• Benjamin, S. G., , and N. L. Seaman, 1985: A simple scheme for objective analysis in curved flow. Mon. Wea. Rev., 113 , 11841198.

  • Berry, E. X., , and R. L. Reinhardt, 1974: An analysis of cloud drop growth by collection. Part IV: A new parameterization. J. Atmos. Sci., 31 , 21272135.

    • Search Google Scholar
    • Export Citation
  • Brown, P. R. A., , and H. A. Swann, 1997: Evaluation of key microphysical parameters in three-dimensional cloud-model simulation using aircraft and multiparameter radar data. Quart. J. Roy. Meteor. Soc., 123 , 22452275.

    • Search Google Scholar
    • Export Citation
  • Bruintjes, R. T., , T. L. Clark, , and W. D. Hall, 1994: Interactions between topographic airflow and cloud/precipitation development during the passage of a winter storm in Arizona. J. Atmos. Sci., 51 , 4867.

    • Search Google Scholar
    • Export Citation
  • Burrows, D. A., 1992: Evaluation of a two-dimensional kinematic cloud model using data from a central Sierra Nevada orographic cloud system. J. Appl. Meteor., 31 , 5163.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., 2004: Sensitivity of orographic precipitation to changing ambient conditions and terrain geometries: An idealized modeling perspective. J. Atmos. Sci., 61 , 588606.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and C. F. Mass, 2000: The 5–9 February 1996 flooding event over the Pacific Northwest: Sensitivity studies and evaluation of the MM5 precipitation forecasts. Mon. Wea. Rev., 128 , 593617.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and Y. Zeng, 2004a: Bulk microphysical sensitivities and pathways 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., , and Y. Zeng, 2004b: Bulk microphysical sensitivities and pathways within the MM5 for orographic precipitation. Part II: Impact of different bulk schemes, barrier width, and freezing level. Mon. Wea. Rev., 132 , 28022815.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , K. J. Westrick, , and C. F. Mass, 1999: Evaluation of the MM5 and Eta-10 precipitation forecasts over the Pacific Northwest during the cool season. Wea. Forecasting, 14 , 137154.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , C. F. Mass, , and K. J. Westrick, 2000: MM5 precipitation verification over the Pacific Northwest during the 1997–1999 cool seasons. Wea. Forecasting, 15 , 730744.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , J. B. Olson, , and J. S. Tongue, 2003: Multi-season verification of the MM5. Part II: Evaluation of high resolution precipitation forecasts over the northeast U.S. Wea. Forecasting, 18 , 458480.

    • Search Google Scholar
    • Export Citation
  • Cox, G. P., 1988: Modeling precipitation in frontal rainbands. Quart. J. Roy. Meteor. Soc., 114 , 115127.

  • Cox, J. A. W., , W. J. Steenburgh, , D. E. Kingsmill, , J. C. Shafer, , B. A. Colle, , O. Bousquet, , B. F. Smull, , and H. Cai, 2005: The kinematic structure of a Wasatch Mountain winter storm during IPEX IOP3. Mon. Wea. Rev., 133 , 521542.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46 , 30773107.

    • 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.

  • Fletcher, N. H., 1962: Physics of Rain Clouds. Cambridge University Press, 386 pp.

  • Garvert, M. F., , B. A. Colle, , and C. F. Mass, 2005a: The 13–14 December 2001 IMPROVE-2 event. Part I: Synoptic and mesoscale evolution and comparison with a mesoscale model simulation. J. Atmos. Sci., 62 , 34743492.

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

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

  • Grell, G. A., , J. Dudhia, , and D. R. Stauffer, 1994: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 138 pp. [Available from National Center for Atmospheric Research, PO Box 3000, Boulder, CO 80307.].

  • Hallett, M. F., , and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249 , 2628.

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

    • Search Google Scholar
    • Export Citation
  • Houze Jr., R. A., , P. V. Hobbs, , and P. H. Herzegh, 1979: Size distributions of precipitation particles in frontal clouds. J. Atmos. Sci., 36 , 156162.

    • Search Google Scholar
    • Export Citation
  • Hsie, E. Y., , R. A. Anthes, , and D. Keyser, 1984: Numerical simulation of frontogenesis in a moist atmosphere. J. Atmos. Sci., 41 , 25812594.

    • Search Google Scholar
    • Export Citation
  • Kessler, E., 1969: On the Distribution and Continuity of Water Substance in Atmospheric Circulations. Meteor. Monogr., No. 32, Amer. Meteor. Soc., 84 pp.

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

  • Medina, S., , and R. A. Houze Jr., 2003: Air motions and precipitation growth in Alpine storms. Quart. J. Roy. Meteor. Soc., 129 , 345371.

    • Search Google Scholar
    • Export Citation
  • Medina, S., , B. R. Smull, , and R. A. Houze Jr., 2005: Cross-barrier flow during orographic precipitation events: Results from MAP and IMPROVE. J. Atmos. Sci., 62 , 35803598.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., , and W. R. Cotton, 1992: Evaluation of the potential for wintertime quantitative precipitation forecasting over mountainous terrain with an explicit cloud model. Part I: Two-dimensional sensitivity experiments. J. Appl. Meteor., 31 , 2650.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., , P. J. DeMott, , and W. R. Cotton, 1992: New primary ice-nucleation parameterization in an explicit cloud model. J. Appl. Meteor., 31 , 708721.

    • Search Google Scholar
    • Export Citation
  • Murakami, M., 1990: Numerical modeling of dynamical and microphysical evolution of an isolated convective cloud—The 19 July 1981 CCOPE cloud. J. Meteor. Soc. Japan, 68 , 107128.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, R. M., and Coauthors, 1992: Winter Icing and Storm Projects (WISP). Bull. Amer. Meteor. Soc., 73 , 951974.

  • Rasmussen, R. M., , I. Geresdi, , G. Thompson, , K. Manning, , and E. Karplus, 2002: Freezing drizzle formation in stably stratified layer clouds: The role of radiative cooling of cloud droplets, cloud condensation nuclei, and ice initiation. J. Atmos. Sci., 59 , 837860.

    • Search Google Scholar
    • Export Citation
  • Rauber, R. M., 1992: Microphysical structure and evolution of a Sierra Nevada shallow orographic cloud system. J. Appl. Meteor., 31 , 324.

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

    • Search Google Scholar
    • Export Citation
  • Reynolds, D. W., , and A. S. Dennis, 1986: A review of the Sierra Cooperative Pilot Project. Bull. Amer. Meteor. Soc., 67 , 513523.

  • Rotunno, R., , and R. Ferretti, 2001: Mechanisms of intense Alpine rainfall. J. Atmos. Sci., 58 , 17321749.

  • Rutledge, S. A., , and P. V. Hobbs, 1983: The mesoscale and microscale structure and organization of clouds and precipitation in mid-latitude cyclones: A model for the “seeder-feeder” process in warm-frontal rainbands. J. Atmos. Sci., 40 , 11851206.

    • Search Google Scholar
    • Export Citation
  • Smith, R. B., , and I. Barstad, 2004: A linear theory of orographic precipitation. J. Atmos. Sci., 61 , 13771391.

  • Stauffer, D. R., , and N. L. Seaman, 1990: Use of four-dimensional data assimilation in a limited area mesoscale model. Part I: Experiments with synoptic-scale data. Mon. Wea. Rev., 118 , 12501277.

    • Search Google Scholar
    • Export Citation
  • Stauffer, D. R., , N. L. Seaman, , and F. S. Binkowski, 1991: Use of four-dimensional data assimilation in a limited-area mesoscale model. Part II: Effects of data assimilation within the planetary boundary layer. Mon. Wea. Rev., 119 , 734754.

    • Search Google Scholar
    • Export Citation
  • Stoelinga, M., and Coauthors, 2003: Improvement of Microphysical Parameterization through Observational Verification Experiments. Bull. Amer. Meteor. Soc., 84 , 18071826.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., , R. M. 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
  • 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
  • View in gallery

    Model precipitation (every 5 mm) from the 1.33-km domain between 2300 UTC 13 Dec to 0100 UTC 14 Dec 2001 (23–25 h). Terrain is gray shaded every 500 m using the inset scale. The bold boxed region represents the area of the microphysical budget. The dashed box represents the location of the average cross section. Line segments C1–C4 are the NOAA P-3 aircraft tracks used for the cloud water verification in section 5b.

  • View in gallery

    Microphysical flowchart for the Reisner-2 scheme. The circles represent the various water species (water vapor, cloud water, cloud ice, rain, snow, and graupel), and the arrows are the processes that link the species (see the appendix for the list of processes).

  • View in gallery

    (a) Average cross section (dashed box in Fig. 1) for the CTL (run 1) showing snow (gray) and graupel (black) every 0.20 g kg−1, rain (thick black) every 0.10 g kg−1, and circulation vectors in the cross section. (b) Same simulation as (a) except for cloud water (black) every 0.10 g kg−1 and cloud ice (gray) every 0.02 g kg−1.

  • View in gallery

    Flowchart of the microphysical processes between 23 and 25 h (2300–0100 UTC) of the control run for the solid box in Fig. 1. The values shown are the ratio of each average microphysical process rate to the total WVL rate (cond + sdep + gdep + idsn + idep) within the box. For example, cond = 73.21 means that 73.21% of the WVL results in condensation, and gsacw = 6.47 means that the movement of cloud water mass to graupel is equivalent to 6.47% of the total WVL rate. The processes are listed in the appendix. The sum of all the microphysical process tendencies for each species is given by (wv, cw, r, ci, g, and s). This sum does not include horizontal advection and diffusion/divergence, which are labeled as hadv and other, respectively. The fallout tendency of rain (rprc), snow (sprc), graupel (gprc), and cloud ice (iprc) are also shown. Microphysical processes greater than 4% of the WVL rate are in bold.

  • View in gallery

    Cross section of the spatial distribution of the major microphysical processes contributing to snow production, showing the ratio of (a) sdep (gray) and icns (black), (b) saci (gray) and ssacw (black) to the total snow production rate from these processes at each point contoured every 20%. (c), (d) The same as (a) except for graupel production showing (c) ggacw (black) and scng (gray) every 5%, and (d) gsacw (black) and gacr (gray) every 20%. (e), (f) The same as (a) except rain production showing racw (gray), gmlt (thick black), smlt (black), and (f) gmlt (black) every 20%. See the appendix for process abbreviations. The 0° and −15°C isotherms are shown in (a) for reference.

