• Ackerman, T. P., and G. Stokes, 2003: The Atmospheric Radiation Measurement Program. Phys. Today, 56, 3845, doi:10.1063/1.1554135.

  • Atlas, D., C. W. Ulbrich, F. D. Marks Jr., R. A. Black, E. Amitai, P. T. Willis, and C. E. Samsury, 2000: Partitioning tropical oceanic convective and stratiform rains by draft strength. J. Geophys. Res., 105, 22592267, doi:10.1029/1999JD901009.

    • Search Google Scholar
    • Export Citation
  • Awaka, J., T. Iguchi, H. Kumagai, and K. Okamoto, 1997: Rain type classification algorithm for TRMM Precipitation Radar. Proc. IEEE 1997 Int. Geoscience Remote Sensing Symp., Singapore, IEEE, 16361638.

  • Bringi, V. N., C. R. Williams, M. Thurai, and P. T. May, 2009: Using dual-polarized radar and dual-frequency profiler for DSD characterization: A case study from Darwin, Australia. J. Atmos. Oceanic Technol., 26, 21072122, doi:10.1175/2009JTECHA1258.1.

    • Search Google Scholar
    • Export Citation
  • Clothiaux, E. E., T. P. Ackerman, G. G. Mace, K. P. Moran, R. T. Marchand, M. Miller, and B. E. Martner, 2000: Objective determination of cloud heights and radar reflectivities using a combination of active remote sensors at the ARM CART sites. J. Appl. Meteor., 39, 645665, doi:10.1175/1520-0450(2000)039<0645:ODOCHA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Clothiaux, E. E., and Coauthors, 2001: The ARM millimeter wave cloud radars (MMCRs) and the active remote sensing of clouds (ARSCL) value added product (VAP). DOE Tech. Memo. ARM VAP-002.1, 38 pp.

  • Doelling, I. G., J. Joss, and J. Riedl, 1998: Systematic variations of Z–R relationships from drop size distributions measured in northern Germany during seven years. Atmos. Res., 47–48, 635649, doi:10.1016/S0169-8095(98)00043-X.

    • Search Google Scholar
    • Export Citation
  • Donner, L. J., C. J. Seman, and R. S. Hemler, 2001: A cumulus parameterization including mass fluxes, convective vertical velocities, and mesoscale effects: Thermodynamic and hydrological aspects in a general circulation model. J. Climate, 14, 34443463, doi:10.1175/1520-0442(2001)014<3444:ACPIMF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ellis, S. M., and J. Vivekanandan, 2010: Water vapor estimates using simultaneous dual-wavelength radar observations. Radio Sci., 45, RS5002, doi:10.1029/2009RS004280.

    • Search Google Scholar
    • Export Citation
  • Ellis, S. M., and J. Vivekanandan, 2011: Liquid water content estimates using simultaneous S and Ka band radar measurements. Radio Sci., 46, RS2010, doi:10.1029/2010RS004361.

    • Search Google Scholar
    • Export Citation
  • Feng, Z., X. Dong, and B. Xi, 2009: A method to merge WSR-88D data with ARM SGP millimeter cloud radar data by studying deep convective systems. J. Atmos. Oceanic Technol., 26, 958971, doi:10.1175/2008JTECHA1190.1.

    • Search Google Scholar
    • Export Citation
  • Feng, Z., S. A. McFarlane, C. Schumacher, S. Ellis, J. Comstock, and N. Bharadwaj, 2014: Constructing a merged cloud-precipitation radar dataset for tropical convective clouds during the DYNAMO/AMIE Experiment at Addu Atoll. J. Atmos. Oceanic Technol., 31, 1021–1042, doi:10.1175/JTECH-D-13-00132.1.

    • Search Google Scholar
    • Export Citation
  • Fliegel, J. M., and C. Schumacher, 2012: Quality control and census of SMART-R observations from the DYNAMO/CINCY2011 field campaign. M.S. thesis, Dept. of Atmospheric Science, Texas A&M University, 84 pp.

  • Gamache, J. F., and R. A. Houze Jr., 1983: Water budget of a mesoscale convective system in the tropics. J. Atmos. Sci., 40, 18351850, doi:10.1175/1520-0469(1983)040<1835:WBOAMC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Geerts, B., and Y. Dawei, 2004: Classification and characterization of tropical precipitation based on high-resolution airborne vertical incidence radar. Part I: Classification. J. Appl. Meteor., 43, 15541566, doi:10.1175/JAM2158.1.

    • Search Google Scholar
    • Export Citation
  • Giangrande, S. E., M. J. Bartholomew, M. Pope, S. Collis, and M. P. Jensen, 2014: A summary of precipitation characteristics from the 2006–11 northern Australian wet seasons as revealed by ARM disdrometer research facilities (Darwin, Australia). J. Appl. Meteor. Climatol., 53, 1213–1231, doi:10.1175/JAMC-D-13-0222.1.

    • Search Google Scholar
    • Export Citation
  • Gu, J.-Y., A. Ryzhkov, P. Zhang, P. Neilley, M. Knight, B. Wolf, and D.-I. Lee, 2011: Polarimetric attenuation correction in heavy rain at C band. J. Appl. Meteor. Climatol, 50, 3958, doi:10.1175/2010JAMC2258.1.

    • Search Google Scholar
    • Export Citation
  • 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, 113121, doi:10.1175/1520-0469(1984)041<0113:SIOTMC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Henderson, P. W., and R. Pincus, 2009: Multiyear evaluations of a cloud model using ARM data. J. Atmos. Sci., 66, 29252936, doi:10.1175/2009JAS2957.1.

    • Search Google Scholar
    • Export Citation
  • Houze, R., 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc., 115, 425461, doi:10.1002/qj.49711548702.

    • Search Google Scholar
    • Export Citation
  • Houze, R., 1993: Cloud Dynamics. Academic Press, 573 pp.

  • Houze, R., 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc., 78, 21792196, doi:10.1175/1520-0477(1997)078<2179:SPIROC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Houze, R., S. Brodzik, C. Schumacher, S. E. Yuter, and C. R. Williams, 2004: Uncertainties in oceanic radar rain maps at Kwajalein and implications for satellite validation. J. Appl. Meteor., 43, 11141132, doi:10.1175/1520-0450(2004)043<1114:UIORRM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., T. Kozu, R. Meneghini, J. Awaka, and K. Okamoto, 2000: Rain-profiling algorithm for the TRMM Precipitation Radar. J. Appl. Meteor., 39, 20382052, doi:10.1175/1520-0450(2001)040<2038:RPAFTT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., T. M. Rickenbach, S. A. Rutledge, P. E. Ciesielski, and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12, 23972418, doi:10.1175/1520-0442(1999)012<2397:TCOTC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Keeler, R. J., J. Luts, and J. Vivekanandan, 2000: SPOLKA: NCAR’s polarimetric Doppler research radar. Proc. IEEE 2000 Int. Geoscience and Remote Sensing Symp., IGARSS 2000, Honolulu, HI, IEEE, 1570–1573.

  • Kollias, P., B. A. Albrecht, and F. D. Mark, 2003: Cloud radar observations of vertical drafts and microphysics in convective rain. J. Geophys. Res., 108, 4053, doi:10.1029/2001JD002033.

    • Search Google Scholar
    • Export Citation
  • Kollias, P., E. E. Clothiaux, B. A. Albrecht, M. A. Miller, K. P. Moran, and K. L. Johnson, 2005: The Atmospheric Radiation Measurement Program cloud profiling radars: An evaluation of signal processing and sampling strategies. J. Atmos. Oceanic Technol., 22, 930948, doi:10.1175/JTECH1749.1.

    • Search Google Scholar
    • Export Citation
  • Kollias, P., E. E. Clothiaux, M. A. Miller, B. A. Albrecht, G. L. Stephens, and T. P. Ackerman, 2007: Millimeter-wavelength radars: New frontier in atmospheric cloud and precipitation research. Bull. Amer. Meteor. Soc., 88, 16081624, doi:10.1175/BAMS-88-10-1608.

    • Search Google Scholar
    • Export Citation
  • Lhermitte, R., 1990: Attenuation and scattering of millimeter wavelength radiation by cloud and precipitation. J. Atmos. Oceanic Technol., 7, 464479, doi:10.1175/1520-0426(1990)007<0464:AASOMW>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci.,28, 702–708, doi:10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.

  • Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci.,29, 1109–1123, doi:10.1175/1520-0469(1972)029<1109:DOGSCC>2.0.CO;2.

  • Matrosov, S. Y., 2005: Attenuation-based estimates of rainfall rates aloft with vertically pointing Ka-band radars. J. Atmos. Oceanic Technol., 22, 4354, doi:10.1175/JTECH-1677.1.

    • Search Google Scholar
    • Export Citation
  • Mie, G., 1908: Beiträge zur Optik trüber Medien, speziell kolloidaler Metallösungen (Contribution to the optics of suspended media, specifically colloidal metal suspensions). Ann. Phys., 330, 377445, doi:10.1002/andp.19083300302.

    • Search Google Scholar
    • Export Citation
  • Ovtchinnikov, M., T. Ackerman, and R. Marchand, 2006: Evaluation of the multiscale framework using data from the Atmospheric Radiation Measurement Program. J. Climate, 19, 17161729, doi:10.1175/JCLI3699.1.

    • Search Google Scholar
    • Export Citation
  • Penide, G., V. V. Kumar, A. Protat, and P. T. May, 2013: Statistics of drop size distribution parameters and rain rates for stratiform and convective precipitation during the north Australian wet season. Mon. Wea. Rev., 141, 3222–3237, doi:10.1175/MWR-D-12-00262.1.

    • Search Google Scholar
    • Export Citation
  • Riihimaki, L. D., and C. N. Long, 2014: Spatial variability of surface irradiance measurements at the Manus ARM site. J. Geophys. Res. Atmos., 119, 54755491, doi:10.1002/2013JD021187.

    • Search Google Scholar
    • Export Citation
  • Rondanelli, R., and R. S. Lindzen, 2008: Observed variations in convective precipitation fraction and stratiform area with sea surface temperature. J. Geophys. Res., 113, D16119, doi:10.1029/2008JD010064.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., and E. Amitai, 1998: Comparison of WPMM versus regression for evaluating Z–R relationships. J. Appl. Meteor., 37, 12411249, doi:10.1175/1520-0450(1998)037<1241:COWVRF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., and C. W. Ulbrich, 2003: Cloud microphysical properties, processes, and rainfall estimation opportunities. Meteor. Monogr., No. 30, Amer. Meteor. Soc., 237–237, doi:10.1175/0065-9401(2003)030<0237:CMPPAR>2.0.CO;2.

  • 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, 26402656, doi:10.1175/1520-0469(1987)044<2640:ADMSOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., J. R. Campbell, J. Zhu, P. Kollias, M. Shupe, and C. Williams, 2005: Lidar and triple-wavelength Doppler radar measurements of the melting layer: A revised model for dark and brightband phenomena. J. Appl. Meteor., 44, 301312, doi:10.1175/JAM-2197.1.

