• Ackerman, T. P., and G. M. Stokes, 2003: The Atmospheric Radiation Measurement Program. Phys. Today, 56, 3844.

  • Alexander, M. J., and Coauthors, 2010: Recent developments in gravity-wave effects in climate models and the global distribution of gravity wave momentum flux from observations and models. Quart. J. Roy. Meteor. Soc., 136, 11031124.

    • Search Google Scholar
    • Export Citation
  • Beres, J. H., M. J. Alexander, and J. R. Holton, 2002: Effects of tropospheric wind shear on the spectrum of convectively generated gravity waves. J. Atmos. Sci., 59, 18051824.

    • Search Google Scholar
    • Export Citation
  • Bouniol, D., A. J. Illingworth, and R. J. Hogan, 2003: Deriving turbulent kinetic energy dissipation rate within clouds using ground based 94 GHz radar. Preprints, 31st Int. Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 4A.5. [Available online at http://ams.confex.com/ams/pdfpapers/63826.pdf.]

  • Clothiaux, E. E., T. P. Ackerman, G. G. Mace, K. P. Moran, R. T. Marchand, M. A. 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.

    • Search Google Scholar
    • Export Citation
  • Comstock, J. M., and C. Jakob, 2004: Evaluation of tropical cirrus cloud properties derived from ECMWF model output and ground based measurements over Nauru Island. Geophys. Res. Lett., 31, L10106, doi:10.1029/2004GL019539.

    • Search Google Scholar
    • Export Citation
  • Comstock, J. M., T. P. Ackerman, and G. G. Mace, 2002: Ground-based lidar and radar remote sensing of tropical cirrus clouds at Nauru Island: Cloud statistics and radiative impacts. J. Geophys. Res., 107, 4714, doi:10.1029/2002JD002203.

    • Search Google Scholar
    • Export Citation
  • Daubechies, I., 1990: The wavelet transform time-frequency localization and signal analysis. IEEE Trans. Inf. Theory, 36, 9611004.

  • Dean, S. M., B. N. Lawrence, R. G. Grainger, and D. N. Heuff, 2005: Orographic cloud in a GCM: The missing cirrus. Climate Dyn., 24, 771780.

    • Search Google Scholar
    • Export Citation
  • Delanoë, J., A. Protat, D. Bouniol, A. Heymsfield, A. Bansemer, and P. Brown, 2007: The characterization of ice cloud properties from Doppler radar measurements. J. Appl. Meteor. Climatol., 46, 16821698.

    • Search Google Scholar
    • Export Citation
  • Deng, M., and G. G. Mace, 2006: Cirrus microphysical properties and air motion statistics using cloud radar Doppler moments. Part I: Algorithm description. J. Appl. Meteor. Climatol., 45, 16901709.

    • Search Google Scholar
    • Export Citation
  • Deng, M., and G. G. Mace, 2008: Cirrus microphysical properties and air motion statistics using cloud radar Doppler moments. Part II: Climatology. J Appl. Meteor. Climatol., 47, 32213235.

    • Search Google Scholar
    • Export Citation
  • Dufresne, J. L., and S. Bony, 2008: An assessment of the primary sources of spread of global warming estimates from coupled atmosphere–ocean models. J. Climate, 21, 51355144.

    • Search Google Scholar
    • Export Citation
  • Foote, G. B., and P. S. Du Toit, 1969: Terminal velocity of raindrops aloft. J. Appl. Meteor., 8, 249253.

  • Fritts, D. C., and M. J. Alexander, 2003: Gravity wave dynamics and effects in the middle atmosphere. Rev. Geophys., 41, 1003, doi:10.1029/2001RG000106.

    • Search Google Scholar
    • Export Citation
  • Gultepe, I., and D. O. Starr, 1995: Dynamical structure and turbulence in cirrus clouds: Aircraft observations during the FIRE. J. Atmos. Sci., 52, 41594182.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., and L. M. Miloshevich, 2003: Parameterizations for the cross-sectional area and extinction of cirrus and stratiform ice cloud particles. J. Atmos. Sci., 60, 936956.

    • Search Google Scholar
    • Export Citation
  • Hoffmann, L., and M. J. Alexander, 2010: Occurrence frequency of convective gravity waves during the North American thunderstorm season. J. Geophys. Res., 115, D20111, doi:10.1029/2010JD014401.

    • Search Google Scholar
    • Export Citation
  • Kärcher, B., and U. Lohmann, 2002: A parameterization of cirrus cloud formation: Homogeneous freezing including effects of aerosol size. J. Geophys. Res., 107, 4698, doi:10.1029/2001JD001429.

    • Search Google Scholar
    • Export Citation
  • Kärcher, B., and J. Ström, 2003: The roles of dynamical variability and aerosols in cirrus cloud formation. Atmos. Chem. Phys., 3, 823838.

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

    • Search Google Scholar
    • Export Citation
  • Kollias, P., E. E. Clothiaux, M. A. Miller, E. P. Luke, K. L. Johnson, K. P. Moran, K. B. Widener, and B. A. Albrecht, 2007b: The Atmospheric Radiation Measurement Program cloud profiling radars: Second-generation sampling strategies, processing, and cloud data products. J. Atmos. Oceanic Technol., 24, 11991214.

    • Search Google Scholar
    • Export Citation
  • Kollias, P., G. Tselioudis, and B. A. Albrecht, 2007c: Cloud climatology at the Southern Great Plains and the layer structure, drizzle, and atmospheric modes of continental stratus. J. Geophys. Res., 112, D09116, doi:10.1029/2006JD007307.

    • Search Google Scholar
    • Export Citation
  • Liou, K. N., 1986: Influence of cirrus clouds on weather and climate processes: A global perspective. Mon. Wea. Rev., 114, 11671199.

  • Lynch, D., K. Sassen, D. Starr, and G. Stephens, Eds., 2002: Cirrus. Oxford University Press, 480 pp.

  • Mace, G. G., and S. Benson, 2008: The vertical structure of cloud occurrence and radiative forcing at the SGP ARM site as revealed by 8 years of continuous data. J. Climate, 21, 25912610.

    • Search Google Scholar
    • Export Citation
  • Mace, G. G., S. Benson, and E. Vernon, 2006: Cirrus clouds and the large-scale atmospheric state: Relationships revealed by six years of ground-based data. J. Climate, 19, 32573278.

    • Search Google Scholar
    • Export Citation
  • Matrosov, S. Y., B. W. Orr, R. A. Kropfli, and J. B. Snider, 1994: Retrieval of vertical profiles of cirrus cloud microphysical parameters from Doppler radar and infrared radiometer measurements. J. Appl. Meteor., 33, 617626.

    • Search Google Scholar
    • Export Citation
  • Mitchell, D. L., P. Rasch, D. Ivanova, G. McFarquhar, and T. Nousiainen, 2008: Impact of small ice crystal assumptions on ice sedimentation rates in cirrus clouds and GCM simulations. Geophys. Res. Lett., 35, L09806, doi:10.1029/2008GL033552.

    • Search Google Scholar
    • Export Citation
  • Moran, K. P., B. E. Martner, M. J. Post, R. A. Kropfli, D. C. Welsh, and K. B. Widener, 1998: An unattended cloud-profiling radar for use in climate research. Bull. Amer. Meteor. Soc., 79, 443455.

    • Search Google Scholar
    • Export Citation
  • Orr, B. W., and R. A. Kropfli, 1999: A method for estimating particle fall velocities from vertically pointing Doppler radar. J. Atmos. Oceanic Technol., 16, 2937.

    • Search Google Scholar
    • Export Citation
  • Pincus, R., C. P. Batstone, R. J. P. Hofmann, K. E. Taylor, and P. J. Glecker, 2008: Evaluating the present-day simulation of clouds, precipitation, and radiation in climate models. J. Geophys. Res., 113, D14209, doi:10.1029/2007JD009334.

    • Search Google Scholar
    • Export Citation
  • Plana-Fattori, A., A. Protat, and J. Delanoë, 2010: Observing ice clouds with a Doppler cloud radar. C. R. Phys., 11, 96103.

  • Platt, C. M. R., J. C. Scott, and A. C. Dilley, 1987: Remote sensing of high clouds. Part VI: Optical properties of midlatitude and tropical cirrus. J. Atmos. Sci., 44, 729744.

    • Search Google Scholar
    • Export Citation
  • Preusse, P., S. D. Eckermann, and M. Ern, 2008: Transparency of the atmosphere to short horizontal wavelength gravity waves. J. Geophys. Res., 113, D24104, doi:10.1029/2007JD009682.

    • Search Google Scholar
    • Export Citation
  • Protat, A., and C. R. Williams, 2011: The accuracy of radar estimates of ice terminal fall speed from vertically pointing Doppler radar measurements. J. Appl. Meteor. Climatol., 50, 21202138.

    • Search Google Scholar
    • Export Citation
  • Protat, A., Y. Lemaître, and D. Bouniol, 2003: Terminal fall velocity and the FASTEX cyclones. Quart. J. Roy. Meteor. Soc., 129, 15131535.

    • Search Google Scholar
    • Export Citation
  • Quante, M., 2006: Turbulenz in Cirruswolken mittlerer Breiten. Ph.D. thesis, Universität Hamburg, 168 pp.

  • Randall, D. A., and Coauthors, 2007: Climate models and their evaluation. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 589–662.

  • Sanderson, B. M., C. Piani, W. J. Ingram, D. A. Stone, and M. R. Allen, 2008: Towards constraining climate sensitivity by linear analysis of feedback patterns in thousands of perturbed-physics GCM simulations. Climate Dyn., 30, 175190.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., Z. Wang, and D. Liu, 2008: Global distribution of cirrus clouds from CloudSat/Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) measurements. J. Geophys. Res., 113, D00A12, doi:10.1029/2008JD009972.

    • Search Google Scholar
    • Export Citation
  • Smith, S. A., and P. R. Jonas, 1997: Wavelet analysis of turbulence in cirrus clouds. Ann. Geophys., 15, 14471456.

  • Smith, S. A., and A. D. DelGenio, 2001: Analysis of aircraft, radiosonde, and radar observations in cirrus clouds observed during FIRE II: The interactions between environmental structure, turbulence, and cloud microphysical properties. J. Atmos. Sci., 58, 451461.

