• Bélair, S., L-P. Crevier, J. Mailhot, B. Bilodeau, and Y. Delage, 2003a: Operational implementation of the ISBA land surface scheme in the Canadian regional weather forecast model. Part I: Warm season results. J. Hydrometeor., 4 , 352370.

    • Search Google Scholar
    • Export Citation
  • Bélair, S., R. Brown, J. Mailhot, B. Bilodeau, and L-P. Crevier, 2003b: Operational implementation of the ISBA land surface scheme in the Canadian regional weather forecast model. Part II: Cold season results. J. Hydrometeor., 4 , 371386.

    • Search Google Scholar
    • Export Citation
  • Bélair, S., J. Mailhot, C. Girard, and P. Vaillancourt, 2005: Boundary layer and shallow cumulus clouds in a medium-range forecast of a large-scale weather system. Mon. Wea. Rev., 133 , 19381960.

    • Search Google Scholar
    • Export Citation
  • Benoit, R., J. Côté, and J. Mailhot, 1989: Inclusion of a TKE boundary layer parameterization in the Canadian regional finite-element model. Mon. Wea. Rev., 117 , 17261750.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., and Y. Zeng, 2004: Bulk microphysical sensitivities within the MM5 for orographic precipitation. Part I: The Sierra 1986 event. Mon. Wea. Rev., 132 , 27802801.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., M. F. Garvert, J. B. Wolfe, C. F. Mass, and C. Woods, 2005: The 13–14 December 2001 IMPROVE-2 event. Part III: Simulated microphysical budgets and sensitivity studies. J. Atmos. Sci., 62 , 35353558.

    • Search Google Scholar
    • Export Citation
  • Côté, J., S. Gravel, A. Méthot, A. Patoine, M. Roch, and A. Staniforth, 1998: The operation CMC-MRB Global Environmental Multiscale (GEM) model. Part I: Design considerations and formulation. Mon. Wea. Rev., 126 , 13731395.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., G. J. Tripoli, R. M. Rauber, and E. A. Mulvhill, 1986: Numerical simulation of the effects of varying ice crystal nucleation rates and aggregation processes on orographic snowfall. J. Climate Appl. Meteor., 25 , 16581680.

    • Search Google Scholar
    • Export Citation
  • Ferrier, B. S., 1994: A double-moment multiple-phase four-class bulk ice scheme. Part I: Description. J. Atmos. Sci., 51 , 249280.

  • Field, P. R., R. J. Hogan, P. R. A. Brown, A. J. Illingworth, T. W. Choularton, and R. J. Cotton, 2005: Parameterization of ice-particle size distributions for mid-latitude stratiform clouds. Quart. J. Roy. Meteor. Soc., 131 , 15071513.

    • Search Google Scholar
    • Export Citation
  • Fouquart, Y., and B. Bonnel, 1980: Computations of solar heating of the earth’s atmosphere: A new parameterization. Contrib. Atmos. Phys., 53 , 3562.

    • Search Google Scholar
    • Export Citation
  • Garand, L., and J. Mailhot, 1990: The influences of infrared radiation on numerical weather forecasts. Preprints. Seventh Conf. on Atmospheric Radiation, San Francisco, CA, Amer. Meteor. Soc., J146–J151.

    • Search Google Scholar
    • Export Citation
  • Garvert, M. F., B. A. Colle, and C. F. Mass, 2005a: The 13–14 December 2001 IMPROVE-2 event. Part I: Evolution and comparison with a mesoscale model simulation. J. Atmos. Sci., 62 , 34723492.

    • Search Google Scholar
    • Export Citation
  • Garvert, M. F., C. P. Woods, B. A. Colle, C. F. Mass, P. V. Hobbs, M. T. Stoelinga, and J. B. Wolfe, 2005b: The 13–14 December 2001 IMPROVE-2 event. Part II: Comparisons of MM5 model simulations of clouds and precipitation with observations. J. Atmos. Sci., 62 , 35203534.

    • Search Google Scholar
    • Export Citation
  • Garvert, M. F., B. Smull, and C. Mass, 2007: Multiscale mountain waves influencing a major orographic precipitation event. J. Atmos. Sci., 64 , 711737.

    • Search Google Scholar
    • Export Citation
  • Geresdi, I., 1998: Idealized simulation of the Colorado hailstorm case: Comparison of bulk and detailed microphysics. Atmos. Res., 45 , 237252.

    • Search Google Scholar
    • Export Citation
  • Grabowski, W. W., 2006: Sixth WMO International Cloud Modeling Workshop. Bull. Amer. Meteor. Soc., 87 , 639642.

  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev., 121 , 764787.

  • Heymsfield, A. J., and J. Iaquinta, 2000: Cirrus crystal terminal velocities. J. Atmos. Sci., 57 , 916938.

  • Hong, S-Y., and H. L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-rage forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Houze Jr., R. A., and S. Medina, 2005: Turbulence as a mechanism for orographic precipitation enhancement. J. Atmos. Sci., 62 , 35993623.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 165–177.

    • Search Google Scholar
    • Export Citation
  • Kogan, Y., 1991: The simulation of a convective cloud in a 3D model with explicit microphysics. Part I: Model description and sensitivity experiments. J. Atmos. Sci., 48 , 11601189.

    • Search Google Scholar
    • Export Citation
  • Kong, F., and M. K. Yau, 1997: An explicit approach to microphysics in MC2. Atmos.–Ocean, 35 , 257291.

  • Lin, Y. L., R. Farley, and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor., 22 , 10651092.

    • Search Google Scholar
    • Export Citation
  • Lo, K. K., and R. E. Passarelli, 1982: The growth of snow in winter storms: An airborne observational study. J. Atmos. Sci., 39 , 697706.

    • Search Google Scholar
    • Export Citation
  • Locatelli, J. D., and P. V. Hobbs, 1974: Fall speeds and masses of solid precipitation particles. J. Geophys. Res., 79 , 21852197.

  • Magono, C., and C. W. Lee, 1966: Meteorological classification of natural snow crystals. J. Faculty Sci. Hokkaido Univ., 2 , 321335.

  • Mailhot, J., and Coauthors, 1998: Scientific description of RPN physics library—Version 3.6. Recherche en prévision numérique, Canadian Meteorological Centre, 188 pp.

  • Marshall, J. S., and W. M. K. Palmer, 1948: The distribution of raindrops with size. J. Meteor., 5 , 165166.

  • Meyers, M. P., R. L. Walko, J. Y. Harrington, and W. R. Cotton, 1997: New RAMS cloud microphysics. Part II: The two-moment scheme. Atmos. Res., 45 , 339.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., and M. K. Yau, 2005a: A multimoment bulk microphysics parameterization. Part I: Analysis of the role of the spectral shape parameter. J. Atmos. Sci., 62 , 30513064.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., and M. K. Yau, 2005b: A multimoment bulk microphysics parameterization. Part II: A proposed three-moment closure and scheme description. J. Atmos. Sci., 62 , 30653081.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., and M. K. Yau, 2006: A multimoment bulk microphysics parameterization. Part III: Control simulation of a hailstorm. J. Atmos. Sci., 63 , 31143136.

    • Search Google Scholar
    • Export Citation
  • Morrison, H., J. A. Curry, and V. I. Khvorostyanov, 2005: A new double-moment microphysics parameterization for application in cloud and climate models. J. Atmos. Sci., 62 , 16651677.

    • Search Google Scholar
    • Export Citation
  • Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. Kluwer Academic, 914 pp.

  • Reisin, T., Z. Levin, and S. Tzivion, 1996: Rain production in convective clouds as simulated in an axisymmetric model with detailed microphysics. Part I: Description of the model. J. Atmos. Sci., 53 , 497520.

    • Search Google Scholar
    • Export Citation
  • Reisner, J., R. M. Rasmussen, and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124 , 10711107.

    • Search Google Scholar
    • Export Citation
  • Seifert, A., and K. D. Beheng, 2001: A double-moment parameterization for simulating autoconversion, accretion and self-collection. Atmos. Res., 59–60 , 265292.

    • Search Google Scholar
    • Export Citation
  • Smith, P. L., 1984: Equivalent radar reflectivity factors for snow and ice particles. J. Climate Appl. Meteor., 23 , 12581260.

  • Stoelinga, M., and Coauthors, 2003: Improvement of microphysical parameterization through observational verification experiment. Bull. Amer. Meteor. Soc., 84 , 18071826.

    • Search Google Scholar
    • Export Citation
  • Ström, J., B. Strauss, F. Shröder, T. Anderson, J. Heintzenberg, and P. Wendling, 1997: In situ observations of the microphysical properties of young cirrus clouds. J. Atmos. Sci., 54 , 25422553.

    • Search Google Scholar
    • Export Citation
  • Sundqvist, H., E. Berge, and J. E. Kristjánsson, 1989: Condensation and cloud parameterization studies with a mesoscale numerical weather prediction model. Mon. Wea. Rev., 117 , 16411657.

    • Search Google Scholar
    • Export Citation
  • Szyrmer, W., S. Laroche, and I. Zawadzki, 2005: A microphysical bulk formulation based on scaling normalization of the particle size distribution. Part I: Description. J. Atmos. Sci., 62 , 42064221.

    • Search Google Scholar
    • Export Citation
  • Telford, J. W., and P. B. Wagner, 1981: Observations of condensation growth determined by entity type mixing. Pure Appl. Geophys., 119 , 934965.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., R. R. Rasmussen, and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132 , 519542.

    • Search Google Scholar
    • Export Citation
  • Walko, R. L., W. R. Cotton, M. P. Meyers, and J. Y. Harrington, 1995: New RAMS cloud microphysics. Part I: The one-moment scheme. Atmos. Res., 38 , 2962.

    • Search Google Scholar
    • Export Citation
  • Woods, C. P., M. T. Stoelinga, J. D. Locatelli, and P. V. Hobbs, 2005: Microphysical processes and synergistic interaction between frontal and orographic forcing of precipitation during the 13 December 2001 IMPROVE-2 event over the Oregon Cascades. J. Atmos. Sci., 62 , 34933519.

    • Search Google Scholar
    • Export Citation
  • Young, K. C., 1993: Microphysical Processes in Clouds. Oxford University Press, 427 pp.

  • View in gallery

    Nested GEM model grids. Inset indicates nesting times for each grid.

  • View in gallery

    Model orography (shading) for (a) 4-km grid (subdomain) and (b) full 1-km grid. Contours denote elevations of 1500, 2000, and 2500 m. Cross hairs depict model locations of rain gauges.

  • View in gallery

    Infrared images from (left) GOES and (right) OLR from the 12-km GEM simulation at (a), (b) 1200 UTC 13 Dec 2001; (c), (d) 0000 UTC 14 Dec 2001; and (e), (f) 1200 UTC 14 Dec 2001.