  • View in gallery

    (a) Snow deposition (every 10 × 105 g kg−1 s−1) and winds (1 full barb = 10 kt) averaged between 450 and 550 mb using 15-min microphysical output from the 1.33-km domain for 2300–0100 UTC 14 Dec. (b) Same as (a) except for the accretion of cloud water by snow averaged between 500 and 600 mb. Terrain from the 1.33-km domain is shaded for reference

  • View in gallery

    Same as Fig. 6, except for the 600–700-mb layer showing graupel growth via (a) accretion of snow by cloud water (every 10 × 105 g kg−1 s−1) and (b) accretion of cloud water by graupel.

  • View in gallery

    Same as Fig. 6, except for the 850–750-mb layer showing rain growth via (a) accretion of rain by cloud water (every 10 × 105 g kg−1 s−1), (b) snowmelt, and (c) graupel melt. A transact of rain gauge locations are shown across the Cascades, with model precipitation within 30% of the observed shown by a X, greater than 130% of observed by an o, and greater than 180% of observed by an O for the 2000 UTC 13 Dec to 0200 UTC 14 Dec period.

  • View in gallery

    (a) Cross section of differences between fixed Nos experiment (run 2) and the CTL (run 1) showing snow (gray) every 0.10 g kg−1, graupel (thick solid) every 0.05 g kg−1, and cloud water (thin solid) every 0.05 g kg−1. (b) Same as (a), except for the NOSQS (run 3) minus CTL.

  • View in gallery

    (a) Same as Fig. 4, except for run 2 (fixed Nos). (b) Same as (a), except for run 3 (NOSQS).

  • View in gallery

    (a) Storm total precipitation (every 15 mm) for the 1.33-km domain between 1400 UTC 13 Dec and 0800 UTC 14 Dec 2001 and 1.33-km terrain is shaded for reference. (b) Difference between the NOSF (run 2) and the CTL experiment every 10 mm starting at ±5 mm, with negative values dashed. (c) Same as (b), except for the Nosq (run 3). (d) Same as (b), except for the COX (run 4). (e) Same as (b), except for the PSACW (run 5). (f) Same as (b), except for the KESS (run 6). (g) Same as (b), except for the SICE and contoured intervals are every 20 mm. (h) Same as (g), except for the warm rain scheme.

  • View in gallery

    Same as Fig. 9, except for the COX (run 4) minus CTL. Snow trajectories are also shown for the CTL (black) and COX (gray).

  • View in gallery

    (a) Same as Fig. 9, except for PSACW–CTL. (b) Same as (a), except for KESS–CTL.

  • View in gallery

    (a) Average cross section for the SICE (run 7) showing snow (gray) every 0.20 g kg−1, and rain (thick black) every 0.10 g kg−1, and cloud ice (thin black) every 0.05 g kg−1. (b) Same as (a), except for the warm rain scheme (run 8) and cloud water is shown rather than cloud ice.

  • View in gallery

    (a) Cloud water verification averaged for each of the five north–south NOAA P-3 legs (C1–C5) shown in Fig. 1 between 2300 UTC 13 Dec and 01 UTC 14 Dec. The observed cloud water amounts (g m−3) is shown by the solid black line and each of the Nos experiments are listed in the inset. (b) Same as (a) except for the KESS, COX, and PSACW experiments. The average N–S altitude (m) of the P-3 for each leg is given by the large open circles in meters, and the average N–S topography height is shaded below in meters.

  • View in gallery

    (a) Ice number concentrations from the Convair aircraft at 6.0 km ASL over the windward slope (black open circles and black-dashed best fit) and MM5 number concentrations derived using the fixed Nos (gray dashed), NosT (solid black), and NOSQS (gray solid) experiments. (b) Same as (a), except for the PSACW (gray solid), Cox (gray dashed), and simple ice (short dashed). (c) Same as (a), except for 4.9 km ASL. (d) Same as (b), except for 4.9 km ASL.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 65 65 13
PDF Downloads 38 38 3

The 13–14 December 2001 IMPROVE-2 Event. Part III: Simulated Microphysical Budgets and Sensitivity Studies

View More View Less
  • 1 Institute for Terrestrial and Planetary Atmospheres, State University of New York at Stony Brook, Stony Brook, New York
  • | 2 Department of Atmospheric Sciences, University of Washington, Seattle, Washington
  • | 3 Institute for Terrestrial and Planetary Atmospheres, State University of New York at Stony Brook, Stony Brook, New York
  • | 4 Department of Atmospheric Sciences, University of Washington, Seattle, Washington
© Get Permissions
Full access

Abstract

This paper investigates the microphysical pathways and sensitivities within the Reisner-2 bulk microphysical parameterization (BMP) of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) for the Improvement of Microphysical Parameterization through Observational Verification Experiment (IMPROVE)-2 field experiment on 13–14 December 2001. A microphysical budget over the windward slope at 1.33-km horizontal grid spacing was calculated, in which the importance of each microphysical process was quantified relative to the water vapor loss (WVL) rate. Over the windward Cascades, the largest water vapor loss was associated with condensation (73% of WVL) and snow deposition (24%), and the windward surface precipitation resulted primarily from accretion of cloud water by rain (27% of WVL), graupel fallout and melt (19%), and snowmelt (6%). Two-thirds of the snow generated aloft spilled over into the lee in an area of model overprediction, resulting in windward precipitation efficiency of only 50%. Even with the large amount of precipitation spillover, the windward precipitation was still overpredicted in many locations.

A series of experiments were completed using different snowfall speeds, cloud water autoconversion, threshold riming values for snow to graupel autoconversion, and slope intercepts for snow. The surface precipitation was most sensitive to those parameters associated with the snow size distribution and fall speed, while decreasing the riming threshold for snow to graupel conversion had the greatest positive impact on the precipitation forecast. All simulations overpredicted cloud water over the lower windward slopes, had too little cloud water over the crest, and had too much ice at moderate-to-large sizes aloft. Riming processes were important, since without supercooled water there were bull’s-eyes of spurious snow spillover over the lee slopes.

Corresponding author address: Dr. B. A. Colle, Marine Sciences Research Center, State University of New York at Stony Brook, Stony Brook, NY 11794-5000. Email: bcolle@notes.cc.sunysb.edu

Abstract

This paper investigates the microphysical pathways and sensitivities within the Reisner-2 bulk microphysical parameterization (BMP) of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) for the Improvement of Microphysical Parameterization through Observational Verification Experiment (IMPROVE)-2 field experiment on 13–14 December 2001. A microphysical budget over the windward slope at 1.33-km horizontal grid spacing was calculated, in which the importance of each microphysical process was quantified relative to the water vapor loss (WVL) rate. Over the windward Cascades, the largest water vapor loss was associated with condensation (73% of WVL) and snow deposition (24%), and the windward surface precipitation resulted primarily from accretion of cloud water by rain (27% of WVL), graupel fallout and melt (19%), and snowmelt (6%). Two-thirds of the snow generated aloft spilled over into the lee in an area of model overprediction, resulting in windward precipitation efficiency of only 50%. Even with the large amount of precipitation spillover, the windward precipitation was still overpredicted in many locations.

A series of experiments were completed using different snowfall speeds, cloud water autoconversion, threshold riming values for snow to graupel autoconversion, and slope intercepts for snow. The surface precipitation was most sensitive to those parameters associated with the snow size distribution and fall speed, while decreasing the riming threshold for snow to graupel conversion had the greatest positive impact on the precipitation forecast. All simulations overpredicted cloud water over the lower windward slopes, had too little cloud water over the crest, and had too much ice at moderate-to-large sizes aloft. Riming processes were important, since without supercooled water there were bull’s-eyes of spurious snow spillover over the lee slopes.

Corresponding author address: Dr. B. A. Colle, Marine Sciences Research Center, State University of New York at Stony Brook, Stony Brook, NY 11794-5000. Email: bcolle@notes.cc.sunysb.edu

1. Introduction

a. Background

This paper is the third in a series of modeling papers on the 13–14 December 2001 precipitation event, which occurred during the second phase of the Improvement of Microphysical Parameterization through Observational Verification Experiment (IMPROVE-2) experiment over the central Oregon Cascades. Garvert et al. (2005a,b) focused on the kinematic and in situ microphysical verification for the control run of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5). Since the distribution of orographic precipitation is controlled by many dynamical and microphysical processes, the objective of this study is to quantify in more detail how the model precipitation was generated during the 13–14 December event through a series of sensitivity experiments and microphysical budgets. These sensitivity studies are motivated by long-term precipitation verification studies (Colle et al. 1999, 2000, 2003), which suggest that deficiencies in model microphysical parameterizations result in too much precipitation along windward slopes. Some of the uncertainties in model microphysics were discussed in Colle and Mass (2000) for a flooding event over the Pacific Northwest, in which the most sophisticated bulk microphysical parameterization (BMP) did not generate the best precipitation distribution.

During the past two decades the BMPs used in mesoscale models, such as the MM5, have become more sophisticated. During the early 1980s, only a warm rain scheme was available, which included cloud and rainwater processes, but no ice processes (Hsie et al. 1984). A simple ice scheme was soon developed to include some snow and ice phase processes (Dudhia 1989), but this scheme does not allow for supercooled water and snow immediately melts above 0°C. The Reisner-1 (or mixed phase Reisner) scheme allows for both of these missing aspects (Grell et al. 1994), but there are no graupel and riming processes. The Reisner-2 scheme includes graupel and prognostic equations for cloud ice number concentration (Reisner et al. 1998; Thompson et al. 2004); however, it is still a single moment scheme since snow and graupel number concentrations are not explicitly calculated.

Progress in BMP development has required field datasets to investigate precipitation processes associated with different types of mesoscale weather phenomena. Recent field experiments, such as IMPROVE (Stoelinga et al. 2003), help verify BMPs and hopefully will suggest avenues of improvement. For example, the Reisner-2 BMP has been recently modified using results from the Winter Ice Storm Project (WISP) over Colorado (Rasmussen et al. 1992; Thompson et al. 2004). Reisner et al. (1998) found that relating the slope intercept for the snow number concentration (Nos) to the snowfall rate slows snow deposition and maintains cloud water mass as compared to a fixed Nos approach. These changes have clearly improved the amount of supercooled water over the Colorado Front Range, but it is not clear how these changes impact the results over other locations.

