in gallery

    Vertical cross section of equivalent radar reflectivity factor (dBZe) for the 6 May 1993 embedded convection case at 0135 UTC. The cross section was taken from (X, Y) coordinates (−139, −53) to (−32, 82) km in Fig. 6. Contours are every 5 dBZe as labeled, with various degrees of shading for values exceeding 35 dBZe. Horizontal tick marks are every 2 km. The traces at the top show the echo classification by the two algorithms as indicated with heavy (light) shading denoting stratiform (convective) rain. The heavy, dashed line in the lower portion of the diagram defines the altitude of the working level used in the SHY95 algorithm

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2004 UTC on 3 Jun 1994. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a)

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at 3.0-km level above mean sea level from the Houston WSR-88D for 0706 UTC on 10 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a)

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 0706 UTC on 10 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a)

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2004 UTC on 3 Jun 1994. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms for the case in which the convective radius in SHY95 has been doubled. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a) using twice the convective radius of the original algorithm. (d) BL classification of the reflectivity field in (a) using (c) as the first guess

  • View in gallery

    (a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2132 UTC on 23 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms for the case in which the convective radius in SHY95 has been doubled. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a) using twice the convective radius of the original algorithm. (d) BL classification of the reflectivity field in (a) using (c) as the first guess

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 142 142 6
PDF Downloads 110 110 5

An Improved Scheme for Convective/Stratiform Echo Classification Using Radar Reflectivity

View More View Less
  • 1 Department of Atmospheric Sciences, Texas A&M University, College Station, Texas
© Get Permissions
Full access

Abstract

An improved algorithm for the partitioning of radar reflectivity into convective and stratiform rain classifications has been developed and tested using data from the Houston, Texas, Weather Surveillance Radar-1988 Doppler. The algorithm starts with output from the current operational version of the Tropical Rainfall Measuring Mission (TRMM) convective/stratiform classification scheme for the ground-based validation sites and corrects the output based on physical characteristics of convective and stratiform rain diagnosed from the three-dimensional structure of the radar reflectivity field. The modified algorithm improved the performance of echo classification by correcting two main sources of error. Heavy stratiform rain, originally classified as convective, and the periphery of convective cores, originally classified as stratiform, were both reclassified by the modified algorithm. When applied to a large dataset of convective storms comprising squall lines, unorganized convection, and embedded convection, it was found that roughly 25% of the total echo area and 14% of the total rain volume were reclassified. The magnitudes of the differences between the original and modified algorithms varied with the morphology of the storm system, suggesting that the quality of current echo classification information supplied by the TRMM program could vary by location depending on the structure of the dominant precipitation systems within a given region. The analysis presented here helps to establish the level of uncertainty in the existing echo classification products available from TRMM.

Corresponding author address: Dr. Mike Biggerstaff, Dept. of Atmospheric Sciences, MS 3150, Texas A&M University, College Station, TX 77843-3150.

mikeb@ariel.met.tamu.edu

Abstract

An improved algorithm for the partitioning of radar reflectivity into convective and stratiform rain classifications has been developed and tested using data from the Houston, Texas, Weather Surveillance Radar-1988 Doppler. The algorithm starts with output from the current operational version of the Tropical Rainfall Measuring Mission (TRMM) convective/stratiform classification scheme for the ground-based validation sites and corrects the output based on physical characteristics of convective and stratiform rain diagnosed from the three-dimensional structure of the radar reflectivity field. The modified algorithm improved the performance of echo classification by correcting two main sources of error. Heavy stratiform rain, originally classified as convective, and the periphery of convective cores, originally classified as stratiform, were both reclassified by the modified algorithm. When applied to a large dataset of convective storms comprising squall lines, unorganized convection, and embedded convection, it was found that roughly 25% of the total echo area and 14% of the total rain volume were reclassified. The magnitudes of the differences between the original and modified algorithms varied with the morphology of the storm system, suggesting that the quality of current echo classification information supplied by the TRMM program could vary by location depending on the structure of the dominant precipitation systems within a given region. The analysis presented here helps to establish the level of uncertainty in the existing echo classification products available from TRMM.

Corresponding author address: Dr. Mike Biggerstaff, Dept. of Atmospheric Sciences, MS 3150, Texas A&M University, College Station, TX 77843-3150.

mikeb@ariel.met.tamu.edu

Introduction

The importance of convective cloud systems to the energy balance in the tropical atmosphere has been recognized for many years (e.g., Riehl and Malkus 1958; Malkus 1962; Riehl and Simpson 1979) and is often expressed in terms of the apparent heat source and the apparent moisture sink (Yanai et al. 1973; Ogura and Cho 1973; Nitta 1975). These quantities represent the net effect of convective cloud systems on the larger-scale environment. At first, only the convective portion of the cloud system was thought to be relevant to the heat and moisture changes in the atmosphere (Ogura and Cho 1973; Lopez 1973; Lewis 1975). Later Houze (1982, 1989) and Johnson and Young (1983) showed that the stratiform rain region also significantly contributed to the derived vertical profile of heating from mature mesoscale convective systems (MCSs). Houze (1982) noted that the latent heating from vapor deposition in the mesoscale updraft at mid- to upper levels resulted in an upward shift in the altitude of the peak heating from an MCS. Moreover, cooling from evaporation (Brown 1979) and melting (Leary and Houze 1979a) in the mesoscale downdraft beneath the midtropospheric cloud base (Zipser 1969, 1977) decreased the heating through mid- to lower levels. Hence, the inclusion of broad regions of stratiform rain appreciably changes the vertical profile of heating diagnosed from a convective cloud system.

Large-scale numerical simulations have illustrated the sensitivity of the atmosphere to the vertical distribution of diabatic heating (Simmons 1982; Hartmann et al. 1984; DeMaria 1985; Lau and Peng 1987). Given this sensitivity, there is a need to partition accurately the heating from an MCS into its convective and stratiform components. However, direct measurement of the apparent heat source is not possible. Fortunately, when radiative effects are neglected, the vertically integrated apparent heat source is approximately given by the net rainfall from the cloud system. Rainfall is a quantity that can be measured either directly by a calibrated gauge or estimated remotely by radar (Ryde 1946) or by satellite (Wilheit et al. 1977; Adler and Negri 1988;Kummerow and Giglio 1994). By partitioning the rainfall into its convective and stratiform components, it is at least possible to partition the vertically integrated apparent heat source into convective and stratiform components.

There have been numerous methods devised to partition the rainfall from precipitating clouds into convective and stratiform components. Many of these originate from studies of rain gauge data (e.g., Austin and Houze 1972; Houze 1973) in which convective classification was assigned whenever the rain rate exceeded some background level by a certain threshold. This background-exceedence technique (BET) generally identifies the core of the convective precipitation. Churchill and Houze (1984) extended BET into two dimensions, using radar reflectivity. Because the technique concentrates on core identification, they assigned a fixed radius of influence to each identified convective core to define the convective area of the core. The radius was large enough to create contiguous zones of convective precipitation whenever convective cores were closely spaced together.

Adler and Negri (1988) applied a variation of BET to distinguish between convective and stratiform rain using infrared satellite data. Instead of searching for local maxima, they looked for local minima in the cloud-top temperatures to denote the location of a convective core. The radius of influence around each core depended on the magnitude of the infrared brightness temperature of the core and was based on a one-dimensional cloud-modeling study (Adler and Mack 1984). Goldenberg et al. (1990) followed a similar method in their analysis of the infrared cloud-top temperatures to separate convective and stratiform regions of a tropical cloud cluster.

In a recent study of rain classification using radar reflectivity, Steiner et al. (1995) reasoned that a fixed convective radius like that used by Churchill and Houze (1984) was insufficient. In an approach somewhat similar to Adler and Negri (1988), they made the radius of influence around each core variable in size. The radius of influence was a function of the magnitude of the area-averaged background reflectivity. In addition, Steiner et al. (1995) made the exceedence threshold dependent on the area-averaged background reflectivity. Although essentially still a background-exceedence technique, their strategy is often referred to as the “peakedness method.”