    • Search Google Scholar
    • Export Citation
  • Schumacher, C., R. A. Houze Jr., and I. Kraucunas, 2004: The tropical dynamical response to latent heating estimates derived from the TRMM Precipitation Radar. J. Atmos. Sci., 61, 13411358, doi:10.1175/1520-0469(2004)061<1341:TTDRTL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., 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, 452462, doi:10.1175/1520-0450(1997)036<0452:SOTEMC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., and J. A. Smith, 2000: Reflectivity, rain rate, and kinetic energy flux relationships based on raindrop spectra. J. Appl. Meteor., 39, 19231940, doi:10.1175/1520-0450(2000)039<1923:RRRAKE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., R. A. Houze Jr., and S. E. Yuter, 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34, 19782007, doi:10.1175/1520-0450(1995)034<1978:CCOTDS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., J. A. Smith, and R. Uijlenhoet, 2004: A microphysical interpretation of radar reflectivity–rain rate relationships. J. Atmos. Sci., 61, 11141131, doi:10.1175/1520-0469(2004)061<1114:AMIORR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Stout, G. E., and E. A. Mueller, 1968: Survey of relationship between rainfall rate and radar reflectivity in the measurement of precipitation. J. Appl. Meteor., 7, 465474, doi:10.1175/1520-0450(1968)007<0465:SORBRR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tao, W. K., 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, 685700.

    • Search Google Scholar
    • Export Citation
  • Thurai, M., V. N. Bringi, and P. T. May, 2010: CPOL radar-derived drop size distribution statistics of stratiform and convective rain for two regimes in Darwin, Australia. J. Atmos. Oceanic Technol., 27, 932942, doi:10.1175/2010JTECHA1349.1.

    • Search Google Scholar
    • Export Citation
  • 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, 355371, doi:10.1175/1520-0450(1996)035<0355:EFTRSO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tokay, A., D. A. Short, C. R. Williams, W. L. Ecklund, and K. S. Gage, 1999: Tropical rainfall associated with convective and stratiform clouds: Intercomparison of disdrometer and profiler measurements. J. Appl. Meteor., 38, 302320, doi:10.1175/1520-0450(1999)038<0302:TRAWCA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Williams, C. R., W. L. Ecklund, and K. S. Gage, 1995: Classification of precipitation clouds in the tropics using 915-MHz wind profilers. J. Atmos. Oceanic Technol., 12, 9961012, doi:10.1175/1520-0426(1995)012<0996:COPCIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Xie, S., M. Zhang, J. S. Boyle, R. T. Cederwall, G. L. Potter, and W. Lin, 2004: Impact of a revised convective triggering mechanism on Community Atmosphere Model, version 2, simulations: Results from short-range weather forecasts. J. Geophys. Res., 109, D14102, doi:10.1029/2004JD004692.

    • Search Google Scholar
    • Export Citation
  • Yoneyana, K., D. Zhang, and C. N. Long, 2013: Tracking pulses of the Madden–Julian oscillation. Bull. Amer. Meteor. Soc.,94, 1871–1891, doi:10.1175/BAMS-D-12-00157.1.

  • Yuter, S. E., and R. A. Houze Jr., 1995: Three-dimensional kinematic and microphysical evolution of Florida cumulonimbus. Part II: Frequency distribution of vertical velocity, reflectivity, and differential reflectivity. Mon. Wea. Rev., 123, 19411963, doi:10.1175/1520-0493(1995)123<1941:TDKAME>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yuter, S. E., and R. A. Houze Jr., 1997: Measurements of raindrop size distribution over the Pacific warm pool and implications for Z–R relations. J. Appl. Meteor., 36, 847867, doi:10.1175/1520-0450(1997)036<0847:MORSDO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zuluaga, M. D., and R. A. Houze Jr., 2013: Evolution of population of precipitation convective systems over the equatorial Indian Ocean in active phases of the Madden–Julian oscillation. J. Atmos. Sci., 70, 27132725, doi:10.1175/JAS-D-12-0311.1.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Doppler moments measured by KAZR on 14 Oct 2011. (a) Doppler spectral width (σd), (b) Doppler velocity (Vd), and (c) radar reflectivity (Ze).

  • View in gallery

    Collocated Ze observed (a) by KAZR and SMART-R at 10-min temporal resolution and (b) by KAZR and S-PolKa at 15-min resolution on 14 Oct 2011.

  • View in gallery

    Comparison of CFADs of Ze observed by (left to right) KAZR, SMART-R, and S-PolKa for different rainfall-rate segments. Red lines are normalized data fractions at a given height. Black lines are mean Ze at a given height.

  • View in gallery

    Differences between ETH observed by SMART-R and KAZR as a function of (a) S-PolKa-derived rain rate and (b) surface-measured rain rate. Differences between ETH observed by KAZR and S-PolKa as a function of (c) S-PolKa-derived rain rate and (d) surface-measured rain rate. Gray triangles are for all collocated data. Blue triangles are for the subset with Pdiff < 30%. Solid black lines and vertical bars are the means and standard deviations of the subset (Pdiff < 30%). The quantity Pdiff is the deviation fraction between surface-measured and precipitation-radar-derived rainfall rates; see the text for more detail about Pdiff.

  • View in gallery

    PDFs of KAZR parameters vs surface-measured precipitation. (a) Near-surface Ze (Ze_sf), (b) GVD between 4 and 5 km, (c) near-surface Doppler velocity (Vd_sf), and (d) GAZ below 1 km. In (a), the KAZR rainfall rate is fitted from Ze_sf in a Z–R power-law relation and is shown as black lines for convective and stratiform rain profiles; the fitting coefficients are listed in Table 3. The red line is S-PolKa Z–R relation [R= (aZ)1/b, where a = 0.027 366 and b = 1.44].

  • View in gallery

    Vertical profiles of (a) Ze and (b) cumulative Ze from cloud tops measured by KAZR on 14 Oct 2011. Legends include the time, gradients of Ze below 1 km (G_Ze), GAZ below 1 km, and surface-measured precipitation rate (Pmet).

  • View in gallery

    Joint PDFs of (a) KAZR near-surface Ze (Ze_sf) and GAZ below 1 km, (b) GVD between 4 and 5 km and GAZ below 1 km, and (c) GVD between 4 and 5 km and Ze_sf.

  • View in gallery

    Grid mean (a) surface rain rate and (c) median deviation and (b) two-parameter fitted surface rain rate as a function of KAZR near-surface Ze (Ze_sf) and GAZ below 1 km. Fitting equation and coefficients are listed in Table 4.

  • View in gallery

    (a) Rain rates from surface measurement at AMF-2 (black crosses) and derived from SMART-R (green asterisks) and S-PolKa (red asterisks). (b)–(g) Rain classifications from KAZR, KAZR collocated with SMART-R, SMART-R, KAZR collocated with S-PolKa, and S-PolKa measurements, respectively. Blue represents stratiform rain, and red represents convective rain.

  • View in gallery

    PDF of all surface-measured precipitation (asterisks) and corresponding surface-measured precipitation distribution of convective (solid line; fraction = 18%) and stratiform (diamonds; fraction = 82%) profiles classified by KAZR.

  • View in gallery

    (a) Number distributions of S-PolKa rain rate for all (black), convective (red), and stratiform (green) precipitation profiles classified by S-PolKa (solid line) and collocated KAZR (dashed line). (b) As in (a), but for selected rainy profiles with Pdiff < 30%. (c) As in (a), but for SMART-R (solid line) and collocated KAZR (dashed line). (d) As in (c), but for selected rainy profiles with Pdiff < 30%.

  • View in gallery

    Diagram of rain classifications from (left) collocated KAZR and S-PolKa and (right) collocated KAZR and SMART-R with Pdiff < 30%.

  • View in gallery

    Composite DSD distributions measured by the disdrometer for KAZR-classified convective (62) and stratiform (313) rain profiles for a mesoscale convective system passing over the AMF-2 site on 14–15 Jan 2012.

  • View in gallery

    Composite CFADs of (left) radar reflectivity and (right) Doppler velocity for KAZR-classified (a),(b) convective and (c),(d) stratiform rain profiles during the AMIE/DYNAMO project.

  • View in gallery

    (a) PDF of relative differences between surface-measured rain rates and the fitted rain rate as a function of surface Ze as shown in Fig. 5a. (b) As in (a), but the fitted rain rate is a function of surface Ze and GAZ as shown in Fig. 8b. The color scales are nonlinear. (c) Means (curves) and standard deviations (vertical lines) of relative differences between surface-measured rain rates and the fitted rain rates as a function of surface Ze (black) and as a function of surface Ze and GAZ (red).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 321 302 28
PDF Downloads 204 193 22

Stratiform and Convective Precipitation Observed by Multiple Radars during the DYNAMO/AMIE Experiment

View More View Less
  • 1 * University of Wyoming, Laramie, Wyoming
  • | 2 McGill University, Montreal, Quebec, Canada
  • | 3 Pacific Northwest National Laboratory, Richland, Washington
  • | 4 University of Miami, Miami, Florida
  • | 5 Centre for Australian Weather and Climate Research, Melbourne, Victoria, Australia
© Get Permissions
Full access

Abstract

In this study, methods of convective/stratiform precipitation classification and surface rain-rate estimation based on the Atmospheric Radiation Measurement Program (ARM) cloud radar measurements were developed and evaluated. Simultaneous and collocated observations of the Ka-band ARM zenith radar (KAZR), two scanning precipitation radars [NCAR S-band/Ka-band Dual Polarization, Dual Wavelength Doppler Radar (S-PolKa) and Texas A&M University Shared Mobile Atmospheric Research and Teaching Radar (SMART-R)], and surface precipitation during the Dynamics of the Madden–Julian Oscillation/ARM MJO Investigation Experiment (DYNAMO/AMIE) field campaign were used. The motivation of this study is to apply the unique long-term ARM cloud radar observations without accompanying precipitation radars to the study of cloud life cycle and precipitation features under different weather and climate regimes. The resulting convective/stratiform classification from KAZR was evaluated against precipitation radars. Precipitation occurrence and classified convective/stratiform rain fractions from KAZR compared favorably to the collocated SMART-R and S-PolKa observations. Both KAZR and S-PolKa radars observed about 5% precipitation occurrence. The convective (stratiform) precipitation fraction is about 18% (82%). Collocated disdrometer observations of two days showed an increased number concentration of small and large raindrops in convective rain relative to dominant small raindrops in stratiform rain. The composite distributions of KAZR reflectivity and Doppler velocity also showed distinct structures for convective and stratiform rain. These evidences indicate that the method produces physically consistent results for the two types of rain. A new KAZR-based, two-parameter [the gradient of accumulative radar reflectivity Ze (GAZ) below 1 km and near-surface Ze] rain-rate estimation procedure was developed for both convective and stratiform rain. This estimate was compared with the exponential Z–R (reflectivity–rain rate) relation. The relative difference between the estimated and surface-measured rainfall rates showed that the two-parameter relation can improve rainfall estimation relative to the Z–R relation.

