• Armi, L., , and G. J. Mayr, 2007: Continuously stratified flows across an Alpine crest with a pass: Shallow and deep föhn. Quart. J. Roy. Meteor. Soc., 133 , 459477.

    • Search Google Scholar
    • Export Citation
  • Baines, P. G., 2001: Mixing in flows down gentle slopes into stratified environments. J. Fluid Mech., 443 , 237270.

  • Baines, P. G., 2005: Mixing regimes for the flow of dense fluid down slopes into stratified environments. J. Fluid Mech., 538 , 247267. doi:10.1017/S0022112005005434.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and D. R. Durran, 2002: The dynamics of mountain-wave induced rotors. J. Atmos. Sci., 59 , 186201.

  • Durran, D., , and J. Klemp, 1987: Another look at downslope winds. Part II: Nonlinear amplification beneath wave-overturning layers. J. Atmos. Sci., 44 , 34023412.

    • Search Google Scholar
    • Export Citation
  • Farmer, D. M., , and L. Armi, 1999: Stratified flow over topography: The role of small-scale entrainment and mixing in flow establishment. Proc. Roy. Soc. London, A455 , 32213258.

    • Search Google Scholar
    • Export Citation
  • Gohm, A., , and G. J. Mayr, 2005: Numerical and observational case-study of a deep adriatic bora. Quart. J. Roy. Meteor. Soc., 131 , 13631392.

    • Search Google Scholar
    • Export Citation
  • Gohm, A., , G. J. Mayr, , S. Mobbs, , S. Arnold, , J. Vergeiner, , L. Darby, , R. Banta, , and S. Sandberg, 2000: Foehn flow in the Austrian Alps interrupted by a cold front passage: Part II. Preprints. Ninth Conf. on Mountain Meteorology, Aspen, CO, Amer. Meteor. Soc., 83–86.

    • Search Google Scholar
    • Export Citation
  • Grubišić, V., , and B. J. Billings, 2007: The intense lee-wave rotor event of sierra rotors IOP 8. J. Atmos. Sci., 64 , 41784201.

  • Grubišić, V., and Coauthors, 2008: The terrain-induced rotor experiment: An overview of the field campaign and some highlights of special observations. Bull. Amer. Meteor. Soc., in press.

    • Search Google Scholar
    • Export Citation
  • Holmboe, J., , and H. Klieforth, 1957: Investigation of mountain lee waves and the air flow over the Sierra Nevada. Final Rep. Contract AF 19(604)–728, University of California, Los Angeles, 283 pp.

  • Kuettner, J., 1959: The rotor flow in the lee of mountains. Tech. Rep., Research Notes 6, Air Force Cambridge Research Center G.R.D., 20 pp.

  • Long, R. R., 1953: Some aspects of the flow of stratified fluids. I. A theoretical investigation. Tellus, 5 , 4257.

  • Long, R. R., 1954: Some aspects of the flow of stratified fluids. II. Experiments with a two-fluid system. Tellus, 6 , 97115.

  • Long, R. R., 1955: Some aspects of the flow of stratified fluids. III. Continuous density gradients. Tellus, 7 , 341357.

  • Mayr, G. J., , J. Vergeiner, , and A. Gohm, 2002: An automobile platform for the measurement of Foehn and gap flows. J. Atmos. Oceanic Technol., 19 , 15451556.

    • Search Google Scholar
    • Export Citation
  • Mayr, G. J., and Coauthors, 2007: Gap flows: Results from map and beyond. Quart. J. Roy. Meteor. Soc., 133 , 881896.

  • Nishiyama, R. T., , and A. J. J. Bedard, 1991: A “Quad-Disc” static pressure probe for measurement in adverse atmospheres: With a comparative review of static pressure probe designs. Rev. Sci. Instrum., 62 , 21932204.

    • Search Google Scholar
    • Export Citation
  • Pawlak, G., , and L. Armi, 2000: Mixing and entrainment in developing stratified currents. J. Fluid Mech., 424 , 4573.