In contrast to studies such as Steiner et al. (1995) that use a horizontal two-dimensional BET to classify rainfall, other investigators have focused on the three-dimensional structure of the hydrometeor field to determine whether a particular area is convective or stratiform. DeMott et al. (1995) extended an earlier version of Steiner et al. (1995) into three dimensions by applying the two-dimensional BET at each height level within a volume of radar reflectivity and adjusting the results to remove shallow convection centered near the melting level and to extend the convective classification to echo top. They were concerned that the rain classification at low levels would misclassify rain from convective cells that tilted strongly with height. Using vertical velocities from a dual-Doppler analysis, DeMott et al. (1995) showed that the three-dimensional application improved the accuracy of the rain classification.

Rosenfeld et al. (1995) also based their rain classification scheme on the three-dimensional hydrometeor field. However, their method was different from the BET used by others. Instead of trying to distinguish between convective and stratiform rain types, their work emphasized improving rainfall estimates from radar reflectivity by using the window probability matching method (Rosenfeld et al. 1994). They combined the effects of differing precipitation growth mechanisms within convective and stratiform clouds along with limitations inherent in using radar data (such as beam filling, radar horizon, and attenuation) and separated the rain into differing values of three parameters: radial reflectivity gradients, brightband fraction, and effective efficiency (a measure of cloud depth). If one assumes that convective rainfall tends to have high radial gradients of reflectivity but stratiform rain usually exhibits a well-defined bright band near the freezing level [caused by the melting of snow as it falls to form raindrops; Austin and Bemis (1950)], it would be possible to apply the Rosenfeld et al. scheme to separate rain into convective and stratiform components. Because this method is fundamentally distinct from a two-dimensional BET, it would serve as a useful check on the sensitivity of the rain classification to algorithm design.

The Steiner et al. (1995) technique has been adopted as the current operational algorithm for rain classification by the ground validation program of the Tropical Rainfall Measuring Mission (TRMM; Simpson et al. 1988). The satellite program uses a different algorithm for echo classification that is partly based on the Steiner et al. (1995) method.

TRMM combines both passive and active spaceborne microwave measurements with ground-based radar and rain gauge data to retrieve diabatic heating profiles from convective cloud systems over the Tropics. The rainfall classification must be performed to a high level of accuracy (within 10%) to allow for accurate retrievals of heating from tropical MCSs (Tao et al. 1993).

Using radar data from Darwin, Australia, covering both monsoon and continental convection, Steiner et al. (1995) showed that their technique works well. Further inspection, however, suggests that there were still times when rainfall was inappropriately classified. Given the sensitivity of the heating retrievals to the rain classification, this uncertainty warrants further investigation.

In this paper, we examine the performance of the Steiner et al. (1995) rain classification algorithm and show that there are two main sources of misclassification: convective classification being assigned to heavy stratiform rain, and stratiform classification being assigned to the periphery of convective cores. By applying additional information based on the three-dimensional hydrometeor field inferred from radar reflectivity, we show that the current algorithm can be improved. A description of the modified algorithm is given. Based on analysis of over 2300 volumes of reflectivity data from 29 convective storm systems, about 25% of the echo area and 14% of the total rain volume were reclassified. The actual percentages varied significantly based on the storm morphology. Except for scattered isolated convective cells, the modified algorithm produced more convective area but less convective rain volume than the original Steiner et al. (1995) method.

Data

Radar reflectivity data from the National Weather Service Weather Surveillance Radar-1988 Doppler (WSR-88D; Klazura and Imy 1993) near Houston, Texas (Fig. 1), were obtained from the National Climatic Data Center in Asheville, North Carolina, for use in this study. Houston is one of the four primary ground validation sites for the TRMM project. This area receives precipitation from a variety of storm types and is often dominated by warm, moist, semitropical air from the Gulf of Mexico.

WSR-88D is a 10-cm wavelength Doppler radar with a 1° half-power beamwidth. The radar data consist of volume scans of radar reflectivity, radial velocity, and spectrum width collected in polar coordinates at increasing elevation angles. The radar is operated in a 360° azimuthal volume scan mode with steps in elevation angles from 0.5° to 19.5° during periods of precipitation. The number of elevation steps and temporal resolution of the data depend on the operational mode of the radar. The bin spacing along the radial is 250 m. However, reflectivity is range averaged over four bins to increase the number of independent measurements collected for each recorded value. Hence, reflectivity is recorded at 1-km intervals along the radar beam although the velocity parameters are recorded at 250-m intervals. Only the reflectivity data were used in this study.

Several storm systems sampled during the spring and summer months of 1993, 1994, and 1996 were included in this study. Data from 1993 and 1994 were collected using the nine-elevation scan strategy (0.5°, 1.45°, 2.4°, 3.35°, 4.3°, 6.0°, 9.9°, 14.6°, and 19.5° in elevation). Each volume scan took roughly 6 min to complete. The 1996 data were collected using the 14-elevation scan strategy (0.5°, 1.45°, 2.4°, 3.35°, 4.3°, 5.25°, 6.2°, 7.5°, 8.7°, 10.0°, 12.0°, 14.0°, 16.7°, and 19.5° in elevation). Despite the increased number of elevation steps, these volume scans required only 5 min to complete.

The choice of the spring and summer months was made in recognition of the need to focus on MCSs that would be representative of those observed in the Tropics. Although large-scale forcing is responsible for about 80% of all springtime MCSs in this region (Hashem and Biggerstaff 1997), the structure and mesoscale dynamics of these MCSs are similar to those observed in the Tropics, where less-pronounced forcing is required to initiate convection. Moreover, Fritsch et al. (1986) noted that warm-season MCSs account for approximately 30%–70% of the precipitation that falls between the Rocky Mountains and the Mississippi River. The storms chosen for this study are therefore representative of the types of systems that produce the bulk of the precipitation during the warm season.

Individual storm systems for this study (Table 1) were chosen based on their mesoscale organization. The systems were classified into three categories based on their morphology: squall lines, large areas of stratiform rain with embedded regions of convection, and unorganized convection with little (if any) stratiform rain. This separation was done to examine differences in the performance of rain classification algorithms based on storm-system morphology. Such knowledge would aid in understanding how the algorithms would perform at other locations that may exhibit a different mix of convective events.

Individual volume scans were selected based on there being significant precipitation within 150 km of the radar location. This distance represents the operational range of the TRMM ground validation program. Reflectivity data from each volume scan were examined to remove spurious returns from ground clutter or anomalous propagation caused by a nonstandard index of refraction.

The radar data were interpolated from polar coordinates to Cartesian coordinates using the Sorted Position Radar Interpolation software package (Mohr and Vaughan 1979). The horizontal domain of the grid was 151 × 151 grid points (300 km in the horizontal with 2-km grid spacing). Two vertical spacings were used: 17 grid points (0.5–9.0 km with 0.5-km grid spacing) for the modified algorithm and 2 grid points (1.5–3.0 km with 1.5-km grid spacing) for the Steiner et al. (1995) algorithm. A reflectivity threshold of 10 dBZ was applied to the interpolated dataset to remove most nonprecipitating echo such as birds or insects from clear-air returns. Any value less than 10 dBZ was removed before the data were ingested by the algorithms.

Steiner et al. algorithm description and performance

Description

The Steiner et al. (1995) algorithm (hereinafter referred to as SHY95) uses gridded radar reflectivity data from the “working level” [defined as a 1.5-km constant-altitude plan position indicator (CAPPI) at ranges less than 100 km and a 3.0-km CAPPI between 100 and 150 km] to perform the partitioning. The algorithm has three steps. First, any grid points having reflectivities greater than 40 dBZe are classified as convective. Second, any grid point that exceeds its background reflectivity (defined as the linear average of nonzero points within an 11-km radius centered on the grid point) by an intensity-dependent value is classified as convective. Last, for each grid point classified as convective by either of the two tests, an intensity-dependent radius of influence is used to assign an area around the point as convective. Any remaining nonzero reflectivity areas are then classified as stratiform. No attempt is made to use directly the vertical structure of the reflectivity field in the SHY95 rain classification scheme. Nevertheless, it is important to note that the SHY95 algorithm was tuned for monsoon cloud systems over Darwin, Australia, using the vertical structure of radar reflectivity.