  • View in gallery

    Synoptic fields from the 12-km GEM simulation (solid) and modified NCEP-AVN analysis (dashed) at 0000 UTC 14 Dec 2001. (a) The 300-hPa isotachs (m s−1) with shading for simulation values greater than 40 m s−1, (b) 500-hPa geopotential height (dam), (c) 850-hPa temperature (°C) and winds, (d) 850-hPa geopotential height, (e) 1000-hPa temperature and winds, and (f) 1000-hPa geopotential height. Simulation (analysis) wind vectors (m s−1) are black (gray).

  • View in gallery

    Map of the IMPROVE-2 study area. Shading denotes the model orography (4-km grid), with an interval of 500 m. Heavy contours denote elevations of 1500 and 2000 m.

  • View in gallery

    Soundings from 1-km simulation (solid) and UW radiosonde (dashed) at (a) 2100, (b) 0000, and (c) 0400 UTC 14 Dec 2001. See Fig. 5 for location of UW radiosonde site. Full (half) barbs denote wind speeds of 10 m s−1 (5 m s−1).

  • View in gallery

    Reflectivity from the (left) NCAR S-Pol radar (1.5° PPI), (middle) the Portland radar (0.5° PPI), and (right) the equivalent reflectivity from 4-km simulation (700 hPa) at (top) 1800 UTC 13 Dec, (middle) 0000 UTC 14 Dec, and (bottom) 1400 UTC 14 Dec 2001. Dotted and dashed circles denote the ∼150 km (90 mi) and ∼200 km (120 mi) and range rings, respectively. Center circles denote location of NCAR S-Pol and Portland radars. Gray squares in northwest corners denote shadow zones. (right) A portion of the full 4-km simulation domain images shown in Fig. 8 with range rings and radar locations corresponding to the Portland radar.

  • View in gallery

    Equivalent reflectivity at 700 hPa from 4-km simulation at (a) 1800 UTC 13 Dec, (b) 0000 UTC 14 Dec, and (c) 1400 UTC 14 Dec 2001. Dashed and dotted circles denote the ∼200 km (120 mi) and ∼150 km (90 mi) range rings for the Portland and NCAR S-Pol radars, respectively. Contours denote the 1500- and 2000-m model orographic heights. The rectangle denotes the 1-km domain.

  • View in gallery

    As in Fig. 8, but for 1-km simulation.

  • View in gallery

    Vertical velocity (w, time averaged from 2300 to 0100 UTC, every 15 min) from 1-km simulation along the vertical cross section indicated in Fig. 5. Solid (dashed) contours denote upward (downward) motion with contours every 0.5 m s−1 (0 m s−1 contour not shown) and upward motion shaded.

  • View in gallery

    Vertical motion (w) at 0000 UTC along leg 2 from 1-km simulation. (top) The w vs distance along the flight path at 725 hPa (∼2.5 km). (bottom) A vertical cross section along the flight leg, denoted by the dashed line. Solid (dashed) contours denote upward (downward) motion with contours every 0.5 m s−1 (0 m s−1 contour not shown) and upward motion is shaded.

  • View in gallery

    Observed and simulated 18-h accumulated precipitation (1400 UTC 13 Dec–0800 UTC 14 Dec 2001) for the (a) 4-km and (b) 1-km simulations (subdomains only): (left) 4-km and (right) 1-km simulations (subdomains). Background shading denotes model values and rain gauges values are indicated at the crosshair symbols (same shading intervals). Lighter (darker) model values under a particular gauge indicate underpredicted (overpredicted) precipitation totals from the model. The Contours denote model orographic heights of 1500 and 2000 m. The square in (a) denotes the subdomain shown in (b).

  • View in gallery

    Observed vs simulated (nearest grid point to rain gauge) 18-h accumulated precipitation (1400 UTC 13 Dec–0800 UTC 14 Dec 2001) for 4-km (solid circles) and 1-km (open diamonds) simulations. Note that there are fewer rain gauge observations located within the 1-km domain than the 4-km domain.

  • View in gallery

    Observed vs simulated 18-h accumulated precipitation (1400 UTC 13 Dec–0800 UTC 14 Dec 2001) for G05a, MY05b’s 1.33-km run (crosses) and the 1-km triple-moment run (diamonds). Note that there are fewer rain gauge observations located within G05a,MY05b’s 1.33-km run domain.

  • View in gallery

    Vertical cross section of in situ observations of hydrometeor types and mass contents from the P-3 and Convair-580 between 2300 and 0100 UTC. Temperatures indicated are in degrees Celsius. (Reproduced from G05b, with permission.)

  • View in gallery

    Vertical cross sections of time-averaged (2300–0100 UTC, every 15 min) hydrometeor mass contents (Qx) from 1-km triple-moment simulation for (a) cloud, (b) ice and rain, (c) graupel, and (d) snow along cross section indicated in Fig. 5. Quantities are in grams per meter cubed at the values indicated in the shading scales; dashed contours for ice in (b) are 0.001, 0.01, and 0.1 g m−3. The vertical axis is pressure (hPa). North–south flight legs of the P-3 are indicated in (a); flight legs from the Convair-580 and the return flight of the P-3 are indicated in (d). The rectangle in (c) denotes the area of the vertical cross section outlined by the red box in Fig. 15.

  • View in gallery

    As in Fig. 16, but for mean-mass diameters Dx. Quantities are in units of micrometers for (a), micrometers (ice) and millimeters (rain) for (b), and millimeters for (c) and (d) with magnitudes indicated in the shading scales. Contours for ice in (b) are 10, 25, 50, and 75 μm.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 88 83 8
PDF Downloads 52 49 4

Simulation of an Orographic Precipitation Event during IMPROVE-2. Part I: Evaluation of the Control Run Using a Triple-Moment Bulk Microphysics Scheme

View More View Less
  • 1 Meteorological Research Division, Recherche en Prévision Numérique, Dorval, Quebec, Canada
  • | 2 McGill University, Montreal, Quebec, Canada
  • | 3 Meteorological Research Division, Recherche en Prévision Numérique, Dorval, Quebec, Canada
© Get Permissions
Full access

Abstract

This paper reports the first evaluation of the Milbrandt–Yau multimoment bulk microphysics scheme against in situ microphysical measurements. The full triple-moment version of the scheme was used to simulate a case of orographically enhanced precipitation with a 3D mesoscale model at high resolution (4- and 1-km grid spacings). The simulations described in this paper also serve as the control runs for the sensitivity experiments that will be examined in Part II of this series. The 13–14 December 2001 case of heavy orographically enhanced precipitation, which occurred over the Oregon Cascades, was selected since it was well observed during the second Improvement of Microphysical Parameterization through Observational Verification Experiment (IMPROVE-2) observational campaign. The simulated fields were compared with observed radar reflectivity, vertical velocity, precipitation quantities from rain gauges, and microphysical quantities measured in situ by two instrumented aircraft. The simulated reflectivity structure and values compared favorably to radar observations during the various precipitation stages of the event. The vertical motion field in the simulations corresponded reasonably well to the mountain-wave pattern obtained from in situ and dual-Doppler radar inferred measurements, indicating that biases in the simulations can be attributed in part to the microphysics scheme. The patterns of 18-h accumulated precipitation showed that the model correctly simulated the bulk of the precipitation to accumulate along the coastal mountains and along the windward slope of the Cascades, with reduced precipitation on the lee side of the crest. However, both the 4- and 1-km simulations exhibited a general overprediction of precipitation quantities. The model also exhibited a distinct bias toward overprediction of the snow mass concentration aloft and underprediction of the mass and vertical extent of the pockets of cloud liquid water on the windward side of the Cascades. Nevertheless, the overall spatial distribution of the hydrometeor fields was simulated realistically, including the mean-mass particle diameters for each category and the observed trend of larger snow sizes to be located at lower altitudes.

Corresponding author address: Dr. Jason Milbrandt, 2121 Trans-Canada Highway, 5th floor, Dorval, QC H9P 1J3, Canada. Email: jason.milbrandt@ec.gc.ca

Abstract

This paper reports the first evaluation of the Milbrandt–Yau multimoment bulk microphysics scheme against in situ microphysical measurements. The full triple-moment version of the scheme was used to simulate a case of orographically enhanced precipitation with a 3D mesoscale model at high resolution (4- and 1-km grid spacings). The simulations described in this paper also serve as the control runs for the sensitivity experiments that will be examined in Part II of this series. The 13–14 December 2001 case of heavy orographically enhanced precipitation, which occurred over the Oregon Cascades, was selected since it was well observed during the second Improvement of Microphysical Parameterization through Observational Verification Experiment (IMPROVE-2) observational campaign. The simulated fields were compared with observed radar reflectivity, vertical velocity, precipitation quantities from rain gauges, and microphysical quantities measured in situ by two instrumented aircraft. The simulated reflectivity structure and values compared favorably to radar observations during the various precipitation stages of the event. The vertical motion field in the simulations corresponded reasonably well to the mountain-wave pattern obtained from in situ and dual-Doppler radar inferred measurements, indicating that biases in the simulations can be attributed in part to the microphysics scheme. The patterns of 18-h accumulated precipitation showed that the model correctly simulated the bulk of the precipitation to accumulate along the coastal mountains and along the windward slope of the Cascades, with reduced precipitation on the lee side of the crest. However, both the 4- and 1-km simulations exhibited a general overprediction of precipitation quantities. The model also exhibited a distinct bias toward overprediction of the snow mass concentration aloft and underprediction of the mass and vertical extent of the pockets of cloud liquid water on the windward side of the Cascades. Nevertheless, the overall spatial distribution of the hydrometeor fields was simulated realistically, including the mean-mass particle diameters for each category and the observed trend of larger snow sizes to be located at lower altitudes.

Corresponding author address: Dr. Jason Milbrandt, 2121 Trans-Canada Highway, 5th floor, Dorval, QC H9P 1J3, Canada. Email: jason.milbrandt@ec.gc.ca

1. Introduction

The considerable increase in computer power in recent years allows for the use of more sophisticated cloud microphysics schemes in high-resolution atmospheric models. By modeling a complex set of cloud microphysical processes, these schemes have proven useful in forecasting the type and quantity of precipitation at the surface and in providing the correct latent heat release and changes in radiative forcing. An important function of a microphysics scheme is the prediction of the hydrometeor size spectra. There are two basic approaches. The first is the fully explicit bin-resolving method, in which the size spectra are divided into a number of bins (e.g., Kogan 1991; Reisin et al. 1996; Geresdi 1998). In general, this approach is prohibitively expensive for use in 3D models, particularly with regard to numerical weather prediction (NWP). The second approach is referred to as the bulk method, in which the complete spectrum of hydrometeors is partitioned into various categories sharing certain bulk characteristics such as density, shape, or mode of origin (e.g., Lin et al. 1983; Cotton et al. 1986; Ferrier 1994; Reisner et al. 1998). For each category in a bulk microphysics scheme (BMS), the size spectrum is represented by an analytic function characterized by a number of parameters that are functions of the moments of the distribution. The evolution of the size spectra is then obtained from solving the predictive equations for one or more moments of the distribution. Since BMSs generally require far fewer prognostic variables than bin-resolving schemes, bulk schemes are generally preferred in NWP and mesoscale research models.