  • Prandtl, L., 1944: Zwei Flüssigkeiten verschiedener Dichte. Führer durch die Strömungslehre. 2nd ed., Friedr. Vieweg & Sohn, 326–330.

    • Search Google Scholar
    • Export Citation
  • Smith, R. B., 1979: The influence of mountains on the atmosphere. Advances in Geophysics, B. Saltzman, Ed., Academic Press, 87–230.

  • Smith, R. B., 1985: On severe downslope winds. J. Atmos. Sci., 42 , 25972603.

  • Vergeiner, I., 1983: Dynamik alpiner Windsysteme. Tech. Rep., Institut für Meteorologie und Geophysik, Universität Innsbruck, 129 pp.

  • Wood, I. R., 1968: Selective withdrawal from a stably stratified fluid. J. Fluid Mech., 32 , 209223.

  • View in gallery

    (a) Topographic map (m) of California; the rectangle marks the area of the Sierra Nevada and OV. (b) Detail with surveyed roads. Bishop (BI) was the operations center and IN is a small village in the center of OV; LE, MGLASS, and MAPR are radio sounding sites. (c) The road leg between OV, Seven Pines (7P), IN, and MC.

  • View in gallery

    The measurement platform mounted on a custom roof rack is an aluminum beam carrying the wind, temperature, and humidity sensors and the quadplate pressure port. The GPS receiver was put into the case of a PC mouse. The attachment for the DGPS unit is also visible. The small insert shows a detail of the radiation shield.

  • View in gallery

    ECMWF model topography (shading interval is 200 m). The actual topographic map of this region is shown in Fig. 1. The lines show the sections A and B, which are aligned with the synoptic flow between the chosen sounding locations during the individual cases. Upstream sounding location A is at 38.4°N, 121.2°W; B is at 37.6°N, 120.8°W. Downstream sounding location is at 38.4°N, 118.0°W.

  • View in gallery

    The θυ profiles from ECMWF analysis with 40-km horizontal mesh width, upstream (thick) of the model’s representation of the Sierra Nevada and downstream (thin) and upstream wind (barbs; short barb: 5 kt; long barb: 10 kt; triangle: 50 kt) for 1300 LST (a) 18 Apr, (b) 15 Apr, (c) 20 Apr, and (d) 17 Apr 2004. Upstream sounding location for (a), (b), and (d) is at B and for (c) is at A in Fig. 3.

  • View in gallery

    Virtual potential temperature profiles for across-valley drives between (right) IN and (left) OV (cf. Fig. 1c) annotated with their starting times for 18 Apr 2004. The gray parts of the curves indicate the region where the road passes through highly structured terrain, which induces pronounced small-scale variations not representative for the slope-scale flow.

  • View in gallery

    Across-valley drive between IN (E) and OV (W) at (a) 0907 and (b) 1636 LST 18 Apr 2004. (panel top curves) Virtual potential temperature (thick) and mixing ratio (thin); (middle) pressure reduced to starting point (thick) and dynamic pressure (thin); and (bottom) elevation (thick) and horizontal wind (same as Fig. 4, averaged for improved legibility).

  • View in gallery

    As in Fig. 5, but for 15 Apr 2004 except that the leg extends now to the eastern part of OV (MC).

  • View in gallery

    Vertical atmospheric structure at (a) 1116 LST at LE and 1000 LST at IN (MAPR) and (b) 1700 LST at LE and 1600 and 1900 LST 15 Apr 2004 at IN (MAPR). (left) Virtual potential temperature; (middle) mixing ratio and relative humidity (dashed); and (right) cross-barrier component of wind. Line SC marks the height of the Sierra Nevada crest, and KP is the height of Kearsarge Pass. Thick lines represent the upstream, and thin lines represent the downstream sounding. The 1900 LST MAPR sounding is dotted.

  • View in gallery

    Measurement drive MC to OV at 1640 LST 15 Apr 2004. (a) As in Fig. 6, except that the leg extends now to the eastern end of OV (MC), and (b) horizontal wind direction (thick) and speed over the upper part of the leg.

  • View in gallery

    As in Fig. 9a, but at 1803 LST 15 Apr 2004.