Field studies have also led to a better understanding of the precipitation and associated kinematic flow processes over topography. For example, upstream flow blocking by the European Alps and the Wasatch Mountains of Utah has been shown to be important in enhancing the precipitation well upstream of the crest (Rotunno and Ferretti 2001; Medina et al. 2005; Cox et al. 2005). In contrast, a less stable air mass favors stronger upslope flow and results in increased riming and accretion over the steep windward slopes of the Alps (Medina et al. 2005). Gravity waves generated by topography have also been shown to influence the structure of the orographic cloud aloft (Bruintjes et al. 1994) and possibly snow aggregation (Medina and Houze 2003). Recent idealized modeling results suggest that mountain waves may have an important contribution to the structure of the vertical motion over the windward slope and resulting precipitation distribution across the barrier (Colle 2004; Smith and Barstad 2004). Overall, there has been limited work relating the spatial distribution of microphysical processes in model simulations for such orographically forced phenomena.

b. Previous model microphysical results

Most previous model microphysical sensitivity results over terrain have utilized two-dimensional models and simplified flows. Several studies have qualitatively compared observations for the 12 February 1986 event over the Sierra Nevada Mountains of California during the Sierra Cooperative Pilot Project (Reynolds and Dennis 1986; Rauber 1992) with either the Regional Atmospheric Modeling System (RAMS; Burrows 1992; Meyers and Cotton 1992; Meyers et al. 1992) or the MM5 (Colle and Zeng 2004a). Meyers and Cotton (1992) found that an increase in graupel density (from 450 to 900 kg m−3) decreased the residence time of graupel aloft, which in turn reduced the graupel mass and total precipitation over the barrier. They also found that a reduction in cloud condensation nuclei (CCN) led to slightly more warm rain precipitation over the barrier, while more CCN resulted in more suspended cloud water, which promoted riming of snow and graupel. Meyers et al. (1992) showed that the pristine ice fields were very sensitive to the changes in the ice nucleation formulation, with the Fletcher (1962) formulation overpredicting ice amounts at cold temperatures (<−30°C) while underpredicting at warm temperatures (−5° to −20°C). In contrast, Colle and Zeng (2004a) found that these large sensitivities to cloud ice aloft resulted in only small variations in the surface precipitation for this event.

Using two-dimensional MM5 simulations of the Sierra 1986 event, Colle and Zeng (2004a) showed that the orographic precipitation distribution in the MM5 Reisner-2 scheme was more sensitive to the slope intercept for snow number concentration and fall speeds for snow and graupel rather than cloud water to rain autoconversion and cloud ice processes. This is consistent with their microphysical budget over the windward slopes, which showed that over 75% of the water vapor loss rate went toward snow and graupel generation. Colle and Zeng (2004b) illustrated using idealized simulations that large sensitivities to cloud water accretion processes occur as the freezing level is raised well above crest level or for a much narrower (10-km half-width) barrier than the Sierras.

Other recent modeling studies have also highlighted specific microphysical sensitivities within BMPs. Colle and Mass (2000) tested the snowfall speed sensitivity for a flooding event over the Pacific Northwest on 5–9 February 1996. They found that using the Cox (1988) snow aggregate fall speed, which is 20% smaller than the standard Rutledge and Hobbs (1983) fall speed, shifted precipitation from the windward side to the lee of the barrier. The slope intercept for snow number concentration (Nos) can vary by two orders of magnitude from 0° to −20°C (Houze et al. 1979); therefore, Thompson et al. (2004) and Colle and Zeng (2004a) tested a temperature-dependent form of Nos in Reisner-2 and found that it increased the amount of snow at cold temperatures (<−20°C). Brown and Swann (1997) found for a convective cloud simulated over southern England that the Hallett–Mossop (H–M) process (Hallett and Mossop 1974) contributed to a majority of ice crystal concentrations in areas of riming between −3° and −8°C. Meanwhile, they found that a reduction of graupel fall speed and in the graupel slope intercept (Nog) threshold reduced the accumulated surface precipitation and decreased the accretion of snow by graupel.

c. Motivation and goals

Most model microphysical studies have either used two-dimensional models to investigate sensitivities within BMPs or simply focused on the differences in surface precipitation and mixing ratios aloft using different BMPs. The IMPROVE field study offers an opportunity to revisit the sensitivities within BMPs using a more comprehensive dataset that was collected for the sole reason of evaluating how fundamental changes in the BMPs affect microphysical processes. Woods et al. (2005) present the observational evolution and Garvert et al. (2005a,b) and describe the kinematic and microphysical verification from a high-resolution control run of MM5.

This paper continues the model investigation of the 13–14 December 2001 event, in which several parameters within the Reisner-2 parameterization are systematically evaluated using a microphysical water/ice budget, and some of the results are verified using aircraft data. Insight in the orographic precipitation processes are also obtained. To our knowledge, this is the first study to complete a full three-dimensional microphysical budget for a simulated orographic precipitation event using a complex BMP. This paper addresses several important questions:

  • What is the 3D microphysical budget for a period of moderate to heavy stratiform precipitation during 13–14 December 2001?
  • How are the model microphysical processes linked to various terrain features and the model precipitation errors at the rain gauge locations?
  • What are the most sensitive microphysical processes for this event?
  • How do the process sensitivities change the microphysical verification?

Section 2 provides a detailed description of the model, experimental design, and analysis methods. Section 3 presents the microphysical budget for the control run averaged over the intensive observation period (IOP) area and plotted for an average cross section across the Cascades. Select microphysical sensitivities and their budgets are presented in section 4. Section 5 verifies these additional microphysical sensitivity simulations using in situ surface and aircraft observations. Summary and conclusions are presented in section 6.

2. Experimental design and MM5 validation

a. MM5 description

The PSU–NCAR mesoscale model was used in a nonhydrostatic mode to simulate this case. Garvert et al. (2005a) describes the detailed setup of the MM5 simulations for the 13–14 December event, which included a 36-km grid covering the eastern Pacific and western North American and nested down to a 1.33-km grid over the central Oregon Cascades (Fig. 1). The 6-h National Centers for Environmental Prediction (NCEP) analysis and subsequent forecasts from the Global Forecast System (GFS) model at 0000 UTC 13 December served as the initialization and lateral boundary conditions for the 36-km MM5. The interpolated GFS data was improved by incorporating surface and upper air observations using a Cressman-type analysis scheme (Benjamin and Seaman 1985). To minimize large-scale forecast drift, four-dimensional data assimilation was utilized on the 36-km MM5 grid during the first 12-h of the forecast (Stauffer and Seaman 1990; Stauffer et al. 1991). After the 36- and 12-km MM5 forecast was integrated, initial and boundary conditions for a separate 4- and 1.33-km run starting at 0600 UTC 13 December were obtained by using hourly data from the 12-km MM5.

The MM5 had 33 full sigma levels in the vertical, with maximum resolution in the boundary layer. All simulations used the medium-range forecast (MRF) planetary boundary layer (Hong and Pan 1996), and the Grell convective parameterization was applied in the 36- and 12-km domains (Grell 1993). The control (CTL) simulation used the Reisner-2 explicit moisture scheme (Reisner et al. 1998), with further modifications by Thompson et al. (2004).

b. Summary of MM5 validation

Garvert et al. (2005a) presents detailed model comparisons with observations on the synoptic scale and mesoscale for the 13–14 December IOP, which involved the passage of a forward-tilting frontal system over the Cascades. It was shown that the MM5 was able to accurately represent the features associated with this storm, including the amplification of the surface low and 500-mb trough, and the position and strength of the 300-mb jet streak. The model also accurately simulated the moisture profile upstream of the Cascades as well as the timing and orientation of the main precipitation band. A low-level jet with wind speeds approaching 40 m s−1 was present, but the model winds were 5–10 m s−1 too weak between 850 and 650 mb. Nevertheless, Garvert et al. (2005a) showed that the 1.33-km grid realistically produced 1–2 m s−1 vertical motions over the windward ridges of the Cascades, while the vertical velocities from the 4-km grid were significantly weaker than observed.

Even with a relatively accurate kinematic and moisture field over the Cascades, Garvert et al. (2005a) showed that the 1.33-km domain still overpredicted precipitation by 50%–100% at many surface sites over the Cascades, especially in the immediate lee of the crest. To further investigate the model microphysics associated with this overprediction, the model analysis herein is restricted to the 1.33-km domain (Fig. 1).

c. Microphysical budget setup

One of the important tools in diagnosing different microphysical parameters is the calculation of a microphysical budget for a volume upstream of the crest (Fig. 1). Colle and Zeng (2004a) showed that by quantifying the relationship among water species, one can determine which microphysical process contributes most to the production and destruction of a specific hydrometeors. Figure 2 and Eqs. (1)(6) show how the various Reisner-2 processes described in the appendix are related through the water vapor (qυ), cloud water (qc), cloud ice (qi), rain (qr), snow (qs), and graupel (qg) mixing ratios. The prognostic equations for each hydrometeor species described in Reisner et al. (1998) are
i1520-0469-62-10-3535-e1
i1520-0469-62-10-3535-e2
i1520-0469-62-10-3535-e3
i1520-0469-62-10-3535-e4
i1520-0469-62-10-3535-e5
i1520-0469-62-10-3535-e6
where p* is the pressure difference between the surface and model top.

There are three contributions to the mixing ratio tendencies in Eqs. (1)(6). The contributions from the microphysical processes are the first set of Pxxxx terms on the right-hand side, while advection (horizontal/vertical), divergence and diffusion are represented by the terms in italics. The horizontal advection term (ADV) are shown separately in the analysis below while the other italic terms are lumped together as other_x, since they are much smaller than ADV. The double-underlined term (Pxxxx) represents the fall out of a precipitable species.