Several deficiencies in BMSs have been identified over the past few years.1 They include the parameterization of autoconversion of cloud water to rain, the treatment of cloud condensation nuclei, the mechanism for ice initiation, the determination of appropriate terminal velocities of ice-phase particles, the form of the assumed particle size distributions, the representation of the effects of snow aggregation, and several others (Stoelinga et al. 2003, hereinafter S03). In addition, bulk schemes vary considerably in their complexity. The number of hydrometeor categories can range from one (e.g., Sundqvist et al. 1989) to seven (e.g., Walko et al. 1995) or more. The number of prognostic variables per category is generally limited to one (e.g., Lin et al. 1983; Kong and Yau 1997), two (e.g., Ferrier 1994; Meyers et al. 1997; Seifert and Beheng 2001; Morrison et al. 2005), or three (Milbrandt and Yau 2005a, b; Szyrmer et al. 2005). Furthermore, the number of microphysical processes that are included, and the details of how each process is parameterized also vary. It is not obvious a priori which aspects of these schemes are of primary importance and which are secondary. Thus, there is a need to resolve these issues to improve microphysics schemes for NWP and other applications.

To investigate some of the problems pertaining to bulk parameterization, researchers from the University of Washington initiated the Improvement of Microphysical Parameterization through Observational Verification Experiment (IMPROVE), which was conducted in collaboration with various research institutions. As described in S03, the main purpose of the experiment was to provide detailed observations for various weather systems with the specific goal of improving quantitative precipitation forecasts (QPFs) by mesoscale models using BMSs. As part of this program, two observational campaigns were conducted during 2001. The second campaign, referred to as IMPROVE-2 or the Oregon Cascades Orographic Field Study, was carried out in November–December on cases of large-scale weather systems with terrain-enhanced precipitation. The orographic forcing depends on low-level winds that can be represented realistically in high-resolution mesoscale models. The forcing also results in the activation of a wide variety of microphysical processes, thus useful for rigorous tests of various schemes. Seventeen intensive observation periods (IOPs) were recorded during IMPROVE-2. Measurements were made from a variety of remotely sensed and in situ instruments, both ground based and aircraft instrumented.

The 13–14 December 2001 case had particularly intense precipitation and was well observed. This was one of the cases examined during the 2004 Sixth World Meteorological Organization (WMO) International Cloud Modeling Workshop in Hamburg, Germany (Grabowski 2006). It was also the subject of several papers in a special section of the Journal of the Atmospheric Sciences (2005, Vol. 62, No. 10). Garvert et al. (2005a, b, hereinafter G05a,b) presented a synoptic and mesoscale examination of this case and performed high-resolution mesoscale model simulations using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (NCAR) Mesoscale Model (MM5). They showed that certain precipitation forecast errors, in particular large quantities of precipitation that spilt over the lee side of the Cascade Mountains, can be attributed directly to biases in the BMS. Colle et al. (2005, hereinafter C05) continued this modeling study with an examination of the microphysical budgets of the processes simulated by the BMS and showed that the surface precipitation was sensitive to the size distribution and fall velocity of snow and to the snow-to-graupel conversion threshold. Woods et al. (2005, hereinafter WSLH) documented several of the microphysical measurements taken from aircraft- and ground-based instruments during this case and examined the interaction between frontal and orographic forcing for the heavy precipitation. More recently, Garvert et al. (2007, hereinafter GSM) examined the kinematic flow structures for this case by combining mesoscale model simulations with airborne Doppler radar observations. They showed that errors in the model representation of a low-level shear layer and the amplitude of vertically propagating mountain waves, which were important in forcing localized convection, were very sensitive to the planetary boundary layer parameterization scheme. Because of the availability of detailed observations and available modeling studies for comparison, the 13–14 December 2001 case is suitable for further examination of the issues pertaining to BMSs.

A multimoment BMS has recently been developed by Milbrandt and Yau (2005a, b, hereinafter MY05a,b). The scheme includes several hydrometeor categories and has options for one, two, or three prognostic moments for each category. The flexibility of the scheme makes it an ideal tool to investigate various BMS issues, such as the importance of the number of moments. Milbrandt and Yau (2006) performed the first major test of the scheme by simulating a supercell hailstorm using a mesoscale model with a 1-km grid spacing. The simulated storm compared favorably to radar observations and the microphysical fields appeared realistic; however, no in situ microphysical measurements were available for direct comparison other than surface precipitation quantities estimated by radar. Furthermore, subtleties in the behavior of a BMS may be masked in simulations of deep convection since the forcing is strong. It is therefore desirable to test the new scheme on other weather systems with detailed in-cloud observations of microphysical quantities.

This paper is the first of a two-part study involving high-resolution simulations of the 13–14 December 2001 IMPROVE-2 case using the Milbrandt and Yau (MY05) scheme. The main objectives of this paper are to evaluate further the skill of the new BMS by comparing simulated precipitation and microphysical fields with observations and to identify potential deficiencies in the scheme’s current configuration. We will show that the MY05 scheme produces a precipitation simulation with a reasonably accurate spatial distribution for this case, though with a systematic overprediction in the quantities. It shall also be shown that there are some distinct biases in the simulated hydrometeor fields. This paper also serves to describe the control simulation used for sensitivity studies in Milbrandt et al. (2008, unpublished manuscript, hereinafter Part II), where the effects of the number of predicted moments are examined.

The paper is organized as follows. Section 2 describes the mesoscale model and the experimental setup. Section 3 provides a brief overview of the case and a comparison of the observed and simulated large-scale features. In section 4 we compare the results from the model with observations of reflectivity, vertical motion, accumulated precipitation, and microphysical fields. A discussion is given in section 5 and the concluding remarks appear in section 6.

2. Model description and experimental setup

The simulations were performed using the limited-area version of the Global Environmental Multiscale (GEM) model, which is based on the fully compressible Euler equations. This configuration of the GEM uses a Mercator projection and is capable of one-way self-nesting. The model dynamics are discussed in detail in (Côté et al. 1998). GEM has a comprehensive physics package (Mailhot et al. 1998), which includes a planetary boundary layer scheme based on turbulent kinetic energy (Benoit et al. 1989), implicit (explicit) vertical (horizontal) diffusion, and a detailed land surface scheme (Bélair et al. 2003a, b). The solar (Fouquart and Bonnel 1980) and infrared (Garand and Mailhot 1990) radiation package are fully interactive with the model clouds.

The experimental setup follows the guidelines described in Grabowski (2006; see links to the case 3 Web site) and utilized by G05a,b who simulated the same case using the MM5 with the Reisner-2 BMS [from Reisner et al. (1998), with modifications described in Thompson et al. (2004)]. Conforming to these guidelines allows for direct comparison of the performance of the MY05 scheme with that of the Reisner-2 scheme and others. The 36-km coarse-resolution simulation was initialized at 0000 UTC 13 December 2001 using the National Centers for Environmental Prediction Aviation Model (NCEP-AVN) analysis, which was modified to include additional surface and upper-air observations (see G05a). The same analyses provided the lateral boundary conditions every 6 h for the full 36-h simulation. The model was successively nested to 12-, 4-, and 1-km grids. Figure 1 depicts the model grids and the nesting times. The orography fields for the 4- and 1-km grids are shown in Fig. 2. Note that except for the 36-km grid, the GEM grids are slightly larger than the MM5 grids in G05a. All of the GEM simulations used 49 vertical levels.

In all runs, shallow convection was treated using the Kuo-transient scheme (Bélair et al. 2005). Subgrid-scale convection in the 36- and 12-km runs were modeled by the Kain and Fritsch (1993) convective parameterization, with settings appropriately specified for the particular grid spacing. The Sundqvist et al. (1989) condensation scheme was used to treat grid-scale clouds in the 36- and 12-km runs. The 4- and 1-km simulations used the triple-moment version of the MY05 scheme. Since comparisons shall be made to G05a,b, we note that their MM5 simulations used 32 vertical sigma levels, the Grell (1993) cumulus parameterization (for the 36- and 12-km grids only), the medium-range forecast planetary boundary layer scheme of Hong and Pan (1996), and the Reisner-2 BMS for grid-scale clouds on all grids.

The MY05 BMS includes six hydrometeor categories: two for the liquid phase—cloud (small, nonsedimenting droplets) and rain (sedimenting drops, including drizzle)—and four for the frozen phase—ice (pristine crystals), snow (large crystals and aggregates), graupel (medium-density rimed crystals), and hail (high-density graupel and frozen drops). The size distribution of each hydrometeor category is represented by a three-parameter complete gamma function.2 The full triple-moment version of the scheme includes prognostic equations for the mass content, total number concentration, and radar reflectivity for each hydrometeor size distribution (except cloud, for which reflectivity is not predicted). Approximately 50 distinct microphysical processes are parameterized. Details of the processes and their formulations are provided in MY05a,MY05b. In this paper, only results from the triple-moment control run are presented. Results from the single- and double-moment versions are discussed in Part II.

Preliminary model simulations of this case using the MY05 scheme produced significant quantities of hail (>0.4 g m−3 at the surface) in regions where the forcing in the model was insufficiently strong to produce hail in the conventional sense, being defined as large (>5 mm in diameter), high-density particles formed from heavy riming (e.g., Young 1993). The simulated hail had mean-mass diameters on the order of 1 mm or less and should thus be interpreted as ice pellets. Nevertheless, it is not aesthetically satisfactory for a microphysics scheme to produce hail under meteorological conditions that do not support its production, which may also lead to other associated simulations errors. Therefore, the conditions for the initiation of hail in the BMS have been made more stringent than are described in MY05b such that hail does not form in weakly forced conditions. The modified conditions are described in the appendix.

The terminology used in the atmospheric sciences to describe hydrometeors is a potential source of confusion between modelers and observationalists. For example, the term “ice” is often used in the descriptions of bulk schemes to refer specifically to the hydrometeor category for pristine crystals only, and excludes larger crystals and other types of ice-phase hydrometeors. To clearly distinguish between references to hydrometeor categories in bulk schemes and references to real particles, all names referring to categories in bulk schemes (real particles) are henceforth written in italics (nonitalics) in this paper.