  • View in gallery

    As in Fig. 7, but for 20 Apr 2004.

  • View in gallery

    (a) As in Fig. 8, but at 1744 LST 20 Apr 2004 at Fresno (MGLASS) and at 1700 LST at Independence (MAPR). (b) Horizontal displacement of the two radiosondes while ascending. Dots are MSL height markers in 1000-m intervals. Also visible are the ground footprints of the radiosondes (gray line).

  • View in gallery

    As in Fig. 9a, but at (a) 1546 LST 20 Apr and (b) 1758 LST 20 Apr 2004.

  • View in gallery

    As in Fig. 9a, but for the drive between MC (E) and 7P (W) at 1714 LST 17 Apr 2004 and leg ending at 7P.

  • View in gallery

    Across-valley drive between IN (E) and 7P (W) at (a) 1806 LST and (b) 1907 LST 17 Apr 2004; as in Fig. 14, except that the leg begins at the eastern end of IN.

  • View in gallery

    (top) Schematic interaction of the downslope windstorm with a cold front in the eastern part of OV. The foehn air closest to the ground rides up the frontal surface while the air aloft still descends. (bottom) Depending on the ratio of the depths of the cold front and descending foehn layer, pressure (reduced to a common altitude) may still decrease past the location where the front intersects the surface and only rise at the hydraulic jump farther downstream.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 31 31 6
PDF Downloads 17 17 2

Hydraulic Interpretation of the Footprints of Sierra Nevada Windstorms Tracked with an Automobile Measurement System

View More View Less
  • 1 Institute of Meteorology and Geophysics, University of Innsbruck, Innsbruck, Austria
© Get Permissions
Full access

Abstract

This article reports results from the Sierra Rotors Project, which took place in the central part of Owens Valley, California, east of the Sierra Nevada in March and April 2004. The aim of the study is to describe the footprints of cross-mountain and downslope airflow by mobile surface measurements and radiosoundings. An instrumented car measured wind, temperature, pressure, and humidity. Four case studies cover the spectrum of forcings behind the foehn-like downslope windstorms. Hydraulic theory as a conceptual model was used to explain the data from the car in combination with radiosoundings. All four cases had a colder air mass on the upstream side, thus creating a hydrostatic pressure forcing. With weak flow parallel to the sierra, no downslope windstorm developed and a valley-slope circulation was documented, which for the first time related continuous pressure measurements to the thermal wind system. A second case with a stronger wind component perpendicular to the sierra caused the flow to plunge to the Owens Valley floor. Signatures indicating supercritical regions with accelerated flow reverting to a subcritical state in a hydraulic jump were found. In the third case, the flow separated from the lee slope and subsequently reattached. In the last case, a downslope windstorm developed ahead of a cold front. The downslope windstorm and cold front coexisted in the valley for several hours, with the latter being confined to its eastern side and the storm riding up over it.

Corresponding author address: Georg Mayr, Institute of Meteorology and Geophysics, University of Innsbruck, Innrain 52, A-6020 Innsbruck, Austria. Email: georg.mayr@uibk.ac.at

This article included in the Terrain-Induced Rotor Experiment (T-Rex) special collection.

Abstract

This article reports results from the Sierra Rotors Project, which took place in the central part of Owens Valley, California, east of the Sierra Nevada in March and April 2004. The aim of the study is to describe the footprints of cross-mountain and downslope airflow by mobile surface measurements and radiosoundings. An instrumented car measured wind, temperature, pressure, and humidity. Four case studies cover the spectrum of forcings behind the foehn-like downslope windstorms. Hydraulic theory as a conceptual model was used to explain the data from the car in combination with radiosoundings. All four cases had a colder air mass on the upstream side, thus creating a hydrostatic pressure forcing. With weak flow parallel to the sierra, no downslope windstorm developed and a valley-slope circulation was documented, which for the first time related continuous pressure measurements to the thermal wind system. A second case with a stronger wind component perpendicular to the sierra caused the flow to plunge to the Owens Valley floor. Signatures indicating supercritical regions with accelerated flow reverting to a subcritical state in a hydraulic jump were found. In the third case, the flow separated from the lee slope and subsequently reattached. In the last case, a downslope windstorm developed ahead of a cold front. The downslope windstorm and cold front coexisted in the valley for several hours, with the latter being confined to its eastern side and the storm riding up over it.