Corresponding author address: Min Deng, Department of Atmospheric Science, University of Wyoming, Dept. 3038, 1000 East University Avenue, Laramie, WY 82071. E-mail: mdeng2@uwyo.edu

Abstract

In this study, methods of convective/stratiform precipitation classification and surface rain-rate estimation based on the Atmospheric Radiation Measurement Program (ARM) cloud radar measurements were developed and evaluated. Simultaneous and collocated observations of the Ka-band ARM zenith radar (KAZR), two scanning precipitation radars [NCAR S-band/Ka-band Dual Polarization, Dual Wavelength Doppler Radar (S-PolKa) and Texas A&M University Shared Mobile Atmospheric Research and Teaching Radar (SMART-R)], and surface precipitation during the Dynamics of the Madden–Julian Oscillation/ARM MJO Investigation Experiment (DYNAMO/AMIE) field campaign were used. The motivation of this study is to apply the unique long-term ARM cloud radar observations without accompanying precipitation radars to the study of cloud life cycle and precipitation features under different weather and climate regimes. The resulting convective/stratiform classification from KAZR was evaluated against precipitation radars. Precipitation occurrence and classified convective/stratiform rain fractions from KAZR compared favorably to the collocated SMART-R and S-PolKa observations. Both KAZR and S-PolKa radars observed about 5% precipitation occurrence. The convective (stratiform) precipitation fraction is about 18% (82%). Collocated disdrometer observations of two days showed an increased number concentration of small and large raindrops in convective rain relative to dominant small raindrops in stratiform rain. The composite distributions of KAZR reflectivity and Doppler velocity also showed distinct structures for convective and stratiform rain. These evidences indicate that the method produces physically consistent results for the two types of rain. A new KAZR-based, two-parameter [the gradient of accumulative radar reflectivity Ze (GAZ) below 1 km and near-surface Ze] rain-rate estimation procedure was developed for both convective and stratiform rain. This estimate was compared with the exponential Z–R (reflectivity–rain rate) relation. The relative difference between the estimated and surface-measured rainfall rates showed that the two-parameter relation can improve rainfall estimation relative to the Z–R relation.

Corresponding author address: Min Deng, Department of Atmospheric Science, University of Wyoming, Dept. 3038, 1000 East University Avenue, Laramie, WY 82071. E-mail: mdeng2@uwyo.edu

1. Introduction

In the mid-1990s the U.S. Department of Energy Atmospheric Radiation Measurement Program (ARM) deployed vertically pointing 35-GHz (λ = 8.6 mm) millimeter Doppler cloud radars (MMCR) as the centerpiece of their observing instruments for monitoring and studying clouds and precipitation processes at several climatologically distinct locations (Ackerman and Stokes 2003). The ARM Program has continuously operated MMCRs at its permanent atmospheric research sites in Alaska, Oklahoma, Darwin (Australia), Manus, and Nauru Islands for many years. Unprecedented long-term time series of cloud and precipitation observations have been collected at these sites. These long-term ARM observations are widely used for a variety of purposes, such as characterizing cloud microphysics and their radiative impacts, improving process-level understanding of cloud life cycles, and model evaluations (Xie et al. 2004; Ovtchinnikov et al. 2006; Henderson and Pincus 2009; Riihimaki and Long 2014).

In the tropics, stratiform and convective components of precipitation in mesoscale convective systems play different roles in the heating and mass transport (Gamache and Houze 1983; Houze 1989; Johnson et al. 1999), both having important impacts on the global general circulation (Hartmann et al. 1984; Schumacher et al. 2004). Heating associated with convective rain is positive throughout the troposphere, while heating profiles in stratiform regions feature heating above the freezing level and cooling below. Accurate convective and stratiform partitioning is thus important to correctly model circulation response to precipitation (Hartmann et al. 1984; Tao et al. 1993; Donner et al. 2001).

Convective precipitation forms mainly through collection of cloud particles (coalescence or riming) in areas where vertical motion w is strong (w > 1 m s−1). In contrast, stratiform precipitation grows mainly by water vapor diffusion on the surface of ice particles that are detrained from the convective region and are characterized by a slower ascent (w < 1 m s−1). Rutledge and Houze (1987) and Houze (1989, 1993, 1997) provided extensive discussions of the terminology, physics, and observational aspects of the classification. The use of dynamic attributes led to a draft magnitude method to separate the rain types as proposed in Atlas et al. (2000).

Observations from scanning precipitation radar show obviously different patterns for convective and stratiform precipitation, given their different dynamical and microphysical attributes. Weak horizontal gradients of reflectivity in stratiform precipitation region are distinct from sharp peaks of the reflectivity core in the most vigorous convective regions (e.g., Steiner et al. 1995). Also, a pronounced layer of enhanced reflectivity and Doppler velocity in a shallow zone just below the 0°C level (i.e., bright band), caused by melting aggregating ice particles, is an unambiguous indicator of the presence of stratiform precipitation (Houze 1997). These features provide the basis for algorithms that use radar reflectivity structures to distinguish convective from stratiform regions (Steiner et al. 1995; Rosenfeld and Amitai 1998). Such algorithms have been widely applied to measurements from scanning precipitation radars. An example is the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar rain-type classifications, produced by merging convective–stratiform separation methods based on the vertical structure, such as brightband identification, echo-top height, and maximum reflectivity (Awaka et al. 1997), and on horizontal variability of the echo, such as the absence of peaks and local echo intensity (Steiner et al. 1995).

A convective/stratiform precipitation classification algorithm together with rain-rate estimates at high temporal resolution is desirable to study convective cloud life cycle and precipitation processes using the long-term MMCR observations. This is especially so because at the tropical ARM sites (e.g., Manus and Nauru), long-term MMCR observations are available but without precipitation radars that can provide convective–stratiform precipitation estimates using the established method discussed above.

Precipitation classification using MMCR measurements alone is, however, challenging because of non-Rayleigh scattering effects from raindrops with diameter larger than ~2 mm and strong signal attenuation (Lhermitte 1990; Kollias et al. 2003; 2007; Matrosov 2005; Feng et al. 2009). Scattering in the non-Rayleigh regime increasingly suppresses the dynamic range of observed radar reflectivity values with decreasing radar wavelengths. Entering the non-Rayleigh scattering regime also affects observed mean Doppler velocities of raindrops from vertically pointing (profiling) radars. The reduced contributions of large particles to the Doppler spectrum with decreasing radar wavelength increases the relative contribution of small raindrops and shifts observed mean Doppler velocities to lower magnitudes. At 35 GHz, attenuation from water vapor is significant, especially in the tropics (Kollias et al. 2007). The specific humidity in the tropical boundary layer can reach values as high as 20–25 g kg−1 and this can cause one-way attenuation of up to 0.35 dB km−1 (10−0.035, or 92%, km−1) at 35 GHz. Signal attenuation induced by hydrometeors, especially in the liquid phase, at these high radar frequencies is even more significant. For example, clouds with 1 g m−3 of liquid water content can cause a two-way attenuation of 1 dB km−1 at 35 GHz. Strong signal attenuation limits the penetration of MMCR signals to a few kilometers into the precipitation layer at high rainfall rates (Kollias et al. 2003), and thus no vertically oriented high-reflectivity structure is observed. Nonetheless, these strong hydrometer interactions offer an alternative way to classify precipitation types and estimate rain rates. The objective of this study is to demonstrate the possibility of classifying precipitation types and estimating rain rates using MMCR observations alone.

To study initiation of the Madden–Julian oscillation (MJO; Madden and Julian 1971, 1972), the Dynamics of the Madden–Julian Oscillation/ARM MJO Investigation Experiment (DYNAMO/AMIE) field campaign was conducted in the tropical Indian Ocean and the surrounding regions from 1 October 2011 through 31 March 2012 (Yoneyana et al. 2013). During this field campaign, three radar systems were deployed at Addu Atoll of the Maldives: the Ka-band ARM zenith-pointing cloud radar (KAZR), the National Center for Atmospheric Research (NCAR) S-band/Ka-band Dual Polarization, Dual Wavelength Doppler Radar (S-PolKa), and the Texas A&M University C-band Shared Mobile Atmospheric Research and Teaching Radar (SMART-R). The three radars took simultaneous and overlapping measurements from 10 October 2011 to 15 January 2012. These radar observations, along with collocated surface rainfall and sounding measurements, provide a unique dataset to explore the possibility of a precipitation classification algorithm and rain-rate estimation based solely on KAZR measurements. This study aims at developing and evaluating such an algorithm using this unique dataset.

This paper is organized as follows. Section 2 describes the three radar systems and other datasets used in this study. Section 3 compares the radar reflectivity features observed by cloud and precipitation radars. In section 4, a precipitation classification algorithm based solely on KAZR observations is introduced and evaluated. Section 5 presents a new KAZR rain-rate estimate, which is evaluated against surface rain measurements. A summary and conclusions are given in section 6.

2. Collocated data description

The ARM KAZR is a 35-GHz (λ = 8.6 mm) vertically pointing Doppler radar, which is a major upgrade to the MMCR. The KAZR replaces the MMCR at Alaska, Oklahoma, Darwin, and Manus since 2011 and utilizes a new digital receiver that provides range-resolved measurements from approximately 30 m to almost 20 km in altitude. The KAZR calibration process includes correcting the reflectivity field by applying corrected radar constants, applying a multiplicative correction to mean Doppler velocity or spectrum width fields (e.g., a sign correction), and correcting any radar parameters that were improperly recorded in the original ingested files. Feng et al. (2014) showed that for nonprecipitating clouds (Ze < 0 dBZ) KAZR reflectivity compares well to well-calibrated S-PolKa reflectivity. A KAZR Active Remote Sensing of Clouds (ARSCL) product (Clothiaux et al. 2000, 2001) was generated from both general and cirrus modes to provide best-estimated cloud boundaries by combining KAZR and lidar measurements with vertical and temporal resolutions of 30 m and 4 s, respectively. In this new product, KAZR observations were also corrected for water vapor attenuation and velocity aliasing. In addition to the KAZR, the ARM Mobile Facility 2 (AMF-2) deployment at the AMIE site of Gan Island (0.7°S, 73.2°E) included the ARM Surface Meteorology System (SMET) that provided 1-min statistics of surface wind speed, wind direction, air temperature, relative humidity, and barometric pressure with mainly conventional in situ sensors as well as rain rate from a Vaisala, Inc., acoustic rain sensor. Balloonborne sounding systems were launched at a minimum rate of eight per day. The KAZR ARSCL data were interpolated in bins of 90 m and 30 s in time in order to compare with surface rain measurements as well as the precipitation radars, which scanned over the KAZR with much lower temporal resolutions as shown later. An example of KAZR Doppler moments is shown in Fig. 1 for 14 October 2011. KAZR observed four precipitating systems during this day. The first three lasted less than 2 h with fall streak structures in the Doppler moments (Fig. 1b). The last system lasted more than 3 h. During 2100–2200 and 2300–0000 UTC, the reflectivity profiles show near-surface radar reflectivity Ze up to 35 dBZ with high vertical and temporal variations in Doppler velocity Vd and Ze. The quantity Ze was attenuated to less than −30 dBZ at about 5 km. During 2200–2300 UTC, Vd and Ze at the melting layer were horizontally smooth, showing the vertical discontinuity or a plateau at about 5 km. The near-surface Ze during this period was relatively weaker.

Fig. 1.
Fig. 1.

Doppler moments measured by KAZR on 14 Oct 2011. (a) Doppler spectral width (σd), (b) Doppler velocity (Vd), and (c) radar reflectivity (Ze).