3. Case overview and simulated large-scale features

To evaluate the BMS against observations, it is important that the observed large-scale flow was sufficiently well simulated since it provides forcing for the microphysics. If the simulated flow field is satisfactory, then differences in the simulated and the observed microphysical fields can be attributed to differences between the BMSs. The synoptic and mesoscale evolution of the case are described in detail in G05a and WSLH. Since the main emphasis in this paper is on the effect of the microphysics scheme, only a brief discussion of the case and the simulated large-scale features is presented.

a. Description of the case

The 13–14 December 2001 storm was characterized by a large-scale baroclinic system with precipitation enhanced by orographic forcing resulting from strong cross-barrier flow in the lower troposphere. On 13 December, the frontal system was located over the northern Pacific Ocean. By 0000 UTC 14 December, the surface cyclone had moved inland. The brightness temperature in infrared satellite imagery indicates a broad cloud shield near the west coast of North America (see left panels in Fig. 3). As the frontal system moved across the Oregon Cascades, several precipitation regimes were observed. Between approximately 2200 UTC 13 December and 0200 UTC 14 December, heavy precipitation occurred in the form of a broad rainband ahead of the cold front. The cross-barrier (southwesterly) flow at 1–2 km MSL was 30–40 m s−1 and resulted in several areas of localized upslope flow and the production of a significant amount of cloud liquid water available for riming. A radar bright band (not shown) was detected at approximately 1.5 km above mean sea level (MSL) throughout the heavy precipitation period.

b. Synoptic-scale features

The simulated large-scale features were similar to those from satellite observations and the NCEP-AVN analyses. Figure 3 depicts the GOES infrared imagery and the simulated outgoing longwave radiation (OLR) from the 12-km run at 1200 UTC 13 December 0000 UTC 14 December, and 1200 UTC 14 December. Although the units for brightness temperature and radiative flux are different, low OLR values (light shading regions in the right panels in Fig. 3) are strongly correlated with high brightness temperatures. The simulated and observed cloud patterns are thus similar in location and areal extent at these times.

The geopotential height, winds, and temperature fields at various levels at 0000 UTC 14 December from the 12-km simulation and the modified NCEP-AVN analysis are shown in Fig. 4. The large-scale mass and flow features are generally well-captured by the model. The trough in the simulation is slightly deeper (with a maximum geopotential height difference of 2.4 dam at 500 hPa) in the region just southwest of Vancouver Island. The fields near Oregon (upstream of the Cascade Mountains) are, however, close to the analysis. For example, the simulated 500-hPa geopotential height is within 1.2 dam, the 850-hPa temperature is ∼1°C colder, and the 850-hPa wind speeds along the Oregon coast are ∼27–30 m s−1. Note that differences in horizontal resolution between the analysis and the model result in differences in mass and flow fields at low levels near high terrain. When compared with the MM5 simulations in G05a,MY05b, the two models show similar large-scale features and similar difference fields between the model and the analysis. The same remark also applies at 1200 UTC 13 December and 14 December (not shown). This result is not surprising because the same initial and boundary conditions are used for both models. However, it is reassuring that differences between the models and the details of the configurations did not give rise to significantly different simulated large-scale fields (e.g., temperature, pressure, and humidity) that drive the microphysics schemes. Vertical velocity, which is a particularly important input parameter to the BMS, is discussed below.

c. Soundings

Special University of Washington (UW) radiosonde soundings near Cresswell, Oregon, provided measurements upstream of the high terrain in the IOP region (see inverted solid triangle in Fig. 5). Figure 6 depicts the observed and model (1 km) soundings at 2100, 0000, and 0400 UTC. The upstream temperature and moisture are important variables affecting the microphysics in the region. Before the period of heavy stratiform precipitation, the temperature and low-level humidity profiles upstream of the Cascades were well simulated (Fig. 6a), although the model sounding is too dry above 700 hPa. The model correctly simulates the strong southwesterly cross-barrier flow at 1–2 km above AMSL (approximately 800–700 hPa). During the stratiform precipitation period (0000 UTC), however, the low-level model winds are too weak (e.g., ∼33 m s−1 at 750 hPa, as compared with the observed 40 m s−1; see Fig. 6b). This underprediction is likely due to a slight phase difference in the arrival of the surface front; at 1 h earlier, the model wind speed at 750 hPa was 37 m s−1, closer to the observed wind speed at 0000 UTC. Also, the simulated profiles are slightly too warm and moist in the layer between approximately 750 and 475 hPa at 0000 UTC. There is fair agreement overall in the postfrontal (0400 UTC) sounding (Fig. 6c), though the model is too warm and moist between 750 and 550 hPa.

The upper-level (i.e., above 750 hPa) moisture biases in the model soundings are potentially problematic in evaluating the performance of the microphysics scheme. We remark, however, that by 1800 UTC, the BMS had already been activated in that region. Thus, the UW sounding comparison cannot be used to evaluate the upper-level precloud moisture supplied to the scheme. The simulated low-level moisture, on the other hand, agreed quite well with the observed soundings. Since much of the forcing for the precipitation for this case comes from an orographically induced pattern of upward motion (discussed below), the moisture in the lowest 1–2 km is an important moisture source. Nevertheless, the overpredicted upper-level moisture, which is consistent with an overprediction in the simulated snow mass contents (discussed below) introduces a complication in evaluating the performance of the BMS for this case due to the uncertainly in the degree to which the moisture supplied to the microphysics scheme was well simulated.

4. Comparison of mesoscale features

a. Radar reflectivity pattern

Figure 7 shows snapshots of observed reflectivity on plan position indicators (PPIs) at a 1.5° elevation angle from the NCAR S-Pol radar, located approximated 150 km to the south, and from the Portland, Oregon, radar, at a 0.5° elevation angle, located in the northern part of the IOP region (see Fig. 5), along with the total equivalent radar reflectivity Ze at 700 hPa from the MY05 scheme in the 4-km simulation over the same area as the Portland radar. Figures 8 and 9 depict, respectively, Ze at 700 hPa from the 4- and 1-km runs. The variable Ze is the sum of the equivalent reflectivities (Zex) from each hydrometeor category x (except for cloud) computed from
i1520-0493-136-10-3873-e1
where |K|2i/|K|2w = 0.176/0.930 = 0.189 is the ratio of the dielectric constants for ice and water (Smith 1984); (ρw/ρi)2 = (1000 kg m−3/917 kg m−3)2 = 1.189 is the square of the ratio of the densities of water and ice; cr = (π/6)ρw (where ρw is the density of water); cx is the coefficient of the mass–diameter relation (see MY05b for values); and Zx, a prognostic variable in the triple-moment scheme, is the sixth moment3 of the size distribution for category x. The times for comparison are 1800, 0000, and 0400 UTC, which correspond to the prefrontal, stratiform, and postfrontal periods respectively.

In general, the high-resolution mesoscale features of this storm at various stages, as observed by radar, were realistically simulated by the model. At 1800 UTC, the Portland radar indicates a broad region of stratiform precipitation to its north and along the coast (Fig. 7b). The early stages of the prefrontal showers along the Cascades are apparent in the S-Pol radar (Fig. 7a). Note that the distinct echo-free region to the northeast is a radar shadow, not a gap in the precipitation. The simulations (Figs. 7c, 8a and 9a) capture both the broad echo region along the coast as well as the showers along the mountains, the latter being more pronounced in the 1-km run.

The second row of panels in Figs. 7 –9 correspond to 0000 UTC, approximately midway through the period of heavy stratiform precipitation. The radar images clearly show the passage of the frontal rainband (Figs. 7d,e) and observed reflectivity values greater than 30 dBZ can be found over an approximately 100 × 40 km2 area to the south and east of the radar. There are also some small patches with reflectivity greater than 40 dBZ. The location and areal extent of the relatively high reflectivity region are well captured by the 4- and 1-km runs. The peak reflectivity values (>40 dBZ) are somewhat better simulated in the 1-km run. The back edge of the long reflectivity band over the ocean, south-west of the radar location in the 4-km simulation (Fig. 8b), also corresponds quite well to the back edge of the cloud pattern from the Geostationary Operational Environmental Satellite-Infrared (GOES-IR) image at that time (Fig. 3c).

The banded patterns of the postfrontal showers along the Cascades observed at 0400 UTC (Figs. 7g,h) were captured by both the 4- and 1-km runs (Figs. 7i, 8c and 9c). The narrow bands, oriented southwest-to-northeast, were reproduced particularly well with the 1-km grid, including the large radar reflectivity values (>45 dBZ), though the peak-simulated values in the 1-km run are too large.

b. Vertical motion

One of the most fundamental controlling parameters driving a microphysics scheme is vertical velocity (w). In the simulations, the strong cross-barrier flow across the Cascades set up a quasi-stationary mountain-wave pattern. Figure 10 depicts a 2-h time average (2300–0000 UTC, every 15 min) of vertical motion from the 1-km simulation along the vertical cross section indicated in Fig. 5. This pattern was quasistationary during the stratiform period. The instantaneous (0000 UTC) vertical motion along leg 2 of the P-3 flight, parallel to the foothills of the Cascades, is shown in Fig. 11. The patterns of upward and downward motion were strongly tied to regions of strong gradients in the orography and extend to the surface. In situ measurements of w were taken during the P-3 flight, based on vertical wind estimated from pressure sensors mounted on the aircraft, and are shown in Figs. 12 and 13 in G05a along with simulated vertical motion fields from the MM5 runs. GSM analyzed airborne dual-Doppler measurements and showed these estimates of w along leg 2 of the P-3 (see their Figs. 13 and 14). Note that the largest values of upward and downward motion were observed along leg 2, except for the strong downward lee wave immediately east of the mountain crest along leg 4.