Corresponding author address: Georg Mayr, Institute of Meteorology and Geophysics, University of Innsbruck, Innrain 52, A-6020 Innsbruck, Austria. Email: georg.mayr@uibk.ac.at

This article included in the Terrain-Induced Rotor Experiment (T-Rex) special collection.

1. Introduction

Downslope windstorms occur in all parts of the world. Upstream of an obstacle, the layer flowing toward and eventually across is relatively thick and slow. It accelerates as it crosses the obstacle and continues its acceleration as it plunges down the lee slope, thinning in the process for continuity reasons. Simple conceptual models for downslope windstorms use linearized gravity wave theory (e.g., Smith 1979) or hydraulics as introduced by Prandtl (1944) and Long (1953, 1954). The latter is nonlinear and therefore better suited for strong downslope windstorms. To be able to solve the equations analytically, steady state is assumed and often integral averages of fluid properties in the vertical are taken (e.g., constant potential temperature for air and constant density for a liquid, respectively). In the most strongly idealized case, there is a single, neutrally stratified flowing layer with an inert layer of different potential temperature (density) on top. This special case is termed “single-layer hydraulics.” Further assumptions are for the flow to be hydrostatic and incompressible in the integrated layer. Analytical solutions also exist for a continuously stratified upstream flow under the assumption of neutrally stratified fluid above the flow. Long’s (1955) seminal laboratory and analytical studies found an accelerated flow passing over the crest of an obstacle and descending as a jet beneath a slow recirculation zone. Smith (1985) developed a special solution of Long’s equation for continuously stratified flow with uniform speed far upstream of the barrier. He prescribed a stagnant wedge of air above the leeward side, ahead of which the streamlines split. Durran and Klemp (1987, cf. their Fig. 8) and Farmer and Armi (1999, their Fig. 3) showed that this flow configuration, despite its continuous stratification, is qualitatively similar to the one for single-layer hydraulic flow, especially when the depth of the continuously stratified system is less than one-half of the vertical wavelength of the steady internal wave supported by that layer (Durran and Klemp 1987). At and downstream of the obstacle, turbulent mixing near the surface and in hydraulic jumps brings the windstorm layer toward a neutral stratification (i.e., the single-layer hydraulics case). Importantly, detrainment at the top of this layer forms a wedge of nearly neutrally stratified and stagnant air aloft (e.g., Farmer and Armi 1999; Baines 2001, 2005; Armi and Mayr 2007), which decouples the downslope windstorm from the upper airflow. Another analytical hydraulic solution for an arbitrary stratification was derived by Wood (1968) for the case of self-similar flow through a gap. Armi and Mayr (2007) observed this solution even in deep flow across a crest, not just in shallow flow through a gap.

The hydraulic characteristics of the downslope windstorm can be exploited to deduce flow states from measurements along the surface. When air plunges down the lee slope, it has entered a supercritical (with respect to the speed of internal gravity waves) state at the upstream location of control, which usually is the crest or the gap. It accelerates and the thickness of the moving layer decreases. At the underlying surface, wind speed increases and pressure (reduced to a common altitude) falls. Farther downstream of the obstacle, conditions in the preexisting air mass are usually subcritical. The supercritical windstorm adjusts to the downstream conditions in a turbulent hydraulic jump, where some of the kinetic energy is dissipated, the layer depth increases, and isentropes slope upward. The steepness of the jump, the level of turbulence, and the horizontal extent of the turbulent flow all depend on the Froude number of the supercritical flow. At the surface, reduced pressure increases or at least stops decreasing in the case of the Froude number upstream just being slightly above critical and downstream just slightly below critical, and wind speeds weaken. Turbulent mixing along the upper and lower boundary of the windstorm layer contribute to an increase of potential temperature along the surface. This increase is then even more pronounced at and downstream of a strongly turbulent hydraulic jump.

