• American Meteorological Society, 1992: Planned and inadvertent weather modification, a policy statement of the American Meteorological Society. Bull. Amer. Meteor. Soc.,73, 331–337.

  • Banta, R. M., 1990: The role of mountain flows in making clouds. Atmospheric Processes in Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 229–283.

  • Benjamin, S. G., K. A. Brewster, R. Brummer, B. F. Jewett, and T. W. Schlatter, 1991: An isentropic three-hourly data assimilation system using CARS aircraft observations. Mon. Wea. Rev.,119, 888–906.

  • Bradley, M. M., 1985: The numerical simulation of orographic storms. Ph.D. dissertation, University of Illinois, Urbana–Champaign, 263 pp.

  • Bruintjes, R., T. Clark, and W. Hall, 1994: Interaction between topographic airflow and cloud and precipitation development during the passage of a winter storm in Arizona. J. Atmos. Sci.,51, 48–67.

  • Bruintjes, R., T. Clark, and W. Hall, 1995: The dispersion of tracer plumes in mountainous regions in central Arizona: Comparisons between observations and modeling results. J. Appl. Meteor.,34, 971–988.

  • Clark, T. L., 1977: A small-scale numerical model using a terrain following coordinate transformation. J. Comput. Phys.,24, 186–215.

  • Clark, T. L., and W. D. Hall, 1991: Multidomain simulations of the time-dependant Navier–Stokes equation: Benchmark error analyses of nesting procedures. J. Comput. Phys.,92, 456–481.

  • Coen, J. L., R. T. Bruintjes, W. D. Hall, T. L. Clark, and R. F. Reinking, 1996a: Numerical simulation of wintertime precipitation development in gravity-wave and upslope flow in Arizona’s Verde Valley. Proc. 12th Int. Conf. on Clouds and Precipitation, Zurich, Switzerland, ICCP/IAMAS, 778–781.

  • Coen, J. L., R. T. Bruintjes, R. F. Reinking, and G. Thompson, 1996b: Summary of the Arizona winter storm case: 5–6 March 1995. Fourth Int. Cloud Modeling Workshop, Clermont-Ferrand, France, WMO/TD-No. 901, 37–44.

  • Heggli, M. F., and R. M. Rauber, 1988: The characteristics and evolution of supercooled liquid water in wintertime storms over the Sierra Nevada: A summary of microwave radiometric measurements taken during the Sierra Cooperative Pilot Project. J. Appl. Meteor.,27, 989–1015.

  • Heimbach, J. A., Jr., and W. D. Hall, 1994: Applications of the Clark model to winter storms over the Wasatch Plateau. J. Wea. Mod.,26, 1–11.

  • Hogg, D. C., F. O. Guiraud, J. B. Snider, M. T. Decker, and E. R. Westwater, 1983: A steerable dual-channel microwave radiometer for measurement of water vapor and liquid in the troposphere. J. Climate Appl. Meteor.,22, 789–806.

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

  • Klimowski, B. A., R. Becker, E. A. Betterton, R. Bruintjes, T. L. Clark, W. D. Hall, B. W. Orr, R. A. Kropfli, P. Piironen, R. F. Reinking, D. Sundie, and T. Uttal, 1998: The 1995 Arizona Program: Toward a better understanding of winter storm precipitation development in mountainous terrain. Bull. Amer. Meteor. Soc.,79, 799–813.

  • Koenig, L. R., and F. W. Murray, 1976: Ice-bearing cumulus cloud evolution: Numerical simulation and general comparison against observations. J. Appl. Meteor.,15, 747–762.

  • Kropfli, R. A., S. Y. Matrosov, T. Uttal, B. W. Orr, A. S. Frisch, K. A. Clark, B. W. Bartram, R. F. Reinking, J. B. Snider, and B. E. Martner, 1995: Cloud physics studies with 8-mm-wavelength radar. Atmos. Res.,35, 299–313.

  • Matrosov, S. Y., R. F. Reinking, R. A. Kropfli, and B. W. Bartram, 1995: Identification of ice hydrometeor types from elliptical polarization radar measurements. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 539–541.

  • Matrosov, S. Y., R. F. Reinking, R. A. Kropfli, and B. W. Bartram, 1996: Estimation of ice hydrometeor types and shapes from radar polarization measurements. J. Atmos. Oceanic Technol.,13, 85–96.

  • Reinking, R. F., 1995: An approach to remote sensing and numerical modeling of orographic clouds and precipitation for climatic water resources assessment. Atmos. Res.,35, 349–367.

  • Reinking, R. F., S. Y. Matrosov, R. T. Bruintjes, B. E. Martner, and R. A. Kropfli, 1995: Further comparison of experimental and theoretical radar polarization signatures due to ice hydrometeor growth habit. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 47–49.

  • Reinking, R. F., S. Y. Matrosov, and R. T. Bruintjes, 1996: Hydrometeor identification with elliptical polarization radar: Applications to glaciogenic cloud seeding. J. Wea. Mod.,28, 6–18.

  • Reinking, R. F., S. Y. Matrosov, R. T. Bruintjes, and B. E. Martner, 1997: Identification of hydrometeors with elliptical and linear polarization Ka-band radar. J. Appl. Meteor.,36, 322–339.

  • Reynolds, D. W., 1988: A report on winter snowpack augmentation. Bull. Amer. Meteor. Soc.,69, 1290–1300.

  • Reynolds, D. W., 1996: The effects of mountain lee waves on transport of liquid propane-generated ice crystals. J. Appl. Meteor.,35, 1435–1456.

  • Smith, R., J. Paegle, T. Clark, W. Cotton, D. Curran, G. Forbes, J. Marwitz, C. Mass, J. McGinley, H.-L. Pan, and M. Ralph, 1997:Local and remote effects of mountains on weather: Research needs and opportunities. Bull. Amer. Meteor. Soc.,78, 877–892.

  • Super, A. B., and B. A. Boe, 1988: Wintertime cloud liquid water observations over the Mogollon Rim of Arizona. J. Wea. Mod.,20, 1–7.

  • Super, A. B., and E. W. Holroyd III, 1989: Temporal variations of cloud liquid water during winter storms over the Mogollon Rim of Arizona. J. Wea. Mod.,21, 35–40.

  • Westwater, E. R., 1972: Microwave emission from clouds. NOAA TR ERL 219-WPL 18, 43 pp. [Available from Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402.].

  • View in gallery

    Topographic map of the AP95 project area in northern Arizona (elevation in meters MSL, 120-km square, 75-m contour interval). Key sites marked as open squares are Prescott (PRC, 1375 m MSL) just northeast of the Bradshaw Mountains, Mingus Mountain (MNG, 2100 m) at the highest point of the Black Hills, Cottonwood in the Verde Valley (CTW, 1040 m), Sedona (SDA, 1275 m), Schnebly Road Curve (SRC, 1400 m), and site T6 (2025 m). The city of Flagstaff is noted for reference.

  • View in gallery

    (a) 500-mb circulation over North America at 0000 UTC 4 Mar 1995. Synoptic waves in succession are labeled as 3, 4, and 5–6 Mar 1995 for their dates of main effect in the AP95 project area. (b) The 0200 UTC 4 Mar 1995 GOES-7 satellite image of the corresponding succession of eastward-moving cloud bands that affected the project area. A foehn trough is faintly visible in the lee of the Black Hills, as a diffuse grey line just northeast and parallel to the black line, which marks the southwest edge of the project area.

  • View in gallery

    Time-series plots for 2–6 Mar 1995 (UTC) of (a) vertical column-integrated atmospheric water vapor (Pw, cm) and liquid water (LW, mm) over CTW; (b) precipitation rate (R, cm h−1); and (c) surface temperature (T, °C) at site T6 on the Mogollon Rim.

  • View in gallery

    Time series of accumulated precipitation (cm of liquid) from 2 to 6 Mar 1995 (UTC) for three sites in a SW–NW line: Cottonwood in the Verde Valley (CTW), windward slope of the Mogollon Rim (SRC), top of the Mogollon Rim (T6).

  • View in gallery

    The flooded Oak Creek, a few kilometers upstream from Sedona, still overflowing and turbulent as it began to subside on 7 Mar 1995.

  • View in gallery

    (a) Cloud reflectivity Ze (dBZ) from a CTW radar over-the-top RHI scan from azimuth 0° (north, at right) through zenith to south, slightly across the 240° cloud-level wind; showing wave clouds between 4 and 8 km AGL, overriding clouds developed in direct orographic lifting upstream at lower altitudes over the Black Hills. MNG peak is left of center at 8-km range; 0156 UTC 3 Mar 1995, 5-km range rings. (b) Subsequent radar RHI (Ze, dBZ) from azimuth 60° (at right), approximately along the wind; 0239 UTC. (c) Zenith-normalized, path-integrated CLW (mm) corresponding to (b); the indicated azimuth (60°) is that where the SMR RHI scan began at antenna elevation angle β = 0°; β equates to horizontal position, relative to β = 90° (zenith, over CTW).

  • View in gallery

    Wind direction as a function of altitude AGL directly over CTW during gravity wave activity on 4 Mar 1995, from a radar VAD scan, 0149–0152 UTC.

  • View in gallery

    Enlargement of the 0000–1200 UTC 4 Mar 1995 time section of zenith-normalized Pw (cm) and LW (mm) over CTW, and R (cm h−1) at T6, from Figs. 2a,b. Shadowing indicates time by which R lags secondary maxima and minima in Pw and LW, which are superimposed on the broad peak of moisture.

  • View in gallery

    Zenith-normalized CLW (mm) from pairs of SMR RHI scans, as in Fig. 6c, at (a) a peak, (b) an intermediate point, and (c) the trough of a pulsation in Pw and CLW, from times marked respectively as a, b, and c in Fig. 8.

  • View in gallery

    Cloud reflectivity Ze (color scale, dBZ) from radar RHI scans at (a) 0246, (b) 0430, and (c) 0501 UTC 4 Mar 1995, from azimuths 60°, 70°, and 70° (at right), with 5-, 2-, and 2-km range rings, respectively; corresponding respectively to Figs. 9a–c. The temperature was ∼−3°C at 700 mb (range ∼2 km at zenith) and ∼−18°C at 500 mb (range ∼5 km at zenith).

  • View in gallery

    EDR (color scale, dB) corresponding to (a), (b), and (c) in Figs. 9a–c and 10a–c. The melting layer is indicated in (a) and (b).

  • View in gallery

    Cloud at onset of wave regeneration and renewed release of precipitation. (a) RHI scan of Ze (color scale, dBZ) from azimuth 70° (at right), 0631 UTC 4 Mar, 2-km range rings; (b) pair of corresponding SMR RHI scans, 0639–0649 UTC, details as in Fig. 6c.

  • View in gallery

    Sequence of liquid water measurements from SMR RHI scans, 2331–1054 UTC 5–6 Mar 1995. Azimuth 90° is at right in each frame.

  • View in gallery

    (a) Horizontal wind direction (°) and (b) corresponding vertical motion (m s−1) as a function of altitude (km) above CTW, from radar VAD scans at 2331–2332 UTC (labeled “wave suppression”) and 0101–0104 UTC (labeled “wave enhancement”) 5–6 Mar 1995. Cloud top is indicated by the top of each profile.

  • View in gallery

    Cloud reflectivity Ze (color scale, dBZ) of the precipitating wave cloud near maximum development (0535 UTC 6 Mar 1995); over-the-top RHI scan from azimuth 60° (at right); 5-km range rings. The temperature was ∼+1°C at 700 mb (range ∼2 km at zenith, slightly below the wave-cloud base) and ∼−15°C at 500 mb (range ∼5 km at zenith, somewhat below the wave-cloud top).

  • View in gallery

    Measurements from a west–east aircraft spiral descent intersecting the foehn trough, wave crest, and upslope cloud between 0555 and 0635 UTC 6 Mar 1995: (a) cloud-droplet concentration (CONC, cm−3, bar scale, superimposed on flight track plotted as altitude in kilometers MSL vs distance relative to CTW); with insets as a function of time approximately equivalent to distance scale, of (b) LWC (g m−3) from the King probe (light line) and the FSSP (dark line); and (c) air temperature (K).

  • View in gallery

    Sequence from the numerical simulation of the gravity wave and upslope clouds, simulation time 0500–1000 UTC 6 Mar 1995, grid spacing 2.7 km, showing vectors from the horizontal and vertical winds, CLW (g kg−1, line contours, 0.02 interval) and IMC (g kg−1, shaded contours and scale). Altitude Z is in kilometers MSL. The east–west (left–right) vertical cross section extends across the Black Hills near Mingus Mountain, over the Verde Valley near CTW, and over the Mogollon Rim. CTW is in lee (to right of) the Black Hills, near distance S = 35 km.