    • Search Google Scholar
    • Export Citation
  • Smith, S. R., D. M. Legler, M. J. Remigio, and J. J. O'Brien, 1999: Comparison of 1997–98 U.S. temperature and precipitation anomalies to historical ENSO warm phases. J. Climate, 12, 35073515.

    • Search Google Scholar
    • Export Citation
  • Stähelin, J., T. G. Shepherd, and T. Peter, 2011: Report on the 18th Session of the SPARC Scientific Steering Group, 2–5 February 2011, Pune, India. SPARC Newsletter, No. 37, Stratospheric Process and Their Role in Climate, Zurich, Switzerland, 1–5.

  • Stubenrauch, C. J., A. Chédin, G. Rädel, N. A. Scott, and S. Serrar, 2006: Cloud properties and their seasonal and diurnal variability from TOVS Path-B. J. Climate, 19, 55315553.

    • Search Google Scholar
    • Export Citation
  • Szyrmer, W., A. Tatarevic, and P. Kollias, 2012: Ice clouds microphysical retrieval using 94-GHz Doppler radar observations: Basic relations within the retrieval framework. J. Geophys. Res., 117, D14203, doi:10.1029/2011JD016675.

    • Search Google Scholar
    • Export Citation
  • Taylor, J. R., W. J. Randel, and E. J. Jensen, 2011: Cirrus cloud-temperature interactions in the tropical tropopause layer: A case study. Atmos. Chem. Phys., 11, 10 08510 095.

    • Search Google Scholar
    • Export Citation
  • Torrence, C., and G. P. Compo, 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79, 6178.

  • Vinnichenko, N. K., 1970: The kinetic energy spectrum in the free atmosphere—1 second to 5 years. Tellus, 22, 158166.

  • Webb, M. J., F. H. Lambert, and J. M. Gregory, 2012: Origins of differences in climate sensitivity, forcing and feedback in climate models. Climate Dyn., 40, 677–707, doi:10.1007/s00382-012-1336-x.

    • Search Google Scholar
    • Export Citation
  • Wyngaard, J. C., 1973: On surface-layer turbulence. Workshop on Micrometeorology, D. A. Haugen, Ed., Amer. Meteor. Soc., 101–149.

  • Zhang, S. D., F. Yi, C. M. Huang, and K. M. Huang, 2012: High vertical resolution analyses of gravity waves and turbulence at a midlatitude station. J. Geophys. Res., 117, D02103, doi:10.1029/2011JD016587.

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

    (a) Monthly availability of MMCR observations at SGP and (b) monthly occurrence of high clouds at SGP subdivided into four categories: all observed high clouds, high clouds for which the hydrometeor fraction per hour > 95%, high clouds that are persistent for 2 h, and high clouds for which GW are detected.

  • View in gallery
    Fig. 2.

    (a) Monthly availability of MMCR observations at Manus and (b) monthly occurrence of high clouds at Manus subdivided into four categories: all observed high clouds, high clouds for which the hydrometeor fraction per hour > 95%, high clouds which are persistent for 2 h, and high clouds for which GW are detected.

  • View in gallery
    Fig. 3.

    Example of Doppler velocity Vd decomposition into particle terminal fall velocity Vt and vertical air motion w on 8 Dec 2004 at SGP site. Positive velocity values indicate downward motion. (a) Radar reflectivity at 10-s and 45-m resolution, (b) Vd at 10-s and 45-m resolution, (c) Vt at 10-s and 135-m resolution, and (d) w at 10-s and 135-m resolution.

  • View in gallery
    Fig. 4.

    Hourly air motion statistics for high cloud case study on 8 Dec 2004 at SGP site: (a) mean, (b) standard deviation, (c) skewness, and (d) updraft fraction.

  • View in gallery
    Fig. 5.

    Time series of monthly mean inertial subrange slopes of the vertical air motion power density spectra determined for (a) the SGP site and (b) the Manus site.

  • View in gallery
    Fig. 6.

    Hourly results based on FFT on vertical air motion for the high cloud case study on 8 Dec 2004 at SGP site. (a) Fraction of turbulent energy, (b) inertial subrange slope of the power density spectrum, and (c) log10 of the turbulent kinetic dissipation rate ɛ.

  • View in gallery
    Fig. 7.

    Histogram of high cloud properties at SGP and Manus for the subset of clouds for which no gravity waves were detected. (a) Radar reflectivity Z, (b) Doppler velocity Vd, (c) reflectivity-weighted particle velocity Vt, (d) reflectivity-weighted particle velocity Vt, corrected by air density, (e) vertical air motion w, (f) in-cloud temperature, (g) cloud height, and (h) cloud thickness.

  • View in gallery
    Fig. 8.

    As in Fig. 7, but for the subset of clouds for which gravity waves were detected.

  • View in gallery
    Fig. 9.

    Profile of high cloud occurrences in percent at (a) SGP and (b) Manus. Total high cloud occurrence vs MMCR availability (dark gray), persistent high clouds with 2-h longevity vs total high cloud occurrence (medium gray), and occurrence of high clouds for which GW were detected vs total high cloud occurrence (light gray).

  • View in gallery
    Fig. 10.

    Mean profiles of (a) GW period and (b) GW wavelength at the SGP site for all months (black), summer season (JJA; dark gray), and winter season (DJF; light gray).

  • View in gallery
    Fig. 11.

    As in Fig. 10, but for Manus.

  • View in gallery
    Fig. 12.

    Histograms of gravity wave periods and wavelengths detected at SGP and Manus.

  • View in gallery
    Fig. 13.

    Histograms of (a) vertical velocity inertial subrange slopes for which ɛ were determined, (b) turbulent kinetic energy dissipation rates ɛ, and (c) fraction of energy contained in the inertial subrange for the high cloud subset for which no gravity waves were detected at the SGP and Manus sites.

  • View in gallery
    Fig. 14.

    As in Fig. 13, but for the high cloud subset for which gravity waves were detected.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 346 146 11
PDF Downloads 229 109 6

Climatology of High Cloud Dynamics Using Profiling ARM Doppler Radar Observations

Heike KalesseDepartment of Atmospheric and Oceanic Sciences, McGill University, Montreal, Quebec, Canada

Search for other papers by Heike Kalesse in
Current site
Google Scholar
PubMed
Close
and
Pavlos KolliasDepartment of Atmospheric and Oceanic Sciences, McGill University, Montreal, Quebec, Canada

Search for other papers by Pavlos Kollias in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Ice cloud properties are influenced by cloud-scale vertical air motion. Dynamical properties of ice clouds can be determined via Doppler measurements from ground-based, profiling cloud radars. Here, the decomposition of the Doppler velocities into reflectivity-weighted particle velocity Vt and vertical air motion w is described. The methodology is applied to high clouds observations from 35-GHz profiling millimeter wavelength radars at the Atmospheric Radiation Measurement Program (ARM) Southern Great Plains (SGP) climate research facility in Oklahoma (January 1997–December 2010) and the ARM Tropical Western Pacific (TWP) site in Manus (July 1999–December 2010). The Doppler velocity measurements are used to detect gravity waves (GW), whose correlation with high cloud macrophysical properties is investigated. Cloud turbulence is studied in the absence and presence of GW. High clouds are less turbulent when GW are observed. Probability density functions of Vt, w, and high cloud macrophysical properties for the two cloud subsets (with and without GW) are presented. Air-density-corrected Vt for high clouds for which GW (no GW) were detected amounted to hourly means and standard deviations of 0.89 ± 0.52 m s−1 (0.8 ± 0.48 m s−1) and 1.03 ± 0.41 m s−1 (0.86 ± 0.49 m s−1) at SGP and Manus, respectively. The error of w at one standard deviation was estimated as 0.15 m s−1. Hourly means of w averaged around 0 m s−1 with standard deviations of ±0.27 (SGP) and ±0.29 m s−1 (Manus) for high clouds without GW and ±0.22 m s−1 (both sites) for high clouds with GW. The midlatitude site showed stronger seasonality in detected high cloud properties.

Corresponding author address: Dr. Heike Kalesse, Department of Atmospheric and Oceanic Sciences, McGill University, 805 Sherbrooke Street West, Burnside Hall, Room 945, Montreal QC H3A 0B9, Canada. E-mail: heike.kalesse@mail.mcgill.ca

Abstract

Ice cloud properties are influenced by cloud-scale vertical air motion. Dynamical properties of ice clouds can be determined via Doppler measurements from ground-based, profiling cloud radars. Here, the decomposition of the Doppler velocities into reflectivity-weighted particle velocity Vt and vertical air motion w is described. The methodology is applied to high clouds observations from 35-GHz profiling millimeter wavelength radars at the Atmospheric Radiation Measurement Program (ARM) Southern Great Plains (SGP) climate research facility in Oklahoma (January 1997–December 2010) and the ARM Tropical Western Pacific (TWP) site in Manus (July 1999–December 2010). The Doppler velocity measurements are used to detect gravity waves (GW), whose correlation with high cloud macrophysical properties is investigated. Cloud turbulence is studied in the absence and presence of GW. High clouds are less turbulent when GW are observed. Probability density functions of Vt, w, and high cloud macrophysical properties for the two cloud subsets (with and without GW) are presented. Air-density-corrected Vt for high clouds for which GW (no GW) were detected amounted to hourly means and standard deviations of 0.89 ± 0.52 m s−1 (0.8 ± 0.48 m s−1) and 1.03 ± 0.41 m s−1 (0.86 ± 0.49 m s−1) at SGP and Manus, respectively. The error of w at one standard deviation was estimated as 0.15 m s−1. Hourly means of w averaged around 0 m s−1 with standard deviations of ±0.27 (SGP) and ±0.29 m s−1 (Manus) for high clouds without GW and ±0.22 m s−1 (both sites) for high clouds with GW. The midlatitude site showed stronger seasonality in detected high cloud properties.

Corresponding author address: Dr. Heike Kalesse, Department of Atmospheric and Oceanic Sciences, McGill University, 805 Sherbrooke Street West, Burnside Hall, Room 945, Montreal QC H3A 0B9, Canada. E-mail: heike.kalesse@mail.mcgill.ca

1. Introduction

Ice clouds are globally distributed and thus play an important role in the earth–atmosphere system. By reflecting and absorbing solar radiation and by absorbing and emitting thermal infrared radiation they have a strong effect on the local and global radiation energy budget (e.g., Liou 1986; Lynch et al. 2002).