The vertical motion pattern along leg 2 from the 1-km GEM simulation can be compared directly to the observed vertical motion as well as that from the 1.33-km MM5 simulation. The cross section from the GEM simulation (Fig. 11) is along approximately the same path as GSM’s Fig. 14 cross section depicting the dual-Doppler-derived w. The patterns of upward and downward motion are very similar in terms of their locations relative to the underlying orography. As discussed in GSM, the updrafts and downdrafts inferred from the dual-Doppler may be unrealistically shallow due to the w = 0 upper boundary condition placed on the retrieval, when in fact upward motion was measured at the echo tops at 5 km. Also, the magnitudes derived from radar are less than those from the in situ observations (Fig. 13 of GSM). The GEM-simulated magnitudes of w at 725 hPa (2.5 km MSL) along leg 2, shown in Fig. 11a, are reasonable but slightly smaller than those from the in situ observations (e.g., Fig. 14 in G05a). For example, the peak values range from +1.6 to −1.6 m s−1 in the simulation compared to approximately +3 to −3 m s−1 from the P-3 (estimated from G05a,MY05b’s figures). However, the values from the GEM simulation were very similar to G05a,MY05b’s 1.33-km simulation values. Also, the downward wind velocity on the immediate lee of the main Cascade crest was estimated to be ∼−3.0 m s−1 at 4 km from the in situ measurements (along leg 4, see Fig. 13 in G05a). At the similar point in time and space, the 1-km GEM simulation had vertical motion of ∼−3.5 m s−1. The 1.33-km MM5 simulation also closely captured the strong downdraft magnitude (Fig. 12 in G05a). Thus, we may infer that the vertical motion in the 1-km GEM simulation was realistic, though somewhat lower than the observed, and it was very close to that of the G05a,MY05b simulation.

c. Precipitation

The overall QPF of the high-resolution simulations were compared with the observed 18-h (1400–0800 UTC) cumulative precipitation from rain gauges. The accumulated values were available for 145 gauges on the 4-km grid. A subset of this (65 gauges) lies within the 1-km grid. The comparison excludes the first 2 h of the 4-km simulation and includes the entire 1-km simulation. Errors due to model spinup during the early part of the simulations are expected to be small since hydrometeor and vertical motion fields from the corresponding driving grids were used to initialize the 4- and 1-km runs. Figure 12 depicts simulated and observed rain gauge values. Figure 13 shows scatterplots between the observations and simulations for the observation points in Fig. 12. As shown, the overall spatial distribution of precipitation for the 4-km run appears to compare favorably to the observations. There is a region of relatively high values (>40 mm) along the coast, lower values in the Willamette valley, higher values along the windward side of the Cascades (>50 mm), and low values (<20 mm) throughout the east leeside region. Overall, there is a systematic bias toward overprediction of the precipitation quantities, as is evident in the scatterplots.

For the 1-km simulation, the overall QPF is qualitatively and quantitatively similar to that of the 4-km run. The regions of relatively higher and lower precipitation totals were similar and the scatterplot indicates a similar trend of a systematic moist bias compared to the observations. There is more detail in the 1-km precipitation pattern on account of the higher-resolution orography. However, due probably to the relatively small number of rain gauges in these regions of complex terrain, the agreement between the simulated and observed values did not change notably.

It is of interest to compare the QPFs from the GEM simulations with those from G05a’s MM5 simulations. Both the 4- and 1.33-km MM5 simulations had a general overprediction of precipitation amounts, particularly along the Cascades and on the lee side of the crest (e.g., Figs. 15 and 17 in G05a). For direct comparison with the GEM results, we have constructed a scatterplot of gauge-versus-model precipitation totals for the 1.33-km MM5 run using the data reported in G05a’s Figs. 16 and 17a. This is shown in our Fig. 14 along with the corresponding results from the 1-km GEM run (for all of the observations within the 1-km GEM domain). The two models exhibit very similar biases toward overprediction and have a similar spread in the data. One notable difference in the precipitation patterns, however, is that the GEM simulations (Fig. 12) do not suffer from the pronounced overprediction on the immediate lee side of the Cascade crest that appears in the MM5 runs (see also C05’s Fig. 1 and GSM’s Fig. 16 for further illustration of this leeside overprediction). Despite the fact that different mesoscale models with somewhat different configurations were used, the fact that both models had the same initial and boundary conditions and produced similar forcing (e.g., vertical velocity), it is likely that the differences in the QPFs can be attributed primarily to the different microphysics schemes. This becomes more evident in Part II, which presents simulations using a different BMS in GEM that result in a similar leeside overprediction problem as in the MM5 runs with the Reisner-2 scheme.

d. Hydrometeor fields

To further evaluate the MY05 BMS, we compare the simulated hydrometeor fields with in situ measurements during the period of intense stratiform precipitation. During a 2-h period (2300–0100 UTC), two instrumented aircraft made continuous measurements of in-cloud hydrometeors. The National Oceanographic and Atmospheric Administration (NOAA) P-3 flew a lawn-mower pattern with five north–south legs approximately 130 km long and spaced 40 km apart at constant altitudes (see Fig. 5). The University of Washington’s Convair-580 flew back and forth along a southwest to northeast path across the Willamette valley and the windward side of the Cascades. Both aircraft recorded imagery of liquid and solid cloud and precipitation particles as well as size spectra and number concentrations (see S03 and WSLH for details).

There are some intrinsic difficulties in comparing in situ hydrometeor measurements and the corresponding predicted fields from a BMS. First, there is a mismatch in scale between the size of the probe and the size of a model grid box. Second, despite the quasi-steady orographic forcing, the considerable variability in the hydrometeor fields in time and space, both horizontally and vertically, makes comparison difficult. Thus, even if comparisons were made at the same observation and model points, small phase errors in predicting the arrival of the surface front or magnitude errors in the strength of the low-level cross-barrier winds could affect the goodness of comparison in the fields of hydrometeors which may not be related to the BMS. We will therefore attempt first a qualitative evaluation of the comparison between the simulated and in situ–measured microphysical fields.

1) Overview of observed and modeled fields

The aircraft observations between 2300 and 0100 UTC were reported in detail in WSLH and G05b. The Convair-580 observed only single ice crystals for altitudes higher than 4.5 km MSL. They consist mainly of assemblages of sectors and side planes as well as some hexagonal plates. Dendrites were observed between 3 and 4 km MSL and columns and aggregates between 2 and 4 km MSL. Along the windward side of the Cascades, pockets of cloud liquid water (CLW) were observed as high as 4 km and graupel was observed between 3 and 4 km. The observations of graupel by the aircraft were supported by the NCAR polarimetric radar measurements, which indicated notable quantities of graupel and/or heavily rimed aggregates over the melting layer along the windward slopes and the crest (Houze and Medina 2005). Below the melting layer (approximately 2 km MSL), rain was observed by the S-POL radar all along the windward slope.

The in situ microphysical observations are summarized in Fig. 15, which is a reproduction of G05b’s Fig. 4c. G05b’s Figs. 4a, b also show the hydrometeor mass fields from the MM5 simulations. The vertical cross section is taken along a portion of the arrow indicated in Fig. 5. Its location was recommended by the working group for this case during the WMO Cloud Modeling Workshop (in 2004) since it runs along the flight path of the Convair-580 and through the north–south legs of the P-3 flight. Figure 16 shows vertical cross sections of time-averaged (2300–0100 UTC) hydrometeor mass contents from the 1-km GEM simulation along the full arrow indicated in Fig. 5. Note that the area of the red square in Fig. 15 is depicted by the gray rectangle in Fig. 16c to facilitate comparison. Figure 17 shows the mean-mass diameter Dx for each hydrometeor category x, computed by
i1520-0493-136-10-3873-e2
where ρ is the air density, cx and dx are parameters for the mass–diameter relations (see MY05a for specific values), qx is the mixing ratio, and NTx is the total number concentration. The variables qx and NTx are explicitly predicted by the triple-moment scheme.

As shown, the overall spatial distribution of the simulated hydrometeor mass fields is fairly consistent with the observations. Pristine ice was present aloft, mostly between 200 and 300 hPa with some at approximately 650 hPa, and was located mainly near the coast (Fig. 16b). There was a deep snow layer extending from the 0°C isotherm—which ranged from 725 to 800 hPa—to 275 hPa, with most of the mass concentrated between approximately 775 and 500 hPa (Fig. 16d). There were pockets of cloud water extending above the 0°C isotherm to 650 hPa located near local orographic peaks (Fig. 16a), collocated with graupel between 825 and 650 hPa (Fig. 16c) and higher along the coast. Rain was present along much of the cross section below the 0°C isotherm (see Fig. 16d) with the largest quantities at the surface located along the coastal mountains and the valley as well as the windward side of the Cascades (Fig. 16b). Note that only small quantities of rain were present on the lee side of the crest. Only trace quantities of hail (not shown), or ice pellets, were present near the mountain crests.

2) Cloud liquid water

The measurements of the CLW taken by the P-3 along its north–south legs at various altitudes are described in detail in G05b and WSLH. The observed mean and peak CLW values along each leg are summarized in Table 1 along with the corresponding values of cloud mass from the 1-km simulation. The peak values are estimated from the time series of CLW in Fig. 4 of WSLH. The model values are taken from the instantaneous, not time-averaged, cloud water fields at the nearest 15 min to the midpoint times of each of the 30-min flight legs along the approximate paths of the legs.

At the lower levels (legs 1 and 2), the peak values in the model are approximately two-thirds of the observed values while the leg-averaged values are notably lower, particularly for leg 2, which was flown 500 m higher. For the other legs, which were all at higher altitudes, the simulated CLW was nearly 0, even for leg 3 which had an observed mean CLW of 0.20 g m−3. Thus, while the low-level values were reasonable, the scheme clearly exhibited a bias toward too-small values of Qc at higher elevations as is evident by the underestimation of the vertical extent of the cloud pockets on the windward side of the Cascades during the stratiform period (Fig. 16a). The underestimation of cloud water at the lower levels is consistent with the simulated vertical motion being too small at those levels. Even during the postfrontal convection, CLW values as high as 0.40 g m−3 were observed by the Convair-580 during its the second flight (at 0526 UTC) near the Cascade crest while the model simulated Qc values of 0.35 g m−3 at that level (900 hPa) at 0500 UTC.

3) Ice-particle mass

The mass concentrations from the in situ ice crystal data obtained from the two aircraft were determined by the following method, described in detail in WSLH. Each of the images from the 2D probes was visually inspected and subjectively classified according to the Magono and Lee (1966) classification scheme. Using the particle diameters estimated from the images and assumed mass–diameter relations for the given particle types, the mass of each particle was determined and summed appropriately to compute mass concentrations.

The Convair-580 flew four legs at constant altitudes between 4 and 6 km AMSL during its first flight (see our Fig. 5 for the flight path and Fig. 14 in WSLH for locations of specific legs). The observed average values, reported in WSLH, are listed in Table 2 along with average and peak values of snow from the 1-km simulations for the corresponding locations. The instantaneous model values are at the nearest 15 min in time to the flight legs (obtained from pilot reports). There is a distinct overprediction of snow mass at upper levels—by up to a factor of 7 for leg-average values—during the stratiform period. The time-averaged snow field from the model (Fig. 16d) also indicates a general overprediction at upper levels. Note that snow was the only ice-phase hydrometeor present at those locations in the model.