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

The S-PolKa is an advanced dual-polarimetric precipitation scanning radar (1° beamwidth). Its dual-polarimetric capabilities yield improved precipitation estimates compared to conventional radars, as well as real-time identification of hydrometeor types and humidity-gradient layers, cold pools, and detailed structure and evolution of precipitating deep convection (Keeler et al. 2000; Ellis and Vivekanandan 2010, 2011). During the DYNAMO/AMIE field campaign, it was located on Hithadhoo Island (0.63°S, 73.10°E), 8.62 km northwest from the KAZR (see Fig. 1 in Feng et al. 2014). Its surveillance scans were followed by a set of range–height indicator (RHI) scans. The entire sequence of scanning patterns was repeated at a 15-min interval. During DYNAMO/AMIE, the S-PolKa operated at a range gate resolution of 150 m and the RHI scans over the AMF-2 site were recorded at steps of 0.5° ranging from an elevation angle of −0.5° to 65°. An objective precipitation classification algorithm based on horizontal radar reflectivity texture (Steiner et al. 1995; Yuter and Houze 1997) was applied to the gridded S-PolKa RHI reflectivity field (with 0.5-km horizontal and vertical resolution) to separate radar echoes into convective and stratiform components, and specifically tuned for the DYNAMO/AMIE experiment (Zuluaga and Houze 2013). The parameters used for the Steiner et al. algorithm are listed in Table 1. For reflectivity profiles, time–height series data from the S-PolKa were reconstructed using its RHI scans in the direction of the KAZR site (141° azimuth), and were linearly interpolated to match the 90-m KAZR vertical grid (Feng et al. 2014). The S-PolKa radar reflectivity data used in this study are the final, quality-controlled data that were fully calibrated for noise correction and atmospheric attenuation (http://www.eol.ucar.edu/projects/dynamo/spol/).

Feng et al. (2014) showed that collocated measurements from the S-PolKa and KAZR agree well in weak to moderately precipitating congestus and deep clouds (rain rate < 5 mm h−1), while KAZR signals are severely attenuated in heavily precipitating convective clouds. Since attenuation from liquid hydrometeors at S band is small and negligible, in this case convective/stratiform rain partitioning from the S-PolKa provides valuable targets for evaluating a KAZR-based convective–stratiform classification. In this study, rain rates derived from S-PolKa were computed using a “hybrid” polarimetrically tuned version of the Z–R relation with differential reflectivity (ZDR) and specific differential phase (KDP). (Details of the algorithm and parameters used for the rain-rate estimation are available at http://www.eol.ucar.edu/projects/dynamo/spol/parameters/rain_rate/rain_rates.html.)

The SMART-R was located on Hithadhoo Island (0.61°S, 73.09°E), 11 km from the KAZR during the DYNAMO/AMIE field campaign. With a 1.5° beamwidth, it was operated on a 10-min scan cycle. Three RHI scans were directed over the AMF-2 site at the beginning of a scan cycle, lasting for about 10 s. For convective–stratiform precipitation classification, the same Steiner et al. algorithm is applied to the gridded SMART-R reflectivity (with 3-km horizontal and 0.5-km vertical resolution) at 3-km height, with slightly lower thresholds for convective echoes (Table 1) to account for various differences in radar beamwidth, wavelength, and gridded dataset. A more detailed description of the convective–stratiform modification and processing methods for version 1 of the SMART-R dataset can be found in Fliegel and Schumacher (2012).

Table 1.

Parameters used for Steiner et al. (1995) convective–stratiform partitioning algorithm on S-PolKa and SMART-R data. Z: reflectivity.

Table 1.

The time–height series using SMART-R RHI reflectivity data were reconstructed centered on the KAZR, using the same approach as done with the S-PolKa data. Rain rates were calculated from SMART-R reflectivity following the Z–R relationship derived from the Mirai Indian Ocean cruise for the Study of the Madden–Julian Oscillation Convection Onset (MISMO) experiment (C. Schumacher 2013, personal communication).

Figure 2 shows an example of collocated Ze between the KAZR and SMART-R (Fig. 2a) and the KAZR and S-PolKa (Fig. 2b) on 14 October 2011. First of all, these cloud and precipitation radars all detected the four precipitating systems. For the last system, KAZR reflectivity is heavily attenuated with almost no echo return above 10 km where precipitation radar echo is up to 10 dBZ. The detailed cloud variation is blurred relative to Fig. 1 because of the lower temporal resolutions. As compared with the S-PolKa, the SMART-R suffers attenuation in heavy rain and is also affected by the resonant scattering from raindrops larger than 5 mm (Gu et al. 2011). This example highlights a unique aspect of KAZR observations for cloud and precipitation study with its high temporal resolution.

Fig. 2.
Fig. 2.

Collocated Ze observed (a) by KAZR and SMART-R at 10-min temporal resolution and (b) by KAZR and S-PolKa at 15-min resolution on 14 Oct 2011.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

3. Vertical distributions of Ze observed from cloud and precipitation radars

To explore KAZR’s potentials for stratiform and convective classification and rain-rate estimation, we compared the vertical structures of Ze measured by the KAZR, SMART-R, and S-PolKa. As previously mentioned, KAZR signals are affected by two-way attenuation from water vapor and liquid hydrometers. While water vapor attenuation has been corrected in the KAZR data (it is negligible for C- and S-band precipitation radar), the difference in Ze at the same height level between the precipitation radars and KAZR is mainly caused by precipitation attenuation in addition to other factors, such as differences in radar sensitivities, fields of view, ground clutter echoes, non-Rayleigh scattering at Ka band for particles larger than about 2 mm, and Bragg scattering at S band.

The contoured-frequency-by-altitude diagrams (CFADs) of radar parameters such as reflectivity and vertical velocity are often used to show the statistical behavior of radar signals (Yuter and Houze 1995). CFADs of Ze from the KAZR, SMART-R, and S-PolKa for different rain-rate segments are compared in Fig. 3. For surface rain rates larger than 10 mm h−1 (top panels in Fig. 3), near-surface (at about 200 m) KAZR returns (32 ± 5 dBZ) are weaker than those of the SMART-R and S-PolKa (40 ± 5 dBZ). The mean returns of the precipitation radars remain about 40 dBZ up to 3 km with larger variation at a given height when compared with the KAZR. The attenuation by rain causes KAZR signals to decrease with height with an averaged gradient of about 8 dB km−1. A normalized data fraction is defined as the ratio of rainy to total radar observation profiles at each range gate. It is about 40% for the KAZR and about 85% for the SMART-R and S-PolKa at 5 km for rain rates greater than 10 mm h−1. The mean altitude of the 0°C level is about 4.8 km according to the sounding temperature profiles. This means for high rain rates, about 50% of the time the KAZR cannot provide melting layer information such as the bright band. The radar reflectivity attenuation due to rain has been utilized to retrieve rain rates by taking advantage of the nearly linear relation between specific attenuation and rain rates (Matrosov 2005).

Fig. 3.
Fig. 3.

Comparison of CFADs of Ze observed by (left to right) KAZR, SMART-R, and S-PolKa for different rainfall-rate segments. Red lines are normalized data fractions at a given height. Black lines are mean Ze at a given height.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

For rain rates of 5–10 mm h−1, the precipitation radars show obvious brightband signatures around the melting layer and a relatively vertical uniform radar return below that. The KAZR shows two modes below 5 km: one with a linear attenuation gradient of about 5 dB km−1 and the other with a smaller gradient and a radar reflectivity plateau at the melting layer.

For rain rates of 1–5 mm h−1, the KAZR profiles show a clear Ze increase from the top to bottom of the melting layer forming a radar reflectivity plateau. Bright bands in the SMART-R and S-PolKa are still observed with a relative uniform layer of radar return below the melting layer, as for rain rates of 5–10 mm h−1.

For rain rates less than 1 mm h−1, vertical distributions of Ze from the KAZR and precipitation radars look more similar than those for the higher rain rates. The modes of the near-surface echoes from the three radars are close to 20 dBZ. This indicates that the large discrepancy of near-surface Ze among the radars under higher rain rates might be caused by the rain accumulated on the radome as well as the non-Rayleigh scattering of the KAZR. Second, the radar signals decrease slightly from melting layer toward the surface because of evaporation or droplet breakup, which was also observed in Sassen et al. (2005). Such a Ze vertical structure with rain rates less than 1 mm h−1 clearly indicates that precipitation is not vertically uniform. Again, the bright bands of the precipitation radars are more pronounced than that of the KAZR for this rain-rate segment. The vertical distribution also shows that the KAZR and S-PolKa are more sensitive than the SMART-R to detect more hydrometers with signals weaker than −20 dBZ above 10 km. It should also be noted that the SMART-R has a surface clutter issue below 1 km as indicated by the sharp increase in mean reflectivity value close to the surface.

Echo-top heights (ETH) detected by the collocated cloud and precipitation radars in precipitating clouds are expected to be different because of their different sensitivities and attenuation, as shown in Fig. 4. For rain rates less than 1 mm h−1, ETH is 0.3–3 km higher for the KAZR than for the SMART-R; for rain rates larger than 5 mm h−1, ETH is 1–6 km lower for the KAZR than for the SMART-R.

Fig. 4.
Fig. 4.

Differences between ETH observed by SMART-R and KAZR as a function of (a) S-PolKa-derived rain rate and (b) surface-measured rain rate. Differences between ETH observed by KAZR and S-PolKa as a function of (c) S-PolKa-derived rain rate and (d) surface-measured rain rate. Gray triangles are for all collocated data. Blue triangles are for the subset with Pdiff < 30%. Solid black lines and vertical bars are the means and standard deviations of the subset (Pdiff < 30%). The quantity Pdiff is the deviation fraction between surface-measured and precipitation-radar-derived rainfall rates; see the text for more detail about Pdiff.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

Because of the differences in radar sample volumes discussed in the previous sections, some discrepancy is expected. To alleviate the discrepancy due to instrument sensitivities and fields of view and other reasons mentioned above, we defined the derived–measured rain-rate difference Pdif to select a subdataset of the collocated rainy profiles:
eq1
where Pmet is the rain rate measured by rain gauges, and Pradar estimated by the precipitation radar. The maximum value of Pdiff is 50% when either Pmet or Pradar is zero, which means either the precipitation radars have false alarms or miss the precipitating events. This may be the inherent difference between a point-measured rain versus areal rain from the precipitation radar at any given instant. Proper Pdiff can ensure the KAZR and scanning radars observe the same precipitating event. Besides, Pdiff is independent of the KAZR, which is an appropriate condition for KAZR classification evaluation with the precipitation radars in the following sections.

With Pdiff less than 30%, the ETH differences (blue triangles) in Fig. 4 tend to have a higher correlation with surface-measured or radar-derived rain rates. Therefore, the threshold of Pdiff is chosen as 30% so that the KAZR and scanning radars observe the same precipitating events without losing too much collocated rainy profiles. In comparison with the S-PolKa (see Figs. 4c and 4d), the KAZR ETH underestimation for high rain rate (>5 mm h−1) is evident. However, for weaker precipitation, the mean ETH difference is close to 0 km.

4. KAZR convective/stratiform precipitation classification

a. KAZR parameters

A convective/stratiform classification algorithm based on single profile radar moments of the KAZR is desired. The draft magnitude method (Atlas et al. 2000) cannot be used because there is no direct and simultaneous air motion measurement. The KAZR observes cloud passing over in an Euler framework. The internal cloud variation and associated temporal variation make it almost impossible to apply the horizontal gradient method derived from scanning precipitation radars.