The large spatial and temporal variability of ice cloud microphysical (e.g., particle habit, particle size, particle density) and macrophysical (e.g., cloud height, thickness, location) properties makes the accurate representation of ice clouds and their effects in global climate models (GCM) an outstanding challenge (e.g., Randall et al. 2007; Mitchell et al. 2008; Sanderson et al. 2008). Furthermore, recent studies have demonstrated that the largest uncertainty in future climate projections is due to cloud parameterization and related cloud feedbacks (e.g., Dufresne and Bony 2008; Pincus et al. 2008; Webb et al. 2012). In that context, especially the representation of (ice) cloud life cycles—which is strongly influenced by the ambient vertical air motion as well as the terminal fall speed of particles—is challenging. As highlighted in Kärcher and Ström (2003), ice cloud microphysical properties are strongly controlled by vertical air velocities and their mesoscale variability. It has been argued that an adequate parameterization of ice clouds in global climate models requires cloud-scale vertical velocity probability density functions (Kärcher and Lohmann 2002). Protat and Williams (2011) also emphasize that long-term observations of ice clouds and their statistical properties including time–height variability and their relationship to the large-scale environment will help to improve the representation of ice cloud life cycles in GCM. They also pointed out the need for a better characterization of the vertical variability (i.e., a variability as a function of ambient temperature) of particle fall speed.

To address these issues, long-term observations from profiling ground-based Doppler cloud radars are well suited (Kollias et al. 2007a). The dynamical properties of ice clouds over a long period of time can be determined via decomposition of the Doppler velocities into reflectivity-weighted particle velocity Vt and vertical air motion w (e.g., Orr and Kropfli 1999; Deng and Mace 2006; Delanoë et al. 2007; Protat and Williams 2011).

The objective of this work is to derive a cloud Doppler radar–based high cloud dynamics climatology with probability density functions of reflectivity-weighted particle velocity Vt and vertical air motion w, as well as turbulence analysis in high clouds. In addition, the Doppler radar measurements are used to detect gravity waves whose correlation with macrophysical properties such as cloud height and cloud thickness is investigated. The European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis model output is used to relate the high clouds dynamics to large-scale dynamics. The high cloud properties and derived cloud-scale air motions are analyzed in terms of seasonal variations and on the influence of the presence of gravity waves.

Section 2 of the study is focusing on the detailed description of the two datasets and the methodology. Section 3 presents the high cloud dynamics climatology of the two analyzed sites, with one being in the midlatitudes and the other one being in the tropics. Section 4 provides a summary and outlook.

2. Methodology

a. Data overview

Here, long-term surface observations of the Atmospheric Radiation Measurement Program (ARM; Ackerman and Stokes 2003) are used. More precisely, the Active Remotely-Sensed Cloud Locations product (ARSCL; Clothiaux et al. 2000) is the basis of this work. The ARSCL product combines millimeter wavelength cloud radar (MMCR), ceilometer, and micropulse lidar (MPL) observations to give an objective representation of clouds (Clothiaux et al. 2000). It should be noted, however, that high thin cirrus below the detection limit of the MMCR are excluded from this study because it only employs MMCR Doppler velocity observations. The study presented here is thus referring to a limited subset of high clouds that are sufficiently thick and long lasting, and results are not representative for thin high cirrus.

Data collected at two ARM fixed sites are used. One dataset is from the Southern Great Plains (SGP) site (36°36′18.0″N, 97°29′6.0″W) in north-central Oklahoma, and the other one is from Manus (2°3′ 39.64″S, 147°25′31.43″E) at the Tropical Western Pacific (TWP) site. The SGP site is characterized by strong seasonality and the available long-term cloud observations span from January 1997 to December 2010 (14 yr). Manus is situated in the tropical western Pacific warm pool, and ARSCL data are available from July 1999 to December 2010 but have periods of inoperability (in total, about 7 yr of data are available). In addition to the ARSCL observations, 1-h-resolution ECWMF model output data (*ecmwfvarX1.c1- files) for the ARM sites is utilized in this study. The ECWMF model output data are used to examine the atmospheric synoptic-scale conditions present during high cloud observations.

Profiles of MMCR radar reflectivity with a temporal and vertical resolution of 10 s and 45 m, respectively, are used to determine a high-resolution hydrometeor location mask. Based on that hydrometeor mask, high cloud identification follows the criterion given in Kollias et al. (2007c) where ice clouds are defined as hydrometeor layers with bases above 6 km. The second criterion is an ECMWF temperature-based one: only clouds observed in an environment of <−10°C are classified as cold clouds. Since mixed-phase clouds can also be present in these conditions, sampled clouds will be referred to as high or cold clouds. Assuming a typical wind speed of about 30 m s−1 at high cloud levels, the 10-s averaging in time corresponds to a 300-m data point spacing, which is comparable to aircraft microphysical sampling (Smith and DelGenio 2001) and assumed to approximate cloud-scale motions.

With a minimum detectable signal of approximately −50 dBZ at 5-km range, the MMCR at the SGP site is about 5 dB more sensitive than the MMCR at the Manus site (Deng and Mace 2008). For comparability of both sites, only signals > −45 dBZ are considered. According to Moran et al. (1998), the calibration uncertainties of the MMCR are 1–2 dB and 0.1 m s−1 for reflectivity and Doppler velocity, respectively. However, in reality, the Doppler velocity error is a function of radar Doppler spectrum width and signal-to-noise ratio (Kollias et al. 2007b). In cirrus clouds, the normalized spectrum width is typically small (below 0.04) and thus, on average, the MMCR Doppler velocity uncertainty is estimated as 0.05 m s−1.

Conditional sampling techniques are applied to all cold cloud layers used in this study. (i) It is required that the ice clouds are far enough away from convective cores so they are not influenced by the dynamics of the deep convection. To exclude high cloud layers directly associated with deep convection, we require the vertical distance to the first cloud top below the high cloud layer to be greater than 0.5 km. (ii) Furthermore, it is required that the 20-min-average hydrometeor fraction at a certain height exceeds 80%. This requirement filters small patches of high clouds, thus making sure the clouds are sufficiently persistent in time and proper time series data lengths are available. Using the 10-s-resolution MMCR data, hourly averages of high cloud base, top, and thickness are estimated. In addition, multilayer cloud scenes were identified.

The cold cloud dynamics climatology is derived from time periods when the MMCR were in good operating status, respectively. For the SGP site, this was the case for 88% of the time within the considered 14-yr period (Fig. 1a). However, it should be noted that the SGP MMCR was upgraded in October 2003 to improve the measurement averaging frequency from 10 to 4 s (Kollias et al. 2007c; Mace and Benson 2008) and that the MMCR operational mode most sensitive to thin cirrus failed from mid-2005 to the end of 2008 (Deng and Mace 2008). The latter led to smaller observed cold cloud frequencies during that period (Fig. 1b). Also, it should be noted that no MMCR data was available for January 2003 at the SGP site.

Fig. 1.
Fig. 1.

(a) Monthly availability of MMCR observations at SGP and (b) monthly occurrence of high clouds at SGP subdivided into four categories: all observed high clouds, high clouds for which the hydrometeor fraction per hour > 95%, high clouds that are persistent for 2 h, and high clouds for which GW are detected.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Among the 14 yr of MMCR observations at the SGP site, high clouds were detected during 26 730 h, which corresponds to a mean occurrence of 24.2% of the time, which is in agreement with findings by Mace et al. (2006). However, as shown in Fig. 1b, high cloud occurrence has a strong year-to-year variability as well as a seasonal dependence. With an average occurrence of 28%, it is highest in winter (January–February) when the SGP site is under the influence of synoptic-scale weather patterns and in late spring to early summer (May–June). With synoptic systems remaining well north of the Southern Great Plains in summer, the primary high cloud generation mechanism in that season is through deep convection, which is less strong in the end of summer–beginning of fall, leading to a minimum in cold cloud occurrence in September (18% occurrence). The high cold cloud frequency in winter 1997/98 (35% in January–February 1998) corresponds to a major El Niño event that had a significant influence on weather patterns in the south-central United States (Smith et al. 1999).

At the Manus site, the MMCR provided useful data for 67% of the time between July 1999 and December 2010 (Fig. 2a). The MMCR was inoperable or not operating properly for a few periods lasting several months between November 2000 and February 2003, June 2005 and May 2006, and December 2008 and January 2009. The upgrade to a higher measurement averaging frequency from 10 to 4 s took place in June 2006. In total, the MMCR detected 25 890 h with high clouds being present; this equals 44% of the time the MMCR was operating. In a seasonal view, the occurrence of MMCR-detectable (i.e., sufficiently thick) high clouds peaks in January (59%) and is lowest in May–June (39%). In the tropics, high cloud generation is either linked to deep convection (anvil cirrus) or detached from convection. The former is generally geometrically and optically thicker and has lower cloud bases and more visible structure within the cloud than the upper-tropospheric cirrus, which is possibly formed by large-scale ascent or cold temperature perturbations (Comstock and Jakob 2004).

Fig. 2.
Fig. 2.

(a) Monthly availability of MMCR observations at Manus and (b) monthly occurrence of high clouds at Manus subdivided into four categories: all observed high clouds, high clouds for which the hydrometeor fraction per hour > 95%, high clouds which are persistent for 2 h, and high clouds for which GW are detected.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

b. Ice particle terminal velocity retrieval

The decomposition of the observed mean Doppler velocity Vd from profiling radars into the vertical air motion w component and the reflectivity-weighted particle fall velocity Vt has been the topic of extensive research in the past. Several statistical techniques have been proposed (e.g., Matrosov et al. 1994; Orr and Kropfli 1999; Delanoë et al. 2007; Plana-Fattori et al. 2010) that are based on the premise that, for a long dwell period when considering the cloud as a whole, mean vertical air motions are small in comparison to ice particle terminal fall speeds (and average to zero) and that the latter are much less fluctuating. Under this approximation, the empirical power-law relationship between radar reflectivity Z (in mm6 m−3) and Vd (Vd = aZb) can directly be used to retrieve terminal fall velocity Vt (Vt = aZb). A recent study by Protat and Williams (2011) provided an excellent review of existing techniques. They also proposed a new technique in which Vd measurements for a particular reflectivity–height pair (to eliminate air-density effects) were averaged over the entire observation length of the cloud to retrieve Vt. This method was found to outperform other statistical techniques and had errors in the retrieved Vt of less than 10 cm s−1 when careful screening of the data was assured. This error magnitude is considered to be sufficiently small so that retrieved Vt can be used in microphysical cloud retrievals such as the one of Szyrmer et al. (2012).