The conversion rate for each term in Eqs. (1)(6) was output every 15 min between hours 16 and 20 of the control simulation (2200 UTC 13 December to 0200 UTC 14 December); however, most of the average budget analysis below is restricted to the 17–19-h period for the sensitivity runs since this is the period of heaviest orographic precipitation (Woods et al. 2005). Following Colle and Zeng (2004a) the data were averaged for a box upstream of the crest (Fig. 1). To determine the relative importance of each process in moving water mass from one hydrometeor category to another, each process was normalized by the integrated water vapor loss within that same box using
i1520-0469-62-10-3535-e7
where Pqqqq(i, j, k) is the conversion rate of a specific microphysical process averaged for the two adjacent sigma levels, WVL is the water vapor loss rate (Cond + Pidsn + Pidep + Psdep + Pgdep), and Δσ is the sigma level difference.

3. Control microphysical budget

a. Domain-averaged analysis

Figure 1 shows the 2-h precipitation from the 1.33-km domain during the period of heavy frontal precipitation between 2300 UTC 13 December and 0100 UTC 14 December while both the National Oceanic and Atmospheric Administration (NOAA) P-3 and Convair research aircrafts completed their first series of flight legs over the Cascades (Woods et al. 2005). There is precipitation enhancement (>20 mm in 2 h) over the windward ridges and localized minima (<10 mm) in the valleys. Precipitation shadowing (<5 mm) is also present to the east of the coastal range over the Willamette Valley and downwind of the Cascades. Interestingly, the greatest precipitation amounts (>25 mm in 2 h) are in the immediate lee of the Cascades because of spillover of hydrometeors over the crest. Garvert et al. (2005a) verified the 1.33-km precipitation against all surface rain gauge locations and found model overprediction by 50%–70% over some of the higher terrain areas and to the lee of the barrier (their Fig. 17a).

Figure 3 shows a west–east cross section of the mixing ratios averaged for 2300–0100 UTC 13–14 December over the dashed boxed region in Fig. 1. During this period there was a deep orographic cloud, with snow extending above 400 mb and graupel between 800 and 600 mb (Fig. 3a). Snow amounts are largest (1.1 g kg−1) over the crest at 600 mb, with the axis of maximum snow aloft extending downward into the immediate lee with the mountain wave circulation. Graupel has a maximum (0.50 g kg−1) at 750 mb over the steep windward slopes, with large values extending downward toward the Cascade crest (Fig. 3a). Cloud water is largest (0.60 g kg−1) at 750 mb over the windward peaks (Fig. 3b). The cloud water extends to 450 mb upstream of the Cascades, with the top of this layer lowering to approximately 600 mb over the crest in an area where snow amounts are increasing. This suggests that the upper-level cloud water near the crest is being depleted at the expense of the snow growth. Cloud ice (0.08 g kg−1) is only found above 400 mb (Fig. 3b). Cloud ice is limited to such high levels and the amount is relatively small since Reisner-2 autoconverts cloud ice to snow at relatively small sizes (100 μm).

Figure 4 shows the dominant microphysical processes averaged horizontally and vertically within the solid boxed region of Fig. 1 between 2300 and 0100 UTC. Each average process tendency was normalized by the WVL rate within the volume, with process values greater than 4% of the WVL rate highlighted in bold. The pathway to cloud water via condensation is the largest sink of water vapor over the windward slope (cond = 73% of WVL). Snow deposition (sdep) accounts for most of the remainder of the WVL (24%). Although ice initiation and ice deposition are small in terms of water mass (<3% of WVL), they are still important since cloud ice autoconverts to snow (icns) at relatively small sizes, and the snow grows rapidly via deposition and riming. A large fraction of cloud water produced results in accretion by rain (racw = 27% of WVL), but there is little cloud water autoconversion (ccnr <1%). Accretion by rain accounts for more than half of the rain fallout over the windward slope, with the fallout scaling as 48% of WVL.

Besides deposition, accretion of cloud water by snow (ssacw) is also a major contributor to snow growth via riming (9% of WVL). Of the snow generated over the windward slope, 16% (6% of WVL) melts to rain (smlt), 11% (4% of WVL) autoconverts to graupel during riming (scng), while 73% (28% of WVL) advects over to the lee of the barrier (hadv_s). The limited snow fallout over the windward slopes (sprc <0.5% WVL) is consistent with the relatively high freezing level (775 mb) and the strong cross-barrier winds (>30 m s−1) advecting hydrometeors into the lee. Since the model winds are 5–10 m s−1 weaker than observed at this time over the crest (Garvert et al. 2005a), the larger observed cross-barrier flow would have favored more snow advection into the lee in the model than shown in Fig. 1.

Half of the snow accretion of cloud water (ssacw) results in snow growth via riming, while the remainder contributes to 50% of the graupel production over the windward slope (gsacw scales as 9% of WVL). The other half of graupel production originates from riming of graupel by rain (gacr) and conversion of snow to graupel (scng). Nearly all the graupel that is produced melts and falls as rain over the windward slope (gmlt scales as 17% of WVL). Graupel melt is the second most common source of rain and it is 3 times larger than snowmelt over the windward slope; therefore, riming processes were important in windward precipitation generation. Woods et al. (2005) showed using aircraft data and rain gauges that riming within the windward feeder cloud was important in enhancing the surface precipitation. The large pathway to surface precipitation via graupel melt is also consistent with the limited flow blocking observed by aircraft during this event (Garvert et al. 2005a), which has been shown to favor more graupel production over the steep windward slope of barriers (Medina et al. 2005).

The windward precipitation efficiency (PE) is defined as the total amount of fallout of hydrometeors over the windward divided by the total WVL within this same area. The windward PE can be obtained by adding the fallout terms in the microphysical budget (rpsc, sprc, and gprc in Fig. 4), since the fallout is normalized by the WVL. The windward PE for the control run is 50%, so only half of the windward WVL leads to surface precipitation upstream of the crest. Over half of the condensate is lost by snow advection into the lee (hadv_s = 28%), while the remainder is lost via evaporation (evap = 23%). The simulated PE for this case is lower than that recently diagnosed for the Sierras (PE = 80%) using MM5 (Colle and Zeng 2004a), since the Cascades are narrower and the cross-barrier flow during this IMPROVE event was twice as strong.

It is important to quantify the spatial distribution of the microphysical processes over the windward slope. Figure 5 shows the percentage contribution that a microphysical process has on the total production rate for a particular hydrometeor species at each point. For snow growth (Figs. 5a,b), conversion of cloud ice to snow (icns) and accretion of cloud ice by snow (saci) dominates above 400 mb and down to 700 mb in the lee of the Cascades. As these crystals fall out between 500 and 400 mb over the windward slope they grow primarily through deposition (sdep), especially over the crest and near the −15°C isotherm. Collection of cloud water by snow (ssacw) occurs between 700 and 450 mb upwind of the Cascades in the layer of supercooled water over this region (cf. Fig. 3b). As a result, snow growth around 600 mb is split evenly between depositional and riming processes upstream of the barrier. Approaching the crest around 600 mb, depositional growth becomes about twice as large as riming (60% versus 34%).

Figures 5c,d show the cross section of graupel production. Graupel growth is small above 400 mb, where only a small amount of depositional growth occurs (not shown). There is also only a small amount of graupel generated over the central Willamette Valley between 750 and 450 mb in an area of collection of cloud water by graupel (ggacw). Graupel growth increases dramatically over the windward slope as the collection of cloud water by snow (gsacw), conversion of snow to graupel (sncg), and riming of graupel by rain (gacr) become large. The gsacw, ggacw, and scng terms are largest where there is enhanced vertical motion over some of the windward peaks, with gsacw and sncg accounting for nearly 50% of the graupel growth around 700 mb. The conversion to graupel represents nearly all of the snow sink between 700 and 600 mb (not shown).

Figures 5e,f show the processes responsible for rain growth in the cross section. Above the melting band (∼750 mb) nearly all the rain growth is from accretion of cloud water by rain (racw), although the total rain growth above 600 mb is relatively small (cf. Fig. 3a). Accretion accounts for nearly all growth below the melting band over the Willamette Valley and the lower windward slope of the Cascades. Accretion and melting graupel have similar contributions within the melting band and over the Willamette Valley and the lower windward slope, while melting of snow only accounts for less than 10% of the rain generation. Along the upper windward slope and crest, graupel melting account for 80% of the rain growth. In contrast, the spillover of snow into the immediate lee results in snowmelt (smlt) and is the largest contributor to rain production (55%) in this region. There is little autoconversion of cloud water to rain (= 1%) across the domain.

b. Horizontal microphysical budgets

The previous section showed the relative contribution of each microphysical process over the windward slope averaged in the north–south direction; however, to illustrate the small-scale variability over the narrow ridges and valleys across the Cascades, select processes are plotted horizontally and averaged for a 100-mb vertical layer centered around where the snow, graupel, and rain growth are a maximum (Figs. 6 –8).

The snow deposition and riming processes for the 1.33-km domain were averaged between the 450- and 550-mb and 500- and 600-mb layers, respectively, for the average period of 2300–0100 UTC 14 December. Snow deposition is largest over the middle and upper windward slopes (50 × 105 kg s−1), with localized areas of 50% less production in the lee of some windward ridges (Fig. 6a). The depositional growth increases 20–30 km upstream of the windward slope since the upward motion increases aloft approaching the terrain (Fig. 3a). This is consistent with the idealized two-dimensional results of Colle (2004), in which a relatively wide barrier (50-km half-width) under stable flow can generate an orographic cloud aloft that extends upstream of the barrier in response to the upstream tilt of the mountain gravity wave. The collection of cloud water by snow (riming) also increases snow growth 20–30 km upstream of the barrier, and it is maximized (>40 × 105 g kg−1 s−1) between 500 and 600 mb over some of the steep windward slopes (Fig. 6b). This suggests that gravity waves off the narrow ridges can have a profound impact on the model microphysical processes well above the crest.

Most of the riming of snow leads to graupel production below 600 mb (Fig. 5d). Figure 7 shows the graupel production terms averaged between 700 and 600 mb. The largest graupel source is the collection of cloud water by snow (gsacw) over the windward slopes (Fig. 7a). There are large areas of gsacw upstream of the crest (40–60 × 10−5 g kg−1 s−1), with the greatest area (80–100 × 105 g kg−1 s−1) immediately upstream of the Cascade crest. Meanwhile, the autoconversion of snow to graupel (scng) is specified in Reisner-2 to be exactly half of gsacw (not shown), and it is the second largest contribution to graupel. The collection cloud water by graupel (gacw) is largest over the lower windward slope and has magnitudes between 10–20 × 105 g kg−1 s−1 (Fig. 7b). The collection of rain by graupel occurs over a similar location as ggacw (not shown), but about half in magnitude.