Important parameters for rain detection from a vertically pointing radar include the near-surface radar reflectivity, signal attenuation, and the gradient of Doppler velocity (GVD) at the melting layer. The first two are related to the volume concentration of hydrometeors. Ze and rain-relations (Z–R) have been widely used for rain estimation based on precipitation radar measurements. The associated drop size difference in stratiform and convective rain leads to statistically significant difference in Z–R relations (Yuter and Houze 1997; Rosenfeld and Ulbrich 2003). The distribution of near-surface Ze of all profiles as a function of surface-measured rain rate (Fig. 5a) shows two distinct modes. For weak rain (<1 mm h −1), Ze has a high correlation with the rain rate. When the rain rate is larger than 1 mm h−1, the reduction of radar reflectivity due to non-Rayleigh scattering, as well as the attenuation of water accumulated on the KAZR radome, is evident. When the rain rate is larger than 5 mm h−1, precipitation attenuation further reduces the slope relative to that in light rain. Matrosov (2005) also showed that the attenuation effects dominate in the observed Ze values for rainfall rates of about 7 mm h−1 and greater.

Fig. 5.
Fig. 5.

PDFs of KAZR parameters vs surface-measured precipitation. (a) Near-surface Ze (Ze_sf), (b) GVD between 4 and 5 km, (c) near-surface Doppler velocity (Vd_sf), and (d) GAZ below 1 km. In (a), the KAZR rainfall rate is fitted from Ze_sf in a Z–R power-law relation and is shown as black lines for convective and stratiform rain profiles; the fitting coefficients are listed in Table 3. The red line is S-PolKa Z–R relation [R= (aZ)1/b, where a = 0.027 366 and b = 1.44].

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

Williams et al. (1995) developed a technique of stratiform rain detection based on a strong bright band as well as a high melting-layer GVD due to phase transformation for the 915-MHz radar measurements. Detection of a high GVD is a useful indicator of melting and the stratiform nature of precipitation. This technique was applied by Geerts and Dawei (2004) to downward-pointing airborne X-band radar without much attenuation and without being obscured by precipitation. The distribution of GVD at 4–5 km from the KAZR observations is presented as a function of the surface-measured rain rate in Fig. 5b. A large portion of the data is clustering near zero GVD, which means that the KAZR signal is either attenuated below the melting layer in convective rain or it rains below the melting layer. For those profiles with radar echo extending above the melting level, GVD does have a tendency to increase with the rain rates, similar to the results in Geerts and Dawei (2004). The KAZR GVD distribution also indicates two modes: the focal is at 5.0 m s−1 km−1 for the light rain mode and about 7.0 m s−1 km−1 for rain rates larger than 1 mm h−1. However, the absence of GVD data at the melting layer and the large variation of GVD in relation to rain rate suggest that GVD alone is not a good indicator for KAZR precipitation classification.

Also presented in Fig. 5 is near-surface Doppler velocity Vd_sf at about 200 m, which is mainly determined by hydrometer size and vertical air motions. Figure 5c shows that Vd_sf is generally larger than 2 m s−1 and has a slight tendency to increase with rain rate. For light rain, Vd_sf increases from about 3 to 6 m s−1. For heavier rain Vd has a mode at about 6.5 m s−1, while it levels off as the rain rate increases further. This is because contributions of hydrometers with diameters larger than 2 mm are suppressed by non-Rayleigh scattering (e.g., Mie 1908; Kollias et al. 2005).

Besides the near-surface measurements of Ze and Vd, the Ze vertical structure should also be taken into account in the classification algorithm. Several vertical profiles of Ze on 14 October 2011 are shown in Fig. 6. At 1690 UTC (profile B) and 2182 UTC (profile C), Ze seems to decrease linearly with height at a gradient of −14.7 and −13.1 dB km−1 below 1 km, respectively. However, the collocated rain rate at 2182 UTC from the surface measurements is about 22 mm h−1, while the surface rain rate at 1690 UTC is less than 1 mm h−1. This indicates that the large gradient at 1690 UTC is more or less related to the internal cloud variation with height, while the large gradient at 2182 UTC is mainly due to heavy rain attenuation. We did find that surface Ze at 1690 UTC is 18 dBZ, which is much smaller than that at 2182 UTC. This comparison indicates that near-surface Ze is an important parameter to separate these two types of conditions.

Fig. 6.
Fig. 6.

Vertical profiles of (a) Ze and (b) cumulative Ze from cloud tops measured by KAZR on 14 Oct 2011. Legends include the time, gradients of Ze below 1 km (G_Ze), GAZ below 1 km, and surface-measured precipitation rate (Pmet).

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

At 2241 UTC (profile D) and 2319 UTC (profile E), Ze shows a brightband structure with a plateau at around 5 km. Ze slightly increases with height below 1 km in profile D, while it remains almost constant at about 32 dBZ from the surface up to 3 km in profile E, resulting in Ze gradients of 4.9 and 0 dB km−1, respectively. The collocated surface rain rates are 0.8 and 5.4 mm h−1, respectively. These two profiles, as well as profiles A, B, and C, indicate that Ze gradients are not unambiguously linearly related to rain rates, while near-surface Ze seems to be more or less closely related to rain rates. Additionally, gradients are expected to be negative because of attenuation but actually can be negative or positive such as in profiles A, D, and E because of natural cloud variations. To avoid this situation, we propose a new parameter: accumulative Ze, defined as the accumulative sum of Ze from the echo top to the current range gate. We calculate the gradient of accumulative Ze (GAZ) below 1 km using linear fitting. For profiles A and D, where Ze increase with height, GAZs below 1 km are close to 0. In contrast, for profiles B and C, Ze are strongly attenuated, and GAZs are very large. GAZ are always negative and mostly monotonically related to rain rate. For simplicity, we use the absolute value of GAZ hereafter.

In Fig. 5d, the distribution of GAZ is plotted as a function of the surface rain rate. The GAZ increases almost log-linearly with rain rates larger than 1 mm h−1 from 1 to about 20 dB km−1. For lower rain rates, the GAZs mainly concentrate around 0.3 dB km−1, but GAZ larger than 1 dB km−1 are also possible. These profiles with GAZ larger than 1 dB km −1 but small near-surface Ze are more or less similar to profile B in Fig. 6.

b. Stratiform and convective rain classification criteria and results

Among all parameters shown in Fig. 5, near-surface Ze and GAZ have the best correlations with surface-measured rainfall rates. The joint probability density function (PDF) distribution and grid mean surface rain rates as a function of KAZR near-surface Ze and the GAZ are shown in Figs. 7a and 8a. The first impression is that Fig. 7a is very similar to Fig. 5d. That is because near-surface Ze has a good correlation with the surface rain rate. Near-surface Ze for precipitating clouds (Fig. 7a) ranges from 5 to about 35 dBZ, which indicates that the lowest near-surface Ze for KAZR precipitation identification can be set at around 5 dBZ.

Fig. 7.
Fig. 7.

Joint PDFs of (a) KAZR near-surface Ze (Ze_sf) and GAZ below 1 km, (b) GVD between 4 and 5 km and GAZ below 1 km, and (c) GVD between 4 and 5 km and Ze_sf.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

Fig. 8.
Fig. 8.

Grid mean (a) surface rain rate and (c) median deviation and (b) two-parameter fitted surface rain rate as a function of KAZR near-surface Ze (Ze_sf) and GAZ below 1 km. Fitting equation and coefficients are listed in Table 4.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

The joint PDF and grid mean surface precipitation as a function of GAZ and near-surface Ze in Figs. 7a and 8a shows three modes. The light precipitation mode, where near-surface Ze is less than 25 dBZ, corresponds to mean surface rain rates less than 2 mm h−1 (Fig. 8a). For this mode, the GAZ ranges from 0.1 to 10 dB km−1. However, the rain-rate contours are almost parallel with GAZ but vary strongly with near-surface Ze. This means that GAZ for light rain are more or less associated with internal cloud and precipitation variation rather than precipitation attenuation.

When near-surface Ze is larger than 25 dBZ, the rain rate increases with both near-surface Ze and GAZ. The heavy precipitation mode is at the upper-right corner of Figs. 7a and 8a, where GAZ is larger than 8 dB km−1 and near-surface Ze is larger than about 32 dBZ. Its grid mean rain rate is larger than 10 mm h−1. In between is the moderate precipitation mode associated with GAZ between 0.5 and 8 dB km−1 and near-surface Ze between 25 and 32 dBZ. This mode has a focal point at 30 dBZ near-surface Ze and 1 dB km−1 GAZ as indicated in Fig. 7a.

The surface rain rate is not necessarily low for stratiform rain or high for convective rain. Therefore, we also take advantage of the GVD at melting layer. The joint PDF of near-surface Ze and GAZ versus GVD between 4 and 5 km are plotted in Figs. 7b and 7c, respectively. It is clear that for the heavy precipitation mode with near-surface Ze larger than 32 dBZ and GAZ larger than 8 dBZ km−1, GVD is zero because KAZR signals are generally attenuated just as seen in Fig. 3. For the moderate precipitation mode, however, GVD starts to increase from 0 to about 6 m s−1 km−1 as GAZ decrease from 8 to 1 dB km−1 (Fig. 7b). This clearly shows that there is a portion of stratiform rain profiles with large GVD and a significant bright band. There is also a fraction of profiles in the light precipitation mode with near-surface Ze less than 25 dBZ but GVD larger than 4 m s−1 km−1 (Fig. 7c).

Given the distinct features of the KAZR variables described above and shown in Fig. 7, we identify potential tropical convective rain from the KAZR measurements if one of the following criteria is met:

  1. surface Ze > 28 dBZ and GAZ > 1.3 dB km−1;

  2. 27 dBZ < surface Ze ≤ 28 dBZ and GAZ > 1.5 dB km−1;

  3. 26 dBZ < surface Ze ≤ 27 dBZ and GAZ > 2.0 dB km−1; or

  4. surface Ze < 25 dBZ and GAZ > 6.0 dB km−1.

Such criteria select cases with rain rates larger than 2.5 mm h−1 depicted in yellow color in Fig. 8a. Then GVD between 4 and 5 km is used to further separate tropical convective and stratiform rain among identified potential convective rain profiles. If GVD is larger than 3.5 m s−1 km−1, which indicates a significant bright band exists, then these profiles are classified as stratiform. Otherwise, they are classified as convective rain profiles.

This algorithm is applied to KAZR observations during October 2011–February 2012. An example of the classification results from KAZR observations on 14 October 2011 is presented in Fig. 9. With 30-s temporal resolution, the KAZR shows higher variability of precipitation classification even within one precipitation system (Fig. 9b). Such a high variability is consistent with that of surface rain rates (Fig. 9a). Collocated classifications from the SMART-R–KAZR and S-PolKa–KAZR are also plotted in Figs. 9c–g. The SMART-R, with 10-min resolution, tends to show a higher variability of precipitation types than the S-PolKa, but certain instantaneous profiles show different classifications from the KAZR as expected because of completely different methodology applied (horizontal texture versus vertical structure). Nevertheless, the three radars captured the large transition between convective and stratiform classifications.

Fig. 9.
Fig. 9.