Here, a modified version of the Doppler velocity separation technique developed by Protat and Williams (2011) is applied for the retrieval of Vt and w and is briefly described now. First, instead of considering the cloud as a whole, a finite time segment (20–60 min) of Vd measurements at a particular height is used. Second, reflectivities were separated into bins of 2-dBZ width. For any given high cloud (to which the filtering criteria as mentioned in section 2a were applied), the number-weighted Vd averages over the fixed time interval were determined for each height H and each reflectivity bin and assumed to be equal to the mean Vt at this ZH pair. Since the MMCR observations have a vertical resolution of 45 m, observations from three consecutive range gates (135 m) and over a time interval of 20 min are used to construct the ZH pairs of VtZ. This triples the number of available observations within a 20-min period without compromising the condition for constant air density. To assure a good quality of the log-space Vt = aZb fit, the linear regression coefficients a and b were only determined if there were at least four reflectivity bins in the 20-min averaging period for which reflectivity-weighted Vd were found. Only reflectivity bins that were populated with at least 12 Vd data points were included to assure a representative number-weighted average (and to not emphasize outliers). Next, the fit coefficients a and b found for a particular 20-min time period were used along with the measured 10-s-resolution reflectivity at every third range gate to determine Vt = aZb at 10-s time resolution and 135-m height resolution within that particular 20-min time period. The vertical air motion at 10-s and 135-m resolution was then determined as residual of VdVt(Z) (at every third range gate). If, however, there were less than four reflectivity bins populated with Vd in a particular 20-min time span, the Vt at 10-s resolution was assumed to be independent of the radar reflectivity and equal to the 20-min mean Vd. This scenario is observed in case of highly homogeneous conditions where the radar reflectivity field varies very little. Subsequently, for these cases the w time series are again calculated as the difference between Vd and Vt. In these two ways, best estimates of Vt and w were determined for each cloud pixel at a time resolution of 10 s and a vertical resolution of 135 m. In contrast to the 20-min running-mean technique proposed by Delanoë et al. (2007), the advantage of the used method is that small-scale cloud features can be resolved as shown in Fig. 3. For this thick ice cloud, which was observed between 6- and 11-km height, radar reflectivities increased from about −40 dBZ at cloud top to values of −15 to 0 dBZ in the lower parts of the cloud, mirroring the increase of particle size toward cloud base (cf. Fig. 3a). Though Vd, which mostly ranged between −0.05 and 1 m s−1, were less variable with height, a general increase of Vd toward cloud base is observed (cf. Fig. 3b). In the retrieved Vt the 20-min averaging time periods are obvious, but high-frequency fluctuations of Vt can also be seen, especially near cloud base (cf. Fig. 3c). Highest Vt were retrieved for the areas in which big particles are present near cloud base. Retrieved vertical air motions w are more fluctuating, especially at cloud base; 50% of the w are positive and 50% are negative (cf. Fig. 3d). Columns of updrafts and downdrafts are observed.

Fig. 3.
Fig. 3.

Example of Doppler velocity Vd decomposition into particle terminal fall velocity Vt and vertical air motion w on 8 Dec 2004 at SGP site. Positive velocity values indicate downward motion. (a) Radar reflectivity at 10-s and 45-m resolution, (b) Vd at 10-s and 45-m resolution, (c) Vt at 10-s and 135-m resolution, and (d) w at 10-s and 135-m resolution.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

To check if the 20-min time averaging window is an appropriate choice, sensitivity tests of averaging time variations were performed for two random months of the SGP site time series (June 2007 and January 2008). As estimation of the quality of the Vt = aZb fit, the coefficient of determination R2 was determined. Chosen averaging times were 5, 10, 20, 30, and 60 min. The R2 values were found to decrease with increasing averaging times from 0.57 at 5 min to 0.48 at 30 and 60 min (0.57, 0.54, 0.51, 0.48, and 0.48). Thus, as also stated in Protat and Williams (2011), increasing the averaging times to more than 20 min is not desired, as not only the statistical performance does not improve but also small-scale features are progressively lost. However, using smaller averaging times than 10–20 min might lead to insufficient filtering of vertical air motion. As a consequence, an averaging time of 20 min was considered to lead to best estimates of Vt and w.

The vertical air motion w is determined as difference between two larger values Vd and Vt: . The uncertainty of w is thus composed of measurement uncertainties in Vd (called ) as well as uncertainties in Vt (called ), which arise from the fit Vt = aZb. Assuming that the error contributions are not correlated, the uncertainty of the vertical air motion is determined as . The standard deviation of the fit is determined from the sum of the squared residuals from the regression. Mean were on the order of 0.14 m s−1 (about 20% relative error). With the measurement error of Vd estimated at 0.05 m s−1, the uncertainty in vertical air motion was on the order of 0.15 m s−1 on average.

Considering all sampled high clouds, at SGP, the means and standard deviations of the a and b coefficient amount to 0.69 ± 0.56 and 0.09 ± 0.1, respectively. At Manus, the mean and standard deviation of the fit coefficient a are 0.68 ± 0.55; for the fit coefficient b, they are 0.11 ± 0.12.

To allow comparisons of the Vt retrieved at different heights, an air-density correction factor f(H) = [ρ(H)/ρ0]0.4 [with ρ(H) being the dry air density at a certain height H and ρ0 being the dry air density at the first radar gate] was applied to derive Vt,0, the true reflectivity-weighted terminal fall speed at a given height H, as Vt,0 = Vt(H)/f(H) (Foote and Du Toit 1969; Protat et al. 2003). Air densities were calculated from temperature and pressure given in the ECMWF file. The correction of Vt has a big influence: Vt,0 at 10-km height were found to be about 50% higher than retrieved Vt.

c. Vertical air motion analysis

Based on the retrieved vertical air motion, the dynamical structure of the high clouds was studied. To assure that gravity waves with relatively long periods can be resolved, the following analysis was only made for high clouds for which—at a certain height—the hydrometeor fraction for two consecutive hours was 100%. This was accomplished by selecting 2-h time spans for which the hydrometeor fraction was >95% (corresponding to a total duration of cloud gaps in 2 h shorter than 6 min). Initial data screening showed that persistent high clouds—determined as high clouds having less than 3-min gaps in total during 1 h—were observed frequently at the SGP and Manus sites (on average 21% and 39%, respectively, of the time for which MMCR data was available per month; cf. Fig. 1). If cloud gaps for these selected clouds were <1 min, they were interpolated to increase the amount of data suitable for wavelet and FFT analysis. High clouds, which after the interpolation of small cloud gaps were present for 2 h at a particular height, on average occurred 6.1% (6710 h) and 9.3% (5660 h) of the time for which MMCR data was available per month at the SGP site and the Manus site, respectively.

For each high cloud, the analysis of w was split into three parts. First, a basic statistical analysis for each 1-h time span was made. Second, wavelet analysis was used to detect gravity waves. Finally, Fourier analysis was used to analyze cloud turbulence scales and derive the eddy dissipation rate. With a 1-h overlap in the 2-h time series analysis, all results from this section have an hourly resolution. Results of the SGP site case study example of 8 December 2004 are used to illustrate the methodology.

1) Statistical analysis of w

Based on 2-h time spans for which continuous high cloud observations at a particular height were detected, the hourly mean, standard deviation, and skewness of w as well as the updraft fraction (w < 0 m s−1) were determined. Figure 4 shows the result for the SGP site case study example on 8 December 2004. For this case mean hourly w ranged between −0.01 and 0.03 m s−1. Maximum standard deviations in w were observed at cloud bottom indicating strong turbulence in this part of the cloud. At cloud bottom (1000–1600 UTC), the skewness was found to be slightly negative, referring to narrow but strong downdrafts and weaker but broader updrafts. For this time period, updraft fractions were less than 50% as shown in Fig. 4d.

Fig. 4.
Fig. 4.

Hourly air motion statistics for high cloud case study on 8 Dec 2004 at SGP site: (a) mean, (b) standard deviation, (c) skewness, and (d) updraft fraction.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

2) Gravity wave detection via wavelet analysis

For each 2-h time span and each height, wavelet analysis was performed on the retrieved vertical air motion time series (10-s resolution) for each hour to detect gravity waves. The analysis followed the excellent description of this technique given in Torrence and Compo (1998) and made use of their publicly available code. The advantage of wavelet analysis over FFT is that is does not only return dominant frequencies of variability of a data time series but also how those frequencies vary in time. Since wavelet bases provide varying window sizes (with low-frequency wavelets being long and high-frequency wavelets being short), small and large events can be isolated (Smith and Jonas 1997). Wavelet analysis is thus highly appropriate when analyzing time series in which a wide range of dominant frequencies occur (Daubechies 1990).

If a peak in the wavelet power spectrum is significantly above the background spectrum (here red noise is chosen as background spectrum), it can be assumed to be a true feature with a certain percent of confidence: in our case 95%. Since smoothing in time enhances confidence in regions of significant power (for details, see Torrence and Compo 1998), the global wavelet spectrum (GWS) was determined as hourly time-averaged local wavelet spectrum. The period of the dominant gravity wave is determined as the maximum GWS peak over the 95% significance level between periods of 300 and 1800 s (5 and 30 min). The amount of energy of it over the 95% confidence is a measure of the gravity wave amplitude. With the assumption that gravity waves are oriented in the horizontal wind direction, their wavelengths λ were calculated as product of gravity wave period and ECMWF horizontal wind speed Ū. With resolved gravity wave periods of 5–30 min, retrieved wavelengths ranged from about 3 to 100 km. Quante (2006) reported 2–5 km as threshold wavelength between turbulence and waves; thus, our smallest detected waves are at the lower end of the gravity wave scale.