Philosophy regarding rain classifications

Application of SHY95 to two different kinds of MCSs observed over southeast Texas is shown in Fig. 2. One storm system (Fig. 2a) is a classic leading-line trailing-stratiform squall line system of the type studied by Newton (1950), Fujita (1955), and Ligda (1956). Following the terminology of Houze et al. (1990), this storm would be strongly classifiable. Biggerstaff and Houze (1991) calculated the vertical air motions and precipitation trajectories within the region behind the deep convective zone for this kind of storm system. A conceptual model based on their findings is presented in Fig. 3. Precipitation trajectories suggested that the stratiform-region rain could be traced back through the mid- to upper portions of the transition zone and into the upper portions of the back of the convective line where small ice crystals would have been detrained from the convective cores and carried rearward by the storm-relative flow.

Using the terminology of Houghton (1968), the rain in the trailing anvil behind the transition zone could be classified as “stratiform” because the vertical air motions in the precipitation growth region were weaker than the fall speeds of the precipitation particles. This condition would limit the precipitation mass growth mechanism to vapor deposition. Rutledge and Houze (1987) used a two-dimensional model to show that the majority of the mass of the stratiform precipitation was added through vapor deposition in the mesoscale updraft at mid- to upper levels. In an environment that supports supercooled liquid (i.e., in and near the convective region where strong vertical motions exist), the precipitation would grow mainly through riming and accretion of cloud water.

Biggerstaff and Houze (1993) showed that the transition zone immediately behind the convective cores in their squall line was a region of deep subsidence that likely aided in the development of a reflectivity minimum at lower levels. Transition zones are often marked by deep subsidence (Houze and Rappaport 1984; Srivastava et al. 1986; Smull and Houze 1987; Chalon et al. 1988; Rutledge et al. 1988) and a reflectivity minimum, or trough, at lower levels (Ligda 1956; Houze 1977; Sommeria and Testud 1984; Smull and Houze 1985; Roux 1988). It should be noted that Rutledge and Houze’s model did not include a transition zone downdraft. Given that the precipitation trajectories from Biggerstaff and Houze (1991) suggest that the rain in the transition zone came from the midlevels of the convective line and that subsidence in the transition zone would only serve to reduce the precipitation mass, it would be appropriate to classify the rainfall in this region as convective. In a recent study of reflectivity–rain-rate relations for tropical cloud systems, Atlas et al. (1999) found that the transition-zone precipitation was more like convective rainfall than like stratiform rainfall. Steiner et al. (1999) found a similar result for a midlatitude squall line they examined. Likewise, the rainfall along the forward portion of the convective cores would also be more appropriately classified as convective.

This classification is consistent with the definition of rain types by Houghton (1968) that compares the vertical air motions to the precipitation particle fall speeds, as long as the definition is applied to the precipitation growth region. Because the rain will be used to estimate the vertically integrated diabatic heating from MCSs as part of the TRMM project (e.g., Simpson et al. 1988), it is important to assign the rain the classification associated with the type of microphysical and dynamic process that contributed to the majority of the mass production.

The importance of echo classification on the determination of heating within MCSs was further discussed by Houze (1997). Rather than apply the echo classification to the precipitation growth region, Houze (1997) suggests that the classification be based on the instantaneous local profile of divergence associated with the precipitation. In this manner the rain within the transition zone would not be classified convective, because the transition-zone divergence profile generally would not exhibit the low-level convergent–upper-level divergent pattern typical of convective updrafts. In the case of deep subsidence with two downdraft maxima, such as the squall line studied by Biggerstaff and Houze (1993), the transition zone would also not qualify for a stratiform rain assignment. A third echo class would be needed to identify correctly the divergence pattern associated with such transition zones. This complication can be avoided by applying the Houghton (1968) definition of precipitation type to the precipitation growth region and allowing for hydrometeor advection. It is this concept that we follow in discussing the performance of the SHY95 algorithm.

Performance of the Steiner et al. method

For the squall line case shown in Fig. 2a, SHY95 correctly identifies the rain in cores of deep convective cells as convective rainfall. The cores are spaced close enough so that a contiguous region of convection is indicated within the leading portion of the squall line system. However, the intensity-dependent convective radius was not large enough to capture the forward edge of the convective line. The algorithm, therefore, labels the reflectivity along the forward portion of the convective line as stratiform. This classification is inappropriate because the precipitation mass of the rain in this region must have developed in the convective updrafts and been detrained from the nearby convective cores. A similar misclassification exists around the periphery of the isolated convective cells ahead of the squall line.

The back edge of the SHY95 convective region closely follows the 40-dBZe contour, leaving the strong horizontal reflectivity gradient region between the 40-dBZe echo and the reflectivity minimum in the transition zone classified as stratiform rain. Because there is a pronounced reflectivity trough, there likely is deep subsidence in the transition zone of this squall line system like that observed by Biggerstaff and Houze (1993). It would be more appropriate to classify the rain in this region as convective, because the precipitation mass would have been developed within the region of strong vertical drafts associated with the convective cores. Although the exact boundary between the convective and stratiform rain region would require detailed trajectories of the precipitation particles and their growth along those trajectories, it seems that the current SHY95 algorithm is conservative in defining the width of the rain produced by the convective cells.

The SHY95 rain classification method does extend the convective classification to other areas of the squall line system. Reflectivity from the trailing anvil region that exceeds 40 dBZe at the working level is automatically assigned a convective classification, and some area around the center of that enhanced reflectivity is also given a convective classification. Likewise, local maxima of enhanced reflectivity within the trailing stratiform area occasionally exceed the background-exceedence threshold required for convective classification. These two artifacts lead to the appearance of embedded convection within the otherwise broad region of stratiform rain. However, these false convective regions vary greatly in location with time as the specific locations of local maxima and grid points exceeding the 40-dBZ threshold vary from one volume scan to the next. The lack of temporal continuity of the convectively classified areas in the broader stratiform rain region indicates that the rain in this area is not associated with deep convection and should be classified as stratiform.

In the unorganized MCS comprising scattered convective cells (Fig. 2b), the SHY95 algorithm again correctly identifies the cores of the convective cells. As in the case of the convective region of the squall line system, however, the intensity-dependent radius is not large enough to enclose all the reflectivity around the core of the cells. The convective cores are ringed by echo that is classified as stratiform. Although it is likely that the vertical air motions along the edges of the convective cores are much weaker than the peak updrafts and downdrafts within the cores, the precipitation mass must have developed within the convective updrafts and, hence, all the rain should be classified convective.

Inspection of the SHY95 performance as a function of time for the scattered convection case revealed that new cells were initially classified as stratiform rain and dissipating convection was also classified as stratiform rain once its low-level reflectivity field became uniform. The misclassification of developing cells would represent a minor source of error in rain-volume partitioning, because the algorithm quickly switched these cells to convective classification once the reflectivity values reached moderate strength. This misclassification could, however, have a significant effect on retrievals of latent heating, because developing cells would exhibit a convective rather than a stratiform heating profile (Houze 1982). Of more concern for rain-volume partitioning is the tendency of the SHY95 method to classify dissipating convection as stratiform rain in these unorganized systems.