During its return flight, the P-3 made a short leg on the lee of the Cascade crest followed by two cross-mountain legs on the windward side (see our Fig. 5 for P-3 track and Fig. 15 in WSLH for locations of specific legs). The ice-particle mass concentrations measured by the P-3 are reported in WSLH; the average and peak values are summarized in Table 3 along with the model values. The mass contents for the model are the sum of the snow and graupel fields, but only in the second windward leg was the graupel mass non 0. For this particular leg, the average (peak) model values were 0.02 (0.06 g m−3) for graupel and 1.28 (1.77 g m−3) for snow. The ice and hail categories contributed no mass at the levels of the flight legs. Therefore, as indicated in Table 3, there is also a distinct overprediction in the snow mass content by the model at lower levels as well.

4) Particle sizes

The aircraft observations of snow particle size distributions show a clear trend of increasing sizes with decreasing altitude. As shown in WSLH and G05b, higher concentrations of larger particles were found at lower levels. Moreover, when the size distributions are approximated by inverse-exponential functions, the slopes of the distributions decrease with altitude, which implies that the mean diameters increase with decreasing altitude. The vertical cross section of the mean-mass snow diameter (Ds) from the simulation (Fig. 16d) indicates that this trend of increasing mean size with decreasing altitude is also well reproduced by the model. Furthermore, the values of Ds appear to be reasonable compared to the in situ observations. For example, G05b show observed snow size distributions (see their Fig. 8) for the first three legs of the Convair flight and reported corresponding size distribution parameters for best fit (inverse exponential) curves. Based on the slope parameters for these size distributions (Table 2 in G05b), Ds is computed4 to be 0.68, 0.79, and 0.71 mm at altitudes of 6.0, 5.4, and 4.9 km, respectively. WSLH showed samples of the images from the 2D probes of ice particles collected along the flight legs (see Fig. 13 of WSLH) of both aircrafts. Estimating from the images, there were particle sizes ranging from approximately 0.1 to 0.6 mm at 5.4 km, 0.4 to 1.5 mm (with some as large as 5 mm) at 3.3 km, and 0.2 to 4 mm from 3.3 to 2.0 km. The corresponding Ds values from the 1-km simulation at these levels and locations were ∼0.8 mm (at 450 hPa, ∼6.0 km), ∼1.1 mm (at 500 hPa, ∼5.5 km), ∼1.3 mm (at 550 hPa, ∼5.0 km), ∼1.0–1.6 mm (at 650 hPa, ∼3.5 km), and 1.7–2.7 mm (at 775 hPa, ∼2.0 km). This is not intended to be a highly quantitative comparison—the observed particle sizes mentioned are estimates from best-fit inverse-exponential curves to size distributions and from size estimates of the sample of images shown in WSLH—but it indicates that the mean snow sizes produced by the BMS are realistic and that the observed trend of decreasing snow sizes with height is correctly captured. Note that size sorting in the simulation also produced a profile of increasing mean diameters of ice with decreasing altitude (Fig. 17b). This increase of ice crystal sizes below the cloud top has been observed for cirrus clouds (Heymsfield and Iaquinta 2000).

Sizes of other types of particles for this case, if observed, have not been documented. However, the mean-mass diameter of the hydrometeor categories included in the MY05 BMS, for which vertical cross sections are shown in Fig. 17, are realistic in comparison with observed sizes for similar weather systems reported in the literature. For cloud droplets, the Dc values ranged from 5 to 25 μm, which is comparable to observations of maritime clouds (e.g., Telford and Wagner 1981). Cloud droplets from continental air masses tend to have higher number concentrations, due to the higher concentrations of aerosols capable of serving as cloud condensation nuclei, and thus smaller mean diameters (Pruppacher and Klett 1997). Mean-mass diameters of rain, Dr, range from 0.2 to 1.3 mm. For stratiform rain with a Marshall–Palmer (1948) drop size distribution, rainfall rates of 1 to 8 mm h−1 (which is the range of simulated precipitation rates during this period over the Cascades) correspond to mean drop diameters of 0.2 to 0.4 mm, respectively. Ice (pristine crystals) have Di values of 15–40 μm at 250 hPa (approximately −55°C), which is reasonable for cirrus clouds (e.g., Ström et al. 1997). The values of Dg (for graupel), ranged from 0.3 to 2.7 mm; within the same order of the Locatelli and Hobbs (1974) measured graupel diameters of between 0.5 and 3.0 mm for winter storms in the Cascade Mountains.

5. Discussion

The MY05 simulations exhibited a distinct bias toward an underprediction of the vertical extent of the pockets of cloud water along the valley and windward side of the Cascades and an excessive mass concentration of snow. Although the upstream soundings generally indicate that the quantity of low-level moisture being advected toward the Cascades was correct and that the orographic forcing due to the low-level cross-barrier winds was reasonable, the low-level vertical velocity fields in the simulations were too weak compared to in situ observations. This may partially account for the underpredicted cloud water. The simulations also overpredicted the amount of upper-level moisture, which is consistent with the excessive snow. Despite these shortcomings in the forcing from the model, the bias toward too much snow and too little cloud in the GEM simulations may also be partly due to deficiencies in the current configuration of the MY05 scheme.

Previous modeling studies have indicated considerable sensitivity of the growth rates of snow and the resulting mass fields of snow, graupel, and cloud water (and ultimately on the precipitation) to the details in the treatment of the snow category in a BMS. For example, Reisner et al. (1998), using a BMS with an inverse-exponential size distribution for snow, found that the treatment of the intercept parameter (N0s) significantly affected the depositional growth rate and the resulting snow and cloud water fields for the simulation of a winter storm with supercooled cloud water present. Similarly, Colle and Zeng (2004) found that the simulation of orographic precipitation was sensitive to the snow size distribution and fall velocity parameters. For the MM5 simulations of the 13–14 December 2001 case, C05 also found that different treatments of N0s as well as values of the fall speed parameters and the snow-to-graupel conversion threshold notably affected the simulated fields of snow, cloud water, and precipitation. It was shown through sensitivity tests in Part II that by modifying the snow size distribution and fall velocity parameters in the MY05 scheme the snow-cloud bias can be altered (improved) dramatically due to changes in the depositional growth rate and/or the residence time of the snow in the growth zone.

It is therefore tempting to suspect that improving the overall simulation for this case may be a simple matter of appropriately tuning the snow distribution parameters. However, the set of simultaneously occurring microphysical processes in a BMS are complex and the processes interact nonlinearly. Indeed, C05 found that for their ice-phase sensitivity experiments, no run resulted in dramatic improvements to the simulated precipitation, all of which suffered from the problem of overprediction on the lee side of the crest. The MY05 simulations, on the other hand, did not exhibit the problem of lee-side precipitation overprediction. The reasons for this are examined in Part II. It will also be shown that the overall surface precipitation quantities for this case are relatively insensitive to variations in the scheme configuration that lead to notable changes in the snow and cloud water mass fields aloft.

Based on observations in midlatitude stratiform clouds, Field et al. (2005) showed that the size distributions of snow (ice crystals larger than 100 μm in diameter) can be accurately described using two moments. Furthermore, they showed that one of the moments can be estimated from a single (reference) moment based on temperature. The implication of this is that a BMS may require only one prognostic moment of the snow distribution in order to describe its evolution. This concept has recently been applied to the representation of the snow category in BMS now generally referred to as the Thompson scheme (Thompson et al. 2004), which originated as the Reisner-2 scheme (G. Thompson 2007, personal communication). This temperature-dependent size distribution has the advantage over other single-moment approaches—where the N0s parameter (assuming an inverse-exponential distribution) is either constant or diagnosed from the mass mixing ratio—that the effect of increasing mean particle diameters with decreasing altitude (i.e., increasing temperature) is partially represented.5 However, this type of single-moment parameterization is based on a representation of an average of many size distributions and does not incorporate the different physics between processes such as diffusional growth and aggregation, where the evolution of the different moments are not linked in the same manner. The lack of monotonic correlation between the size distribution parameters of observed snow spectra that results from the different physical processes is nicely illustrated in the observational study of Lo and Passarelli (1982).

Despite the overprediction of the snow mass content in the simulations with the MY05 scheme, it is not simply fortuitous that the model simulated realistic values of the mean-mass diameters nor is it a coincidence that the observed tendency toward increasing snow sizes with decreasing altitude was well simulated. Rather, this indicates that the ratio of mass-to-number density was treated realistically by the BMS [see Eq. (2)]. Since a multimoment scheme can predict these quantities independently, the ratio of mass-to-number—and thus the mean-particle diameter—evolves more realistically for growth processes such as riming and deposition, where mass increases but the total number remains constant. It also means that the process of snow aggregation, where the number concentration decreases and the total mass content remains constant, thereby increasing the mean-particle size, can be explicitly incorporated. Further, gravitational size sorting, which results in a redistribution of mean-particle sizes in the vertical, can be well modeled in a BMS that is at least double moment (see MY05a). Nevertheless, the distinct overprediction of snow mass in the simulations presented cannot be overlooked. It is possible that the single-moment approach proposed by Field et al. (2005), for example, may be appropriate for the bulk parameterization of snow, despite the advantages of the multimoment approach discussed above. The effects of the number of prognostic moments for this case and the overprediction of snow mass by the MY05 scheme are addressed in Part II.

6. Conclusions

The 13–14 December 2001 IMPROVE-2 case of orographically enhanced precipitation over the Oregon Cascades was simulated using a high-resolution mesoscale model with the triple-moment version of the MY05 BMS. Comparisons of the simulation were made with observations from radar, rain gauges, and in situ aircraft measurements of cloud microphysical fields and vertical air motion. The large-scale meteorological features of the simulation compared favorably to satellite imagery, the NCEP-AVN gridded analyses, and special soundings upstream of the mountains. The mountain-wave pattern of vertical velocity anchored to the Cascades was correctly simulated by the model, though the magnitudes were slightly lower than those from the in situ aircraft measurements. Although the forcing in the model may have been slightly too weak, it appears that errors in the simulated hydrometeor fields—too much snow mass and too little cloud mass—can also be attributed in part to the BMS.

The MY05 scheme had previously been tested only for a case of strong convection and never against in situ microphysical measurements. This case study has proven very useful for evaluating the new BMS. Several positive aspects of the scheme have been demonstrated including its ability to produce a reasonable quantitative precipitation simulation for a case with orographic forcing and a complex set of microphysical processes. Certain deficiencies in the current version of the scheme have also been revealed. In view of these results, it appears that certain restrictions on the evolution of the snow size spectrum (i.e., the allowable range of the distribution prognostic parameters) may in fact be required to better model the ice-phase and cloud water fields. Other assumptions about the snow category, such as the assumed crystal habit that affect the diffusional growth rate, may also be related to the overabundance of snow mass in the simulations. Future studies on the improvement of the MY05 scheme will address these and other aspects.