Gravity waves in the upper troposphere (UT) have wavelengths of a few kilometers to several hundreds of kilometers. They have a multitude of sources such as flow over topography, convection, wind shear, jet imbalance, fronts, and frontogenesis (Fritts and Alexander 2003). Deep convection, for example, is an important source of gravity waves at tropical latitudes and to a lesser extent also at midlatitudes (Hoffmann and Alexander 2010). Gravity wave sources are not distributed uniformly in the horizontal and have significant seasonal variations (Alexander et al. 2010). Zhang et al. (2012) suggested that features of tropospheric gravity waves are likely controlled by their sources. The different gravity wave formation mechanisms lead to a wide scale of wavelengths. For example, gravity waves from shear generation are reported to have wavelengths of a few to a few tens of kilometers (Fritts and Alexander 2003). Convection is thought to lead to the formation of gravity waves with small horizontal wavelengths of 10–200 km and short periods of 10–100 min (Beres et al. 2002). For orographically generated gravity waves, all horizontal scales from a few kilometers to several hundred kilometers can be forced depending on details of the topography (Preusse et al. 2008). Typical vertical scales of inertia waves at synoptic scales arising from geostrophic adjustment are a few kilometers or more, with horizontal scales along the direction of propagation of 10–100 times larger (Fritts and Alexander 2003). Gravity waves do not only influence the large-scale circulation but are also important factors in the formation and properties of (tropical) cirrus (Stähelin et al. 2011). As emphasized by Lynch et al. (2002), gravity waves may coexist and interact with turbulence or convective circulations, making it difficult to distinguish the individual signals in power spectral analysis. They also stated that the interpretation of results is even further complicated by the tendency of gravity waves to occur intermittently in small groups of variable frequency.

As indicated above, using a 2-h time series of vertical velocity data in the wavelet analysis limits our detection to short gravity waves of about 3–100-km wavelength (or 5–30-min periods), which should be kept in mind in the subsequent analysis. However, this is not to be considered a shortcoming, as discussed in the review paper of Alexander et al. (2010): different observation methods are sensitive to different parts of the gravity wave spectrum; with our methodology, we cover the shortest gravity wave wavelengths.

On average, gravity waves were only detected for 4.9% of the time for which MMCR data were available per month (or 5345 h in total) at the SGP site; this is mainly due to the fact that high clouds being persistent for 2 h (at a certain height) were not observed very frequently [only 6.1% of the time (6710 h), as mentioned above]. At the Manus site, gravity waves were detected during 8.5% of the time (4960 h) that the MMCR detected high clouds.

3) Turbulence analysis via FFT

As a third analysis step of the vertical air motion, Fourier analysis is applied to the 2-h periods of vertical velocities in continuous high clouds with 720 sample points at each height H to identify small-scale turbulence.

The turbulent kinetic energy (TKE) dissipation rate ɛ represents the rate at which TKE is dissipated by viscosity and is a good measure of the energy at turbulent scales (Gultepe and Starr 1995). As stated in Lynch et al. (2002), it can be deduced from the vertical velocity power density spectra, if the inertial subrange with a slope of − (Wyngaard 1973) is clearly established. In this case, ɛ can be estimated from the power spectral density for isotropic turbulence in the inertial subrange S(k) = α ɛ2/3 k−5/3, where α is the Kolmogorov constant of 1.62 (Bouniol et al. 2003). We allowed the inertial subrange slope to vary between − ± 20% in our analysis.

It should be mentioned that, to determine the TKE dissipation rate ɛ, the power spectral density S derived via FFT in the frequency f domain was converted to wavenumber k domain via the relations k = 2πf/Ū and S(k) = Ū S(f)/(2π) (Vinnichenko 1970). The term Ū is the hourly mean horizontal wind speed derived from the ECMWF file. In our study, the inertial subrange was defined as range between frequencies of s = 0.0056 Hz (corresponding to 3-min eddies) and s = 0.05 Hz (being the Nyquist frequency).

Lynch et al. (2002) mention that—at least for airborne measurements—sampling rates below 1 Hz affect the related estimation of dissipation rates. We are aware that sampling rate is a critical issue in terms of accuracy of turbulence analysis. For that reason, our turbulence study is limited to data gathered after the measurement averaging frequency of the MMCRs were increased from 10 to 4 s (Kollias et al. 2007b), which was in October 2003 (SGP) and June 2006 (Manus). As shown in the time series of vertical velocity power density spectra slopes between 0.0056 and 0.05 Hz (Fig. 5), the increase of MMCR sampling rate resulted in capturing more energy contained at higher frequencies, thus decreasing the average power density spectra slopes.

Fig. 5.
Fig. 5.

Time series of monthly mean inertial subrange slopes of the vertical air motion power density spectra determined for (a) the SGP site and (b) the Manus site.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

The inertial subrange slope of S was determined with a power-law fit. The fraction of turbulent energy (TEF)—meaning the energy contained in the inertial subrange—within the entire energy spectrum was also determined.

The FFT-derived results for the case study example on 8 December 2004 are presented in Fig. 6. The fraction of turbulent energy was found to be highest at cloud bottom; there, up to 80% of the total variance in w was attributed to frequencies between 0.05 and 0.0056 Hz. The strong turbulence at cloud bottom is mirrored in the high standard deviation in Fig. 4b. Also, the inertial subrange slope is close to − (Fig. 6b) at cloud bottom but varies between 0 and −3 throughout the cloud, with steeper slopes in areas of higher fractions of turbulence. As shown in Fig. 6c, the highest TKE dissipation rates (of around 10−4–10−6 m2 s−3) are observed at cloud bottom and thus are in agreement with the strong turbulence.

Fig. 6.
Fig. 6.

Hourly results based on FFT on vertical air motion for the high cloud case study on 8 Dec 2004 at SGP site. (a) Fraction of turbulent energy, (b) inertial subrange slope of the power density spectrum, and (c) log10 of the turbulent kinetic dissipation rate ɛ.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

3. Results and discussion

a. High cloud occurrence at SGP and Manus

The methodology described previously was employed for the two extensive datasets. Subsequently, both datasets are compared. The comparisons are made for two subsets of high clouds: those for which no gravity waves (GW) were detected and those for which gravity waves were detected. Histograms of high cloud properties in the midlatitudes and the tropics are shown in Fig. 7 (no GW) and Fig. 8 (with GW). Profiles of high cloud occurrences are shown in Fig. 9a (midlatitudes) and Fig. 9b (tropics), and means and standard deviations for all years and all seasons for both locations and both high cloud subsets are presented in Table 1 (midlatitudes) and Table 2 (tropics). As mentioned in section 2a, high clouds are observed more frequently in the tropics (44% of the time the MMCR was operating) than in the midlatitudes (24% of the time), which is in accordance with values reported in literature (e.g., Comstock et al. 2002; Mace et al. 2006; Stubenrauch et al. 2006; Sassen et al. 2008).

Fig. 7.
Fig. 7.

Histogram of high cloud properties at SGP and Manus for the subset of clouds for which no gravity waves were detected. (a) Radar reflectivity Z, (b) Doppler velocity Vd, (c) reflectivity-weighted particle velocity Vt, (d) reflectivity-weighted particle velocity Vt, corrected by air density, (e) vertical air motion w, (f) in-cloud temperature, (g) cloud height, and (h) cloud thickness.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Fig. 8.
Fig. 8.

As in Fig. 7, but for the subset of clouds for which gravity waves were detected.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Fig. 9.
Fig. 9.

Profile of high cloud occurrences in percent at (a) SGP and (b) Manus. Total high cloud occurrence vs MMCR availability (dark gray), persistent high clouds with 2-h longevity vs total high cloud occurrence (medium gray), and occurrence of high clouds for which GW were detected vs total high cloud occurrence (light gray).

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Table 1.

Climatology of derived high cloud properties at the SGP site (January 1997–December 2010) based on hourly means. The upper part of the table refers to the subset of high clouds for which no GW were detected, and the lower one refers to the subset for which GW were detected. Means and standard deviations are presented for all seasons and each season. Thickness is a mean over all observed high cloud layers (up to four detected).

Table 1.
Table 2.

Climatology of high cloud properties at the Manus site (July 1999–December 2010) based on hourly means.

Table 2.

As shown in Fig. 9, the MMCR detected high clouds at the SGP and Manus site between 6 and 14 km and between 6 and 16 km, respectively. Following the geographically different troposphere depths, the maximum MMCR high cloud detection at the SGP site is at 8–10 km (10% of the time); in Manus, it is higher, at 11–12 km (18% of the time). In Fig. 9, it is interesting to note that, at the SGP site, the maximum of persistent high clouds with minimum 2-h longevity and no cloud gaps (see description in section 2) were detected at the same altitude as the maximum detection of high clouds of any longevity, while at Manus persistent high clouds were mostly observed at lower altitudes (8–10 km) than the maximum occurrence of high clouds of any longevity (11–12 km). Thus, gravity waves at the Manus sites were observed most frequently at 8–10-km height, which is 1–3 km below the height of maximum high cloud occurrence of any longevity.

The profiles in Fig. 9 also show that, at the SGP site, for only 60%–70% of the cases when persistent high clouds were present, gravity waves were detected while at the Manus site this happened for 80%–90%. This might be explained by the fact that GW generated by (deep) convection—the dominant generation mechanism of radar-detectable high clouds in the tropics—have short periods of 5–30 min, which can be resolved well by our technique (cf. section 2). Throughout a whole year, high cloud generation in the midlatitudes might be less often associated with GW formation in general.

b. Gravity wave activity

Figures 10 and 11 give more insight into the seasonal gravity wave characteristics by displaying profiles of GW periods and wavelengths for the whole year as well as the summer [June–August (JJA)] and winter [December–February (DJF)] seasons at both sites. The tropical site is characterized by a lack of seasonal dependence in GW period; found GW periods amount to about 19 ± 5 min in summer and 18 ± 6 min in winter (cf. Fig. 11 and Table 2). Also, when excluding the uppermost cirrus levels where only few GW were detected, there seems to be no significant height dependence of GW period at the Manus site. As described in section 2, GW wavelengths were determined from GW period and ECMWF horizontal wind speed Ū. At the tropical site, Ū does not have a strong seasonality; thus, found GW wavelengths are 9 ± 6 km in summer and winter. GW wavelengths are increasing with height because of the increase of Ū with height. The midlatitude site is characterized by a strong seasonality in GW period and wavelength. In their study based on satellite observation, Hoffmann and Alexander (2010) found that, during thunderstorm season (May–August), more than 95% of the observed GW over the Great Plains are related to deep convective clouds. However, it should be noted that their algorithm detects GW in the stratosphere. Here, it is found that, in the summer when convection is the dominant high cloud generating process, short GW with periods of 16 ± 7 min are detected. In winter, high clouds at the SGP site are associated with synoptic weather patterns and found GW periods are longer with values of 19 ± 7 min (cf. Fig. 10 and Table 1). The profile of GW period shows that, with cold clouds generally reaching higher altitudes at the SGP site in summer, GW are also detected at higher altitudes during JJA. GW periods are nearly constant with height up to an altitude of 11 km and decrease with height above that altitude. In Fig. 10b, the strong seasonality of midlatitude horizontal wind speed and thus GW wavelength are displayed. In summer, found GW wavelengths at the SGP site are 13 ± 11 km; in winter they are 43 ± 24 km. The strong increase of Ū with height in winter is mirrored in the GW wavelength profile.