(a) Rain rates from surface measurement at AMF-2 (black crosses) and derived from SMART-R (green asterisks) and S-PolKa (red asterisks). (b)–(g) Rain classifications from KAZR, KAZR collocated with SMART-R, SMART-R, KAZR collocated with S-PolKa, and S-PolKa measurements, respectively. Blue represents stratiform rain, and red represents convective rain.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

In total, the KAZR detected 5% precipitation occurrence, among which, 18% (82%) is convective (stratiform). Figure 10 shows the distribution of surface rain rates of the KAZR-classified stratiform and convective rain profiles. All rain rates measured at the surface show two modes: a dominant mode at about 0.3 mm h−1 and secondary mode at about 3–5 mm h−1. The KAZR-classified convective rain profiles seem to correspond to the mode peaked at 3–5 mm h−1 and those larger than 10 mm h−1, while the profiles classified as stratiform rain correspond to the primary mode at 0.3–0.5 mm h−1.

Fig. 10.
Fig. 10.

PDF of all surface-measured precipitation (asterisks) and corresponding surface-measured precipitation distribution of convective (solid line; fraction = 18%) and stratiform (diamonds; fraction = 82%) profiles classified by KAZR.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

c. Comparison with precipitation radar classification

The KAZR classification results are compared with those from the precipitation radars (Fig. 11 and Table 2). Among 11 712 collocated S-PolKa –KAZR profiles, the S-PolKa detected 576 profiles as precipitation (5% precipitation occurrence), of which 24% is convective and 76% stratiform. The KAZR detected 584 profiles as precipitation, with 16% and 84% as convective and stratiform rain, respectively. For the subset of the data with Pdiff less than 30% (Table 3), the S-PolKa and KAZR detected 394 precipitating cases (about 33% data reduction). The convective (stratiform) rain fractions are 25% (75%) and 17% (83%) for the S-PolKa and KAZR, respectively.

Fig. 11.
Fig. 11.

(a) Number distributions of S-PolKa rain rate for all (black), convective (red), and stratiform (green) precipitation profiles classified by S-PolKa (solid line) and collocated KAZR (dashed line). (b) As in (a), but for selected rainy profiles with Pdiff < 30%. (c) As in (a), but for SMART-R (solid line) and collocated KAZR (dashed line). (d) As in (c), but for selected rainy profiles with Pdiff < 30%.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

Table 2.

The partitioned fraction of convective/stratiform rain from collocated S-PolKa and KAZR, and collocated SMART-R and KAZR in corresponding to Figs. 11a,c.

Table 2.
Table 3.

As in Table 2, but for selected rainy profiles with the Pdiff < 30% corresponding to Figs. 11b,d.

Table 3.

Among 17 568 collocated SMART-R–KAZR profiles, the SMART-R detected 1532 precipitation profiles (9% precipitation occurrence), and convective (stratiform) rain accounts for 14% (86%) of them. The KAZR detected only 960 as precipitation profiles (5% precipitation occurrence), which means that the SMART-R identified about 500 (4% in occurrence) more rainy instances than the KAZR. The SMART-R rain-rate distributions for the SMART-R–KAZR collocated profiles are shown in Fig. 11c. For classified convective profiles the KAZR has a rain-rate distribution very similar to that from the SMART-R. For stratiform rain, however, the SMART-R detected significantly more instances with rate rates less than 1 mm h −1 than the KAZR. For the subset of the data with Pdiff less than 30% in Fig. 11d, the SMART-R and KAZR detected 771 precipitating profiles. The data reduction percentages for the SMART-R and KAZR are 50% and 20%, respectively. The convective and stratiform rain fractions in the subsets are 16% (17%) and 84% (83%) for the SMART-R (KAZR) classification, respectively. The rain-rate distribution in Fig. 11d for the subset data indicates that the SMART-R and KAZR classifications are remarkably similar; discrepancies between the SMART-R and KAZR distributions especially decreased for rain rates < 1 mm h−1. This suggests that the 500-instance difference in the original collocated data comparison may be caused by SMART-R false alarm, which could be due to strong surface clustering signals, resulting in 4% overestimation of precipitation identification.

The overlapped range of rain rates between convective and stratiform profiles for the S-PolKa, SMART-R, and KAZR in Fig. 11 is between 1 and 10 mm h−1. Such an overlap range is consistent with TRMM and surface precipitation radar observations (Rondanelli and Lindzen 2008).

Figure 12 shows profile-by-profile comparisons of collocated data to illustrate how many cases are hit by both the KAZR and precipitation radars and how many are missed by either of them. For the collocated KAZR and SMART-R subset data, 590 and 78 profiles are hits by both for stratiform and convective, respectively. About 100 profiles (15%) are identified into different types by the KAZR and SMART-R. For the KAZR and S-PolKa, 282 and 58 profiles are hits by both radars for stratiform and convective, respectively; 54 profiles (15%) are identified into different classes by KAZR and S-PolKa. Those mismatches might be related to the high variability of precipitation itself and the different temporal and spatial resolutions of the radars.

Fig. 12.
Fig. 12.

Diagram of rain classifications from (left) collocated KAZR and S-PolKa and (right) collocated KAZR and SMART-R with Pdiff < 30%.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

From the above comparison of stratiform and convective rain classification of the KAZR and precipitation radars, we can conclude that statistically the KAZR is able to effectively detect precipitating profiles with about 5% occurrence during the study period. The resulting fractions of convective and stratiform rain are comparable to those from the collocated precipitation radar measurements.

Recently, an emerging technique to classify precipitation type based on in situ measurement of drop size distribution (DSD) has been developed (Bringi et al. 2009; Thurai et al. 2010, Giangrande et al. 2014). From 10 to 28 January 2012 during DYNAMO/AMIE, a 2D video disdrometer (RD-80) from Nagoya University, Japan, was set up within a few meters of the KAZR. It offers 1-min integrated data for 20 size bins with drop diameter ranging from 0.3 to 5 mm. The final data have been quality controlled (http://www.arm.gov/campaigns/amf2012ancdis). A mesoscale convective system passed over the AMF-2 site on 14–15 January 2012. Among those two days, there are 367 KAZR-classified rainy profiles with disdrometer measurements. The corresponding DSDs for KAZR-classified convective (58) and stratiform (309) profiles are shown in Fig. 13. For both convective and stratiform rain, the composite DSD follows a gamma distribution. The number concentrations of convective rain are one order of magnitude larger than those of stratiform rain, which increases the D0 and N0 parameters when fitting the DSD as a gamma distribution. Such a distribution change in DSD observations (i.e., from dominance of small raindrops in stratiform rain to high number concentration of small and large raindrops in convective rain) is consistent with the underlining assumption to separate convective and stratiform rain from disdrometer measurements (Tokay and Short 1996; Tokay et al. 1999; Penide et al. 2013), suggesting the KAZR-based precipitation classification method produces physically consistent result with droplet distributions.

Fig. 13.
Fig. 13.

Composite DSD distributions measured by the disdrometer for KAZR-classified convective (62) and stratiform (313) rain profiles for a mesoscale convective system passing over the AMF-2 site on 14–15 Jan 2012.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

The composite CFADs of KAZR reflectivity and Doppler velocity for KAZR-classified convective and stratiform rain are shown in Fig. 14. For convective rainy profiles, reflectivity decreases with increasing altitude because of attenuation. In contrast, stratiform rain profiles have either uniform vertical reflectivity below the melting level or slightly increasing reflectivity with increasing altitude. A moderate reflectivity plateau rather than a well-defined bright band is evident between 4 and 5 km. This Ze pattern is compared with Fig. 8 in Houze et al. (2004) and Fig. 1 in Yuter and Houze (1995). It shows that for convective rain, the mean Ze of a cloud radar decreases with height with a larger gradient than those of a precipitation radar. For stratiform rain, Ze from both cloud and precipitation radars slightly increase with height from near surface up to the melting layer.

Fig. 14.
Fig. 14.

Composite CFADs of (left) radar reflectivity and (right) Doppler velocity for KAZR-classified (a),(b) convective and (c),(d) stratiform rain profiles during the AMIE/DYNAMO project.

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

Doppler velocity profiles for stratiform and convective rain appear to be very similar above the melting level with values of about 1 m s−1. For stratiform rain, the mean Doppler velocity shows a weak increase with decreasing height above 5 km followed by a rapid increase at the melting level between 5 and 4 km and a weak decrease below 4 km, which was also observed by Tokay et al. (1999) using a wind profiler data. They indicated that the slight increase of mean Doppler velocity from 0.5 to 1.5 m s−1 at levels with temperature lower than −5°C is attributed to increases in the mass of ice crystals mainly due to vapor deposition. A further rapid increase from 1.5 to 6 m s−1 is caused by the phase transition at and below melting layer. A decrease below is partially due to the atmospheric density effect on terminal velocities of raindrops such that raindrops fall faster aloft than near the surface, and it is also due to microphysical processes, namely, evaporation and collision–coalescence.

For convective rain, the mean Doppler velocity shows little variation above 5 km with a monotonic increase with decreasing height. In contrast to stratiform rain, the variation of the Doppler velocity in convective rain is significantly influenced by vertical air motion, which is higher in convective systems. The rapid increase of Doppler velocity between 5 and 3.5 km is again caused by phase transition. The Doppler velocity below 4 km in convective rain tends to be higher than that in stratiform rain. In addition, the vertical gradient of the Doppler velocity below 4 km is opposite for these two types of rain, which was also observed in Tokay et al. (1999). Such opposite gradients are associated with effects of air density and evaporation on the stratiform size distribution and coalescence growth on the convective size distribution.

5. Surface rain-rate estimation with KAZR measurements

Traditional hourly surface station measurements are not able to match high temporal variations of cloud and precipitation observed by the MMCR or KAZR. Along with its cloud observation capability, the cloud radar would be a complementary tool to provide better temporal variability for studies on the convection life cycle. Since the early 1950s, a number of Z–R relationships have been developed based on either a direct approach of measuring rainfall rates by rain gauges and Ze by precipitation radars or an indirect approach of calculating both parameters from the measured raindrop spectra (Stout and Mueller 1968; Doelling et al. 1998; Iguchi et al. 2000; Steiner and Houze 1997; Steiner and Smith 2000). Rosenfeld and Ulbrich (2003) discussed differences in Z–R relationships between maritime and continental, convective, transition, stratiform, and orographic precipitation. Yuter and Houze (1997) argued that the variability of drop sizes in each of the precipitation types precludes a distinct separation of Z–R relation by types. It seems that variations in the microphysical properties under the effects of evaporation, accretion, and coalescence attribute to the differences in the Z–R relation by types. The microphysical aspects of the Z–R relations were examined by Steiner et al. (2004). The exponential coefficients of Z–R relation range from 1 to 3 depending on the raindrop size distribution (Stout and Mueller 1968; Steiner et al. 2004).

For cloud radars, such as the KAZR, non-Rayleigh scattering of drops with diameters larger than 2 mm and attenuation by heavy precipitation also deform the possible Z–R relation in convective and stratiform rain as shown in Fig. 5a. For the first-order approximation, we fitted the Z–R relations for the KAZR-classified two rain types using surface rain-rate measurements as shown in solid black lines in Fig. 5a. The fitting result for KAZR stratiform rain (Table 4) is very close to the S-PolKa Z–R relation (red line in Fig. 5a). The PDF of relative differences of rain rates between the KAZR-fitting and surface measurements are plotted as a function of the surface-measured rain rate in Fig. 15a. The mode of relative difference of light rain is around zero, but the scattering shows that Z–R relation tends to overestimate some light rain, which is related to the more than second-order scattering of rain rates for certain Ze in Fig. 5a. The majority of data is distributed in the relative different region of less than ±70%. For convective rain, the fitted exponential coefficient in the Z–R relation is less than 1 because of attenuation.