The conspicuous difference in the precipitation from our simulations of this case relative to those of G05a,MY05b merits further investigation. Despite the fact that different mesoscale models were used, it appears very likely that the large differences in the simulated fields are strongly linked to the use of different BMSs. It is not obvious a priori whether the differences in the QPFs between the triple-moment version of the MY05 scheme and the single-moment Reisner-2 scheme are due to the number of predicted moments or to other differences in the formulations of the microphysical processes. In Part II of this study, the effects of changing the number of predicted moments for the simulation of this case and the bias of excessive snow quantities in the MY05 scheme are examined.

Acknowledgments

We thank Dr. Mathew Garvert and Dr. Socorro Medina for providing observational data. We also thank the two anonymous reviewers for their careful reviews and helpful comments. This research was funded by the Modelling of Clouds and Climate (MOC2) project through the Canadian Foundation for Climate and Atmospheric Sciences (CFCAS) and the Natural Science and Engineering Research Council of Canada (NSERC).

REFERENCES

  • Bélair, S., L-P. Crevier, J. Mailhot, B. Bilodeau, and Y. Delage, 2003a: Operational implementation of the ISBA land surface scheme in the Canadian regional weather forecast model. Part I: Warm season results. J. Hydrometeor., 4 , 352370.

    • Search Google Scholar
    • Export Citation
  • Bélair, S., R. Brown, J. Mailhot, B. Bilodeau, and L-P. Crevier, 2003b: Operational implementation of the ISBA land surface scheme in the Canadian regional weather forecast model. Part II: Cold season results. J. Hydrometeor., 4 , 371386.

    • Search Google Scholar
    • Export Citation
  • Bélair, S., J. Mailhot, C. Girard, and P. Vaillancourt, 2005: Boundary layer and shallow cumulus clouds in a medium-range forecast of a large-scale weather system. Mon. Wea. Rev., 133 , 19381960.

    • Search Google Scholar
    • Export Citation
  • Benoit, R., J. Côté, and J. Mailhot, 1989: Inclusion of a TKE boundary layer parameterization in the Canadian regional finite-element model. Mon. Wea. Rev., 117 , 17261750.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., and Y. Zeng, 2004: Bulk microphysical sensitivities within the MM5 for orographic precipitation. Part I: The Sierra 1986 event. Mon. Wea. Rev., 132 , 27802801.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., M. F. Garvert, J. B. Wolfe, C. F. Mass, and C. Woods, 2005: The 13–14 December 2001 IMPROVE-2 event. Part III: Simulated microphysical budgets and sensitivity studies. J. Atmos. Sci., 62 , 35353558.

    • Search Google Scholar
    • Export Citation
  • Côté, J., S. Gravel, A. Méthot, A. Patoine, M. Roch, and A. Staniforth, 1998: The operation CMC-MRB Global Environmental Multiscale (GEM) model. Part I: Design considerations and formulation. Mon. Wea. Rev., 126 , 13731395.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., G. J. Tripoli, R. M. Rauber, and E. A. Mulvhill, 1986: Numerical simulation of the effects of varying ice crystal nucleation rates and aggregation processes on orographic snowfall. J. Climate Appl. Meteor., 25 , 16581680.

    • Search Google Scholar
    • Export Citation
  • Ferrier, B. S., 1994: A double-moment multiple-phase four-class bulk ice scheme. Part I: Description. J. Atmos. Sci., 51 , 249280.

  • Field, P. R., R. J. Hogan, P. R. A. Brown, A. J. Illingworth, T. W. Choularton, and R. J. Cotton, 2005: Parameterization of ice-particle size distributions for mid-latitude stratiform clouds. Quart. J. Roy. Meteor. Soc., 131 , 15071513.

    • Search Google Scholar
    • Export Citation
  • Fouquart, Y., and B. Bonnel, 1980: Computations of solar heating of the earth’s atmosphere: A new parameterization. Contrib. Atmos. Phys., 53 , 3562.

    • Search Google Scholar
    • Export Citation
  • Garand, L., and J. Mailhot, 1990: The influences of infrared radiation on numerical weather forecasts. Preprints. Seventh Conf. on Atmospheric Radiation, San Francisco, CA, Amer. Meteor. Soc., J146–J151.

    • Search Google Scholar
    • Export Citation
  • Garvert, M. F., B. A. Colle, and C. F. Mass, 2005a: The 13–14 December 2001 IMPROVE-2 event. Part I: Evolution and comparison with a mesoscale model simulation. J. Atmos. Sci., 62 , 34723492.

    • Search Google Scholar
    • Export Citation
  • Garvert, M. F., C. P. Woods, B. A. Colle, C. F. Mass, P. V. Hobbs, M. T. Stoelinga, and J. B. Wolfe, 2005b: The 13–14 December 2001 IMPROVE-2 event. Part II: Comparisons of MM5 model simulations of clouds and precipitation with observations. J. Atmos. Sci., 62 , 35203534.

    • Search Google Scholar
    • Export Citation
  • Garvert, M. F., B. Smull, and C. Mass, 2007: Multiscale mountain waves influencing a major orographic precipitation event. J. Atmos. Sci., 64 , 711737.

    • Search Google Scholar
    • Export Citation
  • Geresdi, I., 1998: Idealized simulation of the Colorado hailstorm case: Comparison of bulk and detailed microphysics. Atmos. Res., 45 , 237252.

    • Search Google Scholar
    • Export Citation
  • Grabowski, W. W., 2006: Sixth WMO International Cloud Modeling Workshop. Bull. Amer. Meteor. Soc., 87 , 639642.

  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev., 121 , 764787.

  • Heymsfield, A. J., and J. Iaquinta, 2000: Cirrus crystal terminal velocities. J. Atmos. Sci., 57 , 916938.

  • Hong, S-Y., and H. L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-rage forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Houze Jr., R. A., and S. Medina, 2005: Turbulence as a mechanism for orographic precipitation enhancement. J. Atmos. Sci., 62 , 35993623.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 165–177.

    • Search Google Scholar
    • Export Citation
  • Kogan, Y., 1991: The simulation of a convective cloud in a 3D model with explicit microphysics. Part I: Model description and sensitivity experiments. J. Atmos. Sci., 48 , 11601189.

    • Search Google Scholar
    • Export Citation
  • Kong, F., and M. K. Yau, 1997: An explicit approach to microphysics in MC2. Atmos.–Ocean, 35 , 257291.

  • Lin, Y. L., R. Farley, and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor., 22 , 10651092.

    • Search Google Scholar
    • Export Citation
  • Lo, K. K., and R. E. Passarelli, 1982: The growth of snow in winter storms: An airborne observational study. J. Atmos. Sci., 39 , 697706.

    • Search Google Scholar
    • Export Citation
  • Locatelli, J. D., and P. V. Hobbs, 1974: Fall speeds and masses of solid precipitation particles. J. Geophys. Res., 79 , 21852197.

  • Magono, C., and C. W. Lee, 1966: Meteorological classification of natural snow crystals. J. Faculty Sci. Hokkaido Univ., 2 , 321335.

  • Mailhot, J., and Coauthors, 1998: Scientific description of RPN physics library—Version 3.6. Recherche en prévision numérique, Canadian Meteorological Centre, 188 pp.

  • Marshall, J. S., and W. M. K. Palmer, 1948: The distribution of raindrops with size. J. Meteor., 5 , 165166.

  • Meyers, M. P., R. L. Walko, J. Y. Harrington, and W. R. Cotton, 1997: New RAMS cloud microphysics. Part II: The two-moment scheme. Atmos. Res., 45 , 339.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., and M. K. Yau, 2005a: A multimoment bulk microphysics parameterization. Part I: Analysis of the role of the spectral shape parameter. J. Atmos. Sci., 62 , 30513064.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., and M. K. Yau, 2005b: A multimoment bulk microphysics parameterization. Part II: A proposed three-moment closure and scheme description. J. Atmos. Sci., 62 , 30653081.

    • Search Google Scholar
    • Export Citation
  • Milbrandt, J. A., and M. K. Yau, 2006: A multimoment bulk microphysics parameterization. Part III: Control simulation of a hailstorm. J. Atmos. Sci., 63 , 31143136.

    • Search Google Scholar
    • Export Citation
  • Morrison, H., J. A. Curry, and V. I. Khvorostyanov, 2005: A new double-moment microphysics parameterization for application in cloud and climate models. J. Atmos. Sci., 62 , 16651677.

    • Search Google Scholar
    • Export Citation
  • Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. Kluwer Academic, 914 pp.

  • Reisin, T., Z. Levin, and S. Tzivion, 1996: Rain production in convective clouds as simulated in an axisymmetric model with detailed microphysics. Part I: Description of the model. J. Atmos. Sci., 53 , 497520.

    • Search Google Scholar
    • Export Citation
  • Reisner, J., R. M. Rasmussen, and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124 , 10711107.

    • Search Google Scholar
    • Export Citation
  • Seifert, A., and K. D. Beheng, 2001: A double-moment parameterization for simulating autoconversion, accretion and self-collection. Atmos. Res., 59–60 , 265292.

    • Search Google Scholar
    • Export Citation
  • Smith, P. L., 1984: Equivalent radar reflectivity factors for snow and ice particles. J. Climate Appl. Meteor., 23 , 12581260.

  • Stoelinga, M., and Coauthors, 2003: Improvement of microphysical parameterization through observational verification experiment. Bull. Amer. Meteor. Soc., 84 , 18071826.

    • Search Google Scholar
    • Export Citation
  • Ström, J., B. Strauss, F. Shröder, T. Anderson, J. Heintzenberg, and P. Wendling, 1997: In situ observations of the microphysical properties of young cirrus clouds. J. Atmos. Sci., 54 , 25422553.

    • Search Google Scholar
    • Export Citation
  • Sundqvist, H., E. Berge, and J. E. Kristjánsson, 1989: Condensation and cloud parameterization studies with a mesoscale numerical weather prediction model. Mon. Wea. Rev., 117 , 16411657.

    • Search Google Scholar
    • Export Citation
  • Szyrmer, W., S. Laroche, and I. Zawadzki, 2005: A microphysical bulk formulation based on scaling normalization of the particle size distribution. Part I: Description. J. Atmos. Sci., 62 , 42064221.

    • Search Google Scholar
    • Export Citation
  • Telford, J. W., and P. B. Wagner, 1981: Observations of condensation growth determined by entity type mixing. Pure Appl. Geophys., 119 , 934965.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., R. R. Rasmussen, and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132 , 519542.