Based on studies of airmass thunderstorms over Florida and Ohio, Byers and Braham (1949) developed a detailed conceptual model of the vertical motion within unorganized multicell convective clusters. They found that, during the developing stages, the majority of the cloud contained strong updrafts. With time, precipitation that developed in the updraft produced a drag on the rising air and helped to initiate a convective downdraft. The downdraft was further aided by evaporational cooling as dry midlevel air was entrained into the precipitation core at mid- to lower levels. Because their studies were of multicell thunderstorms in a weak shear environment, the downdraft typically undercut the air feeding the updraft. This undercutting led to the dissipation of the convective cell. During the dissipation stage, there was residual upward motion in the upper portions of the cell and weakening downward motions elsewhere. For isolated convective cells and small multicellular clusters, the precipitation rapidly diminished once the growth of new convective cells within the cluster had ceased. Given the typical depth of convective cells (∼10 km) and an average value for terminal velocity of hydrometeors (∼5 m s−1), it would take approximately 30 min for the majority of the convectively grown precipitation to fall to the surface. It would seem appropriate, therefore, for the rain from isolated and small multicellular clusters to be classified as convective unless the rainfall lasted significantly beyond 30 min from the dissipation of the convective cell that had produced it. The current SHY95 algorithm changes the classification from convective to stratiform immediately after the reflectivity at lower levels starts to become uniform.

If the multicellular convective cluster contained numerous dissipating cells, the resulting precipitation region could approach the spatial scale needed to maintain a mesoscale updraft that would contribute significantly to the mass production of the rain. Small multicellular clusters frequently merge to form larger regions of rain (e.g., Tao and Simpson 1984) that could exceed the size needed for significant stratiform rain development. Such a convective cluster was recently analyzed by Yuter and Houze (1995). In their case, the transition to stratiform classification would be warranted, though perhaps not as early as the current algorithm dictates.

In both types of storm systems, the SHY95 algorithm tended to be conservative in the assignment of convective area. However, the effect of limiting the size of the convective region on the determination of total convective rain volume from the MCS is not obvious, because the most intense portions of the stratiform rain regions of squall line systems were misclassified as convective rainfall. It should be noted that increasing the size of the intensity-dependent convective radius would help to mitigate the underestimate of convective area. This mitigation would occur at the expense of further expanding the false convective areas in the intense portions of the stratiform rain regions, however.

Description of the modified algorithm

As shown in the previous section, the SHY95 rain classification method does correctly identify the majority of the rain field. The errors that are observed do not justify complete revision of the basic technique. Instead, the rain classification derived from SHY95 should be further analyzed using more of the reflectivity information that is routinely available from weather radars, namely the vertical structure of the hydrometeor field.

The modified algorithm (hereinafter referred to as BL) is an automated method for convective/stratiform partitioning using radar reflectivity. The algorithm uses results from the SHY95 algorithm in addition to gridded radar reflectivity to perform the partitioning. Depending on the initial classification by SHY95, different tests are performed to determine if the radar reflectivity was correctly classified. These tests require that several parameters be calculated at each grid point.

First, the vertical lapse rate of reflectivity (zZ), defined as the decrease in reflectivity with increasing height above ground in the 3-km layer above the height of maximum reflectivity, is calculated. For a well-developed stratiform region, this rate will be the mixed-phase lapse rate (between 0°C and −20°C) as defined by Zipser and Lutz (1994). Because the height of the freezing level can change over the system’s lifetime, lapse rates for a 3-km layer above the height of the maximum reflectivity were used instead of a fixed height calculation. Toracinta et al. (1996) studied several MCSs in southeast Texas and found a median lapse rate in the convective cores of near 2.0 dB km−1, similar to those found by Zipser and Lutz (1994) for midlatitude continental convection. For the BL algorithm, lapse rates for convective (stratiform) precipitation were defined as having values less (greater) than 3.5 dB km−1.

Second, a brightband fraction (BBF), similar to that described by Rosenfeld et al. (1995), was calculated at each grid point. In contrast to Rosenfeld et al. (1995) in which any reflectivity maximum near the melting levels was taken as an indication of a bright band, we add the additional criterion that the reflectivity lapse rate in the grid column be indicative of stratiform rain (>3.5 dB km−1). Thus, in BL, a grid column is considered to contain a bright band if and only if the reflectivity maximum is within ±1.5 km of the melting level and the reflectivity above the maximum decreases substantially with height. The second criterion was added to account for the possibility of convection having a maximum in reflectivity near the melting level. This modification to Rosenfeld et al.’s (1995) definition of BBF helps to reduce the chance that convective areas would be misclassified. Minor changes to the definition of the size of the region inspected for each BBF calculation were made to account for the use of Cartesian coordinates in this study as opposed to the polar coordinate system used in Rosenfeld et al. (1995).

Last, the magnitude of the two-dimensional horizontal gradient of radar reflectivity (|HZ|), computed using log-scale reflectivity (dBZ), was determined at each grid point. After testing several values of horizontal gradient, a 3.0-dB km−1 threshold was chosen, with stratiform (convective) echo associated with gradients less (greater) than this value. This threshold is smaller than the value used by Rosenfeld et al. (1995), who calculated only the radial reflectivity gradient in their polar-coordinate dataset. The difference is likely due to the smoothing of reflectivity gradients by converting from high-resolution polar to lower-resolution Cartesian space.

Once the parameters were computed, the SHY95 algorithm was executed without the 40 dBZe intensity criterion. It was found that not using this threshold decreased the amount of brightband contamination, which further improved the output from BL. A series of tests were performed to determine if the initial echo classification assigned by SHY95 needed to be changed. The reclassification criteria are based on the physical differences in the structure of convective and stratiform precipitation. If the SHY95 algorithm classified the grid point as convective, the BL algorithm reclassified it if the grid point had

  1. a weak horizontal reflectivity gradient (|HZ| < 3 dB km−1), a steep reflectivity lapse rate (zZ > 3.5 dB km−1), and low reflectivity at the working level (<35 dBZe); or
  2. low reflectivity (<28 dBZe) aloft (defined as roughly twice the altitude of the reflectivity maxima), a weak horizontal reflectivity gradient (|HZ| < 3 dB km−1), and a high BBF (>0.60).
Likewise, if the SHY95 algorithm classified the grid point as stratiform, it was reclassified if the grid point had
  1. a strong horizontal reflectivity gradient (|HZ| > 3 dB km−1), or
  2. a weaker horizontal gradient (|HZ| > 2 dB km−1) and a low BBF (<0.40).
The weaker gradient in criterion 2 helped to account for “smearing” of the reflectivity gradients caused by the width of the radar beam increasing with distance from the radar. The initial SHY95 classification was retained for grid points that failed to meet the conditions for reclassification.

Based on BL, a new convective/stratiform map was created. Examining the results revealed that the modified algorithm still did not reclassify all of the brightband contamination. In addition, the extreme forward edge of convective lines in squall line systems was still misclassified as stratiform echo. To reduce further the misclassified areas, each data point was put through a 17 × 17 gridpoint window, and the fraction of classified data points that had a classification different from the grid point under consideration was computed. If this fraction was greater than 0.55, which corresponds to more than 55% of the surrounding points having a different classification, the grid point was reclassified. The windowing step accounted for nearly 30% of the total area reclassified. Sensitivity tests were performed using different window sizes and thresholds, with the 17 × 17 gridpoint size showing optimal results. Smaller windows tended not to reclassify all of the brightband contamination, and larger windows reclassified areas that were truly convective. Tests also showed that the windowing procedure worked best after applying BL. Such a procedure applied directly to the output from SHY95 expanded the regions affected by brightband contamination and did not wholly reclassify the leading edge of squall lines as convective.

In essence, the windowing step applies a certain scale to echo classifications. Stratiform rain is limited to mesoscale regions—consistent with the concept that this precipitation is associated with the heating and divergence profile from a precipitation growth region that contains a mesoscale updraft above a mesoscale downdraft. Convective regions can still be small, especially if the convection is isolated.

The windowing technique does have one negative consequence: thin lines of convection surrounded by large areas of stratiform rain tend to be reclassified as stratiform echo. For the Houston dataset this tendency is unimportant, because the magnitude of the environmental low-level wind shear is typically too large to support narrow convective lines. In low-shear environments, like the Tropics, convective cell diameters can be small (LeMone and Zipser 1980) and narrow convective lines surrounded by large stratiform regions are possible. For these environments the window size would have to be reduced or a more sophisticated reclassification scheme would have to be developed to identify thin embedded lines of convection and keep them from being reclassified.