  • View in gallery

    Simulated vertical cross section of atmospheric temperature (K) at 0530 UTC 6 Mar 1995 as a function of altitude (Z, km MSL), corresponding the cross section used in Fig. 17. The melting level (273 K) over the Mogollon Rim is indicated by the bold dashed line. (Horizontal distance X is 140 km).

  • View in gallery

    Sequence of vertical cross sections at 0500, 0600, 0700, and 0800 UTC from the simulation, illustrating variations of graupel mass content (GMC, g kg−1, shaded contours and scale) and rain mass content (RMC, g kg−1, line contours, 0.02 interval); grid interval, 2.7 km.

  • View in gallery

    Histograms of wave CLW from (a) 4 Mar and (b) 5–6 Mar, excluding CLW ≥ 2.0 mm, and (c) 6 Mar 1995, including all values.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 185 185 6
PDF Downloads 200 200 2

Influences of Storm-Embedded Orographic Gravity Waves on Cloud Liquid Water and Precipitation

View More View Less
  • a NOAA/Environmental Technology Laboratory, Boulder, Colorado
  • | b CIRES, University of Colorado, and NOAA/Environmental Technology Laboratory, Boulder, Colorado
  • | c MMM Division, National Center for Atmospheric Research, Boulder, Colorado
© Get Permissions
Full access

Abstract

This study illustrates opportunities for much improved orographic quantitative precipitation forecasting, determination of orographic cloud seedability, and flash flood prediction through state-of-the-art remote sensing and numerical modeling of gravity wave clouds. Wintertime field observations with multiple remote sensors, corroborated in this and related papers with a mesoscale–cloud scale numerical simulation, confirm that storm-embedded gravity waves can have a strong and persistent influence on orographic cloud liquid water (CLW) and precipitation. Where parallel mountain ridges dominate the landscape, an upwind ridge can force the wave action, and a downwind ridge can receive the precipitation. The 1995 Arizona Program was conducted in such terrain. In the scenario examined, traveling waves cyclically caused prefrontal cross-barrier winds that produced gravity waves. Significant cloud bands associated with the waves carried substantial moisture to the area. With the passage and waning of the cloud bands, vapor influxes (precipitable water Pw) cycled through large changes in magnitude, and prefrontal peaks in Pw coincided with the gravity waves in a succession of episodes during a five-day period. Thus, the cyclic trend in Pw and the magnitudes of peak Pw were simple indicators of wave cloud development. The first two cycles, with minor peak Pw, were precursors. Significant wave clouds first appeared during the second episode. During the final two episodes with large vapor influxes, very deep, precipitating wave clouds were coupled with underlying clouds formed in flow up the mountain slopes to create the prefrontal storms. Rain fell on an existing snowpack on the main recipient ridge and, in the end, produced rapid runoff and flash flooding.

The gravity waves persistently condensed CLW that averaged 0.5 mm and reached 1.0 mm in the first of the main storm episodes, and averaged 1.0 mm and reached 2.0 mm and more in the second (column-integrated values). These values equaled or exceeded the larger of those represented in liquid water climate datasets for orographic cloud systems in other locations in the West, where only the upslope and not the wave component had been examined. The effect of shifts between cross-barrier and barrier-parallel flows was reflected in abrupt buildups and declines in wave CLW, but the gravity wave clouds persisted for a total of 22 h during the two storm periods. In the wave updrafts, the condensation rate regularly exceeded the consumption rate by ice, even though ice was usually present. Conversion to ice consumed and precipitated wave CLW. Pulses of available Pw and wave CLW on a 2- to 4-h timescale, cyclically followed by partial glaciation, produced the precipitation from the wave clouds. Their seeder effect on the upslope feeder clouds was to enhance the total precipitation from the coupled system. Estimates of the liquid water fluxes in comparison with the precipitation rates suggest precipitation efficiencies in the 11%–33% range from the seeder–feeder couplets. The periods of gravity wave forcing contributed some 80% or more of the total precipitation, and trailing fronts produced the remainder.

Several factors derived from the observed availability of CLW determine the potential for precipitation enhancement by seeding wave clouds; these are enumerated. Given demands for improved water supply, the challenge often presented in mountain watersheds of separating seeding opportunities from potential flash flood situations is examined. The results here show that storms that could threaten flash floods can be readily identified by continuous monitoring with polarization radar and in real-time simulations as those with the altitude of the melting level above the elevation of the highest terrain with existing snowpack.

In the sense that orographically generated gravity waves will significantly influence cloud water and precipitation, geographic transferability of the results is indicated by the existence of wave-generating and precipitation-generating parallel ridges in many places throughout the world. The quantitative effects will, of course, depend on particulars of the locale such as nature of the prevalent forcing, available moisture, and physical stature of the ridges.

Corresponding author address: Roger F. Reinking, NOAA/OAR/ETL, R/E/ET6, 325 Broadway, Boulder, CO 80303.

rreinking@etl.noaa.gov

Abstract

This study illustrates opportunities for much improved orographic quantitative precipitation forecasting, determination of orographic cloud seedability, and flash flood prediction through state-of-the-art remote sensing and numerical modeling of gravity wave clouds. Wintertime field observations with multiple remote sensors, corroborated in this and related papers with a mesoscale–cloud scale numerical simulation, confirm that storm-embedded gravity waves can have a strong and persistent influence on orographic cloud liquid water (CLW) and precipitation. Where parallel mountain ridges dominate the landscape, an upwind ridge can force the wave action, and a downwind ridge can receive the precipitation. The 1995 Arizona Program was conducted in such terrain. In the scenario examined, traveling waves cyclically caused prefrontal cross-barrier winds that produced gravity waves. Significant cloud bands associated with the waves carried substantial moisture to the area. With the passage and waning of the cloud bands, vapor influxes (precipitable water Pw) cycled through large changes in magnitude, and prefrontal peaks in Pw coincided with the gravity waves in a succession of episodes during a five-day period. Thus, the cyclic trend in Pw and the magnitudes of peak Pw were simple indicators of wave cloud development. The first two cycles, with minor peak Pw, were precursors. Significant wave clouds first appeared during the second episode. During the final two episodes with large vapor influxes, very deep, precipitating wave clouds were coupled with underlying clouds formed in flow up the mountain slopes to create the prefrontal storms. Rain fell on an existing snowpack on the main recipient ridge and, in the end, produced rapid runoff and flash flooding.

The gravity waves persistently condensed CLW that averaged 0.5 mm and reached 1.0 mm in the first of the main storm episodes, and averaged 1.0 mm and reached 2.0 mm and more in the second (column-integrated values). These values equaled or exceeded the larger of those represented in liquid water climate datasets for orographic cloud systems in other locations in the West, where only the upslope and not the wave component had been examined. The effect of shifts between cross-barrier and barrier-parallel flows was reflected in abrupt buildups and declines in wave CLW, but the gravity wave clouds persisted for a total of 22 h during the two storm periods. In the wave updrafts, the condensation rate regularly exceeded the consumption rate by ice, even though ice was usually present. Conversion to ice consumed and precipitated wave CLW. Pulses of available Pw and wave CLW on a 2- to 4-h timescale, cyclically followed by partial glaciation, produced the precipitation from the wave clouds. Their seeder effect on the upslope feeder clouds was to enhance the total precipitation from the coupled system. Estimates of the liquid water fluxes in comparison with the precipitation rates suggest precipitation efficiencies in the 11%–33% range from the seeder–feeder couplets. The periods of gravity wave forcing contributed some 80% or more of the total precipitation, and trailing fronts produced the remainder.

Several factors derived from the observed availability of CLW determine the potential for precipitation enhancement by seeding wave clouds; these are enumerated. Given demands for improved water supply, the challenge often presented in mountain watersheds of separating seeding opportunities from potential flash flood situations is examined. The results here show that storms that could threaten flash floods can be readily identified by continuous monitoring with polarization radar and in real-time simulations as those with the altitude of the melting level above the elevation of the highest terrain with existing snowpack.

In the sense that orographically generated gravity waves will significantly influence cloud water and precipitation, geographic transferability of the results is indicated by the existence of wave-generating and precipitation-generating parallel ridges in many places throughout the world. The quantitative effects will, of course, depend on particulars of the locale such as nature of the prevalent forcing, available moisture, and physical stature of the ridges.

Corresponding author address: Roger F. Reinking, NOAA/OAR/ETL, R/E/ET6, 325 Broadway, Boulder, CO 80303.

rreinking@etl.noaa.gov

Introduction

An understanding of the cloud properties and processes that lead to orographic precipitation is fundamental to predicting and managing water supplies and hazards from mountain runoff for a large part of the global population. This need is especially acute in Arizona, which is short of water and relies predominantly on winter orographic precipitation. This substantially mountainous state is also susceptible to flash floods, not only from summer convective storms but also from sudden winter runoffs. In the region of the Mogollon Rim, the most important watershed in Arizona, the winter orographic precipitation is strongly influenced by storm-embedded gravity waves induced by flow across quasi-parallel ridges, according to initial studies (Bruintjes et al. 1994, 1995). A need to understand better the cloud physics of such precipitating mountain wave clouds is specifically mentioned in a U.S. Weather Research Program survey (Smith et al. 1997). Building on the initial studies, the 1995 Arizona Program (AP95) was conducted to establish a better understanding of precipitation development and runoff from winter storms in mountainous terrain, particularly as influenced by the gravity waves (Klimowski et al. 1998).

Central to the gravity wave influence on the precipitation process is the wave-generated condensate, that is, the cloud liquid water (CLW) that the waves make available for conversion to precipitation. The conversion may occur by processes proceeding with or without stimulation from cloud seeding. The Arizona Ground Water Management Act (Revised Statue 45-165) authorizes exploration of many water augmentation plans including weather modification. Defining the potential of seeding was therefore an important goal of the AP95. Such precipitation enhancement with existing techniques is in the realm of feasibility (American Meteorological Society 1992). The actual potential is significantly determined by the quantity and persistence of CLW made available to the precipitation process. The need for good seasonal runoff, as well as enhancement of snowpack and runoff by cloud seeding, quite normally go hand in hand with the need for flash flood prediction when the runoff may occur too rapidly. Mountain weather presents such contradictions. In the flash flood situation, two of several important factors are the CLW and the atmospheric conditions that may cause rainfall, rather than snowfall, on existing snowpack. Rain on snowpack, rapid runoff, and flooding occurred on two occasions during AP95.

From the prior atmospheric studies over the Mogollon Rim, which included numerical simulations and aircraft observations, it was possible to formulate the following set of hypotheses:

  1. Storm-embedded, orographically forced gravity waves can develop significant quantities of supercooled CLW.
  2. Conversion to ice will precipitate some of the wave-cloud water, and this precipitation will naturally seed underlying orographic clouds that form directly on the upwind slopes of ridges.
  3. The seeder (wave cloud)–feeder (orographic cloud) couplets will increase the conversion of water condensed in orographic clouds to precipitation, such that total precipitation will be greater than that from either the overriding gravity wave cloud or the orographic cloud alone.
  4. The waves will generate more CLW than naturally formed ice can consume.
  5. Seeding the wave cloud CLW surplus will enhance the couplet and total precipitation.
  6. Persistent, precipitating waves in warm storms may encourage flash floods by adding to rain on existing snowpack.
  7. Key measurements and modeling can separate seeding opportunities from flash flood threats.

This set of hypotheses constitutes the conceptual model for AP95. The concept of hypotheses 1–4 is illustrated in previous publications by a cross section of the ridges and the valley, with simulated wave dynamics and responding cloud physics (Fig. 9 in Bruintjes et al. 1994; Fig. 2 in Klimowski et al. 1998). Measurements to test the hypotheses were made with multiple remote sensors interactively guided with a numerical mesoscale–cloud scale model run in real time and for post-event analysis, all supported by rawinsondes and standard surface instrument systems (Reinking 1995; Klimowski et al. 1998). This paper is focused on the wave-induced CLW, which is the microphysical parameter at the core of the conceptual model. A 2–6 March 1995 episode of wave clouds and storms is analyzed, utilizing the measurements and some intercomparisons with a numerical simulation for 6 March. The storms presented substantial gravity wave forcing and CLW, and eventually caused one of the two flash floods during the project.