The rain production is plotted horizontally for the layer just below the melting level between 850 and 750 mb (Fig. 8). Accretion of cloud water by rain is largest along the steep areas of the lower windward slopes (>150 × 105 g kg−1 s−1) (Fig. 8a). Most of the rain precipitation generated over the middle windward slope is from melting of graupel, with localized maximum (>250 × 105 g kg−1 s−1) in the immediate lee of the windward ridges. Meanwhile, there is a large amount of snowmelt in the lee of the Cascades (>500 × 105 g kg−1 s−1) (Fig. 8b), and there are some localized maxima of snowmelt associated with spillover into some of the windward valleys under strong southwesterly flow.

Garvert et al. (2005a) showed a sharp gradient in model precipitation performance across the central Cascades. One can attempt to relate the gauge verification to the model microphysical processes in order to determine which hydrometeor species may have contributed to the model precipitation errors. For example, further eastward within a major valley on the windward slopes (Fig. 8a), there is a rapid transition within 10–15 km from 120% of observed (X) to 205% (O) and 165% (o) of observed during the 2000 UTC 13 December to 0200 UTC 14 December period. Only the precipitation gauges across the central Cascades were analyzed for this short 6-h period, since they were the best maintained during the field experiment. The furthermost west point has little contribution to rain from snowmelt, while the gauge 10 km to the east has a 40%–50% contribution from snowmelt associated with the spillover over the ridge to the south (Fig. 8b). Further up the windward slope of the Cascades the precipitation is within 130% of observed, and this area is dominated by graupel melt and fallout (Fig. 8c). Meanwhile, there is large overprediction in the lee associated with snowmelt. These results suggest that overprediction occurs when there is a relatively large snow contribution to the precipitation fallout, and that the model produced too much snow aloft, which was shown in Garvert et al. (2005b).

4. Process sensitivity experiments

A set of experiments were completed using different microphysical parameters within Reisner-2 to quantify the process sensitivities for the 13–14 December event. The model microphysics are evaluated during the 2300–0100 UTC period using different slope intercepts for the number concentration of snow, a slower snowfall speed, a decreased threshold for snow riming, a different cloud water autoconversion, and two other simplified BMPs within the MM5. The goal of these sensitivity simulations is not to find a fix to the BMP, but to focus primarily on those processes that are important to snow and graupel growth. To obtain the difference in microphysical budgets between the CTL and each experiment, the average WVL between the two simulations are used. Detailed verification of these sensitivity results will be shown in section 5.

a. Slope intercept for snow number concentration (Nos)

The slope intercept in the Marshall–Palmer distribution of snow influences the snow fallout, riming, deposition, and melting of snow. For this first series of experiments (runs 1–3 in Table 1), the snow intercept parameter (Nos) in the CTL (run 1), which depends on temperature (Thompson et al. 2004)
i1520-0469-62-10-3535-e8
where T0 = 273.15 K, was changed to either a fixed Nos = 2 × 107 m−4 (run 2) as in other well-known BMPs (Lin et al. 1983; Rutledge and Hobbs 1983) or a function of snow mixing ratio (run 3 or NOSQS) as described in Reisner et al. (1998). From Eq. (8), Nos decreases as the temperature increases, which results in more larger snow particles and fewer smaller particles. This NosT approach is designed to parameterize the decreasing number concentration associated with the aggregation process at warmer temperatures (Greg Thompson 2004, personal communication). Using a NosT results in less snow crystals than a fixed Nos for temperatures warmer than −19°C. Meanwhile, a NOSQS typically favors more large snow particles and fewer small particles than the other Nos approaches, and NOSQS was found to be closer to observed over the Colorado Front Range during the WISP experiment (Reisner et al. 1998).

Using a fixed Nos (Fig. 9a), there are large differences compared to the CTL in the amount of snow, cloud water, and graupel for the average west to east cross section during the 2300–0100 UTC period. Since a fixed Nos allows more snow particles than NosT at relatively warm temperatures, snow is increased by 50% in the 500–750-mb layer for a fixed Nos. The increased number of snow particles in this layer also acts to deplete the available cloud water and therefore decreases graupel production, which results in a reduction of these two species by 25% below 500 mb. The snow and graupel differences with the CTL are largest over the barrier around 700 mb, where there are large vertical motions and snow accretional processes (cf. Fig. 5). The additional snow produced aloft by the fixed Nos is in the opposite direction needed to improve the model, which will be highlighted in section 5.

A fixed Nos decreases the condensation contribution to WVL from 73%–66% (Fig. 10a), while snow deposition increases from 24%–27% of WVL. Less available cloud water results in slightly less collection of cloud water by graupel as compared to the CTL run, which in turn results in less graupel melt. Overall, a fixed Nos favors a microphysical pathway involving slightly more snow deposition (wv → sdep → snow) rather than graupel production via riming (wv → cond → ggacw, gsacw → graupel). These changes are consistent with the greater number of snow particles using fixed Nos at cold temperatures, which can more efficiently deplete the available supersaturated water vapor.

The impact of changing Nos to a fixed value can also be evaluated using the storm total (1400 UTC 13 December to 0800 UTC 14 December) surface precipitation shown in Garvert et al. (2005a) and Fig. 11a. Using a fixed Nos reduces the precipitation by 5–10 mm (15%) over many areas of the upper windward slopes due to a decrease of graupel and rain fallout (Fig. 11b). Meanwhile, the increase in snow aloft results in more leeside spillover, where there is 25 mm (20%) more precipitation in the lee than the CTL. Unfortunately, this is in the opposite direction needed to improve the forecast in the lee. Much of the increased snow aloft that replaces the graupel still rimes and falls out before the lee, so there is little windward PE difference with the CTL (Table 1).

Using a NOSQS relationship favors a broader snow distribution, with fewer snow particles at colder temperatures. As a result, snow is reduced by 0.5 g kg−1 over the upper windward slope and crest as compared to the CTL (Fig. 9b). The snow differences with the CTL begin are first evident around 500 mb over the lower windward slope and extend downward toward the crest in a pattern similar to a falling snow trajectory. With reduced snow growth there is 0.1–0.2 g kg−1 more cloud water over this region and 0.15 g kg−1 more graupel over the crest. The decrease in snow is almost completely offset by the increase in cloud water and graupel. This additional cloud water above the crest is needed to improve the verification given the underprediction shown for this region (Garvert et al. 2005b). There is little difference with the CTL upstream of the barrier where the vertical motions are weak.

The changes in the microphysical budget using a variable NOSQS as compared to the CTL are much larger over the barrier than a fixed Nos (Fig. 10b). NOSQS results in a 60% reduction in snow deposition over the barrier. As a result, the amount of snow advection (hadv_s) is reduced by 40% as compared to the control and there is slightly less snowmelt (smlt) and accretion of cloud water by snow (ssacw). Meanwhile, the pathway to cloud water and cloud ice is increased, with 10% more condensation than in the CTL run and over twice as much cloud ice deposition. The increased cloud water results in slightly more rain accretion and fallout as well as accretion by snow to form graupel. The greater graupel and rain than the CTL results in 5–15 mm more precipitation over the much of the windward slope region using the NOSQS (Fig. 11c), while less snow in the lee results in 5–15 mm less in the lee of the Cascades and slightly better verification in this region (section 5a).

b. Snowfall speed (Vs)

Both Colle and Mass (2000) and Colle and Zeng (2004a) showed that precipitation fallout in the MM5 is strongly dependent on the snowfall speed. The CTL uses the Rutledge and Hobbs’ (1983) expression for fall speed, which was derived for an unrimed radiating assemblages of plates, sideplanes, bullets, and columns (Locatelli and Hobbs 1974). This fall speed is 20%–30% larger than for unrimed radiating assemblages of dendrites (Cox 1988; Ferrier 1994). A separate simulation (COX or run 4) was completed using the slower Cox fall speed to demonstrate the sensitivity and ramifications of switching to this relationship.

Figure 12 shows the differences in trajectories for snow in the CTL (black) and COX (gray) runs as well as the difference in mixing ratios between the two simulations for the average cross section. Snow trajectories were calculated every 100 s using interpolated 15-min output of vertical motion and snow mixing ratios. The mixing ratios were used to calculate the snowfall speed along the trajectory, which was added to the model vertical and horizontal air motions at each point to get the total particle motion. The COX trajectories above 600 mb fall out 20–30 km farther downwind than the CTL. There is less snow falling out over the upper windward slope in the COX, which results in less riming and graupel. The slower fall speed results in 5–15 mm less precipitation near the crest and 5–15 mm more storm total precipitation in the lee as compared to the CTL (Fig. 11d). Additional leeside precipitation is a problem for COX, since it already adds to an existing positive bias in this region.

The changes to the microphysical budget were relatively small using the slower Cox fall speed (not shown). Snow deposition decreased from 24% of WVL in the CTL to 22%, since more snow is advected across the barrier above the layer of greatest depositional growth around 525 mb. This reduction in snow growth is compensated by less snow riming below 600 mb (3%–4% of WVL rate), which resulted in less snow converting to graupel. As a result, snow increases slightly below 600 mb while graupel decreases (Fig. 12). The windward PE decreases from 50% in the CTL to 47% given the additional spillover of snow into the lee with the slower Cox fall speed (Table 1).

c. Snow accretion of cloud water

An accurate prediction of the partition between snow and graupel aloft is important since more (less) graupel favors faster (slower) precipitation fallout over the windward slope. The initial generation of graupel in Reisner-2 is determined primarily by the snow to graupel autoconversion (scng), and this process is initiated when the amount of snow riming (psacw) exceeds depositional growth (sdep) by a factor of 2.5 (Thompson et al. 2004). An experiment was completed in which this factor was reduced to 1.0 (Murakami 1990), a threshold that was applied in earlier versions of Reisner-2 (Reisner et al. 1998).

Reducing this threshold favors more snow to graupel autoconversion in the area of riming below 500 mb; therefore, in the average west to east cross section snow was reduced by 0.40 g kg−1 (40%) over the windward slope while graupel was increased 0.20 g kg−1 around 600 mb (Fig. 13a). The reduction in snow results in less depletion of supercooled water around 600 mb, so cloud water was increased by 0.15 g kg−1 over the windward slope. Snow is decreased since the amount of snow deposition and condensation is decreased and increased by 3% of WVL, respectively (not shown). The larger amount of cloud water leads to a 4% WVL increase in accretion of cloud water by rain.