Fig. 10.
Fig. 10.

Mean profiles of (a) GW period and (b) GW wavelength at the SGP site for all months (black), summer season (JJA; dark gray), and winter season (DJF; light gray).

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Fig. 11.
Fig. 11.

As in Fig. 10, but for Manus.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Figure 12 compares the histograms of GW periods and wavelengths of all altitudes and throughout all the years at both sites. In general, the occurrence frequencies of GW with different periods detected at both sites look similar. GW with periods of 26–28 min were detected most frequently (25%); the occurrence of periods between 10 and 23 min was also similar at the midlatitude and the tropical site. However, GW with very short periods of below 10 min were detected more frequently (10%–15%) at the SGP site than at the Manus site (4%–7%). Figure 12b shows again that GW wavelengths detected at the SGP site have a broader spectrum because of the strong seasonality and vertical dependence of Ū. At the SGP site, GW with λ of 5–70 km were observed, with a small peak of occurrence of GW with λ = 10–15 km. At Manus, detected λ ranged between 5 and 40 km, with a pronounced peak: 42% of all GW have λ of 10 km. This is probable yet another indicator of the more uniform formation mechanism of GW in the tropics.

Fig. 12.
Fig. 12.

Histograms of gravity wave periods and wavelengths detected at SGP and Manus.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

As just discussed and shown in Fig. 12, periods of GW were often found to be >20 min. The question if 20 min is thus sufficiently long to assume the vertical air motion averages out to 0 m s−1 for the Vd decomposition can be posed. As a sensitivity check of the methodology, the averaging time in the Doppler velocity decomposition technique (cf. section 2) was increased from 20 min to 1 h. It was found that this led to a capturing of only about 10% more GW with long periods over 25 min (increase from 25% occurrence frequency to 35% occurrence frequency) than in the 20-min averaging case and thus does not have a strong influence.

c. High cloud reflectivity, cloud height, and temperature

Histograms of hourly means of Z, Vd, Vt, Vt,0, w, cloud temperature T, cloud height, and cloud thickness over the entire time series are presented in Fig. 7 (no GW) and Fig. 8 (with GW). For both high cloud subsets, higher reflectivities were found at Manus. This can be attributed to the higher probability of large ice crystals being present in tropical cirrus than in midlatitude cirrus (Lynch et al. 2002; Deng and Mace 2008). The mean radar reflectivity values at SGP and Manus are −21 and −15 dBZ, respectively, for high clouds for which no GW were detected and −16 and −9 dBZ, respectively, for high clouds for which GW were detected. The comparison of the probability density functions (PDF) of in-cloud temperature (Figs. 7f, 8f) indicates that in the case of high clouds without GW the PDF is strongly positively skewed. At Manus, more than 25% of all observed high clouds without GW have mean in-cloud temperatures of <200 K. Equally cold in-cloud temperatures of high clouds in the tropics have also been reported by, for example, Platt et al. (1987) and Taylor et al. (2011). It is possible, that these high cold clouds were not as long lived as the lower warmer counterparts or that they were not captured entirely by the MMCR (cf. profiles in Fig. 9). The average high cloud bases at SGP and Manus were about 0.5 km higher for high clouds for which no GW were detected (8.5 and 10 km, respectively; cf. Tables 1, 2) and with a mean cloud depth of 1.4 km they were about 1.1 km thinner than the high clouds for which GW were detected. These findings are in accordance with Platt et al. (1987), who also reported that, for very low temperatures of high clouds in the tropics (below 203 K), the cloud depth decreases strongly. At SGP, the cloud bases for both high cloud subsets increased and decreased with the seasonal deepening and thinning of the troposphere over this region (cf. Table 1) as also found in similar studies such as Mace et al. (2006) and Deng and Mace (2008).

d. Particle terminal fall velocity

Positive Doppler velocities Vd indicate downward motion. The means and standard deviations of Vd for the high clouds for which no GW were detected were similar for both sites (0.52 ± 0.35 m s−1). However, the mean Vd was higher when GW were detected (0.65 ± 0.31 m s−1 and 0.59 ± 0.37 m s−1) for the Manus and SGP site, respectively. The mean values (Tables 1, 2) of Vd as well as the positive skewness in the Vd histogram are mirrored in the Vt histogram. However, for comparability of Vt, the air-density-corrected reflectivity-weighted particle terminal fall velocities Vt,0 should be looked at. When doing that, the positive skewness in the histogram is removed (cf. Figs. 7d, 8d) and, because of different cloud height distributions at both sites, differences between in Vt,0 for SGP and Manus are higher than for Vt. For the subset of high clouds for which no GW were observed, Vt,0 amount to 0.8 ± 0.48 m s−1 and 0.86 ± 0.49 m s−1, respectively. For the subset of high clouds for which GW were detected, Vt,0 at SGP and Manus amount to 0.89 ± 0.52 m s−1 and 1.03 ± 0.41 m s−1, respectively. Tables 1 and 2 also show that, while at the tropical site only small seasonal differences in the discussed cloud parameters can be found, the values of cold cloud thickness, Vd, Vt, and Vt,0 at SGP are largest in the summer season and smallest in winter season, which reflects the dependence of high cloud characteristics on their generation mechanism and thus in turn on the large-scale meteorological behavior. Our Vt values of 0.51 ± 0.32 m s−1 (SGP) and 0.52 ± 0.34 m s−1 compare well with the ones reported by Deng and Mace (2008), who compared SGP and Manus ice cloud dynamics based on data between January 1999 and December 2005. They found means and standard deviations of 0.55 ± 0.25 m s−1 for SGP and 0.53 ± 0.33 m s−1 for Manus.

e. Vertical air motion

Mean cloud-scale vertical air motion w (which is shown in Fig. 7e, 8e) amounted to 0 m s−1 at both measurement sites with standard deviations on the order of ±0.27 m s−1 (SGP) and ±0.29 m s−1 (Manus) for the high clouds without GW and ±0.22 m s−1 (both sites) for the high clouds with GW. At the SGP site, the variability of w for both high cloud subsets was found to be highest in spring–summer and lowest in winter. It should be kept in mind that here hourly averages of w over several years are reported. The large standard deviations illustrate that strong cloud-scale updrafts and downdrafts are possible which also have been reported in cirrus dynamics case studies (e.g., Gultepe and Starr 1995). There they found w in midlatitude cirrus to vary strongly with amplitudes of 0.03 m s−1 (in the presence of mesoscale waves of λ = 170 km) and amplitudes of 0.2 m s−1 (when small-scale wave structures with λ = 10–20 km were observed) and reaching values of up to 0.4–0.7 m s−1 in individual small-scale convective updrafts. Exploring the correlation of found GW wavelength to maximum hourly w in our two datasets showed the following: At SGP, w for short, medium, and long GW with λ < 20 km, λ = 20–30 km, and λ > 30 km had maximum values of 2.1, 1.9, and 0.42 m s−1, respectively. At Manus, amplitudes of w for the same GW-classes were 2.9, 0.77, and 0.1 m s−1. Thus, updraft and downdraft strengths were also found to decrease with increasing GW wavelength. Lynch et al. (2002) stress the fact that the intensity and duration of the local vertical circulation strongly influences the nucleation mechanisms and subsequent cloud development. In addition, Mace et al. (2006) point out that, besides cloud-scale vertical motion, large-scale vertical motion on synoptic scales plays an important role in the cirrus cloud structure in the midlatitudes. However, they also point out that, even though ascent is required for nucleation and maintenance of cirrus at some (smaller) scales, large-scale vertical motion is nearly as likely to be descending as ascending at times when cirrus clouds are observed in the midlatitudes. We checked if the large-scale ECMWF vertical velocity Ω (Pa s−1) has an influence on the high cloud occurrence. For both sites, we found that large-scale ascent (negative Ω) was favorable for the occurrence of high clouds: roughly two thirds of all high clouds were found in such conditions. A closer look at the two high cloud subsets showed that this dependence was even more pronounced when high clouds with GW were observed: At SGP and Manus, 0.62% and 0.71% of high clouds with GW were observed during large-scale ascent; for high clouds for which no GW were detected, this was the case in 58% and 65% of the observations, respectively.

f. Turbulence analysis results

Results of the turbulence analysis based on FFT of w are presented in Tables 1 and 2 as well as Figs. 13 and 14. As discussed in section 2, the turbulence analysis was restricted to data points for which the inertial subrange slope of − ± 20% in the power density spectrum of w was clearly established. For verification of this limitation, the histograms of the inertial subrange slopes are also shown and their annual means and seasonal values are presented in Tables 1 and 2.

Fig. 13.
Fig. 13.

Histograms of (a) vertical velocity inertial subrange slopes for which ɛ were determined, (b) turbulent kinetic energy dissipation rates ɛ, and (c) fraction of energy contained in the inertial subrange for the high cloud subset for which no gravity waves were detected at the SGP and Manus sites.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

Fig. 14.
Fig. 14.

As in Fig. 13, but for the high cloud subset for which gravity waves were detected.