    • Search Google Scholar
    • Export Citation
  • Walko, R. L., W. R. Cotton, M. P. Meyers, and J. Y. Harrington, 1995: New RAMS cloud microphysics. Part I: The one-moment scheme. Atmos. Res., 38 , 2962.

    • Search Google Scholar
    • Export Citation
  • Woods, C. P., M. T. Stoelinga, J. D. Locatelli, and P. V. Hobbs, 2005: Microphysical processes and synergistic interaction between frontal and orographic forcing of precipitation during the 13 December 2001 IMPROVE-2 event over the Oregon Cascades. J. Atmos. Sci., 62 , 34933519.

    • Search Google Scholar
    • Export Citation
  • Young, K. C., 1993: Microphysical Processes in Clouds. Oxford University Press, 427 pp.

APPENDIX

Modified Hail-Forming Conditions

For the simulations described in this paper and in Part II, the following modifications have been made to the MY05 scheme as described in MY05a,MY05b in order to suppress the initiation of hail for conditions that are not conducive in nature for the production of real hail. The new conditions were tested such that hail production is inhibited for the case described in this paper but that hail is still allowed to form for cases of strong forcing (i.e., strong vertical motion and/or large drops in cold conditions).

Conversion from graupel to hail

The mean-mass graupel diameter Dg must now exceed 2.5 mm and the vertical air velocity must be greater than 3 m s−1 (upward) for conversion of graupel to hail to be considered. For the single-moment configuration the Dg condition is equivalent to the requirement that the graupel mass content exceed 1.8 g m−3. If these conditions are met, then the minimum hail diameter for wet growth Dh0, given by the Shumann–Ludlam limit as a function of ambient temperature, pressure, and liquid water content, is computed [see MY05b their Eq. (48)]. If the ratio Dg/Dh0 is greater than 0.05 then conversion of graupel to hail will occur.

Three-component freezing

Collisional freezing of rain with ice, snow, or graupel to form hail can now only occur if the mean-mass diameter of rain exceeds 1 mm for the double- and triple-moment configurations of the scheme. In the single-moment configuration, the rain mixing ratio must exceed 0.3 g kg−1. These conditions prohibit hail production from rain where only trace amounts of rain are present.

Probabilistic freezing of rain

For the spontaneous freezing of rain to hail, the ambient air temperature must now be colder than −10°C. Also, in the single-moment configuration the rain mixing ratio must exceed 0.1 g kg−1.

Fig. 1.
Fig. 1.

Nested GEM model grids. Inset indicates nesting times for each grid.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 2.
Fig. 2.

Model orography (shading) for (a) 4-km grid (subdomain) and (b) full 1-km grid. Contours denote elevations of 1500, 2000, and 2500 m. Cross hairs depict model locations of rain gauges.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 3.
Fig. 3.

Infrared images from (left) GOES and (right) OLR from the 12-km GEM simulation at (a), (b) 1200 UTC 13 Dec 2001; (c), (d) 0000 UTC 14 Dec 2001; and (e), (f) 1200 UTC 14 Dec 2001.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 4.
Fig. 4.

Synoptic fields from the 12-km GEM simulation (solid) and modified NCEP-AVN analysis (dashed) at 0000 UTC 14 Dec 2001. (a) The 300-hPa isotachs (m s−1) with shading for simulation values greater than 40 m s−1, (b) 500-hPa geopotential height (dam), (c) 850-hPa temperature (°C) and winds, (d) 850-hPa geopotential height, (e) 1000-hPa temperature and winds, and (f) 1000-hPa geopotential height. Simulation (analysis) wind vectors (m s−1) are black (gray).

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 5.
Fig. 5.

Map of the IMPROVE-2 study area. Shading denotes the model orography (4-km grid), with an interval of 500 m. Heavy contours denote elevations of 1500 and 2000 m.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 6.
Fig. 6.

Soundings from 1-km simulation (solid) and UW radiosonde (dashed) at (a) 2100, (b) 0000, and (c) 0400 UTC 14 Dec 2001. See Fig. 5 for location of UW radiosonde site. Full (half) barbs denote wind speeds of 10 m s−1 (5 m s−1).

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 7.
Fig. 7.

Reflectivity from the (left) NCAR S-Pol radar (1.5° PPI), (middle) the Portland radar (0.5° PPI), and (right) the equivalent reflectivity from 4-km simulation (700 hPa) at (top) 1800 UTC 13 Dec, (middle) 0000 UTC 14 Dec, and (bottom) 1400 UTC 14 Dec 2001. Dotted and dashed circles denote the ∼150 km (90 mi) and ∼200 km (120 mi) and range rings, respectively. Center circles denote location of NCAR S-Pol and Portland radars. Gray squares in northwest corners denote shadow zones. (right) A portion of the full 4-km simulation domain images shown in Fig. 8 with range rings and radar locations corresponding to the Portland radar.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 8.
Fig. 8.

Equivalent reflectivity at 700 hPa from 4-km simulation at (a) 1800 UTC 13 Dec, (b) 0000 UTC 14 Dec, and (c) 1400 UTC 14 Dec 2001. Dashed and dotted circles denote the ∼200 km (120 mi) and ∼150 km (90 mi) range rings for the Portland and NCAR S-Pol radars, respectively. Contours denote the 1500- and 2000-m model orographic heights. The rectangle denotes the 1-km domain.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 9.
Fig. 9.

As in Fig. 8, but for 1-km simulation.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 10.
Fig. 10.

Vertical velocity (w, time averaged from 2300 to 0100 UTC, every 15 min) from 1-km simulation along the vertical cross section indicated in Fig. 5. Solid (dashed) contours denote upward (downward) motion with contours every 0.5 m s−1 (0 m s−1 contour not shown) and upward motion shaded.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 11.
Fig. 11.

Vertical motion (w) at 0000 UTC along leg 2 from 1-km simulation. (top) The w vs distance along the flight path at 725 hPa (∼2.5 km). (bottom) A vertical cross section along the flight leg, denoted by the dashed line. Solid (dashed) contours denote upward (downward) motion with contours every 0.5 m s−1 (0 m s−1 contour not shown) and upward motion is shaded.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 12.
Fig. 12.

Observed and simulated 18-h accumulated precipitation (1400 UTC 13 Dec–0800 UTC 14 Dec 2001) for the (a) 4-km and (b) 1-km simulations (subdomains only): (left) 4-km and (right) 1-km simulations (subdomains). Background shading denotes model values and rain gauges values are indicated at the crosshair symbols (same shading intervals). Lighter (darker) model values under a particular gauge indicate underpredicted (overpredicted) precipitation totals from the model. The Contours denote model orographic heights of 1500 and 2000 m. The square in (a) denotes the subdomain shown in (b).

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 13.
Fig. 13.

Observed vs simulated (nearest grid point to rain gauge) 18-h accumulated precipitation (1400 UTC 13 Dec–0800 UTC 14 Dec 2001) for 4-km (solid circles) and 1-km (open diamonds) simulations. Note that there are fewer rain gauge observations located within the 1-km domain than the 4-km domain.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 14.
Fig. 14.

Observed vs simulated 18-h accumulated precipitation (1400 UTC 13 Dec–0800 UTC 14 Dec 2001) for G05a, MY05b’s 1.33-km run (crosses) and the 1-km triple-moment run (diamonds). Note that there are fewer rain gauge observations located within G05a,MY05b’s 1.33-km run domain.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 15.
Fig. 15.

Vertical cross section of in situ observations of hydrometeor types and mass contents from the P-3 and Convair-580 between 2300 and 0100 UTC. Temperatures indicated are in degrees Celsius. (Reproduced from G05b, with permission.)

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 16.
Fig. 16.

Vertical cross sections of time-averaged (2300–0100 UTC, every 15 min) hydrometeor mass contents (Qx) from 1-km triple-moment simulation for (a) cloud, (b) ice and rain, (c) graupel, and (d) snow along cross section indicated in Fig. 5. Quantities are in grams per meter cubed at the values indicated in the shading scales; dashed contours for ice in (b) are 0.001, 0.01, and 0.1 g m−3. The vertical axis is pressure (hPa). North–south flight legs of the P-3 are indicated in (a); flight legs from the Convair-580 and the return flight of the P-3 are indicated in (d). The rectangle in (c) denotes the area of the vertical cross section outlined by the red box in Fig. 15.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Fig. 17.
Fig. 17.

As in Fig. 16, but for mean-mass diameters Dx. Quantities are in units of micrometers for (a), micrometers (ice) and millimeters (rain) for (b), and millimeters for (c) and (d) with magnitudes indicated in the shading scales. Contours for ice in (b) are 10, 25, 50, and 75 μm.

Citation: Monthly Weather Review 136, 10; 10.1175/2008MWR2197.1

Table 1.

Observed CLW vs model Qc along P-3 flight legs. Model values for the five legs are taken at 2330, 2345, 0000, 0030, and 0045 UTC, respectively. Observation and model numbers not in parentheses (in parentheses) denote mean (peak) values (g m−3) along flight legs.

Table 1.
Table 2.

Observed ice mass content vs model Qs along Convair flight legs. Model values for the four legs are taken at 2315, 0000, 0045, and 0100 UTC, respectively. Observation and model numbers not in parentheses (in parentheses) denote mean (peak) values (g m−3) along flight legs.

Table 2.
Table 3.

Observed ice mass content vs model Qs + Qg along P-3 flight legs. Model values are taken at 0100 UTC. Observation and model numbers not in parentheses (in parentheses) denote mean (peak) values (g m−3) along flight legs.

Table 3.

1

Many of these deficiencies in cloud schemes are also equally important for bin-resolving schemes.

2

In MY05a, the size distribution is expressed as a four-parameter gamma function for generality. Except for cloud, all categories could be equivalently expressed as a three-parameter gamma function.

3

The size distributions for each category x in the MY05 scheme are based on the diameter of spheres with appropriate bulk densities, thus using an exponent (dx) of 3 in the assumed mass–diameter relations. It would also be possible to use a mass–diameter relation for snow with an exponent of 2 (where the “diameter” represents the maximum crystal dimension). In that case, the reflectivity would be proportional to the fourth moment of the size distribution, since in general reflectivity is proportional to the (2dx)th moment.

4

The mean-mass diameter Ds can be computed from (2) with the total number concentration NTs obtained from the size distribution function [e.g. (4) and (5) from MY05a], assuming spherical particles. Note, assuming dx = 2 for the mass–diameter relation reduces the estimated Ds by approximate 20% from that by assuming spherical particles (dx = 3).

5

In the case of an inverse-exponential size distribution, the intercept parameter decreases with increasing temperature. Thus, for given mass content, the mean-particle size increases with increasing temperature (decreasing altitude).

Save