With more autoconversion from snow to graupel there is less snow available for depositional growth and riming, so these terms for snow growth in windward microphysical budget decrease by 3% and 6% of WVL (not shown), respectively, as compared to the CTL. Meanwhile, condensation increases by 3% and there is 1%–3% more accretion of cloud water and rain by graupel. As a result of the increased microphysical pathway to more rain and graupel fallout, the windward precipitation efficiency increases from 50% in the CTL to 53% (Table 1).

Figure 11e shows the surface precipitation differences of this experiment with the CTL run. An increased amount of graupel increases the precipitation by 5–10 mm (5%–10%) over the windward ridges, while the precipitation within some of the windward valleys is decreased by 5–10 mm. This reduction of valley precipitation reduces the overprediction shown in Fig. 8b and Garvert et al. (2004a). The largest impacts are in the immediate lee of the Cascades, with reduced snow spillover resulting in 10–15 mm (10%–15%) less precipitation, which helps reduce some of the overprediction.

d. Cloud water autoconversion

The CTL simulation used the Berry and Reinhardt (1974) method for the cloud to rain autoconversion, which is described in Thompson et al. (2004). For several years the Kessler (1969) approach was utilized in Reisner-2, which uses a threshold (0.35 g kg−1) and a simple Heaviside function to do the conversion. The CTL approach resulted in very little autoconversion in the windward mass budget (cf. Fig. 4); therefore, another simulation was completed using Kessler (KESS). This approach increased the autoconversion, and the amount of cloud water in the average cross section decreased by around 0.06 g kg−1 (Fig. 13b). The reduction of cloud water also reduces the riming growth of snow over the lower windward slope by 0.05 g kg−1 and graupel over the crest by an equivalent amount. As result, this helps somewhat with the snow and cloud water overpredictions noted in Garvert et al. (2005b).

The amount of autoconversion increases from 0.02%–5% of WVL using the Kessler approach for the windward budget (not shown). Meanwhile, cloud water accretion by rain only decreased by 1%, so there is a net increase of rain, which increased precipitation efficiency by 2% compared to the CTL. The reduction in cloud water also leads to slightly less accretional growth of snow and graupel by cloud water. The KESS run has 5–30 mm more precipitation over some of the steeper windward slopes (Fig. 11f); however, the vertical differences in mixing ratio using KESS are smaller than the ice sensitivities shown above.

e. Simple ice and warm rain schemes

To further quantify the importance of riming processes on the surface precipitation, the simulation was rerun without these processes. Specifically, a simulation was completed using the simple ice (SICE) BMP in MM5 (SICE run), which neglects supercooled water, and therefore only includes snow, and cloud ice exists aloft with no riming or graupel processes. The SICE scheme also uses a fixed Nos approach to the snow number concentration; therefore, the results can be compared to the fixed Nos run above.

Figure 14a shows the mixing ratios of snow, rain, and cloud ice using SICE for the same average cross section, which can be compared to the control (cf. Fig. 3). Since liquid water below 0°C cannot exist aloft in SICE, the microphysical pathway is dominated by depositional growth (not shown), resulting in a snow maximum (2.0 g kg−1) immediately over the crest that is nearly double that of the Reisner-2 CTL run. This maximum in snow in SICE gets advected over into the immediate lee, resulting in a well-defined rain maximum (>0.5 g kg−1) in this region and nearly twice as much precipitation as the control (Figs. 11a,g), which makes the leeside overprediction problem much worse. Meanwhile, there is less fallout of rain over the upper windward slope by 20–30 mm (30%–40%) since there is no contribution by graupel melt or cloud water accretional processes aloft.

The combination of a fixed Nos and no supercooled water aloft results much more spillover of precipitation in the SICE run, which is the opposite trend needed to improve upon the CTL run. On the other hand, some ice is important, as demonstrated by a simulation using the warm rain scheme (WRAIN) with no ice processes. This scheme is dominated by condensational and rain accretional processes; therefore, in the average cross section, a broad area of rain exists over the lower windward slope (>0.5 g kg−1) that is 20%–40% greater than the CTL (Figs. 3a, 14b). There are also sharper gradients of precipitation upstream of the Cascades and in the immediate lee in the WRAIN since there is no snow fallout from aloft into these regions. As a result, the surface precipitation is reduced by half in the immediate lee of the Cascades and nearly doubled in many areas of the windward slope.

5. Microphysical verification of experiments

a. Surface precipitation verification

Garvert et al. (2005a) present the verification of the CTL surface precipitation using all precipitation gauges for the 1400 UTC 13 December through 0800 UTC 14 December period. The precipitation for the individual experiments described above was also verified for this period and results separated into three different regions across the barrier: upstream lowlands, windward slope, and lee side (Table 1).

For the upstream lowlands, all runs have slight overprediction. The NOSQS, fixed Nos, and the riming conversion threshold (PSACW) simulations do reduce the mean error (ME) slightly, and the fixed Nos has the lowest mean absolute error (MAE). The fixed Nos performs slightly better since it reduced the cloud water in a region aloft, where the other experiments have a larger cloud water overprediction (see section 5b). As a result, there is less accretional growth upstream using a fixed Nos. In contrast, the KESS run depletes more cloud water through autoconversion, thus increasing the ME and MAEs upstream of the barrier. The SICE run performs worst, as too much ice aloft falls out too efficiently.

Over the windward slope the overprediction is widespread among the various sensitivity runs. Those simulations, which favor less snow growth aloft (NOSQS and PSACW) or more advection into the lee (COX) performed best, with exception of the warm rain run. The large overprediction in the warm rain suggests that ice processes were important in properly simulating the distribution of precipitation across the barrier. In contrast, when too much ice is produced aloft, such as in the SICE and fixed Nos runs, there are more dramatic overpredictions along the windward slope. The PSACW run performs best for the windward slope region since it reduces the amount of snow aloft, and the increased graupel it generated fell out in more correct locations.

In the lee side of the Cascades, there is also positive ME associated with too much precipitation spillover in all runs except the warm rain simulation. In contrast, the warm rain has an underprediction problem, which suggests that having some ice spillover is important. Those simulations with larger amounts of snow aloft relative to the control (fixed Nos, SICE, and COX) have a larger spillover problem in the lee, while runs with less snow aloft (NOSQS and PSACW) tend to reduce the spillover overprediction. The KESS is similar to the CTL since there was little cloud water in the lee to impact the simulation. None of the ice sensitivity runs are able to solve the overprediction in the lee at all stations, so the warm rain scheme actually has a lower MAE than all other simulations.

Overall, the most accurate simulation for all locations and amounts is PSACW, suggesting that the increased riming aloft helped the predictions. It is clear those schemes that introduce more ice aloft (fixed Nos and SICE) worsen the simulation. There is no simulation that dramatically outperformed others on average, and the fact that warm rain has the lowest MAE in the lee suggests that more work is needed to improve the BMPs.

b. Cloud water verification

Garvert et al. (2005b) verified against the control MM5 simulation the cloud water using the NOAA P-3 aircraft between 2300 UTC 13 December and 0100 UTC 14 December as it completed a series of north–south legs over the windward Cascades (cf. Fig. 1). Figure 15a shows the cloud water verification for the Nos Reisner-2 experiments, while the other experiments are shown in Fig. 15b. All experiments overpredict cloud water over the lower windward slope (by 0.2 g m−3) at altitudes of 2–3 km, while there is underprediction (by 0.1 g m−3) above the crest between 3 and 4 km. All simulations are close to the observed 40 km to the east of the crest. Using a fixed Nos reduces the overprediction over the lower windward slope since more water vapor goes to depositional snow growth (cf. Fig. 10a); however, it more dramatically underpredicts at higher altitudes over the crest, since the deposition removes too much available saturated water. In contrast, using a variable NOSQS results in less snow growth (Fig. 10b), so the overpredictions are worse at lower levels and are less above the crest. Reducing the fall speed for snow in the COX run and allowing for more riming in PSACW also increases the overpredictions at lower levels, while there is little change compared to the CTL over the crest and immediate lee. NOSQS, PSACW, and COX all favor less snow aloft, therefore less cloud water is depleted. Using a Kessler autoconversion results in less overprediction at lower levels, but there is little change compared to the CTL near the crest.

c. Snow verification

Garvert et al. (2005b) presents the ice verification from the CTL run using the snow and ice concentrations observed by the NOAA P-3 and Convair aircrafts. The ice verification was revisited in this paper using the Convair data to show the impact of the sensitivity experiments. The methods used to quantify the aircraft number concentrations and mixing ratios are described in Woods et al. (2005) and Garvert et al. (2005b).

Figure 16 and Table 2 show the model ice-number verification results for two Convair flight legs at 6.0 and 4.9 km ASL over the windward slope of the Cascades (see Fig. 2 of Garvert et al. 2005b for locations). As noted in Garvert et al. (2005b), since the back (western) edge of the upper-level precipitation shield was approximately 30 min too fast in the model relative to the Convair over the Willamette Valley, a 16–19-h average was applied to the model data to focus on the prefrontal stratiform precipitation shield. It is encouraging that the observed snow distribution aloft is primarily exponential as parameterized in the model using a Marshall–Palmer distribution (Fig. 16). However, at 6.0 km ASL (Figs. 16a,b), the observed size distribution has a significantly steeper slope than all experiments, and the model mixing ratios are 2–3 times larger than observed (Table 2). The NOSQS experiment has the best snow mixing ratio prediction; however, it has a much worse (broader) slope distribution, which dramatically underpredicts smaller ice particles. Since the NOSQS number concentration is close to the observed in the 2–3-mm diameter range, for which a large portion of the snow mass is located, it generates a better overall snow mixing ratio. Meanwhile, there is little difference between a fixed Nos and NosT (CTL), since both intercept formulations are nearly the same at −18°C. The PSACW and Kessler experiments are also very similar to the CTL, while the SICE and Cox have a slightly broader distribution than the CTL at the larger sizes, resulting in greater snow overpredictions.