Background

Topography, precipitation climate, and key observation sites

The AP95 study area in northern Arizona is characterized by three quasi-parallel mountain ranges reaching slightly above 2 km above mean sea level (MSL) separated by two broad valleys that bottom slightly under 1 km MSL. The orientation of the ridges and valleys is predominantly northwest–southeast. Along a perpendicular transect from the southwest are the Bradshaw Mountains, Prescott Valley, Black Hills with a high point at Mingus Mountain (MNG), Verde Valley, and the Mogollon Rim (Fig. 1). Southwesterly winds approximately perpendicular to the ridges prevail during most storm periods. The ridges create the organized gravity waves and other complex airflows that strongly link cloud processes and the precipitation distribution and quantity to the topography. The influence of direct upslope forcing has been examined elsewhere in many studies, but the influence of overriding gravity waves on CLW and precipitation have received little attention, particularly via observations. Banta (1990) summarized the numerical sensitivity experiments of Bradley (1985), which estimate the effects waves have on orographic precipitation. Bradley determined that the enhanced upward motion provided by waves can produce higher liquid water contents than the simple orographic lifting models, and that the wave-generated liquid can be effective in precipitation production. Bradley’s study focuses on precipitation on the west coast of Wales. Just as in Arizona, much of that precipitation results from moist southwesterly flow in the warm sectors of cyclonic storm systems. Bradley found that the atmosphere was thermodynamically stable in seven out of eight such cases, indicating a prevalent influence of mountain waves. In modeling studies parallel to those of Bruintjes et al (1994) for Arizona, Heimbach and Hall (1994) simulated storms over the Wasatch Plateau of Utah. The results indicated significant generation of CLW in storm-embedded, mountain-forced gravity waves over the valley upwind of the Plateau. Primarily from rawinsonde studies, Reynolds (1996) determined that gravity waves in the lee of the Sierra Nevada were common to most winter storms, and that the consequent strong descent in the lee had a detrimental effect on the growth of ice particles generated (by seeding) on the upwind crest. The wave updrafts beyond the troughs were measured, but not the CLW generated in the updrafts or the microphysical effects of that component of cloud water.

For direct upslope orographic clouds in Utah, Colorado, and California, Heggli and Rauber (1988) and others observed CLW concentrated in the lowest kilometer above the barriers, and Reynolds (1988) concluded that (1) the column-integrated CLW was less than 0.2 mm 85% of the time, but (2) only 0.1 or 0.2 mm is significant to orographic precipitation if persistent. However, case studies from the climatologies such as those reported Heggli and Rauber and others show that storm periods with many hours with CLW between 0.5 and 1.0 mm do occur. A two-month, 1987 climatology of vertically pointing radiometer measurements on the Mogollon Rim shows a maximum value of only 0.55 mm, but it based on 1-h averaging, which masks higher values of shorter durations (Super and Boe 1988). It is noted that three storms provided 75% of the liquid flux over the area during that period; AP95 was similar, and this is likely typical. From CLW measurements made with a vertically pointing microwave radiometer in those 1987 storm cases, Super and Holroyd (1989) found “abundant CLW” (consistently, ∼0.3–0.5 mm, with peaks to ∼1 mm), although the measurement at their site would have been from the upslope cloud component or a combination of upslope and advected wave components. In comparison with such case studies, the following analysis show as much and more CLW in the gravity wave clouds well above the barrier slopes, over the Verde Valley. Notably, plan position indicator (PPI) scanning with the radiometer revealed that the liquid water (LW) developed southwest of the rim 53% of the time (Super and Boe 1988). This is the direction where wave clouds as well as the predominant upslope forcing occur. An obvious key to the new measurements is use of the range-height-indicator (RHI)-scanning scanning microwave radiometer (SMR). None of the other studies used the advantage of RHI scanning to measure the distribution of liquid in a vertical plane from ridge to ridge, through any wave clouds.

In terms of the conceptual model relative to the actual project area topography, prefrontal flow is forced upward over each of the barriers (Fig. 1), inducing the gravity wave motions through a considerable depth of the troposphere. Direct orographic lifting produces CLW on the windward slope of the Black Hills. In the lee of this ridge, the air in the waves first descends abruptly in the foehn trough and then rises abruptly downwind over the Verde Valley to produce the wave cloud and its CLW over the intervening Verde Valley. If this cloud generates precipitation, a seeder (wave cloud)–feeder (upslope cloud) couplet forms with a second upslope cloud to enhance precipitation on the windward slopes of the subsequent ridge and key watershed, the broad Mogollon Rim. This recipient rim annually collects some 250 cm of precipitation (liquid equivalent, 15–30 cm) during the cold season. Much less precipitation, usually rain, falls in the Verde Valley. Cottonwood, Arizona (CTW) (Fig. 1), on the southwestern edge of this valley in the rain shadow of the Black Hills, typically receives under 2 cm monthly from January through March, whereas Sedona (SDA), on the northeastern edge, receives 4.5 cm monthly.

The main remote-sensing site was at CTW, at 1.0 km MSL. Precipitation-recording sites, including CTW, Schnebly Road Curve (SRC), and T6 used in this study (Fig. 1) and others were oriented along two lines, one perpendicular to the terrain extending from MNG northeast through CTW and T6, and one from CTW southeast along the floor of the Verde Valley. CTW is optimally located just downwind of the usual foehn trough, so the storm-embedded wave clouds were monitored with minimal impact of precipitation on the sensors.

Key instruments and numerical model

Most of the remote-sensing data for this study are from two instruments developed by the NOAA Environmental Technology Laboratory (ETL): an SMR (Hogg et al. 1983), and a cloud-sensing, Ka-band (8.66 mm), dual-polarization Doppler radar (Kropfli et al. 1995). The SMR measured the path-integrated values of water vapor (WV) and total liquid water. The radar provided measurements of the cloud structure, dynamics, and microphysics.

At the zenith pointing angle (antenna elevation angle β = 90°), the vapor measurement is the column-integrated precipitable water vapor (Pw). RHI scans were made with the SMR to map the LW across the gravity waves, particularly along the winds at cloud level. These over-the-top scans moved in a plane from ∼7.5° above one horizon, through zenith, to ∼7.5° above the opposite horizon at 0.5° s−1 (∼5.5 min/scan). In processing, each measurement of LW within such a scan, integrated along the slant path of the received radiation, was translated to the value along a corresponding vertical path through cloud depth, and presented in zenith-normalized plots of LW as a function of the SMR’s β. The result, equivalently, is the horizontal distribution of the vertical-column-integrated LW in the overlying cloud, relative to the instrument site under zenith (β = 90°). Liquid water of 1 mm equates to a 1 g m−3 liquid water concentration (LWC) through a 1-km-deep cloud. This measurement is key to this study. It is most accurate when drop sizes are such that their scattering is in the Rayleigh regime. As drop diameters exceed about 100 μm, the radiometric retrieval begins to break down. However, the consequent degradation of the measurement is not necessarily significant until considerably larger drop sizes occur; therefore, some judgment must be exercised. Experience indicates that the SMR accurately measures quantities of LW up to about 2 mm; the measurement is not significantly degraded by drizzle, at range or on the antenna, even though the drops can be several hundred microns in diameter. A uniform water coating or sprinkling of drops on the antenna will not degrade the measurement. There is degradation if sufficient wetting occurs to cause rivulets to run across the reflector. During two 1-h periods during the wave activity precipitation rates at CTW reached 0.05 mm h−1, so light that such problems if any were minimal. Thus, the CTW site, near the valley precipitation minimum, was ideal for the SMR.

The wave-cloud measurements were sometimes influenced by background levels of the water content of light drizzle or rain falling between the cloud and the sensor, which we define as rain liquid water (RLW). The background was normally discernable from the wave CLW. Note that our definitions now include the liquid concentration, LWC, the general path-integrated LW, and that separated into CLW and RLW. As will be demonstrated, an RHI scan either showed minimal or relatively small but measurable background liquid (RLW) through the wave trough and revealed peak LW through the wave crest. The wave condensate was estimated as the difference, peak LW − background RLW = wave CLW. This difference eliminates the effect of RLW during uniform drizzle across the scan, or possibly makes the wave CLW estimate slightly more conservative when drizzle occurred upstream but not through the wave crest portion of the scan. The effect is somewhat diminished because of wide spacing of settling drizzle drops. Another possible effect is that a melting layer will increase scatter and inflate the radiometric LW measurement. For the radiometer frequency used in AP95, an inflation of less than 7% would be expected (Westwater 1972). An error of this order is not significant to the interpretations. Also, over CTW, the observed melting layers associated with drizzle were shallow, and calculation of wave CLW by the differencing technique (above) eliminates the effect or makes the estimate slightly more conservative, just as for the intervening drizzle or rain described above.

Similar to the SMR RHI scans, radar RHI scans from β = 0° to 175°, horizon-to-horizon through zenith, mapped the cloud equivalent reflectivity factor (Ze), radial velocity (Vr), elliptical depolarization ratio (EDR), and a correlation field (CR). To limit velocity folding, the range limits were normally set at a radius of 12.5 or 25 km; at proper azimuths, these extended beyond the crest of the Black Hills at 8 km but only over the windward slopes of the Mogollon Rim, the plateau of which began at 25–40 km (Fig. 1). Radar velocity-azimuth-display (VAD) scans, nominally at β = 75°, were included to obtain profiles of the horizontal wind vector and vertical motion in the clouds over the CTW site. Cloud depth was estimated from CR. The CR is derived from a combination of the signal-to-noise ratio and the coherence of the received signal, so it defines the boundaries between useful and poor return. Cloud liquid water flux (CLWF) was approximated from the cloud depth, and the radiometrically measured wave CLW advected with a mean in-cloud wind vector from a corresponding VAD.

The parameter EDR, measured in radar RHI scans, is used to identify hydrometeor types and distinguish ice from liquid, as detailed by Matrosov et al. (1995, 1996) and Reinking et al. (1996, 1997). The radar beam is depolarized differently by the various classes of hydrometeors, allowing them to be separately identified in distinguishable signatures of EDR as a function of β. Small cloud droplets will normally produce no measurable polarization signal, even though the cloud can be detected in Ze to ∼−35 dBZ. For the radar as configured in Arizona, drizzle drops under about 280-μm diameter (i.e., spheres) produce a baseline signal of EDR ≈ −21 dB that is invariant with β. Ice hydrometeors produce depolarizations that vary measurably with β and/or are offset to EDR > −21 dB. Thus, all-liquid clouds were expected to be distinguishable from those glaciated or of mixed phase.

A high-spectral resolution lidar (HSRL) at CTW, a PBL acoustic sounder lower in the valley near CTW, and a Cheyenne II cloud physics aircraft provided some supporting measurements for this study (Klimowski et al. 1998). Also, rawinsondes provided cloud temperatures for estimating the supercooling, potential nucleation and growth habits of ice, and 0°C level. The melting layer was also lucidly monitored in its radar depolarization signature.

For simulations, the AP95 used the anelastic, nonhydrostatic Clark–Hall/National Center for Atmospheric Research (NCAR) numerical mesoscale–cloud scale model (Clark 1977; Clark and Hall 1991). The model uses Kessler’s (1969) warm rain treatment and the Koenig and Murray (1976) ice microphysics scheme. It carries microphysical variables including mixing ratios of WV, CLW, and RLW, and number concentration and mixing ratio for ice crystals and graupel. For the postevent analyses, this application exploited the model’s two-way interactive grid-nesting capability, coarse-grain parallelization, and vertically stretched terrain-following coordinates (63 levels were used). Rapid update cycle (RUC) large-scale analysis data were used to initialize the outermost domain, and later 3-hourly RUC data were used to update the boundary conditions [RUC is the operational version of the NOAA Forecast Systems Laboratory’s Mesoscale Analyses and Prediction System (MAPS); Benjamin et al. 1991]. Three levels of nesting were used by Coen et al. (1996a,b) for a simulation of the wave episode on 6 March 1995. The large-scale flow in the region was simulated with a 1170-km-square domain with 24.3-km horizontal grid spacing; this domain and a nested domain with 8.1-km grid spacing were initialized at 0300 UTC 6 March when an encroaching cold front was located in the northwest corner of the largest domain. A third, innermost domain nested to 2.7-km focused on the Black Hills, Verde Valley, and the Mogollon Rim; it was initialized at 0330 UTC and continued for 14 h until 1730 UTC. A sample of the results for 6 March extends the interpretation and dimensions of the observations presented in this paper.