Citation: Journal of Climate 26, 17; 10.1175/JCLI-D-12-00695.1

The mean TEF within the entire energy spectrum was found to be higher for the high cloud subset for which no GW were detected (0.44 ± 0.17 at SGP, 0.36 ± 0.14 at Manus) than for the high clouds for which GW were detected (0.27 ± 0.13 at SGP, 0.16 ± 0.11 at Manus). Seasonal differences were more pronounced at SGP than at Manus. At the SGP, the highest TEF were observed in winter, lowest in summer for both high cloud subsets. Remarkable in the histogram of TEF of the high clouds for which GW were detected is the 42% frequency of occurrence of TEF = 0.1–0.2 (cf. Fig. 14c).

Turbulent kinetic energy dissipation rates ɛ were higher in the cold clouds for which no GW were detected. At SGP, ɛ in high clouds without and with GW amounted to (0.31 ± 2.1) × 10−4 m2 s−3 and (0.12 ± 1.5) × 10−4 m2 s−3, respectively. At Manus, values of (0.69 ± 4.8) × 10−4 m2 s−3 and (0.1 ± 1.0) × 10−4 m2 s−3 were observed for the two high cloud subsets (with and without GW), respectively. Values of ɛ are highly variable: the standard deviations of the hourly periods are one magnitude higher than the hourly means. Gultepe and Starr (1995) present an overview of field experiments for which turbulence measurements have been evaluated. They report ɛ values of 0.01 × 10−4 to 10 × 10−4 m2 s−3 in cirrus. Lynch et al. (2002) summarize that, on average, ɛ is one order of magnitude larger in cirrus clouds than in cloud-free regions of the upper troposphere.

Smith and DelGenio (2001) tested if there was a correlation between cloud-average ɛ and cloud-layer-average wind shear. However, no correlation was found that they attributed to the fact that turbulence typically occurs in vertically thin layers that are dynamically decoupled from each other because of intervening layers of high stability. We also checked the relationship between ɛ and vertical shear of wind speed and wind direction (based on ECMWF wind data) but could not find a correlation, either. This result is attributed to the time resolution used in the analysis: when using hourly averages of ɛ and shear, lots of small-scale variation cannot be captured anymore. However, using a higher time resolution is also not feasible in our study because of the size of the dataset.

4. Summary and conclusions

Long-term ground-based cloud radar observations provide the opportunity to develop a high cloud dynamics climatology. In this study, ARSCL observations from the ARM midlatitude site at SGP (January 1997–December 2010) and the tropical site at Manus (July 1999–December 2010) were employed. Based on the vertically pointing 35-GHz Doppler cloud radar MMCR, a decomposition of the Doppler velocities into reflectivity-weighted particle velocity Vt and vertical air motion w was introduced. For high clouds persisting for 2 h, wavelet analysis on the retrieved w was used to detect gravity waves (GW). A turbulence analysis of the persistent high clouds was made via FFT on the vertical air motion. Probability density functions of the hourly mean values of the high cloud parameters such as reflectivity, Vt, w, in-cloud temperature, cloud height, cloud thickness, GW period and wavelength, and turbulent kinetic energy dissipation rates ɛ were presented. The climatology of hourly means and standard deviations over all years and each season for two high cloud subsets—one for which no GW were detected and one for which GW were detected—were constructed. Results presented here are only applicable to the subset of cold clouds which is detectable by the MMCR and thus not to high thin cirrus.

As found in other studies, high clouds were observed less frequently in the midlatitudes (24% of the time) than in the tropics (44% of the time). The observations indicated that maximum occurrence of these clouds in the midlatitudes was at 8–10 km while it was higher in Manus (11–12 km). GW detection at the SGP site had its maximum at the same height as the maximum high cloud occurrence; however, in Manus GW were most often detected at an altitude of 8–10 km, which is 1–3 km below the height of maximum high cloud occurrence of any longevity. It could thus be argued that the presence of gravity waves increases the lifetime of the tropical high clouds.

Seasonal profiles of GW periods and wavelength were studied. For Manus, almost no seasonal dependence in GW period and wavelength was found; in the summer and winter seasons, GW periods were about 19 ± 5 min (9 ± 6 km). The SGP site, however, was characterized by a strong seasonality due to different high cloud and GW generating mechanisms. In summer, deep convection is the dominant high cloud generating process and short GW with periods (wavelengths) of 16 ± 7 min (13 ± 11 km) were observed. In winter, when high clouds at the SGP site are related to synoptic weather patterns, GW periods (wavelengths) were longer, with values of 19 ± 7 min (43 ± 24 km).

Mean MMCR reflectivities in high clouds with GW were 5–6 dBZ higher than for high clouds in which no GW were detected. With radar reflectivity being a strong function of particle size, this suggests that larger particles are observed in high clouds with GW. For both sites, the subset of high clouds for which no GW were detected was characterized by Vd of 0.52 ± 0.35 m s−1. The Vd at both sites are larger in high clouds for which GW were found though. At Manus and SGP, they amount to 0.65 ± 0.31 m s−1 and 0.59 ± 0.37 m s−1, respectively. Ice particle growth is on the one hand attributed to the influence of the GW itself: the updrafts of the ascending parts of the wave lead to supersaturation and thus favor ice particle formation and growth (Dean et al. 2005). Additionally, the altitude of the high cloud and thus the in-cloud temperature plays a role. Ice particle size increases with increasing in-cloud temperature (Heymsfield and Miloshevich 2003); bigger ice particles were thus likely present in the high clouds associated with GW, which had mean in-cloud temperatures of 226 and 223 K at the SGP and Manus sites, respectively, which is on average 2 and 5 K warmer than mean in-cloud temperatures of the high clouds without GW, respectively.

In terms of air-density-corrected reflectivity-weighted particle velocity Vt,0, means and standard deviations over the entire datasets at both stations for the subset of high clouds for which no GW were detected amounted to 0.8 ± 0.48 m s−1 and 0.86 ± 0.49 m s−1 at SGP and Manus, respectively. Differences between both sites are attributed to different high cloud-altitude distributions and thus probably also to related differences in ice particle size. At both sites, higher values of Vt,0 were found for the high cloud subset for which GW were detected.

Large standard deviations in determined cloud-scale vertical air motion w representing strong cloud-scale updrafts and downdrafts of up to ±0.27 m s−1 (SGP) and ±0.29 m s−1 (Manus) were found. Updraft and downdraft amplitudes were also found to decrease with increasing GW wavelength. In addition, the dependence of high cloud occurrence on large-scale vertical motion based on the ECMWF vertical velocities was studied. For the midlatitude and the tropical site, we found that the majority of high clouds (about ⅔) were observed under large-scale ascent conditions.

Turbulence analysis results can be summed up as follows: The presence of GW in high clouds is associated with a decrease in the fraction of energy from turbulence in the entire vertical velocity energy density spectrum as well as to a decrease of turbulent kinetic energy dissipation rate, respectively.

In summary, the study showed differences in high cloud dynamics at the two different geographic locations. The presence of GW was found to have substantial influences on high cloud properties at both sites. However, the high cloud climatology at the midlatitude site was characterized by stronger seasonality than the tropical one.

As an outlook, further research could be done in the direction of diurnal variations of high cloud dynamical properties. The influence of large-scale atmospheric state on the dynamical properties is also of interest and could be determined with an atmospheric state regime classification. Furthermore, the retrieved Vt and the associated radar reflectivity and temperature profiles will be used as input in ice-cloud microphysical retrieval algorithms (e.g., Szyrmer et al. 2012).

Acknowledgments

Support for this research was funded by the Office of Biological and Environmental Research of the Environmental Sciences Division of the U.S. Department of Energy as part of the Atmospheric Systems Research (ASR) program.

APPENDIX

List of Variables and Abbreviations

Table A1.

List of variables used.

Table A1.

REFERENCES

  • Ackerman, T. P., and G. M. Stokes, 2003: The Atmospheric Radiation Measurement Program. Phys. Today, 56, 3844.

  • Alexander, M. J., and Coauthors, 2010: Recent developments in gravity-wave effects in climate models and the global distribution of gravity wave momentum flux from observations and models. Quart. J. Roy. Meteor. Soc., 136, 11031124.

    • Search Google Scholar
    • Export Citation
  • Beres, J. H., M. J. Alexander, and J. R. Holton, 2002: Effects of tropospheric wind shear on the spectrum of convectively generated gravity waves. J. Atmos. Sci., 59, 18051824.

    • Search Google Scholar
    • Export Citation
  • Bouniol, D., A. J. Illingworth, and R. J. Hogan, 2003: Deriving turbulent kinetic energy dissipation rate within clouds using ground based 94 GHz radar. Preprints, 31st Int. Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 4A.5. [Available online at http://ams.confex.com/ams/pdfpapers/63826.pdf.]

  • Clothiaux, E. E., T. P. Ackerman, G. G. Mace, K. P. Moran, R. T. Marchand, M. A. 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.

    • Search Google Scholar
    • Export Citation
  • Comstock, J. M., and C. Jakob, 2004: Evaluation of tropical cirrus cloud properties derived from ECMWF model output and ground based measurements over Nauru Island. Geophys. Res. Lett., 31, L10106, doi:10.1029/2004GL019539.

    • Search Google Scholar
    • Export Citation
  • Comstock, J. M., T. P. Ackerman, and G. G. Mace, 2002: Ground-based lidar and radar remote sensing of tropical cirrus clouds at Nauru Island: Cloud statistics and radiative impacts. J. Geophys. Res., 107, 4714, doi:10.1029/2002JD002203.

    • Search Google Scholar
    • Export Citation
  • Daubechies, I., 1990: The wavelet transform time-frequency localization and signal analysis. IEEE Trans. Inf. Theory, 36, 9611004.

  • Dean, S. M., B. N. Lawrence, R. G. Grainger, and D. N. Heuff, 2005: Orographic cloud in a GCM: The missing cirrus. Climate Dyn., 24, 771780.

    • Search Google Scholar
    • Export Citation
  • Delanoë, J., A. Protat, D. Bouniol, A. Heymsfield, A. Bansemer, and P. Brown, 2007: The characterization of ice cloud properties from Doppler radar measurements. J. Appl. Meteor. Climatol., 46, 16821698.