Table 4.

List of coefficients of fitted KAZR rain rate as a function Ze: R = (aZ)1/b for convective and stratiform rain profiles as shown in Fig. 5a. The Ze is in reflectivity decibels, and R is in millimeters per hour.

Table 4.
Fig. 15.
Fig. 15.

(a) PDF of relative differences between surface-measured rain rates and the fitted rain rate as a function of surface Ze as shown in Fig. 5a. (b) As in (a), but the fitted rain rate is a function of surface Ze and GAZ as shown in Fig. 8b. The color scales are nonlinear. (c) Means (curves) and standard deviations (vertical lines) of relative differences between surface-measured rain rates and the fitted rain rates as a function of surface Ze (black) and as a function of surface Ze and GAZ (red).

Citation: Journal of Applied Meteorology and Climatology 53, 11; 10.1175/JAMC-D-13-0311.1

In addition to near-surface Ze, the surface rain rate is also well correlated with the GAZ for heavy precipitations in Fig. 8a. Figure 8c shows that there are 100% median deviations in the surface-measured rain when Ze_sf < 10 dB and GAZ < 10 dB km−1 while surface Ze is less than 30 dB. But for data heavily distributed grids (PDF larger than 0.5%) in Fig. 7a, the median deviation of rain rates from the grid mean (Fig. 8c) is less than 50%. Such a well-correlated relation suggests that the rain rates can be estimated with GAZ and near-surface Ze based on the KAZR observations:
eq2
where the units are reflectivity decibels for Ze, millimeters per hour for R, and decibels per kilometer for GAZ. It is referred as a two-parameter relation. The fitting results are shown in Fig. 8b. The fitting coefficients are listed in Table 5. The constant c is larger for convective rain than for stratiform rain, indicating better correlation of GAZ with rain rates in heavy precipitation as shown in Figs. 5d and 8a. The PDFs of relative difference between KAZR-fitting and surface-measured rain rates are plotted as a function of surface-measured rain rate in Fig. 15b. First, this two-parameter relation also tends to overestimate some light rain as the Z–R relation. Second, as compared with Fig. 15a, Fig. 15b shows two clear focal points, indicating that rain rates from two-parameter relation are less scattered from the measured rain rates. The frequency from the two-parameter fitting with the relative rain-rate difference of less than 50% are clearly higher than that from the Z–R fitting. The comparison of mean and standard deviation of relative differences between surface-measured and fitted rain rates in Fig. 15c further shows that the two-parameter relation improves the correlation between the fitted and measured rain rates and hence decreases the standard deviation by about 50% in comparison to the Z–R relation. The decrease in the mean bias is most prominent for heavy rain rates by up to 20%. The mean bias for both fitting methods can be larger than 100% when rain rates are less than 0.1 mm h−1, where the radar near-surface Ze is less than 10 dBZ as shown in Figs. 5a and 8c.
Table 5.

List of coefficients of fitted KAZR rain rate as a function Ze and the gradient of accumulative Ze (GAZ) below 1 km: log10(R) = a + b × Ze + c × log10(GAZ) × Ze for convective and stratiform rain profiles as shown in Fig. 8b. The Ze is in reflectivity decibels, R is in millimeters per hour, and GAZ is in decibels per kilometer.

Table 5.

6. Discussion and summary

This study used collocated and simultaneous measurements of cloud and precipitation radars and surface rain measurement on Addu Atoll during the DYNAMO/AMIE field campaign from 1 October 2011 to 15 January 2012. Such a comprehensive dataset provides an excellent opportunity for the development and evaluation of an objective precipitation classification scheme and rain-rate estimation based only on cloud radar measurements.

The comparison of vertical distributions of cloud and precipitation radar measurements shows that for heavy precipitation, the KAZR suffers from strong attenuation with an average Ze gradient 8 dBZ km−1. For rain rates less than 10 mm h−1, SMART-R and S-PolKa echoes show significant bright bands at the melting layer while KAZR echoes show a plateau. For rain rates of less than 1 mm h−1, observed Ze structures by the KAZR and precipitation radar are very similar, indicating the obedience of Rayleigh scattering and probably negligible rain attenuation for the KAZR.

Among many KAZR parameters, near-surface Ze and GAZ below 1 km show the best correlations with surface rain rates. The Doppler velocity gradient associated with a significant bright band at the melting layer provides an extra constraint for the stratiform rain identification. A new KAZR-based precipitation classification method using those three variables is developed. Both KAZR and S-PolKa radars observed about 5% precipitation occurrence during the field campaign while the SMART-R identified 4% more. The profile-by-profile comparisons do show differences among the cloud and precipitation radars and highlight the value of high-temporal-resolution KAZR measurements. For the subset data with Pdiff less than 30%, precipitation occurrence and classified convective–stratiform rain fractions from the KAZR compared favorably to those collocated SMART-R and S-PolKa data. The KAZR-classified convective (stratiform) precipitation fraction is about 18% (82%). Disdrometer observations show an increased number concentration of small and large raindrops in KAZR-identified convective rain compared to dominant small raindrops in stratiform rain. The composite CFADs of KAZR reflectivity and Doppler velocity also show two distinct structures for convective and stratiform rain. These results indicate that the method offers physically consistent results for these two types of rain.

Previous rain estimations of cloud radars (Matrosov 2005) are based on attenuation. In this study, based on collocated surface rain-rate and KAZR measurements, a new KAZR-based, two-parameter (the GAZ below 1 km and near-surface Ze) rain rate estimation is developed for both convective and stratiform rain. It is evaluated using relative differences between estimated and surface-measured rain rates. Results show that the two-parameter method can improve rain-rate estimation by decreasing the mean bias up to 20% in the heavy rain and standard deviation by about 50% when compared with the Z–R relationship. According to S-PolKa domain (150-km radius) averaged precipitation during the entire DYNAMO/AMIE field campaign, the convective–stratiform rain ratio is 2.7. The convective–stratiform rain ratios from surface-measured and the two-parameter estimated rain rates from the KAZR are 2.73 and 2.84, respectively. These results further confirm that the profiling cloud radar is a feasible and alternative tool to study precipitation as well as clouds. However, when applying the technique developed from this study to long-term MMCR or other cloud radar measurements, the criteria and coefficients developed in this study may need to be modified cautiously depending on the wavelengths and specifications of different cloud radar systems. Moreover, the mean bias can be larger than 100% when the rain rates are less than 0.1 mm h−1. Because the results derived from this study are based on measurement in a tropical convective regime, it might not be applicable to warm rain in the midlatitude.

The motivation of the precipitation classification and rain-rate estimation algorithm developed in this study is to utilize long-term MMCR observations for precipitation related studies. Whenever there are precipitation radar observations, the cloud radar would be a useful complementary tool to provide better temporal and vertical sampling of precipitating clouds. For long-term cloud radar observations without accompanying precipitation radar observations or surface rain measurements, the convective–stratiform precipitation classification and rain-rate estimation developed in this study using cloud radar alone would be crucial and useful.

Acknowledgments

This work is supported by the Office of Science of the U.S. Department of Energy as part of the Atmospheric Systems Research Program and uses data from the Atmospheric Radiation Measurement Climate Research Facility. We are grateful for the NCAR S-PolKa team, Texas A&M University SMART-R team, and ARM AMF2 team for providing the quality radar and precipitation data used in this study. The authors thank Drs. Sergey Y. Matrosov from CIRES/University of Colorado and NOAA/ESRL and Samuel Haimov, Dave Leon, and Zhien Wang from University of Wyoming for their suggestion and comments and three anonymous reviewers for their thorough comments and constructive criticisms.

REFERENCES

  • Ackerman, T. P., and G. Stokes, 2003: The Atmospheric Radiation Measurement Program. Phys. Today, 56, 3845, doi:10.1063/1.1554135.

  • Atlas, D., C. W. Ulbrich, F. D. Marks Jr., R. A. Black, E. Amitai, P. T. Willis, and C. E. Samsury, 2000: Partitioning tropical oceanic convective and stratiform rains by draft strength. J. Geophys. Res., 105, 22592267, doi:10.1029/1999JD901009.

    • Search Google Scholar
    • Export Citation
  • Awaka, J., T. Iguchi, H. Kumagai, and K. Okamoto, 1997: Rain type classification algorithm for TRMM Precipitation Radar. Proc. IEEE 1997 Int. Geoscience Remote Sensing Symp., Singapore, IEEE, 16361638.

  • Bringi, V. N., C. R. Williams, M. Thurai, and P. T. May, 2009: Using dual-polarized radar and dual-frequency profiler for DSD characterization: A case study from Darwin, Australia. J. Atmos. Oceanic Technol., 26, 21072122, doi:10.1175/2009JTECHA1258.1.

    • Search Google Scholar
    • Export Citation
  • Clothiaux, E. E., T. P. Ackerman, G. G. Mace, K. P. Moran, R. T. Marchand, M. Miller, and B. E. Martner, 2000: Objective determination of cloud heights and radar reflectivities using a combination of active remote sensors at the ARM CART sites. J. Appl. Meteor., 39, 645665, doi:10.1175/1520-0450(2000)039<0645:ODOCHA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Clothiaux, E. E., and Coauthors, 2001: The ARM millimeter wave cloud radars (MMCRs) and the active remote sensing of clouds (ARSCL) value added product (VAP). DOE Tech. Memo. ARM VAP-002.1, 38 pp.

  • Doelling, I. G., J. Joss, and J. Riedl, 1998: Systematic variations of Z–R relationships from drop size distributions measured in northern Germany during seven years. Atmos. Res., 47–48, 635649, doi:10.1016/S0169-8095(98)00043-X.

    • Search Google Scholar
    • Export Citation
  • Donner, L. J., C. J. Seman, and R. S. Hemler, 2001: A cumulus parameterization including mass fluxes, convective vertical velocities, and mesoscale effects: Thermodynamic and hydrological aspects in a general circulation model. J. Climate, 14, 34443463, doi:10.1175/1520-0442(2001)014<3444:ACPIMF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ellis, S. M., and J. Vivekanandan, 2010: Water vapor estimates using simultaneous dual-wavelength radar observations. Radio Sci., 45, RS5002, doi:10.1029/2009RS004280.

    • Search Google Scholar
    • Export Citation
  • Ellis, S. M., and J. Vivekanandan, 2011: Liquid water content estimates using simultaneous S and Ka band radar measurements. Radio Sci., 46, RS2010, doi:10.1029/2010RS004361.

    • Search Google Scholar
    • Export Citation
  • Feng, Z., X. Dong, and B. Xi, 2009: A method to merge WSR-88D data with ARM SGP millimeter cloud radar data by studying deep convective systems. J. Atmos. Oceanic Technol., 26, 958971, doi:10.1175/2008JTECHA1190.1.

    • Search Google Scholar
    • Export Citation
  • Feng, Z., S. A. McFarlane, C. Schumacher, S. Ellis, J. Comstock, and N. Bharadwaj, 2014: Constructing a merged cloud-precipitation radar dataset for tropical convective clouds during the DYNAMO/AMIE Experiment at Addu Atoll. J. Atmos. Oceanic Technol., 31, 1021–1042, doi:10.1175/JTECH-D-13-00132.1.

    • Search Google Scholar
    • Export Citation
  • Fliegel, J. M., and C. Schumacher, 2012: Quality control and census of SMART-R observations from the DYNAMO/CINCY2011 field campaign. M.S. thesis, Dept. of Atmospheric Science, Texas A&M University, 84 pp.