Weather scenario

A synoptic buildup over a five-day period resulted in cyclically increasing moisture influxes and gravity wave activity. These were driven initially (2–3 March) by weak short-wave impulses traveling inland from the Pacific in otherwise regionally zonal flow. The period began with zonal flow at 500 mb across the lower United States and an Arctic low in place with its minimum pressure over north-central Canada. As an extension of the Arctic low, troughing gradually increased over the Pacific Northwest and dropped southward along the California coast. An accompanying weak surface low soon extended from the Pacific Northwest through the Great Basin and into the project area. Four successive shortwaves emanating from the synoptic system affected the project area. The first, early on 2 March came inland as a weak baroclinic perturbation in the 500-mb zonal flow that traveled under the Arctic low. The next three short waves emanated from the trough that was deepening over the Pacific off the west coast. These three waves affected the project area with successively increasing intensity on 3, 4, and 5–6 March. Figure 2a shows the Arctic low and the 500-mb troughing associated with these three waves as they appeared at 0000 UTC 4 March. The wave that affected the project area on 3 March was then over the Midwest, the system of 4 March was beginning to affect the area, and the system that was to affect the area on 5–6 March was deepening off the west coast. When the latter system moved inland and was affecting the project area, the trough line extended from the Gulf of California through western Wyoming and into north-central Canada, and the surface low had moved southeastward and was centered over the New Mexico–Colorado border, directly east of the project area. Increasingly significant moisture and prefrontal cloud bands, and in the project area, associated southwesterly flows and gravity waves were produced by these three systems. The latter two produced significant precipitation on the Mogollon Rim (Fig. 1). The satellite image in Fig. 2b, from 0600 UTC 4 March, shows the succession of three cloud bands, a weak one over the southeastern United States that had affected the project area on 2–3 March, a stronger cloud band affecting the area at the time of the image, and the third one moving landward from the Pacific that most intensely affected the area on 5–6 March. Surface cold fronts trailed the last two cloud bands.

In comparison, the 15 March 1987 storm analyzed by Bruintjes et al. (1994), which also realized the effects of embedded gravity waves, differed in that the synoptic low was associated with the polar rather than the Arctic front and it evolved to a cut-off circulation over the Great Basin. The prefrontal part of that storm was slightly cooler; the 700-mb temperature was −5°C over the project area, whereas it was −3°C and 0°C during the 4 and 5–6 March 1995 events, respectively. The 1987 event was otherwise similar in that the circulation of a synoptic low developed over California and moved eastward; strong southwesterly flow with the approach of the system brought substantial moisture from the Pacific and the Gulf of California over the area. This influx occurred in well-defined cloud bands with southwest–northeast orientation that moved eastward. These bands carried large amounts of moisture through a deep layer of the atmosphere in flow perpendicular to the mountain barriers and led to substantial precipitation in the area. To conclude that these storms are typical has pitfalls, but they exhibited prominent features that occur frequently in the major winter storms in northern Arizona.

After forcing only a few lenticular clouds on 2 March, and minor waves and cloud development with a trace of precipitation on 2–3 March, the prefrontal flows across the series of ridges set up persistent, large-amplitude, gravity waves within the cloud bands and over the Verde Valley during the two successive main storm events, on 4 and 5–6 March 1995. The periods with gravity wave clouds prevailed for ∼2.5, 6.5, and 15.5 h on 3, 4, and 5–6 March 1995, respectively, as observed by radar and marked in Fig. 3a. Vapor influxes (precipitable water Pw) to the project area built and ebbed with the cycles of passing cloud bands during the 5-day period. Wave-cloud development coincided with the peaks of Pw, which were caused by the moisture in the cloud bands. The continuous record of Pw and LW from zenith measurements with the SMR at CTW in Fig. 3a illustrates the trend of moisture building cyclically from ∼0000 UTC 2 March until ∼1200 UTC 6 March (Fig. 3a). The Pw minima were in the range of 1.0–1.4 cm prior to the final drying to 0.6 cm, whereas the maxima increased from an initial 1.9 cm to a final 2.7 cm. The amplitude of the variation in vapor increased with time, with the final peak rising to 225% of the preceding minimum. This variation and final enhancement in the vapor supply is very significant. Each drying period with the passage of a short wave brought clearing, most definitively with a weak frontal passage at ∼1200 UTC 4 March, and abruptly with a final and definitive frontal passage between 1100 and 1200 UTC 6 March. The five-day LW trend followed the vapor trend. During the periods of wave–cloud activity, the LW (Fig. 3a) commonly reached 0.5–2 mm; these zenith measurements were often somewhat upwind of the wave crests where the largest quantities of CLW were condensed.

Means and standard deviations of Pw and LW for the gravity wave periods (Table 1) were taken from both zenith-pointing measurements and zenith-normalized values taken as one point per 0.5 s from full SMR RHI scans; data from numerous entire scans through the wave cloud layers, not just the points in the wave crests, are therefore included, so the mean values of LW are conservative. Points with LW > 2 mm were excluded to minimize possible non-Rayleigh effects, so the 6 March contribution is further moderated. Nevertheless, the values establish the increasing vapor that was required, and a baseline for the increasing liquid that was condensed, to produce the trend of wave-cloud activity. As specified, the two main events occurred with mean Pw = 2.1–2.4 cm and mean LW = 0.4–0.5 mm.

On 4 March, gravity wave clouds, accompanied by upslope clouds, forced the first substantial precipitation. The precipitation rate R pulsed to 1 cm h−1 and briefly to 2 cm h−1 at site T6 on the Mogollon Rim, where the most precipitation fell (Figs. 1 and 3b). Only traces fell at CTW and elsewhere in the Verde Valley, but this period brought accumulations to 0.9 cm at SRC on the windward slope of the Mogollon Rim and 2.9 cm on top of the rim at T6. The precipitation rates doubled those of 4 March, consistently pulsing to 2 cm h−1 and briefly to 3.5 cm h−1. The 5–6 March storm added 0.15–0.2 cm (liquid or liquid equivalent precipitation) at low valley locations including CTW, 3.1 cm to SRC, and 5.1 cm to T6. For the cross-barrier cross-section through CTW, SRC, and T6, this brought the liquid-equivalent precipitation for the 2–6 March series of synoptic waves to 0.3, 4.0, and 8.0 cm, respectively (Fig. 4). The waves of the two main events dissipated when the fronts or cold-air advection reached the area from the northwest, and orographic forcing ceased. Precipitation from the trailing fronts (the last two pulses on 4 March and last single pulse on 6 March, Fig. 3b) contributed to the accumulations but amounted to only ∼15%–25% of the totals. The precipitation fell predominantly during the gravity-wave periods, thus emphasizing the importance of the combined gravity wave and direct upslope forcing.

The traveling synoptic waves of this series were warm. During periods of peak Pw and gravity wave activity, rawinsondes from the Black Hills at 0200 UTC 3 March and 0137 and 0635 UTC 4 March measured a 0°C level between 1.5 and 1.8 km AGL (CTW), slightly above highest terrain. The 0°C level cyclically lifted to 2.0 km above ground level (AGL) at 1730 UTC 5 March and 2.1 km AGL at 0650 UTC 6 March. From 2 to 6 March, the indicated surface temperature on the rim (at T6, Fig. 3c) ranged from −1° to +7°C; the coldest temperatures were reached after the frontal passages and/or drying periods with cold air advection. The long periods with temperatures several degrees above 0°C necessarily caused melt and pushed the snowpack on the rim toward water saturation. With the near-zero temperatures, the frontal passages marginally produced snow or rain mixed with snow, rather than rain, at the higher elevations. However, temperatures during both of the prefrontal 4 March and 5–6 March gravity wave periods began near 4°C and cooled to near 0°C. Therefore, most of this precipitation fell as rain, which led to further saturation and melting of the snowpack. Planned seeding tests were suspended when forecasts, measurements, and in-place seeding suspension criteria indicated a possible flood hazard. Oak Creek, which drains through Sedona, and other rivers draining from the rim into the valley flooded at record levels in the low areas of the Verde Valley. The photograph in Fig. 5 conveys the force and hazard of the flood.

Features and influences of orographic gravity wave clouds

The following exemplary observations illustrate on 50-m to 50-km scales the cloud dynamical and microphysical features through the evolution of this 2–6 March 1995 storm episode and serve to test the conceptual model of AP95. The cloud bands associated with the synoptic storm circulations extended well upwind from the project area (Fig. 2b). These bands and any precipitation processes they initiated were enhanced by the interaction of the southwesterly–westerly prefrontal flows with the parallel ridges that forced orographic clouds in upslope flow and induced the storm-embedded gravity waves. Thus, the bands reaching the mountains were transformed into a combination of gravity wave and upslope cloud systems. The gravity wave clouds and their influences on precipitation were readily observed. Over the mountain slopes, near the horizons, the wave and upslope clouds usually appeared as one mass so the upslope clouds were only occasionally separately identifiable. The presence and added influence of the upslope clouds on the precipitation was affirmed, however, by aircraft measurements, the simulations, the actual cross-barrier flows, visual observations, and some of the near-horizon remote-sensing measurements.

The strengthening systems: 3 and 4 March 1995

Early development

The first peak of Pw, early on 2 March 1995, was significant as a precursor, although with only minor cloudiness. Progressively more cloud development occurred during the ensuing Pw cycles. Relatively high and cold wave clouds were observed over the Verde Valley early on 3 March, during the second Pw peak, which showed slight increases over the first cycle (Fig. 3a). The first radar and rawinsonde observations found wave clouds in two shallow stable layers (Figs. 6a,b) with base temperatures of ∼−10° and ∼−20°C, respectively, at ∼4 and ∼6 km AGL (CTW). These clouds were initially accompanied by weak upslope cloud development upstream on the Black Hills (0156 UTC, Fig. 6a). They thinned downwind over CTW, and SMR measurements indicated no liquid at this time. CTW HSRL lidar depolarization measurements of hydrometeor depolarization through the same period corroborate that the upper and lower layers merged as a midlevel seeder–feeder couplet and glaciated from ∼0120 to 0230 UTC;lenses of cloud with radar EDR to −17 dB also confirmed the presence of ice during this period. The lidar indicated that the upper layer was consistently ice; before and after this period the lower layer exhibited stratifications, some that were all liquid and some of mixed phase, in agreement with the observations. Although the radar reflectivities of these clouds reached only ∼−2 dBZ, Ze nicely illustrated the enhancement of hydrometeor size and/or concentrations in or trailing downwind from the wave crests, where the wavelengths were only ∼10 km and amplitudes were only ∼300 m (0239 UTC, Fig. 6b). Here, CLW rose from zero in a wave trough to ∼0.2 mm in the crest and fell to zero again in the downwind trough (Fig. 6c). Thus, even in these comparatively simple wave clouds, the CLW and hydrometeor characteristics varied not only through the wave but also with seeder–feeder interactions, and coupling with underlying upslope clouds was observed over the upwind ridge.

Pulsations and the precipitation stage

The first major event of the 5-day-episode occurred on 4 March. The vapor supply was that of the broad third surge in Pw (∼0000–0900 UTC, Fig. 3a), which was significantly increased in quantity and duration over the 3 March episode. In contrast to the shallow lenticulars of the 3 March episode, much more LW was condensed, and the wave clouds were 4–5 km deep and precipitating. Combined with underlying upslope clouds, these were the clouds of the prefrontal storm.

Some details are pertinent. The 0137 UTC sounding found the base of a saturated layer at 2 km AGL (700 mb) and −3°C, 2 km lower and 7°C warmer than the lowest wave–cloud base early on 3 March. When there was significant wave activity, the wind direction was persistently uniform from about 250° to 260° above 2.0–2.5 km AGL (CTW) (Fig. 7). The heavy precipitation fell on the windward slope of the Mogollon Rim from 0300 to 0500 UTC, and on the rim itself from 0130 to 0900 UTC, with an interruption from ∼0500 to 0700 UTC (Fig. 4). The gravity waves added water to influxes of ice from upstream, to enhance crystal growth and the formation of precipitation. The part of the storm with gravity wave forcing (6.5 h within the 0000–0900 UTC period) contributed some 86% and 77% of the total 4 March precipitation at SRC and T6, respectively. These contributions are conservative because even the trailing frontal clouds interacted with a hydraulic jump over CTW, observed by radar. During the wave activity, synchronized wavelike pulsations in both Pw and LW of 2- to 4-h timescales were superimposed on the broad peak of the vapor surge (Fig. 8). Here, Pw pulsated in the range of ∼±0.15 to ±0.35 cm around a substantial mean that slowly declined from ∼2.4 to 2.0 cm. The CTW zenith LW reached maxima between 1 and 2 mm and minima between 0.1 and 0.5 mm. The maxima and minima in R at T6 lagged those in Pw and LW over CTW by approximately 30–45 min (Figs. 3b and 8). If the precipitation formed from the LW in the crest of the gravity waves near CTW, and then fell at ∼1 m s−1 as it was transported by measured in-cloud winds of ∼17–20 m s−1 over the 48-km distance from CTW to T6, it would have fallen to the rim ∼40–47 min after formation, in accord with the observed lag.