Results

Once the BL algorithm was developed, both algorithms were applied on a large independent set of convective storms within the vicinity of the Houston WSR-88D. Table 1 shows a summary of each case used in the study. A total of seven squall line cases (391 volume scans), 11 unorganized convection cases (931 volume scans), and 11 embedded convection cases (995 volume scans) were used. Statistics describing the differences between the two algorithms were calculated. Sensitivity tests were then performed to examine the robustness of the modified algorithm and its difference with the original algorithm.

Individual case studies

Squall line

Figure 4 shows the results from applying both algorithms to the 23 May 1993 squall line at 2132 UTC. The reflectivity field (Fig. 4a) indicates this storm system was a strongly classifiable, symmetric, leading-line trailing-stratiform squall line system (Houze et al. 1990). The broad trailing-stratiform region and relatively low peak reflectivity in the convective line (exceeding 50 dBZ in only a few cells) suggest that at this time the storm system was in its mature stage (Leary and Houze 1979b). The difference between the algorithms (Fig. 4b) reveals that the majority of the original classification has not changed by BL. Nevertheless, significant differences did occur. In particular, SHY95 misclassified the heavy stratiform rain regions behind the convective line (Fig. 4c). In a similar storm system, Biggerstaff and Houze (1991) referred to the heavy stratiform rain region as the secondary band to denote that it was found downwind of the intense portions of the convective line. Matejka and Schuur (1991) showed that this region was characterized by a vertical mesoscale updraft–downdraft couplet. Indeed, the secondary band was downwind of the most intense portion of the mesoscale updraft in the stratiform region. Hence, the secondary band should be classified as stratiform rain as indicated by BL (Fig. 4d).

Another difference between the two algorithms is that the weak reflectivity along the leading edge of the squall line is classified as stratiform rain by SHY95. Vertical cross sections of vertical velocity along the leading portion of the high-reflectivity band in squall line systems (e.g., Houze et al. 1989) have determined this region consists of strong vertical drafts associated with developing convective cells. The lack of high reflectivity at low levels is common for developing convective cells. Because strongly classifiable squall lines are characterized by their leading convective line and trailing stratiform structure, the areas along the leading edge appear to be misclassified by SHY95. In contrast, the BL algorithm (Fig. 4d) classified almost all of the leading edge as convective.

The differences in the classification by the two algorithms are further illustrated by the vertical cross section taken perpendicular to the leading convective line (Fig. 5). The SHY95 algorithm indicated a convective cell was embedded in the stratiform region near X = −94 km. The vertical structure of reflectivity in this area, however, suggests that the echo is actually associated with stratiform rain, as indicated by the BL algorithm.

The forward edge of the precipitation region (−22 < X < −8 km in Fig. 5) is labeled stratiform by the SHY95 algorithm. The vertical reflectivity structure clearly shows that this region is one of developing convection, however. Likewise, the small, isolated precipitation shaft at X = 12 km is classified as stratiform precipitation by SHY95. This echo is associated with a new cell forming ahead of the main band of convection, however. The BL algorithm corrected the classification of these two regions. In summary, the modified algorithm was able to reclassify each of the misclassified areas—those associated with the secondary band in the stratiform region and those associated with the leading convective line.

Stratiform rain with embedded convection

The 6 May 1993 storm system illustrates the differences in the two algorithms for cases with embedded convection. The reflectivity field (Fig. 6a) shows a large area of precipitation with a northwest–southeast-oriented line of enhanced reflectivities about 50–100 km to the west of the radar. In addition, several areas of enhanced reflectivity could also been seen to the southwest of the main line. Differences between the original SHY95 and the modified algorithm covered less area than the squall line case examined in the previous section (cf. Figs. 4b and 6b). In general, however, the area reclassified was greater for embedded systems than for squall lines.

Results from the SHY95 algorithm (Fig. 6c) show that the main band of high reflectivity and several areas to the southwest of the main band were classified as convective. The isolated small cores exhibited poor temporal continuity and were associated with points that barely exceeded the algorithm thresholds for convective classification. Several of the small convective cores to the southwest of the main reflectivity band were reclassified by BL (Fig. 6d). A larger area, centered near (X, Y) coordinates of (−105, −17) in Fig. 6a, was also reclassified.

A vertical cross section of radar reflectivity (Fig. 7) was taken from the southwest to northeast, [from (X, Y) coordinates (−139, −53) to (−32, 82) km in Fig. 6a] through three of the enhanced-reflectivity regions of which one, the center region near (−105, −17), was reclassified. The reflectivity structure for the two regions that were not reclassified exhibits deep reflectivity cores associated with mature-stage convective cells. The region between these two deep convective cores exhibits reflectivity structure that would be associated with either dissipating convective cells (X′ = 64–79 km in the cross section) or with stratiform rain (X′ = 40–64 km in the cross section). The original algorithm classified almost the entire area as convective while the modified algorithm retained the convective classification associated with the dissipating cell and reclassified the area with the pronounced bright band (centered at a X = 54 km in Fig. 6d) and weak reflectivity aloft to stratiform rain. Based on the vertical reflectivity structure, the modified algorithm output would seem to be more appropriate.

Unorganized convection with little stratiform rain

An unorganized convective storm system with little stratiform rain was sampled by the Houston WSR-88D during 3 June 1994. Results from the two algorithms applied to this storm system at 2004 UTC are shown in Fig. 8. This storm system contained several isolated convective cells, many having diameters of less than 10 km (Fig. 8a) and peak reflectivity less than 50 dBZe. Relative to the total area of precipitation, the difference in echo classification was large, with most of the reclassification being from stratiform to convective rain (Fig. 8b).

In many of the rain regions, only the cores of the isolated reflectivity cells were classified by SHY95 as convective, with the lower reflectivity around the core classified as stratiform (Fig. 8c). Some of the misclassified echo is likely nonprecipitating, because reflectivity from an isolated rain shaft would be smeared by the width of the radar beam. Nevertheless, precipitation surrounding these high-reflectivity cores was clearly from hydrometeors that have been swept outward from the convective drafts through horizontal advection and should be classified as convective rain. The modified algorithm did change these regions to convective classification (Fig. 8d). Moreover, areas where the convective cells have dissipated (e.g., X, Y = −20, −80 km in Fig. 8), producing a stratified reflectivity structure, were retained as stratiform rain. Areas with lower lapse rates of reflectivity aloft and high horizontal gradients of reflectivity (e.g., X, Y = −50, −120 km in Fig. 8) were reclassified as convective rain.

Difference statistics for the individual case studies

Table 2 summarizes the difference in echo area between the SHY95 and BL algorithms for the three example cases discussed in section 5a. In this table and subsequent tables, the numbers reflect changes relative to the output from the SHY95 algorithm. The unorganized case [(section 5a(3)] had the greatest percentage of echo area reclassification (53%), with a large percentage (70%) of SHY95 stratiform echo being reclassified to convective rain. This result is also seen in the increase in the convective echo fraction from 0.25 to 0.77. For the squall line case, 21% of the echo area was reclassified. The lack of change in the convective echo fraction is evidence of the fact that both echo associated with the secondary band and echo around the edges of the leading line were reclassified. For the embedded case, only about 8% of the echo area was reclassified.

To illustrate further the effects of the differences in echo classification between the two algorithms, the change in the amount of rain volume associated with each classification was computed (Table 3). Here, reflectivities were converted to rain rates using a standard radar reflectivity–rain rate (Z–R) relationship:
ZR1.4
where Z is taken to be the equivalent radar reflectivity factor (mm6 m−3), and R is rain rate (mm h−1). The rain rate at each grid point was then summed to get an estimation of the total rainfall the system would produce. Although some research (e.g., Tokay and Short 1996) suggests that convective and stratiform rain have different Z–R relationships, the analysis here is intentionally kept simple. Because the vertically integrated diabatic heating produced by a precipitation system can be related to rainfall, changes in the total rainfall could serve as an indicator for changes in the total heating.