    • Search Google Scholar
    • Export Citation
  • Deng, M., and G. G. Mace, 2006: Cirrus microphysical properties and air motion statistics using cloud radar Doppler moments. Part I: Algorithm description. J. Appl. Meteor. Climatol., 45, 16901709.

    • Search Google Scholar
    • Export Citation
  • Deng, M., and G. G. Mace, 2008: Cirrus microphysical properties and air motion statistics using cloud radar Doppler moments. Part II: Climatology. J Appl. Meteor. Climatol., 47, 32213235.

    • Search Google Scholar
    • Export Citation
  • Dufresne, J. L., and S. Bony, 2008: An assessment of the primary sources of spread of global warming estimates from coupled atmosphere–ocean models. J. Climate, 21, 51355144.

    • Search Google Scholar
    • Export Citation
  • Foote, G. B., and P. S. Du Toit, 1969: Terminal velocity of raindrops aloft. J. Appl. Meteor., 8, 249253.

  • Fritts, D. C., and M. J. Alexander, 2003: Gravity wave dynamics and effects in the middle atmosphere. Rev. Geophys., 41, 1003, doi:10.1029/2001RG000106.

    • Search Google Scholar
    • Export Citation
  • Gultepe, I., and D. O. Starr, 1995: Dynamical structure and turbulence in cirrus clouds: Aircraft observations during the FIRE. J. Atmos. Sci., 52, 41594182.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., and L. M. Miloshevich, 2003: Parameterizations for the cross-sectional area and extinction of cirrus and stratiform ice cloud particles. J. Atmos. Sci., 60, 936956.

    • Search Google Scholar
    • Export Citation
  • Hoffmann, L., and M. J. Alexander, 2010: Occurrence frequency of convective gravity waves during the North American thunderstorm season. J. Geophys. Res., 115, D20111, doi:10.1029/2010JD014401.

    • Search Google Scholar
    • Export Citation
  • Kärcher, B., and U. Lohmann, 2002: A parameterization of cirrus cloud formation: Homogeneous freezing including effects of aerosol size. J. Geophys. Res., 107, 4698, doi:10.1029/2001JD001429.

    • Search Google Scholar
    • Export Citation
  • Kärcher, B., and J. Ström, 2003: The roles of dynamical variability and aerosols in cirrus cloud formation. Atmos. Chem. Phys., 3, 823838.

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

    • Search Google Scholar
    • Export Citation
  • Kollias, P., E. E. Clothiaux, M. A. Miller, E. P. Luke, K. L. Johnson, K. P. Moran, K. B. Widener, and B. A. Albrecht, 2007b: The Atmospheric Radiation Measurement Program cloud profiling radars: Second-generation sampling strategies, processing, and cloud data products. J. Atmos. Oceanic Technol., 24, 11991214.

    • Search Google Scholar
    • Export Citation
  • Kollias, P., G. Tselioudis, and B. A. Albrecht, 2007c: Cloud climatology at the Southern Great Plains and the layer structure, drizzle, and atmospheric modes of continental stratus. J. Geophys. Res., 112, D09116, doi:10.1029/2006JD007307.

    • Search Google Scholar
    • Export Citation
  • Liou, K. N., 1986: Influence of cirrus clouds on weather and climate processes: A global perspective. Mon. Wea. Rev., 114, 11671199.

  • Lynch, D., K. Sassen, D. Starr, and G. Stephens, Eds., 2002: Cirrus. Oxford University Press, 480 pp.

  • Mace, G. G., and S. Benson, 2008: The vertical structure of cloud occurrence and radiative forcing at the SGP ARM site as revealed by 8 years of continuous data. J. Climate, 21, 25912610.

    • Search Google Scholar
    • Export Citation
  • Mace, G. G., S. Benson, and E. Vernon, 2006: Cirrus clouds and the large-scale atmospheric state: Relationships revealed by six years of ground-based data. J. Climate, 19, 32573278.

    • Search Google Scholar
    • Export Citation
  • Matrosov, S. Y., B. W. Orr, R. A. Kropfli, and J. B. Snider, 1994: Retrieval of vertical profiles of cirrus cloud microphysical parameters from Doppler radar and infrared radiometer measurements. J. Appl. Meteor., 33, 617626.

    • Search Google Scholar
    • Export Citation
  • Mitchell, D. L., P. Rasch, D. Ivanova, G. McFarquhar, and T. Nousiainen, 2008: Impact of small ice crystal assumptions on ice sedimentation rates in cirrus clouds and GCM simulations. Geophys. Res. Lett., 35, L09806, doi:10.1029/2008GL033552.

    • Search Google Scholar
    • Export Citation
  • Moran, K. P., B. E. Martner, M. J. Post, R. A. Kropfli, D. C. Welsh, and K. B. Widener, 1998: An unattended cloud-profiling radar for use in climate research. Bull. Amer. Meteor. Soc., 79, 443455.

    • Search Google Scholar
    • Export Citation
  • Orr, B. W., and R. A. Kropfli, 1999: A method for estimating particle fall velocities from vertically pointing Doppler radar. J. Atmos. Oceanic Technol., 16, 2937.

    • Search Google Scholar
    • Export Citation
  • Pincus, R., C. P. Batstone, R. J. P. Hofmann, K. E. Taylor, and P. J. Glecker, 2008: Evaluating the present-day simulation of clouds, precipitation, and radiation in climate models. J. Geophys. Res., 113, D14209, doi:10.1029/2007JD009334.

    • Search Google Scholar
    • Export Citation
  • Plana-Fattori, A., A. Protat, and J. Delanoë, 2010: Observing ice clouds with a Doppler cloud radar. C. R. Phys., 11, 96103.

  • Platt, C. M. R., J. C. Scott, and A. C. Dilley, 1987: Remote sensing of high clouds. Part VI: Optical properties of midlatitude and tropical cirrus. J. Atmos. Sci., 44, 729744.

    • Search Google Scholar
    • Export Citation
  • Preusse, P., S. D. Eckermann, and M. Ern, 2008: Transparency of the atmosphere to short horizontal wavelength gravity waves. J. Geophys. Res., 113, D24104, doi:10.1029/2007JD009682.

    • Search Google Scholar
    • Export Citation
  • Protat, A., and C. R. Williams, 2011: The accuracy of radar estimates of ice terminal fall speed from vertically pointing Doppler radar measurements. J. Appl. Meteor. Climatol., 50, 21202138.

    • Search Google Scholar
    • Export Citation
  • Protat, A., Y. Lemaître, and D. Bouniol, 2003: Terminal fall velocity and the FASTEX cyclones. Quart. J. Roy. Meteor. Soc., 129, 15131535.

    • Search Google Scholar
    • Export Citation
  • Quante, M., 2006: Turbulenz in Cirruswolken mittlerer Breiten. Ph.D. thesis, Universität Hamburg, 168 pp.

  • Randall, D. A., and Coauthors, 2007: Climate models and their evaluation. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 589–662.

  • Sanderson, B. M., C. Piani, W. J. Ingram, D. A. Stone, and M. R. Allen, 2008: Towards constraining climate sensitivity by linear analysis of feedback patterns in thousands of perturbed-physics GCM simulations. Climate Dyn., 30, 175190.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., Z. Wang, and D. Liu, 2008: Global distribution of cirrus clouds from CloudSat/Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) measurements. J. Geophys. Res., 113, D00A12, doi:10.1029/2008JD009972.

    • Search Google Scholar
    • Export Citation
  • Smith, S. A., and P. R. Jonas, 1997: Wavelet analysis of turbulence in cirrus clouds. Ann. Geophys., 15, 14471456.

  • Smith, S. A., and A. D. DelGenio, 2001: Analysis of aircraft, radiosonde, and radar observations in cirrus clouds observed during FIRE II: The interactions between environmental structure, turbulence, and cloud microphysical properties. J. Atmos. Sci., 58, 451461.

    • Search Google Scholar
    • Export Citation
  • Smith, S. R., D. M. Legler, M. J. Remigio, and J. J. O'Brien, 1999: Comparison of 1997–98 U.S. temperature and precipitation anomalies to historical ENSO warm phases. J. Climate, 12, 35073515.

    • Search Google Scholar
    • Export Citation
  • Stähelin, J., T. G. Shepherd, and T. Peter, 2011: Report on the 18th Session of the SPARC Scientific Steering Group, 2–5 February 2011, Pune, India. SPARC Newsletter, No. 37, Stratospheric Process and Their Role in Climate, Zurich, Switzerland, 1–5.

  • Stubenrauch, C. J., A. Chédin, G. Rädel, N. A. Scott, and S. Serrar, 2006: Cloud properties and their seasonal and diurnal variability from TOVS Path-B. J. Climate, 19, 55315553.

    • Search Google Scholar
    • Export Citation
  • Szyrmer, W., A. Tatarevic, and P. Kollias, 2012: Ice clouds microphysical retrieval using 94-GHz Doppler radar observations: Basic relations within the retrieval framework. J. Geophys. Res., 117, D14203, doi:10.1029/2011JD016675.

    • Search Google Scholar
    • Export Citation
  • Taylor, J. R., W. J. Randel, and E. J. Jensen, 2011: Cirrus cloud-temperature interactions in the tropical tropopause layer: A case study. Atmos. Chem. Phys., 11, 10 08510 095.

    • Search Google Scholar
    • Export Citation
  • Torrence, C., and G. P. Compo, 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79, 6178.

  • Vinnichenko, N. K., 1970: The kinetic energy spectrum in the free atmosphere—1 second to 5 years. Tellus, 22, 158166.

  • Webb, M. J., F. H. Lambert, and J. M. Gregory, 2012: Origins of differences in climate sensitivity, forcing and feedback in climate models. Climate Dyn., 40, 677–707, doi:10.1007/s00382-012-1336-x.

    • Search Google Scholar
    • Export Citation
  • Wyngaard, J. C., 1973: On surface-layer turbulence. Workshop on Micrometeorology, D. A. Haugen, Ed., Amer. Meteor. Soc., 101–149.

  • Zhang, S. D., F. Yi, C. M. Huang, and K. M. Huang, 2012: High vertical resolution analyses of gravity waves and turbulence at a midlatitude station. J. Geophys. Res., 117, D02103, doi:10.1029/2011JD016587.

    • Search Google Scholar
    • Export Citation
Save