The significance of the gravity waves in generating precipitation can be determined by further exploring the cause and effect of the Pw and LW pulsations. To do this, consider the decline of a pulse marked as a, b, and c in Fig. 8. From a–c, Pw diminished from a peak of ∼2.40 to 2.25 and then to a minimum of 2.05 cm, while the maximum LW dropped from ∼1.0 to 0.5 to 0.1 mm. The corresponding sequence of SMR RHI scans shows the distributions of LW through the wave clouds (Figs. 9a–c, 0237–0509 UTC). Responding transitions in Ze and EDR were evident (Figs. 10a–c and 11a–c, respectively). The simplest illustration of the pronounced wave cloud and its liquid water structure is that from the middle of this sequence (b in Fig. 8; 0430 UTC, and Figs. 9b, 10b, and 11b). At this time, the gravity half-wave length was about 10–12 km (scaled from the trough in the lee of the Black Hills to the crest; Fig. 10b). The Ze structure reflected descending flow to the foehn trough, followed by strong ascent through a wave amplitude of ∼2 km. In the incoming downslope flow to the foehn trough (SMR β < 30°; Fig. 9b), the CLW was depleted to a minimal 0.1 mm, but the cloud was not dissipated. Cloud temperatures were near −3°C at base. Cloud top ranged from −18°C (5 km AGL) to −26°C (6 km AGL). EDR values well above the −21 dB value for spherical drops, confirm a presence of liquid-consuming ice particles (Fig. 11b). In the lower portions of the cloud, temperatures around −5°C, and strong depolarizations with EDR ≈ −16 to −14 dB through the trough, indicate relatively pristine, needlelike growth habits. Given the preexisting cloud band, the cold tops and warm base of wave clouds, and probable enhancement of liquid in the upstream clouds over the Black Hills, the needles likely resulted from a Hallett–Mossop ice crystal multiplication mechanism. Despite this ice influx, renewed condensation in the wave updraft significantly increased the CLW to a substantial peak of about 0.5 mm near or somewhat beyond the CTW zenith. This is a large quantity of available liquid. Downstream of these peaks, 0.3–0.5 mm of CLW (Fig. 9b) was maintained against the presence of transformed ice (EDR ≈ −18 to −19 dB, quite invariant with β, likely rimed, irregular ice forms; Fig. 11b). Thus, while the wave cloud was producing precipitation downwind, beyond the 8-km range (Fig. 10b), a considerable portion of the CLW was not converted to ice and precipitation. Some inefficiency in the precipitation process is indicated.

The precipitation associated with this wave episode was often more intense on the windward slope of the Mogollon Rim than that indicated by the 8-dBZ maximum Ze of the cloud and rainshaft in Fig. 10b. The earliest example (a in the sequence in Fig. 8) shows more widespread precipitation, reaching Ze ∼ 18 dBZ, blowing off the upwind ridge and also falling over the slopes of the downwind rim (left and right, respectively, in Fig. 10a, 0246 UTC). A radar VAD scan at 0234 UTC measured winds in the cloud layer that were uniformly from ∼255°, with speeds near ∼19 m s−1 in the cloud core; this produced a wave cloud with half-wavelength and amplitude similar to the previous example, and drove an updraft to 0.7 m s−1 near the wave crest above CTW. The position of this measurement indicates a stronger updraft a few kilometers upwind. As Pw reached a 2.4-cm peak (a in Fig. 8), the updraft generated LW that peaked at 0.8–1.0 mm. The pattern in the SMR RHIs shows the dominant influence of the wave in generating the liquid (Fig. 9a). The upswings outside the LW minima at 0.1–0.2 mm, in the wings of the horizontal (LW-β) distribution, are influenced by precipitation and likely also by the presence of CLW in underlying, upslope clouds (Fig. 9a). The peak-to-minimum difference is taken as the wave CLW, which after cresting just beyond zenith at ∼0.6–0.8 mm, rapidly diminished downwind at this time. However, the cloud was again maintained (note Ze). An EDR ≈ −17 to −15 dB, with detectable but not pronounced dependence on radar elevation angle (Fig. 11a), suggests efficiently precipitating crystal aggregates. However, the rain rate did not exceed that of the 0430 UTC example, so glaciation must have claimed its share. Together, these factors indicate that the almost complete CLW depletion was mainly due to 1) consumption by ice to form precipitation and 2) glaciation followed by sublimation, with much less to direct evaporation of the CLW. This suggests a precipitation process that was again inefficient but by reason of partial sublimation rather than excess availability of CLW.

In contrast to the first two scenarios, precipitation ceased completely when the wave cloud temporarily lost most of its forcing, Pw advection reached a minimum, and CLW diminished to a mere 0.1 mm (Point c in Fig. 8; Fig. 9c). Remaining was a 3-km-deep layer cloud in weak subsidence and with considerable ice (0501 UTC, Figs. 10c, 11c). The only wind profiles at this time were those measured with the acoustic sounder near CTW, which showed a temporary but significant veering in the surface to 0.9 km AGL layer beginning at ∼0445 UTC, from which a wind shift to a ridge-parallel west-northwest (WNW) direction aloft could be implied, given the differing structure during wave acitvity. By ∼0615 UTC, after the Pw and CLW minima, gradual backing to southerly PBL winds indicated the return to southwesterly flow aloft.

With the forcing flow restored, some scattered low-level clouds accompanied some wave regeneration. The wave and low-level clouds replenished the CLW to 0.4–0.8 mm after the near-glaciation period and led to additional precipitation (Figs. 12a,b). The radar Ze shows complex embedded wave activity on half-wavelength scales of ∼2–10 km; the SMR RHI scans show the irregular influence on the CLW from the low-level clouds superimposed on that of the main wave, which comprised the upper part of cloud. This combination contributed to the brief period of R ≈ 3 mm h−1, the greatest precipitation rate of this event on the rim, about 30 min later (Fig. 8). Near 0900 UTC, this final wave-cloud surge of 4 March ended.

The depiction resulting from this examination of the a–c sequence (Fig. 8) and the subsequent wave rejuvenation is as follows. Substantial levels of vapor and liquid resulted from southwesterly advection and prominent wave forcing. Intermissions in the forcing occurred even within the broad peak in vapor supply, and were sufficient to cause the CLW to plunge to near zero. Lesser depletions and replenishments in vapor and wave CLW occurred during more consistent, but still fluctuating wave forcing, as ice hydrometeor development cyclically responded to these inputs. The ice not only evolved in quantity but also in its forms that, with cycling variations in efficiency, converted vapor and CLW to precipitation, which appropriately lagged the transitions in the clouds. Feeder upslope clouds also formed, and the collective data leave no doubt that the wave clouds were seeders creating couplets to influence the precipitation from the combined cloud system.

The melting layer: Wave-associated rain on snowpack

The melting layer was clearly depicted as the radar bright band in EDR; it is indicated in Figs. 11a,b (where the choice of color scale makes it black in the first of these figures). It occurred near 1.5 km AGL over CTW, equivalently about 0.3 km above the Mogollon Rim, while the surface temperature at T6 varied in the 1.0°–2.5°C range (Fig. 2c). This further affirms that the wave-associated precipitation fell predominantly as rain rather than snow on the windward slopes and as melting snow or rain on the rim, pushing the snowpack, already wet from warm prestorm temperatures, toward saturation. Layer-embedded convection to ∼2.5 km AGL accompanied the weak frontal passage at ∼1200 UTC and added 0.1 cm of precipitation at SRC and 0.6 cm at T6 (Fig. 4). The surface temperature at T6 dropped to 0°–1°C (Fig. 3c), so this precipitation likely also fell as near-melting snow.

Maximum development: 5–6 March 1995

Onset: Wind influence

As Pw made its final strong increase of this storm series, a wave-cloud episode was initiated after 1900 UTC 5 March and lasted ∼15 h. The melting level had risen another 200–300 m compared to the previous episode. The radar-recorded wave clouds in multiple layers forming during the startup hours, as SMR scans consistently measured CLW ≈ 0.6–1.0 mm. The horizontal distribution of this wave CLW was very similar to that in Fig. 9a, but contrary to those observations, associated ice and precipitation were minimal, and stronger upslope cloud components on the slopes of the rim were indicated at the lowest scan angles. Activity temporarily waned as 5 March ended. Then measurable, continuous precipitation began to fall at T6 at ∼0000 UTC 6 March, about 2.5 h after the first major peak in Pw was attained. Highlights of the evolution of the LW throughout this episode, from 2330 to 1100 UTC 5–6 March, are exemplified by the irregularly spaced but sequential examples of SMR RHIs in Figs. 13a–c.

The correlated pulsations of wind direction, wave activity, Pw, CLW, and R of 4 March occurred again, but with larger quantities of vapor and liquid involved (Fig. 3a) and with more persistent gravity waves after 0000 UTC 6 March. Again, the stronger vapor advection was coincident with the cross-barrier flows from south of 270°. The most dramatic variation occurred near 0000 UTC, when the winds above ∼2 km AGL (CTW) shifted significantly from north to south of 270°, that is, from barrier-parallel to cross-barrier flow, and Pw and wave activity responded in accord, all within the broad vapor surge of this event. Wind direction and vertical motion profiles show the change between two VAD scans. First, at 2331 UTC 5 March (Fig. 14a), the winds between 2.0 and 4.5 km AGL (CTW) were strongly from the northwest in a layer that was clearing; the Pw, following a pulse to ∼2.9 cm, declined to a minimum of ∼2.5 cm at most. Descent across the Verde Valley, measured at −1 to −4 m s−1 above CTW, was observed in two nonprecipitating cloud layers of Ze < −10 dBZ that had only remnants of wave activity and no evidence in EDR of ice or large drops. The CLW irregularly reached only ∼0.4 mm and reflected the disorganization of the clouds (Fig. 13a).

As the flow shifted to substantially south of 270° (0101 UTC, Fig. 14a), Pw rose to ∼3.1 cm, and a wave updraft as strong as +5 m s−1 forced a deep cloud layer (2.0–6.4 km AGL) over CTW (0101 UTC 6 March, Fig. 14b). The wave crest was offset slightly downwind of CTW, so the vertical motion measurement was made within the maximum updraft. The 0042–0053 UTC SMR RHIs found peak LW of ∼1.3 mm (Fig. 13b), with an evident wave-forced component that rose above a base value of ∼0.4 mm. The minimum wave contribution, added by the updraft, was the ∼0.9 mm difference between the peak and the base. This wave cloud had generated ice, was drizzling in the valley, and was precipitating more heavily over the slopes of the ridges. Thus, the base value can be attributed to the drizzle component of the total liquid (RLW, section 2), or to a combination of RLW and wave CLW not evaporated in transit through the foehn trough. Wind speeds of ∼20–25 m s−1 within the cloud core were about 15% stronger than those on 4 March. As on 4 March, the peaking Pw and CLW at this time were followed about 30–45 min later by the first precipitation, which increased through R = 1 cm h−1 at ∼0130 UTC and 2 cm h−1 at ∼0145 UTC.

Within the period of less than 1 hr just examined, the abrupt changes in key parameters were substantial. The winds aloft shifted by as much as 100°, Pw increased by 25%, CLW contributed by the wave increased from 0 to 0.9 mm, and R at T6 increased from 0 with subsidence to 2 cm h−1 with gravity wave forcing. Pulsations in these correlated parameters continued throughout the event, but at consistently higher moisture levels (Figs. 3a and 13a–c), because the winds aloft made only minor excursions north of 270°.

Wave persistence; CLW evolution and spatial distribution

The SMR scans indicated a continuous wave CLW signature from ∼0100 UTC to ∼1030 UTC (Figs. 13b–i), except from ∼0700 to 0800 UTC when the wave cloud was monitored on radar, but its CLW signature could not be well distinguished from upslope CLW and heavier rain (Fig. 13h). The wave components of CLW cycled between ∼1 and 2 mm until ∼0630 UTC (Figs. 13b–f). Very representative of the period was the powerful wave at 0535 UTC (Fig. 15), with a ∼10 km half-wavelength, ∼3 km amplitude, and a significant presence of ice, some as pristine crystals (EDR similar to Fig. 11b), against which large quantities of wave CLW were still condensed in updrafts of at least 2–3 m s−1.