For the case with unorganized convection with little stratiform rain, the rain-volume differences (19%) were significantly less than that of the echo area (53%). Although 75% of the SHY95 stratiform rainfall was reclassified to convective rain, in this case the reflectivities associated with the reclassified echo were small and did not contribute as much to the total rain volume. Still, a change in rain classification of about 20% may be important when comparing with other rain estimation methods (e.g., satellite) or when retrieving the area-averaged diabatic heating from a storm system.

For the squall line and embedded convection storm systems, the rain-volume differences were similar to the magnitude of the changes in area classification. In both cases, the fraction of convective rainfall decreased by about 13%.

Changes in area and rain volume for the entire database

The results in section 5a apply to a limited sample of radar data—one volume scan from three different storm systems. To determine the effect of the modified algorithm over a climatological scale, the algorithms have been applied to the entire database of 2317 volume scans collected during 29 different mesoscale convective systems (Table 1). The differences in area are summarized in Table 4, and the differences in rain volume using Eq. (1) are summarized in Table 5.

As with the individual cases, the unorganized convective cases with little stratiform rain had the highest echo-area change (44%). For these cases, a large percentage of SHY95 stratiform points (60%) were reclassified, as seen in the increase in convective fraction from 0.29 to 0.70. Despite the significant increase in convective echo area, the fraction of convective rain volume only increased by 7% for this kind of storm system. The weak reflectivity surrounding the core of the convective cell accounts for a significant amount of total echo area but contributes little to the total rain from these types of storms.

The fraction of convective area increased for squall line and embedded systems, as well. Here, however, the rain volume from weak reflectivity surrounding the convective cores was exchanged with rain volume from high reflectivity associated with enhanced stratiform rain in secondary bands to produce a net decrease in the fraction of convective rain. Hence, the storm-system-averaged effect of the improvements to the spatial distribution of echo classification made by the modified algorithm is an overall increase in convective area but a slight overall decrease in convective rain volume (under the assumption of a single Z–R conversion for both echo types).

Although the systemwide convective fraction of rainfall may have been little changed, the spatial pattern of echo classification was significantly affected. The change in spatial pattern of convective rain could have important consequences on validation of space-based retrievals of rainfall and applications that combine satellite data with cloud models to retrieve heating profiles (e.g., Tao et al. 1993). In addition, studies that use different radar reflectivity–rain rate relationships depending on the classification of the echo would be strongly affected by an improper spatial pattern in echo classification.

The dependence of the differences in echo classification on storm morphology suggests that the quality of current echo classification information supplied by the TRMM program could vary by location. Regions in which the precipitation climate is dominated by monsoon systems (which tend to produce broad areas of rainfall with embedded convection) would have uncertainties in echo classification statistics similar to the differences found for the embedded cases examined here. Other tropical locations are dominated by scattered convection with little stratiform rain and would have uncertainties similar to the unorganized cases. Hence, the analysis presented here helps to establish the level of uncertainty in the existing echo-classification products available from TRMM.

Sensitivity tests

To test the robustness of both algorithms, sensitivity tests were performed on various sample cases. The first test was to run both algorithms on a squall line case using different reflectivity maps: the original working level (defined as the 1.5-km CAPPI to a range of 100 km from the radar and the 3.0-km CAPPI from 100 to 150 km range from the radar) and a 3-km (above mean sea level) CAPPI. Using the CAPPI (Fig. 9) resulted in a larger area of enhanced reflectivities in the stratiform region (from brightband effects) within 50 km of the radar, which SHY95 misclassified as convective (Fig. 9c). When the working level is used (Fig. 10), this enhanced echo was not seen because of the decreased influence of the bright band near the radar. Even though the SHY95 algorithm did not perform as well using the CAPPI, the BL algorithm was able to make the correct reclassification in both cases (cf. Figs. 9d and 10d). This result suggests that the BL algorithm is not as sensitive to the height of the reflectivity map as the SHY95 algorithm is.

One method for possibly improving the SHY95 results would be to modify at least one of the steps in the algorithm, thus eliminating the need for a secondary algorithm. As stated earlier, eliminating the 40-dBZe threshold criterion slightly decreased the amount of misclassified echo associated with the bright band. However, most of the misclassified echo was still present. Another test would be to double the convective radius. When the convective radius criterion was doubled and applied to an unorganized convection case (Fig. 11), more of the echo around the cell’s core was classified as convective (Fig. 11c). SHY95 conducted similar tests and reported the same effect. Thus, the results from the BL algorithm (Fig. 11d) were not as dramatic. However, when applied to the 23 May 1993 squall line case at 2132 UTC (Fig. 12), the doubled convective radius increased the enhanced reflectivity region to the rear of the squall line, which resulted in a larger area that was misclassified by the SHY95 algorithm (Fig. 12c). The BL algorithm reclassified most of the affected area (Fig. 12d). However, the magnitude of the misclassification was too large for BL to correct completely, because BL uses SHY95 as the first guess. Hence, doubling the convective radius may improve SHY95 for unorganized convection cases, but it severely and adversely affects the quality of echo classification in squall line systems.

Conclusions

An improved algorithm for the partitioning of radar data into their convective and stratiform components has been developed and tested using data from the Houston, Texas, WSR-88D. The algorithm starts with output from the Steiner et al. (1995) horizontal two-dimensional background-exceedence echo classification algorithm as the initial solution. Changes are made to the classification depending on the horizontal and vertical structure of the radar reflectivity. Hence, the modified algorithm makes use of the full three-dimensional volume of reflectivity to correct the original algorithm output.

The modified algorithm improved the performance of echo classification by correcting two main sources of error. Heavy stratiform rain, originally classified as convective, and the periphery of convective cores, originally classified as stratiform, were both correctly reclassified by the modified algorithm. Even though the new algorithm showed an improvement in the partitioning, additional work needs to be done to validate the performance of both algorithms further. In particular, comparison against dual-Doppler-derived vertical motions with in situ microphysical data would help to determine the accuracy of either approach.

When applied to a large dataset of convective storms made up of squall lines, unorganized convection, and embedded convection, it was found that roughly 25% of the total echo area and 14% of the total rain volume were reclassified. The magnitudes of the differences between the original and modified algorithms varied with the morphology of the storm system. The echo-area differences ranged from 16% for squall lines to 44% for unorganized convection. For rain-volume differences, the magnitudes varied only slightly from 13% for unorganized convection to 14% for embedded convection. The dependence of the differences in echo classification on storm morphology suggests that the quality of current echo classification information supplied by the TRMM program could vary by location, depending on the structure of the dominant precipitation systems within a given region. The analysis presented here helps to establish the level of uncertainty in the existing echo classification products available from TRMM.

It has been suggested that an uncertainty of 10% in rain classification can have a significant effect on the retrieval of diabatic heating in MCSs (Tao et al. 1993). If so, then the results here suggest that a significant improvement can be obtained by using the modified version of the current TRMM echo classification algorithm.

The importance of echo classification on rain estimation at the TRMM ground validation sites is also an area of concern. Steiner and Houze (1997) found that the estimated monthly convective rain fraction can vary significantly depending on the choice of reflectivity–rain rate relation used to determine rainfall from radar reflectivity. Atlas et al. (1999) found that the convective and stratiform regions of a tropical cloud system had unique reflectivity–rain rate relations. Moreover, they found that the transition zone of their cloud system had a reflectivity–rain rate relation that was closer to convective precipitation than to stratiform precipitation. Hence, the correct partitioning of precipitation is important.

Sensitivity tests showed that simple changes to the original algorithm that preserved the two-dimensional background-exceedence technique were unable to match the quality of the output obtained using the fully modified approach. Hence, for proper classification of precipitation, it is best to use the full volumetric data available from most weather radars.

Acknowledgments

This work was sponsored by the National Aeronautics and Space Administration Tropical Rainfall Measuring Mission under Grant NAG5-4776. We are grateful to Dr. M. Steiner for supplying the source code for the original SHY95 algorithm.

REFERENCES

  • Adler, R. F., and R. A. Mack, 1984: Thunderstorm cloud height–rainfall rate relations for use with satellite rainfall estimation techniques. J. Climate Appl. Meteor.,23, 280–296.