The results are similar when the Convair descended to 4.9 km ASL and the observed snow mixing ratios increased by a factor of 2 (Figs. 16b,c; Table 2). For all experiments, the model slope distributions (λs) became much less steep between 6.0 and 4.9 km ASL, while there was little change in the observed slope. At these warmer temperatures, the fixed Nos now has a slightly steeper distribution than the CTL. The NOSQS approach again produces the best mixing ratio of all experiments, but its size distribution is much too broad; therefore, it obtains a better mixing ratio for the wrong reason. The additional riming in the PSACW experiment was able to reduce the snow number and mass concentrations somewhat relative to the CTL, which is consistent with its better surface precipitation verification noted in section 5a.

Overall, it is apparent that all model experiments overpredict snow number concentrations at moderate to large sizes, resulting in excessive spillover of snow into the immediate lee of the Cascades. Interestingly, the observed slope intercept increases with increasing temperature (Table 2), while the simulated intercept decreases. Therefore, none of the Nos approaches are able to produce this additional snow number concentration at midlevels. The observed slope intercept does decrease below 4.0 km in the P-3 data in an area of snow aggregation (Woods et al. 2005), but model overprediction was still prevalent (Garvert et al. 2005b).

6. Conclusions

This study investigated the 13–14 December 2001 IMPROVE-2 sensitivities of surface precipitation, bulk microphysical processes, and water/ice mixing ratios to selected parameters within the Reisner-2 bulk microphysical parameterization (BMP) of MM5 using a grid spacing of 1.33 km. Unlike previous modeling case studies of orographic precipitation over the Cascades (Colle and Mass 2000), the Reisner-2 BMP was fully diagnosed by completing a detailed microphysical budget. Each microphysical process from Reisner-2 was output for a box over the windward slope of the Cascades and normalized by the water vapor loss (WVL) rate. This approach allows one to quantify the water mass movement between hydrometeor species, thereby allowing diagnosis of how changes made to a BMP can alter the other processes and resulting surface precipitation.

During the 2300 UTC 13 December to 0100 UTC 14 December period, the largest water vapor loss rate over the windward slope of the Cascades in the 1.3-km domain was associated with condensation (73% of WVL) and snow deposition (24%). The cloud water led to accretion of cloud water by rain (scales as 27% of WVL), which resulted in more than half the rain fallout integrated over the windward slope. A large fraction of cloud water also led to riming of snow and graupel over the windward ridges, with graupel fallout and melt (19% of WVL) contributing to the second most important source of windward surface precipitation. The snow deposition was largest over the middle windward slope around 550 mb, and it extended 20–30 km upwind of the barrier in response to the deep mountain circulation. Snow fallout and melt was relatively small on average (6% of WVL) along the windward slope because of the significant spillover into the immediate lee.

Many of the rain gauge overpredictions in the immediate lee of the Cascades as well as some of the windward valleys were associated with localized areas of snowmelt, with subsidence in the lee of windward ridges and the Cascade crest bringing excessive amounts of snow to the surface. This result was also highlighted in Garvert et al. (2005b), which showed that Reisner-2 produced nearly twice as much snow as observed over the windward slope at midlevels.

Several different parameters were evaluated within the Reisner-2 BMP. The surface precipitation was most sensitive to the snow size distribution and fall speed, while decreasing the riming threshold for snow to graupel conversion had the greatest positive impact on the precipitation forecast. The partition between condensation and deposition and the resulting surface precipitation is strongly affected by the method used to define the snow intercept parameter (Nos). Using a fixed Nos approach favors more smaller snow particles and fewer larger ones at warmer temperatures, so deposition of snow aloft was more than twice as large as using a Nos approach that is a function of mixing ratio. A fixed Nos reduced the precipitation by 5–10 mm (15%) over many areas of the upper windward slope because of a decrease of graupel and rain fallout; however, the additional snow aloft resulted in more overprediction in the immediate lee. This is consistent with a separate simulation using a simple ice scheme, which includes no supercooled water and a fixed Nos approach, and this resulted in the worst precipitation simulation. Having some ice and snow aloft is important, since a warm rain run with no ice resulted in too little precipitation in the immediate lee.

All Reisner-2 experiments overpredicted cloud water over the lower windward slopes, produced too little cloud water over the crest, and had too much ice at moderate-to-large sizes aloft. As a result, none of the experiments resulted in dramatic improvements in the precipitation forecast. The experiments do suggest that too much cloud water was being depleted as a result of excessive snow growth aloft over the windward slopes, and there were many more smaller ice particles observed at midlevels than in the model. This suggests that the ice growth rates in the model were greater than observed and the model broadened the size distribution too rapidly; therefore, model growth processes such as deposition, riming, and aggregation need closer analysis in future studies. Simulating the proper ice distribution and riming aloft is also dependent on the Nos parameter for snow, so this parameter may need a more advanced relationship than just a function of temperature or mixing ratio. Finally, additional field cases will be analyzed to determine the generality of the model verification and microphysical sensitivities presented in this study.

Acknowledgments

This research was supported by the National Science Foundation (ATM-0094524). This work benefited from several useful discussions with Greg Thompson on recent changes made to the Reisner-2 MM5 code. Use of the MM5 was made possible by the Microscale and Mesoscale Meteorological (MMM) Division of the National Center for Atmospheric Research (NCAR), which is supported by the National Science Foundation.

REFERENCES

  • Benjamin, S. G., , and N. L. Seaman, 1985: A simple scheme for objective analysis in curved flow. Mon. Wea. Rev., 113 , 11841198.

  • Berry, E. X., , and R. L. Reinhardt, 1974: An analysis of cloud drop growth by collection. Part IV: A new parameterization. J. Atmos. Sci., 31 , 21272135.

    • Search Google Scholar
    • Export Citation
  • Brown, P. R. A., , and H. A. Swann, 1997: Evaluation of key microphysical parameters in three-dimensional cloud-model simulation using aircraft and multiparameter radar data. Quart. J. Roy. Meteor. Soc., 123 , 22452275.

    • Search Google Scholar
    • Export Citation
  • Bruintjes, R. T., , T. L. Clark, , and W. D. Hall, 1994: Interactions between topographic airflow and cloud/precipitation development during the passage of a winter storm in Arizona. J. Atmos. Sci., 51 , 4867.

    • Search Google Scholar
    • Export Citation
  • Burrows, D. A., 1992: Evaluation of a two-dimensional kinematic cloud model using data from a central Sierra Nevada orographic cloud system. J. Appl. Meteor., 31 , 5163.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., 2004: Sensitivity of orographic precipitation to changing ambient conditions and terrain geometries: An idealized modeling perspective. J. Atmos. Sci., 61 , 588606.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and C. F. Mass, 2000: The 5–9 February 1996 flooding event over the Pacific Northwest: Sensitivity studies and evaluation of the MM5 precipitation forecasts. Mon. Wea. Rev., 128 , 593617.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and Y. Zeng, 2004a: Bulk microphysical sensitivities and pathways 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., , and Y. Zeng, 2004b: Bulk microphysical sensitivities and pathways within the MM5 for orographic precipitation. Part II: Impact of different bulk schemes, barrier width, and freezing level. Mon. Wea. Rev., 132 , 28022815.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , K. J. Westrick, , and C. F. Mass, 1999: Evaluation of the MM5 and Eta-10 precipitation forecasts over the Pacific Northwest during the cool season. Wea. Forecasting, 14 , 137154.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , C. F. Mass, , and K. J. Westrick, 2000: MM5 precipitation verification over the Pacific Northwest during the 1997–1999 cool seasons. Wea. Forecasting, 15 , 730744.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , J. B. Olson, , and J. S. Tongue, 2003: Multi-season verification of the MM5. Part II: Evaluation of high resolution precipitation forecasts over the northeast U.S. Wea. Forecasting, 18 , 458480.

    • Search Google Scholar
    • Export Citation
  • Cox, G. P., 1988: Modeling precipitation in frontal rainbands. Quart. J. Roy. Meteor. Soc., 114 , 115127.

  • Cox, J. A. W., , W. J. Steenburgh, , D. E. Kingsmill, , J. C. Shafer, , B. A. Colle, , O. Bousquet, , B. F. Smull, , and H. Cai, 2005: The kinematic structure of a Wasatch Mountain winter storm during IPEX IOP3. Mon. Wea. Rev., 133 , 521542.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46 , 30773107.

    • 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.

  • Fletcher, N. H., 1962: Physics of Rain Clouds. Cambridge University Press, 386 pp.

  • Garvert, M. F., , B. A. Colle, , and C. F. Mass, 2005a: The 13–14 December 2001 IMPROVE-2 event. Part I: Synoptic and mesoscale evolution and comparison with a mesoscale model simulation. J. Atmos. Sci., 62 , 34743492.

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

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

  • Grell, G. A., , J. Dudhia, , and D. R. Stauffer, 1994: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 138 pp. [Available from National Center for Atmospheric Research, PO Box 3000, Boulder, CO 80307.].

  • Hallett, M. F., , and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249 , 2628.

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

    • Search Google Scholar
    • Export Citation
  • Houze Jr., R. A., , P. V. Hobbs, , and P. H. Herzegh, 1979: Size distributions of precipitation particles in frontal clouds. J. Atmos. Sci., 36 , 156162.

    • Search Google Scholar
    • Export Citation
  • Hsie, E. Y., , R. A. Anthes, , and D. Keyser, 1984: Numerical simulation of frontogenesis in a moist atmosphere. J. Atmos. Sci., 41 , 25812594.

    • Search Google Scholar
    • Export Citation
  • Kessler, E., 1969: On the Distribution and Continuity of Water Substance in Atmospheric Circulations. Meteor. Monogr., No. 32, Amer. Meteor. Soc., 84 pp.

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

  • Medina, S., , and R. A. Houze Jr., 2003: Air motions and precipitation growth in Alpine storms. Quart. J. Roy. Meteor. Soc., 129 , 345371.

    • Search Google Scholar
    • Export Citation
  • Medina, S., , B. R. Smull, , and R. A. Houze Jr., 2005: Cross-barrier flow during orographic precipitation events: Results from MAP and IMPROVE. J. Atmos. Sci., 62 , 35803598.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., , and W. R. Cotton, 1992: Evaluation of the potential for wintertime quantitative precipitation forecasting over mountainous terrain with an explicit cloud model. Part I: Two-dimensional sensitivity experiments. J. Appl. Meteor., 31 , 2650.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., , P. J. DeMott, , and W. R. Cotton, 1992: New primary ice-nucleation parameterization in an explicit cloud model. J. Appl. Meteor., 31 , 708721.