Aircraft CLW measurements during a 40-min flight (0555–0635 UTC) through the wave and upslope cloud are consistent with the SMR observations. The in situ measurements (Figs. 16a–c) confirm the 1) spatial variation of CLW through the wave crest and beyond, 2) existence of the upslope cloud with CLW approximately equal to that in the wave, and 3) the coupling between the wave and upslope cloud. Here are the details. The flight path transect relative to CTW ± 30 km (Fig. 16a) can be related to the wave-cloud structure using the range rings to 18 km shown in the radar image in Fig. 15. The aircraft flew in circles advected with the wind, initially at a constant altitude of ∼3.6 km AGL (CTW) through the wave trough, then in the crest, and finally in a slow descent with the wave through 0°C near 2.6 km AGL, ending in the coupled wave and upslope clouds at 2.1 km AGL. Superimposed on the transect are vertical bars that indicate the cloud-droplet concentrations measured with a forward scattering spectrometer probe (FSSP). These concentrations and corresponding LWC from the FSSP and a King probe (Fig. 16b) were near zero in the foehn trough, increased to 100–200 cm−3 (0.3–0.7 g m−3) in the wave crest, declined to 50–100 cm−3 (0.1–0.2 g m−3) in the descending wave, and rose again to 150–300 cm−3 (0.3–0.9 g m−3) where the wave and upslope clouds merged. The low droplet concentrations suggest a maritime influence, in accord with the fetch of the synoptic system and the Pacific origin of the associated cloud bands. This would lead to a broad droplet distribution, and increase the probability of ice crystal multiplication by the Hallett–Mossop mechanism, which requires drops of diameters around 25 μm. The LWC values encompass the range of peak values from the two probes (the FSSP estimates were 10%–40% higher than those from the King probe, and the excursions to near-zero LWC apparently occurred as the aircraft repeatedly circled in and out of the liquid water core of the cloud). Corresponding SMR measurements (0605–0615 UTC, Fig. 13g) show only ∼0.1 mm of CLW in the foehn trough, rising to a peak in the wave of 1.6–2.0 mm, and depleting beyond the crest to ∼0.3 mm. For the ∼4-km wave-cloud depth, the equivalent LWC was ∼0.45 g m−3 through the crest and ∼0.1 g m−3 well beyond the crest, about the same as measured with the aircraft.

The quantities and the fluctuations in CLW increased by 50% or more after ∼0630 UTC (Figs. 13h–j). These measurements above 2 mm are questionable because of SMR limitations, but the horizontal (LW β) distributions make physical sense and suggest semiquantitative validity. Figure 13i, for example, shows very low and certainly valid LW quantities of just 0.1–0.4 mm in the foehn trough region rising downwind to 3.0–3.5 mm, suggesting that the latter values are also valid and indicate wave CLW. Together with the radar observations, the demise of wave activity with frontal passage at ∼1100 UTC is well illustrated in Fig. 13j, where only remnant cloud and/or rain on the ridges, at the wings of the LW-β distribution, is indicated. The prefrontal rain rate at T6, which remained in the 1–3 cm h−1 range for some 9 h, ended at the same time as wave activity (Fig. 3b). This wave period contributed 77% and 87% of the total event precipitation at SRC and T6, respectively. The trailing frontal passage added only slightly (23% and 13%, respectively) to the rain that induced the flash flood.

Simulated features

A sample of the simulation of the 6 March event using the Clark–Hall/NCAR model (section 2b) integrates and substantially extends the interpretation from the observations. Coen et al. (1996a,b) provide more details of the simulation and additional measurement-model intercomparisons.

A sequence of vertical cross sections from the simulation illustrates the nature and evolution of the wave dynamics, wave and upslope cloud LWC (g kg−1), and ice-crystal mass concentration (IMC, g kg−1) over a field of view about twice as wide as that of the radar (Fig. 17a–k). The period, in simulation time, is 0500–1000 UTC 6 March. A vertical cross-section of the temperature field at 0530 UTC is shown in Fig. 18; the altitude of the melting layer above the Mogollon Rim is indicated by the short bold dashed line. In the sequence of Fig. 17, the horizontal and vertical components of the winds in the plane of the cross section, u and w, are indicated by the vectors. At all times in Fig. 17, the simulated mountain-induced gravity wave can be noted as descent followed by strong lifting in the immediate lee of the Black Hills. Downstream from this, the wave is modulated by the rim, and descent occurs well over the rim near S = 90 km. A wave-cloud component of CLW is often evident aloft near the culmination of the updraft, and persistent upslope components of CLW are formed in more gentle uplift on the windward mountain slopes. Ice crystals are consistently present in the simulated wave but highly variable in mass concentration. These features are all as observed with the remote sensors. The altitude and temperature of the core (centroid) of the wave CLW varies through the period between about 7 and 4 km AGL and approximately −25° and −5°C (248–268 K, Fig. 18). The measurements do not indicate the altitude of the liquid within the wave, but the observed wave cloud persisted mainly in the 4- to 6-km-AGL layer, or at ∼−5° to −20°C, indicating that the real wave CLW was at times somewhat lower and warmer than the simulated wave CLW.

This simulation illustrates, more clearly than the qualitative radar observations of ice content, the role of the IMC in cyclically diminishing the vapor and CLW inputs to the wave clouds. Stepping through Figs. 17a–k shows that each buildup of wave CLW is accompanied or soon followed by a buildup in IMC that is also generated in the wave. This is followed by complete or partial, but temporary, wave-cloud glaciation, which leads both to ice precipitating and ice lost to downwind sublimation. When the ice concentration is diminished, the wave CLW rebuilds, and the cycle is repeated. In all, the CLW and IMC both pulse, with glaciation (IMC buildups) lagging liquid buildups in accord with the time required for the phase change. In detail, a wave CLW buildup peaks at ∼0530 UTC (Fig. 17b); this water is glaciated by 0630 (Fig. 17d), and IMC diminishes to a minimum by 0700 (Fig. 17e), just as wave CLW begins to be replenished. The CLW subsequently reaches a renewed maximum at ∼0700–0730 (Figs. 17e,f), where it is split into two cores aloft, temporarily at a low altitude. The wave updraft is strengthened at ∼0800 (Fig. 17g), even as wave CLW partially glaciates (0730–0800, Figs. 17f,g). The IMC reaches a maximum at ∼0830 (Fig. 17h). As those ice crystals settle out and the IMC diminishes, the wave CLW builds to a third maximum at ∼1000 UTC (Fig. 17k) to begin another cycle. The simulated cloud intermittently completely glaciates during wave forcing, whereas the observations indicated that CLW condensation in the updrafts was continuously maintained. Also, the radar observations show a cloud with a base near 2.5 km AGL maintained through the wave trough in the lee of the Black Hills, whereas the simulation shows the cloud at higher altitude, not always but often disconnected at the trough. To be true to the observations, simulated ice transported through the trough would compete with but not overcome liquid condensed in the wave updraft, thus affecting quantities. These factors indicate the known need for improvement in the model microphysics, which is common to all state-of-the-art models. However, the simulated CLW pulses occur on a timescale of the order of 1.5–3 h, in accord with the measured pulsations in CLW (e.g., Figs. 13a–c), which occurred on a timescale of ∼1.2–2.4 h.

Although an hour-to-hour, 1:1 time correspondence is difficult to identify, several additional simulated features can be paired with those observed: Spatially well-isolated volumes of wave CLW in the simulation (e.g., Figs. 17b,f,g,k) were also isolated in the measurements, particularly in Figs. 13b–g. The simulated wave CLW and upslope CLW sometimes merge (most evidently in Figs. 17h,j). Similarly, horizontal CLW distributions, like the SMR RHI measurements in Fig. 13h, suggest a wave component merged with upslope CLW and rain. Also, the simulation sometimes shows a second wave crest, deepening the upslope CLW (e.g., Figs. 17i,k), and some radar and SMR scans showed a second wave crest, as suggested in Fig. 13d. The quantity of wave CLW, and the sum (CLW + IMC), generally increase with time through the period of the simulation (Figs. 17a–k); the radar could only qualitatively measure the presence of ice, but the upward trend in CLW was observed with the SMR (Figs. 13a–c).

The upslope cloud and the effect of the seeder–feeder couplet

In the simulation, the upslope clouds develop more CLW than the wave clouds, although adding the CLW and IMC in the wave cloud brings the total condensed water substance in two clouds close to equality. The aircraft measurements (Figs. 16a,b) indicated approximate equality in the CLW alone, mostly ∼0.5 g m−3, although the flight altitude may have been below the core of the wave cloud and above the core of the upslope cloud, thus underestimating both. From the simulation and aircraft observations of the 15–16 March 1987 storm over the same terrain, Bruintjes et al. (1994) determined that “the CLW regions associated with the waves extended through much deeper layers of the atmosphere and in quantities a factor of 2 larger than those associated with the forced ascent over the ridges.” In that case, the wave CLW exceeded 1.2 g kg−1, and the CLW associated with the low-level uplift reached ∼0.6 g kg−1. The ratio of wave to upslope CLW should be expected to vary either side of 1, as the vapor supply is variably distributed through the atmosphere and differentially advected over the mountains and up the Verde Valley.

Notably, the only ice in the upslope clouds simulated for this 6 March storm was that settled from aloft. Throughout the simulated portion of the 6 March gravity wave period, the ice from the wave cloud is seeding the upslope cloud. This was confirmed by the continuum of radar observations, such as the Ze image in Fig. 15, and aircraft measurements of ice. An abbreviated 3-h sequence of vertical cross sections from the simulation shows graupel mass content (GMC) that formed from some of the ice crystals (IMC), and the rain (RMC) resulting from melting of the total precipitating ice (Fig. 19). The alternating buildups and decays of the wave CLW followed by those in IMC (Fig. 17) carried through to the formation of rain. In the simulation, the formation of the graupel and rain follow the glaciations that create the IMC. At 0500 UTC wave CLW is building (see discussion above), and the graupel and rain contents are significant but diminishing (0500–0600 UTC, Fig. 19). The renewed wave CLW is again glaciated after 0530, leading to another buildup of graupel and rain (0600–0700 UTC, Fig. 19); the wave CLW is again renewed between 0700 and 0730 UTC, and glaciation to ice crystals again ensues (Fig. 17), leading to more still greater graupel and rain concentrations by 0800 UTC (Fig. 19).

The simulation places the melting level near 2.0 km AGL (CTW; 3.0 km MSL) at the beginning of the period shown in Fig. 17 (0530 UTC, Fig. 18), but the simulation also indicates that the settling crystals were at times converted to rain at somewhat higher altitudes (e.g., top of the RMC contours at 0700 UTC, Fig. 19);the aircraft and the radar found the melting level near 2.4 km AGL (CTW; 3.4 km MSL). The aircraft encountered the tops of upslope CLW near −1°C (Fig. 16c), nominally in agreement with the simulated upslope cloud, which is predominantly warmer than 0°C. A 0647 UTC sounding found the −5° and −10°C levels at 3.9 and 4.8 km MSL, respectively, above the simulated upslope cloud, except where deepened by primary or secondary wave crests. The true upslope cloud associated with the ridges was too warm to initiate precipitation by ice processes. The same conclusion was reached by Bruintjes et al. (1994) for the cooler 1987 storm.

The simulated wave cloud as a seeder provides ice particles that collect water by deposition and riming in the upper reaches of the upslope cloud, and then melt to form large drops, which collide and coalesce with the upslope cloud droplets to form precipitation. Could the upslope cloud develop precipitation by drop growth from vapor and collision–coalescence, without seeder ice from the waves? This would require a 30–60 min horizontal transit in the low-level 10–20 m s−1 winds, and therefore could occur only near the trailing edge of the upslope liquid envelope. This was not likely to occur near the leading edge, where substantial precipitation was accumulated during the full wave period at the windward stations such as SRC (∼1.7 cm of liquid, Fig. 4) and in general agreement, the simulation accumulated 1.2 cm of liquid. This is also not likely with the continental condensation nuclei that are normally available in the winter desert atmosphere. Together, the observations and simulation indicate a critical seeder role of the wave cloud in extracting water from the upslope feeder clouds to produce the total orographic precipitation.

Statistical aspects of liquid water in the wave clouds of these case studies

The statistics of wave cloud liquid water from the two main events do not constitute a climatology but do indicate the range and variability possible within individual events. The statistics also provide a basis to estimate the general magnitudes of the liquid fluxes and precipitation efficiencies of these cases. Wave condensation exceeding ice consumption was extremely persistent within the wave updrafts. The signature of wave-generated CLW in the SMR RHI scans, a minimum in the Black Hills foehn trough followed downwind by a maximum emerging near or just beyond the culmination of the updraft, was absent during only ∼2 h within the 22 h of the 4 and 5–6 March wave periods. From each of 141 RHIs from these two periods, the CLW generated by the wave [the difference (peak LW − RLW = wave CLW; section 2b)] was estimated to produce the histograms in Figs. 20a,b. The distribution is much broader for 5–6 March than for 4 March; using only the values under 2 mm, the respective medians are 0.9 and 0.4 mm, and the means are 1.0 and 0.5. Risking the uncertainties in LW ≥ 2 mm by using all measured values adds 28 RHIS to the 5–6 March case and extends the distribution to a maximum wave condensate of nearly 4 mm with a 1.2 mm median and a 1.3 mm mean. This distribution is rational rather than erratic and smoothly declines to the 4-mm value (Fig. 20c). The larger values build up to the potential adiabatic wave-cloud condensate of 6 March, which was approximately 4 mm [∼2.5 g kg−1, determined by lifting a parcel through 2 km, the nominal wave amplitude, from 0°C and 2 km AGL (CTW), the nominal wave-cloud base]. These numbers add more credence to the measurements of CLW ≥ 2 mm, and at least suggest how large the overall liquid water values might become in strongest development. From Table 1, the ratios of mean LW (from wave crests and troughs and RLW) to mean Pw are ∼0.02 for both 4 March and 5–6 March, whereas the ratios of mean wave CLW exclusively, from the histograms, to mean Pw are ∼0.02 for 4 March and ∼0.04 for 5–6 March. (The latter ratio increases to ∼0.05 if all of the wave CLW values are used.) The stronger wave dynamics during the latter event were evidently condensing a doubled percentage of the available vapor, indicating a condensation process that was twice as efficient.