  • ——, and A. J. Negri, 1988: A satellite infrared technique to estimate tropical convective and stratiform rainfall. J. Appl. Meteor.,27, 30–51.

  • Atlas, D., C. W. Ulbrich, F. D. Marks Jr., E. Amitai, and C. R. Williams, 1999: Systematic variation of drop size and radar–rainfall relations. J. Geophys. Res.,104, 6155–6169.

  • Austin, P. M., and A. C. Bemis, 1950: A quantitative study of the“bright band” in radar precipitation echoes. J. Meteor.,7, 145–151.

  • ——, and R. A. Houze Jr., 1972: Analysis of the structure of precipitation patterns in New England. J. Appl. Meteor.,11, 926–935.

  • Biggerstaff, M. I., and R. A. Houze Jr., 1991: Kinematic and precipitation structure of the 10–11 June 1985 squall line. Mon. Wea. Rev.,119, 3034–3065.

  • ——, and ——, 1993: Kinematics and microphysics of the transition zone of the 10–11 June 1985 squall line. J. Atmos. Sci.,50, 3091–3110.

  • Brown, J. M., 1979: Mesoscale unsaturated downdrafts driven by rainfall evaporation: A numerical study. J. Atmos. Sci.,36, 313–338.

  • Byers, H. R., and R. R. Braham Jr., 1949: The Thunderstorm. U.S. Govt. Printing Office, 287 pp.

  • Chalon, J. P., G. Jaubert, F. Roux, and J. P. LaFore, 1988: The West African squall line observed on 23 June 1981 during COPT 81:Mesoscale structure and transports. J. Atmos. Sci.,45, 2744–2763.

  • Churchill, D. D., and R. A. Houze Jr., 1984: Development and structure of winter monsoon cloud clusters on 10 December 1978. J. Atmos. Sci.,41, 933–960.

  • DeMaria, M., 1985: Linear response of a stratified tropical atmosphere to convective forcing. J. Atmos. Sci.,42, 1944–1959.

  • DeMott, C. A., R. Cifelli, and S. A. Rutledge, 1995: An improved method for partitioning radar data into convective and stratiform components. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 233–236.

  • Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contribution of mesoscale convective weather systems to the warm-season precipitation in the United States. J. Climate Appl. Meteor.,25, 1333–1345.

  • Fujita, T. T., 1955: Results of detailed synoptic studies of squall lines. Tellus,7, 405–436.

  • Goldenberg, S. B., R. A. Houze Jr., and D. D. Churchill, 1990: Convective and stratiform components of a winter monsoon cloud cluster determined from geosynchronous infrared satellite data. J. Meteor. Soc. Japan,68, 37–63.

  • Hartmann, D. L., H. H. Hendon, and R. A. Houze Jr., 1984: Some implications of the mesoscale circulations in tropical cloud clusters for large-scale dynamics and climate. J. Atmos. Sci.,41, 113–121.

  • Hashem, M. S., and M. I. Biggerstaff, 1997: Organization of convection in mesoscale systems. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 483–484.

  • Houghton, H. G., 1968: On precipitation mechanisms and their artificial modification. J. Appl. Meteor.,7, 851–859.

  • Houze, R. A., Jr., 1973: A climatological study of vertical transports by cumulus-scale convection. J. Atmos. Sci.,30, 1112–1123.

  • ——, 1977: Structure and dynamics of a tropical squall-line system. Mon. Wea. Rev.,105, 1540–1567.

  • ——, 1982: Cloud clusters and large-scale vertical motions in the tropics. J. Meteor. Soc. Japan,60, 396–409.

  • ——, 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc.,115, 425–461.

  • ——, 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc.,78, 2179–2196.

  • ——, and E. N. Rappaport, 1984: Air motions and precipitation structure of an early summer squall line over the eastern tropical Atlantic. J. Atmos. Sci.,41, 553–574.

  • ——, S. A. Rutledge, M. I. Biggerstaff, and B. F. Smull, 1989: Interpretation of Doppler weather radar displays of midlatitude mesoscale convective systems. Bull. Amer. Meteor. Soc.,70, 608–619.

  • ——, B. F. Smull, and P. Dodge, 1990: Mesoscale organization of springtime rainstorms in Oklahoma. Mon. Wea. Rev.,118, 613–654.

  • Johnson, R. H., and G. S. Young, 1983: Heat and moisture budgets of tropical mesoscale anvil clouds. J. Atmos. Sci.,40, 2138–2147.

  • Klazura, G. E., and D. A. Imy, 1993: A description of the initial set of analysis products available from the NEXRAD WSR-88D system. Bull. Amer. Meteor. Soc.,74, 1293–1311.

  • Kummerow, C., and L. Giglio, 1994: A passive microwave technique for estimating rainfall and vertical structure information from space. Part I: Algorithm description. J. Appl. Meteor.,33, 3–18.

  • Lau, K.-M., and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory. J. Atmos. Sci.,44, 950–972.

  • Leary, C. A., and R. A. Houze Jr., 1979a: Melting and evaporation of hydrometeors in precipitation from the anvil clouds of deep tropical convection. J. Atmos. Sci.,36, 669–679.

  • ——, and ——, 1979b: The structure and evolution of convection in a tropical cloud cluster. J. Atmos. Sci.,36, 437–457.

  • LeMone, M. A., and E. J. Zipser, 1980: Cumulonimbus vertical velocity events in GATE. Part I: Diameter, intensity and mass flux. J. Atmos. Sci.,37, 2444–2457.

  • Lewis, J. M., 1975: Test of the Ogura-Cho Model on a prefrontal squall line case. Mon. Wea. Rev.,103, 764–778.

  • Ligda, M. G. H., 1956: The radar observation of mature prefrontal squall lines in the midwestern United States. Swiss Aero Review, No. 11–12, 1–3.

  • Lopez, R. E., 1973: Cumulus convection and larger scale circulations II. Cumulus and mesoscale interactions. Mon. Wea. Rev.,101, 856–870.

  • Malkus, J. S., 1962: Large-scale interactions. The Sea: Ideas and Observations in Progress in the Study of the Seas, M. N. Hill, Ed., Vol. 1, Interscience Publishers, 88–294.

  • Matejka, T., and T. J. Schuur, 1991: The relation between vertical air motions and the precipitation band in the stratiform region of a squall line. Preprints, 25th Int. Conf. on Radar Meteorology, Paris, France, Amer. Meteor. Soc., 501–504.

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

  • Newton, C. W., 1950: Structure and mechanism of the prefrontal squall line. J. Meteor.,7, 210–222.

  • Nitta, T., 1975: Observational determination of cloud mass flux distributions. J. Atmos. Sci.,32, 73–91.

  • Ogura, Y., and H.-R. Cho, 1973: Diagnostic determination of cumulus cloud populations from observed large-scale variables. J. Atmos. Sci.,30, 1276–1286.

  • Riehl, H., and J. S. Malkus, 1958: On the heat balance in the equatorial trough zone. Geophysica (Helsinki), 6, 503–538.

  • ——, and J. Simpson, 1979: The heat balance of the equatorial trough zone, revisited. Contrib. Atmos. Phys.,52, 287–304.

  • Rosenfeld, D., D. B. Wolff, and E. Amitai, 1994: The window probability matching method for rainfall measurements with radar. J. Appl. Meteor.,33, 682–693.

  • ——, E. Amitai, and D. B. Wolff, 1995: Classification of rain regimes by the three-dimensional properties of reflectivity fields. J. Appl. Meteor.,34, 198–211.

  • Roux, F., 1988: The West African squall line observed on 23 June 1981 during COPT 81: Kinematics and thermodynamics of the convective region. J. Atmos. Sci.,45, 406–426.

  • Rutledge, S. A., and R. A. Houze Jr., 1987: A diagnostic modeling study of the trailing stratiform region of a midlatitude squall line. J. Atmos. Sci.,44, 2640–2656.

  • ——, ——, M. I. Biggerstaff, and T. Matejka, 1988: The Oklahoma–Kansas mesoscale convective system of 10–11 June 1985: Precipitation structure and single-Doppler radar analysis. Mon. Wea. Rev.,116, 1409–1430.