  • Gamache, J. F., and R. A. Houze Jr., 1983: Water budget of a mesoscale convective system in the tropics. J. Atmos. Sci., 40, 18351850, doi:10.1175/1520-0469(1983)040<1835:WBOAMC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Geerts, B., and Y. Dawei, 2004: Classification and characterization of tropical precipitation based on high-resolution airborne vertical incidence radar. Part I: Classification. J. Appl. Meteor., 43, 15541566, doi:10.1175/JAM2158.1.

    • Search Google Scholar
    • Export Citation
  • Giangrande, S. E., M. J. Bartholomew, M. Pope, S. Collis, and M. P. Jensen, 2014: A summary of precipitation characteristics from the 2006–11 northern Australian wet seasons as revealed by ARM disdrometer research facilities (Darwin, Australia). J. Appl. Meteor. Climatol., 53, 1213–1231, doi:10.1175/JAMC-D-13-0222.1.

    • Search Google Scholar
    • Export Citation
  • Gu, J.-Y., A. Ryzhkov, P. Zhang, P. Neilley, M. Knight, B. Wolf, and D.-I. Lee, 2011: Polarimetric attenuation correction in heavy rain at C band. J. Appl. Meteor. Climatol, 50, 3958, doi:10.1175/2010JAMC2258.1.

    • Search Google Scholar
    • Export Citation
  • 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, 113121, doi:10.1175/1520-0469(1984)041<0113:SIOTMC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Henderson, P. W., and R. Pincus, 2009: Multiyear evaluations of a cloud model using ARM data. J. Atmos. Sci., 66, 29252936, doi:10.1175/2009JAS2957.1.

    • Search Google Scholar
    • Export Citation
  • Houze, R., 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc., 115, 425461, doi:10.1002/qj.49711548702.

    • Search Google Scholar
    • Export Citation
  • Houze, R., 1993: Cloud Dynamics. Academic Press, 573 pp.

  • Houze, R., 1997: Stratiform precipitation in regions of convection: A meteorological paradox? Bull. Amer. Meteor. Soc., 78, 21792196, doi:10.1175/1520-0477(1997)078<2179:SPIROC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Houze, R., S. Brodzik, C. Schumacher, S. E. Yuter, and C. R. Williams, 2004: Uncertainties in oceanic radar rain maps at Kwajalein and implications for satellite validation. J. Appl. Meteor., 43, 11141132, doi:10.1175/1520-0450(2004)043<1114:UIORRM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., T. Kozu, R. Meneghini, J. Awaka, and K. Okamoto, 2000: Rain-profiling algorithm for the TRMM Precipitation Radar. J. Appl. Meteor., 39, 20382052, doi:10.1175/1520-0450(2001)040<2038:RPAFTT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., T. M. Rickenbach, S. A. Rutledge, P. E. Ciesielski, and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12, 23972418, doi:10.1175/1520-0442(1999)012<2397:TCOTC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Keeler, R. J., J. Luts, and J. Vivekanandan, 2000: SPOLKA: NCAR’s polarimetric Doppler research radar. Proc. IEEE 2000 Int. Geoscience and Remote Sensing Symp., IGARSS 2000, Honolulu, HI, IEEE, 1570–1573.

  • Kollias, P., B. A. Albrecht, and F. D. Mark, 2003: Cloud radar observations of vertical drafts and microphysics in convective rain. J. Geophys. Res., 108, 4053, doi:10.1029/2001JD002033.

    • Search Google Scholar
    • Export Citation
  • Kollias, P., E. E. Clothiaux, B. A. Albrecht, M. A. Miller, K. P. Moran, and K. L. Johnson, 2005: The Atmospheric Radiation Measurement Program cloud profiling radars: An evaluation of signal processing and sampling strategies. J. Atmos. Oceanic Technol., 22, 930948, doi:10.1175/JTECH1749.1.

    • Search Google Scholar
    • Export Citation
  • Kollias, P., E. E. Clothiaux, M. A. Miller, B. A. Albrecht, G. L. Stephens, and T. P. Ackerman, 2007: Millimeter-wavelength radars: New frontier in atmospheric cloud and precipitation research. Bull. Amer. Meteor. Soc., 88, 16081624, doi:10.1175/BAMS-88-10-1608.

    • Search Google Scholar
    • Export Citation
  • Lhermitte, R., 1990: Attenuation and scattering of millimeter wavelength radiation by cloud and precipitation. J. Atmos. Oceanic Technol., 7, 464479, doi:10.1175/1520-0426(1990)007<0464:AASOMW>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci.,28, 702–708, doi:10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.

  • Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci.,29, 1109–1123, doi:10.1175/1520-0469(1972)029<1109:DOGSCC>2.0.CO;2.

  • Matrosov, S. Y., 2005: Attenuation-based estimates of rainfall rates aloft with vertically pointing Ka-band radars. J. Atmos. Oceanic Technol., 22, 4354, doi:10.1175/JTECH-1677.1.

    • Search Google Scholar
    • Export Citation
  • Mie, G., 1908: Beiträge zur Optik trüber Medien, speziell kolloidaler Metallösungen (Contribution to the optics of suspended media, specifically colloidal metal suspensions). Ann. Phys., 330, 377445, doi:10.1002/andp.19083300302.

    • Search Google Scholar
    • Export Citation
  • Ovtchinnikov, M., T. Ackerman, and R. Marchand, 2006: Evaluation of the multiscale framework using data from the Atmospheric Radiation Measurement Program. J. Climate, 19, 17161729, doi:10.1175/JCLI3699.1.

    • Search Google Scholar
    • Export Citation
  • Penide, G., V. V. Kumar, A. Protat, and P. T. May, 2013: Statistics of drop size distribution parameters and rain rates for stratiform and convective precipitation during the north Australian wet season. Mon. Wea. Rev., 141, 3222–3237, doi:10.1175/MWR-D-12-00262.1.

    • Search Google Scholar
    • Export Citation
  • Riihimaki, L. D., and C. N. Long, 2014: Spatial variability of surface irradiance measurements at the Manus ARM site. J. Geophys. Res. Atmos., 119, 54755491, doi:10.1002/2013JD021187.

    • Search Google Scholar
    • Export Citation
  • Rondanelli, R., and R. S. Lindzen, 2008: Observed variations in convective precipitation fraction and stratiform area with sea surface temperature. J. Geophys. Res., 113, D16119, doi:10.1029/2008JD010064.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., and E. Amitai, 1998: Comparison of WPMM versus regression for evaluating Z–R relationships. J. Appl. Meteor., 37, 12411249, doi:10.1175/1520-0450(1998)037<1241:COWVRF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., and C. W. Ulbrich, 2003: Cloud microphysical properties, processes, and rainfall estimation opportunities. Meteor. Monogr., No. 30, Amer. Meteor. Soc., 237–237, doi:10.1175/0065-9401(2003)030<0237:CMPPAR>2.0.CO;2.

  • 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, 26402656, doi:10.1175/1520-0469(1987)044<2640:ADMSOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., J. R. Campbell, J. Zhu, P. Kollias, M. Shupe, and C. Williams, 2005: Lidar and triple-wavelength Doppler radar measurements of the melting layer: A revised model for dark and brightband phenomena. J. Appl. Meteor., 44, 301312, doi:10.1175/JAM-2197.1.

    • Search Google Scholar
    • Export Citation
  • Schumacher, C., R. A. Houze Jr., and I. Kraucunas, 2004: The tropical dynamical response to latent heating estimates derived from the TRMM Precipitation Radar. J. Atmos. Sci., 61, 13411358, doi:10.1175/1520-0469(2004)061<1341:TTDRTL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., 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, 452462, doi:10.1175/1520-0450(1997)036<0452:SOTEMC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., and J. A. Smith, 2000: Reflectivity, rain rate, and kinetic energy flux relationships based on raindrop spectra. J. Appl. Meteor., 39, 19231940, doi:10.1175/1520-0450(2000)039<1923:RRRAKE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., R. A. Houze Jr., and S. E. Yuter, 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34, 19782007, doi:10.1175/1520-0450(1995)034<1978:CCOTDS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Steiner, M. R., J. A. Smith, and R. Uijlenhoet, 2004: A microphysical interpretation of radar reflectivity–rain rate relationships. J. Atmos. Sci., 61, 11141131, doi:10.1175/1520-0469(2004)061<1114:AMIORR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Stout, G. E., and E. A. Mueller, 1968: Survey of relationship between rainfall rate and radar reflectivity in the measurement of precipitation. J. Appl. Meteor., 7, 465474, doi:10.1175/1520-0450(1968)007<0465:SORBRR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tao, W. K., 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, 685700.

    • Search Google Scholar
    • Export Citation
  • Thurai, M., V. N. Bringi, and P. T. May, 2010: CPOL radar-derived drop size distribution statistics of stratiform and convective rain for two regimes in Darwin, Australia. J. Atmos. Oceanic Technol., 27, 932942, doi:10.1175/2010JTECHA1349.1.

    • Search Google Scholar
    • Export Citation
  • 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, 355371, doi:10.1175/1520-0450(1996)035<0355:EFTRSO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Tokay, A., D. A. Short, C. R. Williams, W. L. Ecklund, and K. S. Gage, 1999: Tropical rainfall associated with convective and stratiform clouds: Intercomparison of disdrometer and profiler measurements. J. Appl. Meteor., 38, 302320, doi:10.1175/1520-0450(1999)038<0302:TRAWCA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Williams, C. R., W. L. Ecklund, and K. S. Gage, 1995: Classification of precipitation clouds in the tropics using 915-MHz wind profilers. J. Atmos. Oceanic Technol., 12, 9961012, doi:10.1175/1520-0426(1995)012<0996:COPCIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Xie, S., M. Zhang, J. S. Boyle, R. T. Cederwall, G. L. Potter, and W. Lin, 2004: Impact of a revised convective triggering mechanism on Community Atmosphere Model, version 2, simulations: Results from short-range weather forecasts. J. Geophys. Res., 109, D14102, doi:10.1029/2004JD004692.

    • Search Google Scholar
    • Export Citation
  • Yoneyana, K., D. Zhang, and C. N. Long, 2013: Tracking pulses of the Madden–Julian oscillation. Bull. Amer. Meteor. Soc.,94, 1871–1891, doi:10.1175/BAMS-D-12-00157.1.

  • Yuter, S. E., and R. A. Houze Jr., 1995: Three-dimensional kinematic and microphysical evolution of Florida cumulonimbus. Part II: Frequency distribution of vertical velocity, reflectivity, and differential reflectivity. Mon. Wea. Rev., 123, 19411963, doi:10.1175/1520-0493(1995)123<1941:TDKAME>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yuter, S. E., and R. A. Houze Jr., 1997: Measurements of raindrop size distribution over the Pacific warm pool and implications for Z–R relations. J. Appl. Meteor., 36, 847867, doi:10.1175/1520-0450(1997)036<0847:MORSDO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zuluaga, M. D., and R. A. Houze Jr., 2013: Evolution of population of precipitation convective systems over the equatorial Indian Ocean in active phases of the Madden–Julian oscillation. J. Atmos. Sci., 70, 27132725, doi:10.1175/JAS-D-12-0311.1.

    • Search Google Scholar
    • Export Citation
Save