To budget cloud water and estimate precipitation efficiency, not only the condensate but also the rate of its production, that is, the wave cloud liquid water flux, CLWF, must be determined. Here we define precipitation efficiency as the ratio of the condensate (cloud liquid water) flux to the precipitation rate. Some estimates of the CLWF were made by combining measurements as outlined in section 2b. The results, excluding LW > 2, are given in Table 2. The measurements are particularly sensitive to the estimates of cloud core depth (δh), and statistical significance cannot be attached to these few samples, but they are sufficient in number and reasonably spaced though the main precipitation periods of 4 and 6 March to be meaningful. The differences in wave CLW between 4 March and 5–6 March are reflected in the fluxes. The estimated wave CLWF ≈ 1.4–3.0 g m−2 s−1 (0.5–1.1 cm h−1) for 4 March, while the R averaged ∼0.25 cm h−1; CLWF ≈ 3.4–9.0 g m−2 s−1 (1.2–3.2 cm h−1) for 6 March, and R averaged ∼0.50 cm h−1 (VAD scans to make such estimates were not available after 0600 UTC 6 March). The simulation of the 1987 wave case indicated that the wave CLW was twice the upslope CLW, but the simulation for 6 March indicates that the wave CLW was often less than the upslope CLW; the brief aircraft measurements on 6 March indicated near equality. The greatest efficiencies are derived under the assumption that the upslope CLW equaled the wave CLW. Allowing this, and reasonably assuming that the upslope in-cloud winds were half as strong as in the wave cloud, the CLWFs from the upslope component were half that of the wave component. If this situation applied, for 4 March, the total CLWF was 0.8–1.7 cm h−1 and the precipitation efficiency was 15%–33%; and for 5–6 March, the total CLWF was 1.8–4.7 cm h−1 and the precipitation efficiency was 11%–28%. The 6 March storm processed more liquid late in the storm, but precipitation rates did not increase, indicating a lower efficiency. However, compared to 4 March, at least twice as much CLW was processed, and twice as much precipitation was produced.

Summary and assessment of the hypotheses

Within the 5-day period examined, successive cloud bands from the Pacific alternating with interludes of drying created the 24–36 h cycles with extreme variations (up to ∼125%) in Pw. The significant final enhancement in vapor supply reached ∼2.7 cm. The cyclic trend in precipitable water was in itself a simple predictor of the wave clouds, which were derived from the interaction of the cloud bands with the mountain ridges. The westerly to southwesterly, prefrontal cross-barrier flows, carrying the moisture of the broad peaks in Pw, provided successively increasing forcing of the wave–cloud activity and precipitation. Even the comparatively simple wave clouds of the early, short event (3 March) illustrated characteristic spatial variations in CLW and hydrometeors, seeder–feeder interactions, and coupling with underlying upslope clouds. During the final two episodes (4 and 5–6 March), storm-driving gravity waves persisted for a total of 22 h. The resulting 4–6 km deep, precipitating wave clouds were the essence of the two storms. They developed very much as indicated by the simulation of part of the last episode.

Despite their persistence, variations in the cross-barrier forcing and vapor influx, superimposed on periods of peak Pw, caused these wave clouds to fluctuate in structure and position, and to substantially pulsate, on a timescale of 2–4 h, in CLW content and the production of ice. The simulation illustrated, more clearly than the radar observations of qualitative ice content, the role of the IMC in cyclically diminishing the vapor and CLW inputs to the wave clouds. The short cycles of strong condensation in the wave updrafts followed by partial glaciation produced enduring albeit pulsing precipitation, which lagged the wave cloud vapor and liquid inputs by 30–45 min, as it should considering the time required for the snow and rain to form and fall out. The percentage of available vapor condensed in the second storm, which had somewhat stronger wave forcing, was twice that of the first storm; and in direct proportion to the mean CLW condensed in the wave clouds, the precipitation totals from the two storms differed by a factor of 2. Precipitation from periods of gravity wave forcing was dominant, accounting for the order of 75%–85% of the totals at the high elevations, and belittling the contributions from trailing fronts. This and the resulting flash flood established the importance of the gravity-wave influence.

The hypotheses of the AP95 study can now be revisited. In contrast to the earlier studies that found 0.2 mm to be the 85th percentile of CLW in clouds from direct upslope forcing, the gravity waves observed in these major events regularly generated 0.5–2 mm and more, even in the common presence of competing ice. These are large quantities of condensate (hypothesis 1). Although extreme values of the measurements, near 4 mm, are to be regarded with suspicion due to limits of the microwave radiometer, they suggested that adiabatic values could be reached if condensation rates overpowered ice consumption rates.

The common presence of ice, along with the new CLW condensed in the wave clouds, was confirmed by the radar polarization measurements. The preexisting synoptic cloud bands initiated cloud processes, but the orographic waves amplified them. Conversion to ice did precipitate some of the wave-cloud water, and this precipitation did create seeder–feeder couplets with the upslope clouds (hypothesis 2). The observations and simulation together indicate that the upslope clouds were unlikely to produce the precipitation distribution or quantity observed without wave-cloud ice to extract the liquid. That the seeder–feeder process increased the conversion of upslope CLW to precipitation necessarily follows from the established couplet and the presence of substantial CLW in the underlying upslope cloud, with the result that the total precipitation was greater than it could have been from either the gravity-wave cloud or the upslope cloud alone. This deduction is reinforced by the simulation, here and in Coen et al. (1996a,b) and Reinking et al. (1997) (hypothesis 3), and by the aircraft measurements.

The waves almost always generated more CLW than the naturally formed ice could consume, within their updrafts (hypothesis 4). The data indicate that this is not always true beyond the updraft, where the liquid condensed in the wave was divided, at times between precipitation and direct downwind losses to evaporation, and at times between precipitation and losses to downwind glaciation followed by sublimation. Thus, inefficiencies evidently occurred due alternately to underconsumption and overconsumption of the wave CLW. This makes it difficult to specify definitively the opportunities for cloud seeding that depend on the availability of CLW. Favorable factors include 1) ice-free wave clouds with considerable CLW observed early in the storm periods; 2) the consistency with which the condensation rate exceeded the ice consumption rate in the strong updrafts, where the cloud was only slightly supercooled and seeding agents would more efficiently nucleate ice than natural ice nuclei, and where even more liquid might be condensed on more hygroscopic condensation nuclei; 3) the intermittent surplus of available CLW beyond the wave crest, indicating at least occasional seedability to improve conversion to precipitation (the intermittence of the CLW surplus and the pulsing glaciation argue equally against good seeding opportunities); 4) predominantly warmer than freezing temperatures and considerable CLW in the upslope clouds, where an enhanced couplet and hygroscopic agents offer an advantage; and 5) the estimated 11%–33% range of precipitation efficiency of the couplet that implies room for improvement. Thus, we have clarified the framework within which useful seeding of the wave cloud to enhance the couplet and total precipitation must be accomplished, but it is improbable that the seeding potential can be better defined without continuous, more quantitative measurements and simulations of the ice, because this is the factor least understood (hypothesis 5).

The persistent, precipitating waves in these warm storms did induce flash flooding by producing rain rather than snow, to melt the existing snowpack (hypothesis 6). Forewarnings of the potential flash flood were extracted from the standard rawinsondes and storm synoptics, but were better derived by continuously monitoring the persistence and distribution of precipitation and its phase (i.e., the altitude of the melting level), as continuously observed in the radar reflectivity and depolarization (bright band) signatures, respectively. Likewise, the simplest key to separating the seeding opportunity from the flash flood threat was to determine the height of the melting level, relative to the elevations of terrain covered with snowpack. The melting level was accurately monitored by dual-polarization radar measurements, and can be well simulated, post facto or in real time, if input data are of sufficient spatial and temporal density (hypothesis 7).

The evident conflict between the potential for flash floods and the possibility for precipitation enhancement represents reality, so why even consider cloud seeding? The answer, of course, lies with the facts that 1) water shortages and flash floods are not always seasonably separable, 2) orographic storms, including those affecting northern Arizona, are commonly colder and do not pose flash flood threats, 3) watersheds throughout the West for decades have been and continue to be operationally seeded to provide needed water and hydroelectric power, and in broader perspective, 4) although scientific investigation of cloud seeding has waned, fresh water supplies are seriously dwindling worldwide, and the demand for viable technologies from current and potential users is increasing, so the need will continue for better definition and prediction of seedability.

Conclusions

The observations, supported by the simulation, together demonstrate the central role that persistent prefrontal, orographic gravity waves can have in production and transformation of cloud liquid and ice, and the substantial influence they can have on precipitation. The relevance of such storm-embedded gravity waves to the prediction of orographic precipitation, the resulting precious water resource, and the initiation of flash floods in complex terrain is made evident. Further, a better basis is established from which to judge the potential for cloud seeding in terms of the availability of wave-cloud liquid water and the manner and efficiency with which it is consumed in natural precipitation processes. The geographic transferability of at least the qualitative aspects of these results is made clear by any topographic map that shows the multiplicity of parallel ridges worldwide and by similarities in simulations of orographically influenced storms in other regions. Storms passing parallel ridges will generate embedded gravity waves that will have the potential to form significant, precipitating clouds and so influence hazardous mountain weather, total storm precipitation, and the fresh water resource. The quantitative effects will, of course, depend on many particular factors such as prevalent forcing, elevations and breadths of ridges, available moisture, and consequent wave cloud–upslope cloud interaction.

Both the measurements and the simulation demonstrate the need for the high spatial and temporal resolution of the mesoscale and cloud scale to capture the dynamic and microphysical transitions associated with the waves. Overall, the cloud ice was not well quantified; this is perhaps the greatest shortcoming of this assessment and current remote sensing and numerical modeling capabilities and the foremost aspect that needs improvement. Significant advancements are nevertheless evident in the detail of the remote sensing measurements and the progressing agreement between the simulations and these measurements. Further progress here will allow us to address the remaining challenges, to determine whether ice can be effectively measured and predicted, whether precipitation therefore can be better measured and predicted, and whether a seeding methodology can be designed to tap the wave-cloud water source. The question of where adequate fresh water resources will come from in the twenty-first century is mounting. Demand will ensure that progress will continue, and eventually full storm water budgets and seedability will be well defined.

Acknowledgments

The NOAA Atmospheric Modification Program funded the field work. The NOAA Office of Global Programs partially funded the analyses. Dennis Sundie, Arizona Department of Water Resources, managed The 1995 Arizona Program, and Eric Betterton, The University of Arizona (U.A.), directed the field operations. Cyrus Jones, U.A., assisted from the data archive. RoelofBruintjes, NCAR, provided the analyzed aircraft data. Brad Orr, NOAA/ETL, assisted with the flux calculations. Engineers Bruce Bartram, Kurt Clark, Duane Hazen, W. B. Madsen, software engineer Jan Gibson, and data analysts Michelle Ryan and Tracey Sutherland of NOAA/ETL made the data collection and processing possible.

REFERENCES

  • American Meteorological Society, 1992: Planned and inadvertent weather modification, a policy statement of the American Meteorological Society. Bull. Amer. Meteor. Soc.,73, 331–337.

  • Banta, R. M., 1990: The role of mountain flows in making clouds. Atmospheric Processes in Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 229–283.

  • Benjamin, S. G., K. A. Brewster, R. Brummer, B. F. Jewett, and T. W. Schlatter, 1991: An isentropic three-hourly data assimilation system using CARS aircraft observations. Mon. Wea. Rev.,119, 888–906.

  • Bradley, M. M., 1985: The numerical simulation of orographic storms. Ph.D. dissertation, University of Illinois, Urbana–Champaign, 263 pp.

  • Bruintjes, R., T. Clark, and W. Hall, 1994: Interaction between topographic airflow and cloud and precipitation development during the passage of a winter storm in Arizona. J. Atmos. Sci.,51, 48–67.

  • Bruintjes, R., T. Clark, and W. Hall, 1995: The dispersion of tracer plumes in mountainous regions in central Arizona: Comparisons between observations and modeling results. J. Appl. Meteor.,34, 971–988.

  • Clark, T. L., 1977: A small-scale numerical model using a terrain following coordinate transformation. J. Comput. Phys.,24, 186–215.