  • Ryde, J. W., 1946: The attenuation and radar echoes produced at centimetre wavelengths by various meteorological phenomena. Meteorological Factors in Radio Propagation, Physical Society, 169–188.

  • Simmons, A. J., 1982: The forcing of stationary wave motion by tropical diabatic heating. Quart. J. Roy. Meteor. Soc.,108, 503–534.

  • Simpson, J., R. F. Adler, and G. R. North, 1988: A proposed Tropical Rainfall Measuring Mission (TRMM) satellite. Bull. Amer. Meteor. Soc.,69, 278–295.

  • Smull, B. F., and R. A. Houze Jr., 1985: A midlatitude squall line with a trailing region of stratiform rain: Radar and satellite observations. Mon. Wea. Rev.,113, 117–133.

  • ——, and ——, 1987: Dual-Doppler radar analysis of a midlatitude squall line with a trailing region of stratiform rain. J. Atmos. Sci.,44, 2128–2148.

  • Sommeria, G., and J. Testud, 1984: COPT 81: A field experiment designed for the study of dynamics and electrical activity of deep convection in continental tropical regions. Bull. Amer. Meteor. Soc.,65, 4–10.

  • Srivastava, R. C., T. J. Matejka, and T. J. Lorello, 1986: Doppler radar study of the trailing anvil region associated with a squall line. J. Atmos. Sci.,43, 356–377.

  • Steiner, M., and R. A. Houze Jr., 1997: Sensitivity of the estimated monthly convective rain fraction to the choice of Z–R relation. J. Appl. Meteor.,36, 452–462.

  • ——, ——, and S. E. Yuter, 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor.,34, 1978–2007.

  • ——, J. A. Smith, S. J. Burges, C. V. Alonso, and R. W. Darden, 1999: Effect of bias adjustment and rain gauge data quality control on radar rainfall estimation. Water Resour. Res.,35, 2487–2503.

  • Tao, W.-K., and J. Simpson, 1984: Cloud interactions and merging: Numerical simulations. J. Atmos. Sci.,41, 2901–2917.

  • ——, S. Lang, J. Simpson, and R. Adler, 1993: Retrieval algorithms for estimating the vertical profiles of latent heat release: Their applications for TRMM. J. Meteor. Soc. Japan,71, 685–700.

  • Tokay, A., and D. A. Short, 1996: Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. J. Appl. Meteor.,35, 355–371.

  • Toracinta, E. R., K. I. Mohr, E. J. Zipser, and R. E. Orville, 1996: A comparison of WSR-88D reflectivities, SSM/I brightness temperatures, and lightning for mesoscale convective systems in Texas. Part I: Radar reflectivity and lightning. J. Appl. Meteor.,35, 902–918.

  • Wilheit, T. T., A. T. C. Chang, M. S. V. Rao, E. B. Rodgers, and J. S. Theon, 1977: A satellite technique for quantitatively mapping rainfall rates over the oceans. J. Appl. Meteor.,16, 551–560.

  • Yanai, M., S. Esbensen, and J.-H. Chu, 1973: Determination of bulk properties of tropical cloud clusters from large-scale heat and moisture budgets. J. Atmos. Sci.,30, 611–627.

  • Yuter, S. E., and R. A. Houze Jr., 1995: Three-dimensional kinematic and microphysical evolution of Florida cumulonimbus. Part II: Frequency distributions of vertical velocity, reflectivity, and differential reflectivity. Mon. Wea. Rev.,123, 1941–1963.

  • Zipser, E. J., 1969: The role of organized unsaturated convective downdrafts in the structure and rapid decay of an equatorial disturbance. J. Appl. Meteor.,8, 799–814.

  • ——, 1977: Mesoscale and convective-scale downdrafts as distinct components of squall-line structure. Mon. Wea. Rev.,105, 1568–1589.

  • ——, and K. Lutz, 1994: The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability? Mon. Wea. Rev.,122, 1751–1759.

Fig. 1.
Fig. 1.

Location of the Houston WSR-88D. The ellipse indicates the 150-km range limit of the data used in this study

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 2.
Fig. 2.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 1052 UTC on 2 May 1993. Intensity according to the grayscale. (b) Same as in (a) but for 1701 UTC on 22 Jun 1996. (c) SHY95 echo classification for (a). Light (dark) shading indicates convective (stratiform) classification. (d) Same as (c) but for the reflectivity field denoted in (b)

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 3.
Fig. 3.

Conceptual model of the precipitation trajectories and mean vertical motions throughout a leading-line trailing-stratiform squall line system (from Biggerstaff and Houze 1993)

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 4.
Fig. 4.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2132 UTC on 23 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a). The heavy, dark line in (a), (c), and (d) denotes the location of the vertical cross section shown in Fig. 5

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 5.
Fig. 5.

Vertical cross section of equivalent radar reflectivity (dBZe) for the 23 May 1993 squall line system at 2132 UTC. The cross section was taken from (X, Y) coordinates (−150, −32) to (50, −32) km in Fig. 4. Contours are every 5 dB as labeled, with various degrees of shading for values exceeding 35 dBZe. Horizontal tick marks are every 2 km. The traces at the top show the echo classification by the two algorithms as indicated with heavy (light) shading denoting stratiform (convective) rain. The heavy, dashed line in the lower portion of the diagram defines the altitude of the working level used in the SHY95 algorithm

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 6.
Fig. 6.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 0135 UTC on 6 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a). The heavy, dark line in (a), (c), and (d) denotes the location of the vertical cross section shown in Fig. 7

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 7.
Fig. 7.

Vertical cross section of equivalent radar reflectivity factor (dBZe) for the 6 May 1993 embedded convection case at 0135 UTC. The cross section was taken from (X, Y) coordinates (−139, −53) to (−32, 82) km in Fig. 6. Contours are every 5 dBZe as labeled, with various degrees of shading for values exceeding 35 dBZe. Horizontal tick marks are every 2 km. The traces at the top show the echo classification by the two algorithms as indicated with heavy (light) shading denoting stratiform (convective) rain. The heavy, dashed line in the lower portion of the diagram defines the altitude of the working level used in the SHY95 algorithm

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 8.
Fig. 8.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2004 UTC on 3 Jun 1994. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a)

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 9.
Fig. 9.

(a) Equivalent radar reflectivity factor (dBZe) at 3.0-km level above mean sea level from the Houston WSR-88D for 0706 UTC on 10 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a)

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 10.
Fig. 10.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 0706 UTC on 10 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a). (d) BL classification of the reflectivity field in (a)

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 11.
Fig. 11.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2004 UTC on 3 Jun 1994. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms for the case in which the convective radius in SHY95 has been doubled. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a) using twice the convective radius of the original algorithm. (d) BL classification of the reflectivity field in (a) using (c) as the first guess

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Fig. 12.
Fig. 12.

(a) Equivalent radar reflectivity factor (dBZe) at the working level from the Houston WSR-88D for 2132 UTC on 23 May 1993. Intensity according to the grayscale. (b) Difference between the SHY95 and BL algorithms for the case in which the convective radius in SHY95 has been doubled. Light (dark) shading indicates reclassification to convective (stratiform) echo. (c) SHY95 classification of the reflectivity field in (a) using twice the convective radius of the original algorithm. (d) BL classification of the reflectivity field in (a) using (c) as the first guess

Citation: Journal of Applied Meteorology 39, 12; 10.1175/1520-0450(2001)040<2129:AISFCS>2.0.CO;2

Table 1.

Summary of cases used

Table 1.
Table 2.

Summary of the differences between SHY95 and BL for the three example cases discussed in section 5a

Table 2.
Table 3.

Summary of the differences in rain-volume classification for SHY95 versus BL for the three example cases discussed in section 5a

Table 3.
Table 4.

Summary of the differences in echo area classification for SHY95 vs BL for all 2317 volume scans used in the study

Table 4.
Table 5.

Summary of the differences in rain-volume classification for SHY95 vs BL for all 2317 volume scans used in the study

Table 5.
Save