    • Search Google Scholar
    • Export Citation
  • Murakami, M., 1990: Numerical modeling of dynamical and microphysical evolution of an isolated convective cloud—The 19 July 1981 CCOPE cloud. J. Meteor. Soc. Japan, 68 , 107128.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, R. M., and Coauthors, 1992: Winter Icing and Storm Projects (WISP). Bull. Amer. Meteor. Soc., 73 , 951974.

  • Rasmussen, R. M., , I. Geresdi, , G. Thompson, , K. Manning, , and E. Karplus, 2002: Freezing drizzle formation in stably stratified layer clouds: The role of radiative cooling of cloud droplets, cloud condensation nuclei, and ice initiation. J. Atmos. Sci., 59 , 837860.

    • Search Google Scholar
    • Export Citation
  • Rauber, R. M., 1992: Microphysical structure and evolution of a Sierra Nevada shallow orographic cloud system. J. Appl. Meteor., 31 , 324.

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

    • Search Google Scholar
    • Export Citation
  • Reynolds, D. W., , and A. S. Dennis, 1986: A review of the Sierra Cooperative Pilot Project. Bull. Amer. Meteor. Soc., 67 , 513523.

  • Rotunno, R., , and R. Ferretti, 2001: Mechanisms of intense Alpine rainfall. J. Atmos. Sci., 58 , 17321749.

  • Rutledge, S. A., , and P. V. Hobbs, 1983: The mesoscale and microscale structure and organization of clouds and precipitation in mid-latitude cyclones: A model for the “seeder-feeder” process in warm-frontal rainbands. J. Atmos. Sci., 40 , 11851206.

    • Search Google Scholar
    • Export Citation
  • Smith, R. B., , and I. Barstad, 2004: A linear theory of orographic precipitation. J. Atmos. Sci., 61 , 13771391.

  • Stauffer, D. R., , and N. L. Seaman, 1990: Use of four-dimensional data assimilation in a limited area mesoscale model. Part I: Experiments with synoptic-scale data. Mon. Wea. Rev., 118 , 12501277.

    • Search Google Scholar
    • Export Citation
  • Stauffer, D. R., , N. L. Seaman, , and F. S. Binkowski, 1991: Use of four-dimensional data assimilation in a limited-area mesoscale model. Part II: Effects of data assimilation within the planetary boundary layer. Mon. Wea. Rev., 119 , 734754.

    • Search Google Scholar
    • Export Citation
  • Stoelinga, M., and Coauthors, 2003: Improvement of Microphysical Parameterization through Observational Verification Experiments. Bull. Amer. Meteor. Soc., 84 , 18071826.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., , R. M. 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
  • 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

APPENDIX

Abbreviation and Description of Each Microphysical Process in the Reisner-2 Scheme

i1520-0469-62-10-3535-ta01

Fig. 1.
Fig. 1.

Model precipitation (every 5 mm) from the 1.33-km domain between 2300 UTC 13 Dec to 0100 UTC 14 Dec 2001 (23–25 h). Terrain is gray shaded every 500 m using the inset scale. The bold boxed region represents the area of the microphysical budget. The dashed box represents the location of the average cross section. Line segments C1–C4 are the NOAA P-3 aircraft tracks used for the cloud water verification in section 5b.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 2.
Fig. 2.

Microphysical flowchart for the Reisner-2 scheme. The circles represent the various water species (water vapor, cloud water, cloud ice, rain, snow, and graupel), and the arrows are the processes that link the species (see the appendix for the list of processes).

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 3.
Fig. 3.

(a) Average cross section (dashed box in Fig. 1) for the CTL (run 1) showing snow (gray) and graupel (black) every 0.20 g kg−1, rain (thick black) every 0.10 g kg−1, and circulation vectors in the cross section. (b) Same simulation as (a) except for cloud water (black) every 0.10 g kg−1 and cloud ice (gray) every 0.02 g kg−1.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 4.
Fig. 4.

Flowchart of the microphysical processes between 23 and 25 h (2300–0100 UTC) of the control run for the solid box in Fig. 1. The values shown are the ratio of each average microphysical process rate to the total WVL rate (cond + sdep + gdep + idsn + idep) within the box. For example, cond = 73.21 means that 73.21% of the WVL results in condensation, and gsacw = 6.47 means that the movement of cloud water mass to graupel is equivalent to 6.47% of the total WVL rate. The processes are listed in the appendix. The sum of all the microphysical process tendencies for each species is given by (wv, cw, r, ci, g, and s). This sum does not include horizontal advection and diffusion/divergence, which are labeled as hadv and other, respectively. The fallout tendency of rain (rprc), snow (sprc), graupel (gprc), and cloud ice (iprc) are also shown. Microphysical processes greater than 4% of the WVL rate are in bold.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 5.
Fig. 5.

Cross section of the spatial distribution of the major microphysical processes contributing to snow production, showing the ratio of (a) sdep (gray) and icns (black), (b) saci (gray) and ssacw (black) to the total snow production rate from these processes at each point contoured every 20%. (c), (d) The same as (a) except for graupel production showing (c) ggacw (black) and scng (gray) every 5%, and (d) gsacw (black) and gacr (gray) every 20%. (e), (f) The same as (a) except rain production showing racw (gray), gmlt (thick black), smlt (black), and (f) gmlt (black) every 20%. See the appendix for process abbreviations. The 0° and −15°C isotherms are shown in (a) for reference.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 6.
Fig. 6.

(a) Snow deposition (every 10 × 105 g kg−1 s−1) and winds (1 full barb = 10 kt) averaged between 450 and 550 mb using 15-min microphysical output from the 1.33-km domain for 2300–0100 UTC 14 Dec. (b) Same as (a) except for the accretion of cloud water by snow averaged between 500 and 600 mb. Terrain from the 1.33-km domain is shaded for reference

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 7.
Fig. 7.

Same as Fig. 6, except for the 600–700-mb layer showing graupel growth via (a) accretion of snow by cloud water (every 10 × 105 g kg−1 s−1) and (b) accretion of cloud water by graupel.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 8.
Fig. 8.

Same as Fig. 6, except for the 850–750-mb layer showing rain growth via (a) accretion of rain by cloud water (every 10 × 105 g kg−1 s−1), (b) snowmelt, and (c) graupel melt. A transact of rain gauge locations are shown across the Cascades, with model precipitation within 30% of the observed shown by a X, greater than 130% of observed by an o, and greater than 180% of observed by an O for the 2000 UTC 13 Dec to 0200 UTC 14 Dec period.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 9.
Fig. 9.

(a) Cross section of differences between fixed Nos experiment (run 2) and the CTL (run 1) showing snow (gray) every 0.10 g kg−1, graupel (thick solid) every 0.05 g kg−1, and cloud water (thin solid) every 0.05 g kg−1. (b) Same as (a), except for the NOSQS (run 3) minus CTL.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 10.
Fig. 10.

(a) Same as Fig. 4, except for run 2 (fixed Nos). (b) Same as (a), except for run 3 (NOSQS).

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 11.
Fig. 11.

(a) Storm total precipitation (every 15 mm) for the 1.33-km domain between 1400 UTC 13 Dec and 0800 UTC 14 Dec 2001 and 1.33-km terrain is shaded for reference. (b) Difference between the NOSF (run 2) and the CTL experiment every 10 mm starting at ±5 mm, with negative values dashed. (c) Same as (b), except for the Nosq (run 3). (d) Same as (b), except for the COX (run 4). (e) Same as (b), except for the PSACW (run 5). (f) Same as (b), except for the KESS (run 6). (g) Same as (b), except for the SICE and contoured intervals are every 20 mm. (h) Same as (g), except for the warm rain scheme.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 12.
Fig. 12.

Same as Fig. 9, except for the COX (run 4) minus CTL. Snow trajectories are also shown for the CTL (black) and COX (gray).

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 13.
Fig. 13.

(a) Same as Fig. 9, except for PSACW–CTL. (b) Same as (a), except for KESS–CTL.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 14.
Fig. 14.

(a) Average cross section for the SICE (run 7) showing snow (gray) every 0.20 g kg−1, and rain (thick black) every 0.10 g kg−1, and cloud ice (thin black) every 0.05 g kg−1. (b) Same as (a), except for the warm rain scheme (run 8) and cloud water is shown rather than cloud ice.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 15.
Fig. 15.

(a) Cloud water verification averaged for each of the five north–south NOAA P-3 legs (C1–C5) shown in Fig. 1 between 2300 UTC 13 Dec and 01 UTC 14 Dec. The observed cloud water amounts (g m−3) is shown by the solid black line and each of the Nos experiments are listed in the inset. (b) Same as (a) except for the KESS, COX, and PSACW experiments. The average N–S altitude (m) of the P-3 for each leg is given by the large open circles in meters, and the average N–S topography height is shaded below in meters.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Fig. 16.
Fig. 16.

(a) Ice number concentrations from the Convair aircraft at 6.0 km ASL over the windward slope (black open circles and black-dashed best fit) and MM5 number concentrations derived using the fixed Nos (gray dashed), NosT (solid black), and NOSQS (gray solid) experiments. (b) Same as (a), except for the PSACW (gray solid), Cox (gray dashed), and simple ice (short dashed). (c) Same as (a), except for 4.9 km ASL. (d) Same as (b), except for 4.9 km ASL.

Citation: Journal of the Atmospheric Sciences 62, 10; 10.1175/JAS3552.1

Table 1.

List of microphysical simulations, mean, and mean absolute precipitation errors (mm) from the 1.33-km domain experiments for the 1400 UTC 13 Dec to 0800 UTC 14 Dec period (14–32 h) for the upstream lowlands, windward slope, lee side, and all locations. The mean observed precipitation (Pobs) is given in the table heading for each region. The simulations with the best mean errors and means absolute errors are bold. The windward precipitation efficiency (%) for the 2300–0100 UTC period is also shown.

Table 1.
Table 2.

A comparison between the observed values for snow mixing ratio, slope intercept, and slope of the snow number concentration from the Convair aircraft vs the sensitivity runs. The observed aircraft values and the error bars for the snow mixing ratios were obtained from Garvert et al. (2005b).

Table 2.
Save