  • Clark, T. L., and W. D. Hall, 1991: Multidomain simulations of the time-dependant Navier–Stokes equation: Benchmark error analyses of nesting procedures. J. Comput. Phys.,92, 456–481.

  • Coen, J. L., R. T. Bruintjes, W. D. Hall, T. L. Clark, and R. F. Reinking, 1996a: Numerical simulation of wintertime precipitation development in gravity-wave and upslope flow in Arizona’s Verde Valley. Proc. 12th Int. Conf. on Clouds and Precipitation, Zurich, Switzerland, ICCP/IAMAS, 778–781.

  • Coen, J. L., R. T. Bruintjes, R. F. Reinking, and G. Thompson, 1996b: Summary of the Arizona winter storm case: 5–6 March 1995. Fourth Int. Cloud Modeling Workshop, Clermont-Ferrand, France, WMO/TD-No. 901, 37–44.

  • Heggli, M. F., and R. M. Rauber, 1988: The characteristics and evolution of supercooled liquid water in wintertime storms over the Sierra Nevada: A summary of microwave radiometric measurements taken during the Sierra Cooperative Pilot Project. J. Appl. Meteor.,27, 989–1015.

  • Heimbach, J. A., Jr., and W. D. Hall, 1994: Applications of the Clark model to winter storms over the Wasatch Plateau. J. Wea. Mod.,26, 1–11.

  • Hogg, D. C., F. O. Guiraud, J. B. Snider, M. T. Decker, and E. R. Westwater, 1983: A steerable dual-channel microwave radiometer for measurement of water vapor and liquid in the troposphere. J. Climate Appl. Meteor.,22, 789–806.

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

  • Klimowski, B. A., R. Becker, E. A. Betterton, R. Bruintjes, T. L. Clark, W. D. Hall, B. W. Orr, R. A. Kropfli, P. Piironen, R. F. Reinking, D. Sundie, and T. Uttal, 1998: The 1995 Arizona Program: Toward a better understanding of winter storm precipitation development in mountainous terrain. Bull. Amer. Meteor. Soc.,79, 799–813.

  • Koenig, L. R., and F. W. Murray, 1976: Ice-bearing cumulus cloud evolution: Numerical simulation and general comparison against observations. J. Appl. Meteor.,15, 747–762.

  • Kropfli, R. A., S. Y. Matrosov, T. Uttal, B. W. Orr, A. S. Frisch, K. A. Clark, B. W. Bartram, R. F. Reinking, J. B. Snider, and B. E. Martner, 1995: Cloud physics studies with 8-mm-wavelength radar. Atmos. Res.,35, 299–313.

  • Matrosov, S. Y., R. F. Reinking, R. A. Kropfli, and B. W. Bartram, 1995: Identification of ice hydrometeor types from elliptical polarization radar measurements. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 539–541.

  • Matrosov, S. Y., R. F. Reinking, R. A. Kropfli, and B. W. Bartram, 1996: Estimation of ice hydrometeor types and shapes from radar polarization measurements. J. Atmos. Oceanic Technol.,13, 85–96.

  • Reinking, R. F., 1995: An approach to remote sensing and numerical modeling of orographic clouds and precipitation for climatic water resources assessment. Atmos. Res.,35, 349–367.

  • Reinking, R. F., S. Y. Matrosov, R. T. Bruintjes, B. E. Martner, and R. A. Kropfli, 1995: Further comparison of experimental and theoretical radar polarization signatures due to ice hydrometeor growth habit. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 47–49.

  • Reinking, R. F., S. Y. Matrosov, and R. T. Bruintjes, 1996: Hydrometeor identification with elliptical polarization radar: Applications to glaciogenic cloud seeding. J. Wea. Mod.,28, 6–18.

  • Reinking, R. F., S. Y. Matrosov, R. T. Bruintjes, and B. E. Martner, 1997: Identification of hydrometeors with elliptical and linear polarization Ka-band radar. J. Appl. Meteor.,36, 322–339.

  • Reynolds, D. W., 1988: A report on winter snowpack augmentation. Bull. Amer. Meteor. Soc.,69, 1290–1300.

  • Reynolds, D. W., 1996: The effects of mountain lee waves on transport of liquid propane-generated ice crystals. J. Appl. Meteor.,35, 1435–1456.

  • Smith, R., J. Paegle, T. Clark, W. Cotton, D. Curran, G. Forbes, J. Marwitz, C. Mass, J. McGinley, H.-L. Pan, and M. Ralph, 1997:Local and remote effects of mountains on weather: Research needs and opportunities. Bull. Amer. Meteor. Soc.,78, 877–892.

  • Super, A. B., and B. A. Boe, 1988: Wintertime cloud liquid water observations over the Mogollon Rim of Arizona. J. Wea. Mod.,20, 1–7.

  • Super, A. B., and E. W. Holroyd III, 1989: Temporal variations of cloud liquid water during winter storms over the Mogollon Rim of Arizona. J. Wea. Mod.,21, 35–40.

  • Westwater, E. R., 1972: Microwave emission from clouds. NOAA TR ERL 219-WPL 18, 43 pp. [Available from Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402.].

Fig. 1.
Fig. 1.

Topographic map of the AP95 project area in northern Arizona (elevation in meters MSL, 120-km square, 75-m contour interval). Key sites marked as open squares are Prescott (PRC, 1375 m MSL) just northeast of the Bradshaw Mountains, Mingus Mountain (MNG, 2100 m) at the highest point of the Black Hills, Cottonwood in the Verde Valley (CTW, 1040 m), Sedona (SDA, 1275 m), Schnebly Road Curve (SRC, 1400 m), and site T6 (2025 m). The city of Flagstaff is noted for reference.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 2.
Fig. 2.

(a) 500-mb circulation over North America at 0000 UTC 4 Mar 1995. Synoptic waves in succession are labeled as 3, 4, and 5–6 Mar 1995 for their dates of main effect in the AP95 project area. (b) The 0200 UTC 4 Mar 1995 GOES-7 satellite image of the corresponding succession of eastward-moving cloud bands that affected the project area. A foehn trough is faintly visible in the lee of the Black Hills, as a diffuse grey line just northeast and parallel to the black line, which marks the southwest edge of the project area.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 3.
Fig. 3.

Time-series plots for 2–6 Mar 1995 (UTC) of (a) vertical column-integrated atmospheric water vapor (Pw, cm) and liquid water (LW, mm) over CTW; (b) precipitation rate (R, cm h−1); and (c) surface temperature (T, °C) at site T6 on the Mogollon Rim.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 4.
Fig. 4.

Time series of accumulated precipitation (cm of liquid) from 2 to 6 Mar 1995 (UTC) for three sites in a SW–NW line: Cottonwood in the Verde Valley (CTW), windward slope of the Mogollon Rim (SRC), top of the Mogollon Rim (T6).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 5.
Fig. 5.

The flooded Oak Creek, a few kilometers upstream from Sedona, still overflowing and turbulent as it began to subside on 7 Mar 1995.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 6.
Fig. 6.

(a) Cloud reflectivity Ze (dBZ) from a CTW radar over-the-top RHI scan from azimuth 0° (north, at right) through zenith to south, slightly across the 240° cloud-level wind; showing wave clouds between 4 and 8 km AGL, overriding clouds developed in direct orographic lifting upstream at lower altitudes over the Black Hills. MNG peak is left of center at 8-km range; 0156 UTC 3 Mar 1995, 5-km range rings. (b) Subsequent radar RHI (Ze, dBZ) from azimuth 60° (at right), approximately along the wind; 0239 UTC. (c) Zenith-normalized, path-integrated CLW (mm) corresponding to (b); the indicated azimuth (60°) is that where the SMR RHI scan began at antenna elevation angle β = 0°; β equates to horizontal position, relative to β = 90° (zenith, over CTW).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 7.
Fig. 7.

Wind direction as a function of altitude AGL directly over CTW during gravity wave activity on 4 Mar 1995, from a radar VAD scan, 0149–0152 UTC.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 8.
Fig. 8.

Enlargement of the 0000–1200 UTC 4 Mar 1995 time section of zenith-normalized Pw (cm) and LW (mm) over CTW, and R (cm h−1) at T6, from Figs. 2a,b. Shadowing indicates time by which R lags secondary maxima and minima in Pw and LW, which are superimposed on the broad peak of moisture.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 9.
Fig. 9.

Zenith-normalized CLW (mm) from pairs of SMR RHI scans, as in Fig. 6c, at (a) a peak, (b) an intermediate point, and (c) the trough of a pulsation in Pw and CLW, from times marked respectively as a, b, and c in Fig. 8.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 10.
Fig. 10.

Cloud reflectivity Ze (color scale, dBZ) from radar RHI scans at (a) 0246, (b) 0430, and (c) 0501 UTC 4 Mar 1995, from azimuths 60°, 70°, and 70° (at right), with 5-, 2-, and 2-km range rings, respectively; corresponding respectively to Figs. 9a–c. The temperature was ∼−3°C at 700 mb (range ∼2 km at zenith) and ∼−18°C at 500 mb (range ∼5 km at zenith).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 11.
Fig. 11.

EDR (color scale, dB) corresponding to (a), (b), and (c) in Figs. 9a–c and 10a–c. The melting layer is indicated in (a) and (b).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 12.
Fig. 12.

Cloud at onset of wave regeneration and renewed release of precipitation. (a) RHI scan of Ze (color scale, dBZ) from azimuth 70° (at right), 0631 UTC 4 Mar, 2-km range rings; (b) pair of corresponding SMR RHI scans, 0639–0649 UTC, details as in Fig. 6c.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 13.
Fig. 13.

Sequence of liquid water measurements from SMR RHI scans, 2331–1054 UTC 5–6 Mar 1995. Azimuth 90° is at right in each frame.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 14.
Fig. 14.

(a) Horizontal wind direction (°) and (b) corresponding vertical motion (m s−1) as a function of altitude (km) above CTW, from radar VAD scans at 2331–2332 UTC (labeled “wave suppression”) and 0101–0104 UTC (labeled “wave enhancement”) 5–6 Mar 1995. Cloud top is indicated by the top of each profile.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 15.
Fig. 15.

Cloud reflectivity Ze (color scale, dBZ) of the precipitating wave cloud near maximum development (0535 UTC 6 Mar 1995); over-the-top RHI scan from azimuth 60° (at right); 5-km range rings. The temperature was ∼+1°C at 700 mb (range ∼2 km at zenith, slightly below the wave-cloud base) and ∼−15°C at 500 mb (range ∼5 km at zenith, somewhat below the wave-cloud top).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 16.
Fig. 16.

Measurements from a west–east aircraft spiral descent intersecting the foehn trough, wave crest, and upslope cloud between 0555 and 0635 UTC 6 Mar 1995: (a) cloud-droplet concentration (CONC, cm−3, bar scale, superimposed on flight track plotted as altitude in kilometers MSL vs distance relative to CTW); with insets as a function of time approximately equivalent to distance scale, of (b) LWC (g m−3) from the King probe (light line) and the FSSP (dark line); and (c) air temperature (K).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 17.
Fig. 17.

Sequence from the numerical simulation of the gravity wave and upslope clouds, simulation time 0500–1000 UTC 6 Mar 1995, grid spacing 2.7 km, showing vectors from the horizontal and vertical winds, CLW (g kg−1, line contours, 0.02 interval) and IMC (g kg−1, shaded contours and scale). Altitude Z is in kilometers MSL. The east–west (left–right) vertical cross section extends across the Black Hills near Mingus Mountain, over the Verde Valley near CTW, and over the Mogollon Rim. CTW is in lee (to right of) the Black Hills, near distance S = 35 km.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 18.
Fig. 18.

Simulated vertical cross section of atmospheric temperature (K) at 0530 UTC 6 Mar 1995 as a function of altitude (Z, km MSL), corresponding the cross section used in Fig. 17. The melting level (273 K) over the Mogollon Rim is indicated by the bold dashed line. (Horizontal distance X is 140 km).

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 19.
Fig. 19.

Sequence of vertical cross sections at 0500, 0600, 0700, and 0800 UTC from the simulation, illustrating variations of graupel mass content (GMC, g kg−1, shaded contours and scale) and rain mass content (RMC, g kg−1, line contours, 0.02 interval); grid interval, 2.7 km.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Fig. 20.
Fig. 20.

Histograms of wave CLW from (a) 4 Mar and (b) 5–6 Mar, excluding CLW ≥ 2.0 mm, and (c) 6 Mar 1995, including all values.

Citation: Journal of Applied Meteorology 39, 6; 10.1175/1520-0450(2000)039<0733:IOSEOG>2.0.CO;2

Table 1.

Mean and standard deviation of Pw and LW during periods of gravity wave activity.*

Table 1.
Table 2.

Estimates of wave cloud liquid water flux (CLWF).

Table 2.
Save