The minority of hydraulic jumps may contain rotors—horizontal vortices. “Rotor” has been used with slightly different meanings in the literature. Earlier literature (Kuettner 1959) and the soaring community apply rotor to a fairly large vortex with a horizontal axis found to the lee of topographic obstacles that persists for longer (order of an hour) times but whose structure may not be visible in individual snapshots. More recent literature (e.g., Doyle and Durran 2002) defines rotor more generally as a vortex with horizontal axis downstream of obstacles. No duration or size constraints are applied. Taking the latter definition to the case of flow separation (e.g., at a sharp edge of the topography), the resulting reversed flow would also have to be called rotor. To avoid misconceptions we have decided to use the more general term “horizontal vortex.”

The aims of the study are to examine measurements along the surface from a weather station on wheels (WOW) and along vertical profiles on the upstream and downstream side from operational and research radiosondes to describe different settings for windstorms down the Sierra Nevada into Owens Valley (OV) in California during the Sierra Rotors Project (SRP; Grubišić and Billings 2007). The measurements are interpreted in the framework of hydraulic theory. Section 2 describes the topography and the instrumentation used here. Four case studies in section 3 show different scenarios of downslope windstorms including one no-show case. Section 4 presents a summary and conclusions.

2. Data and analysis method

The Sierra Rotors Project took place in March and April 2004 as the explorative part of the Terrain-Induced Rotor Experiment (T-REX; Grubišić et al. 2008) in the western United States. The NNW–SSE- oriented Sierra Nevada slowly rises on its western side to above 4 km MSL before it drops almost 3 km into Owens Valley, which is approximately 15 km wide at its bottom. At its eastern edge, the Inyo Mountains rise up to 3 km, and the Whites farther north almost reach back to the altitude of the Sierra crest, and are therefore likely to exert downstream hydraulic control on strong flows down the Sierra slope. The crest-to-crest distance is almost 30 km. Figure 1a shows a topographic map of the target area for both SRP and T-REX. The deep incision of the central Owens Valley in the vicinity of the town of Independence (IN; Fig. 1b) in eastern California was the location of the downstream measurements. While the area was chosen because of the relatively uniform shape of the Sierra Nevada, several gaps are cut into the mountain range. One of them is the Kearsarge Pass (KP) at approximately 3.6 km MSL, 400 m below the surrounding crest (Fig. 1c). The legendary Sierra Wave Project (Holmboe and Klieforth 1957), one of the first large mountain meteorological field campaigns, took place in the same region in the 1950s.

Radiosoundings measured near-vertical profiles on the upstream side of the Sierra Nevada at Lemoore (LE; Naval Air Station) and Fresno [Mobile GPS/Loran Atmospheric Sounding System (M-GLASS) mobile research system from the National Center for Atmospheric Research (NCAR)]. Downstream, radiosondes were launched from a station in the vicinity of Independence (MAPR in Fig. 1b). A 915-MHz wind profiler from NCAR was also deployed at MAPR and another one was located at the Independence airport, just north of the city. One row of automatic weather stations (AWSs) from the Desert Research Institute (DRI) was strung across the valley at Independence and two rows just south of it. Extending an approach pioneered half a century ago in that very same region (Holmboe and Klieforth 1957, p. 9), an instrumented car detected “footprints” of strong downslope events along the roads in Owens Valley, drawn in Fig. 1b. Contrary to the historic method of stopping at discrete intervals to measure atmospheric variables, measurements took place continuously while driving. The most frequently driven stretch was from the beginning of Mazourka Canyon (MC) in the Inyos through Independence to the western end of the canyon-like Onion Valley (2800 m MSL) in the western slope of the Sierra (Fig. 1c). Also shown are distance markers along the leg for reference in the figures. An earlier version of the car platform is described in Mayr et al. (2002). Figure 2 shows the system used during the SRP. This unit was equipped with sensors of higher accuracy and additionally measured horizontal wind, and in addition 3D position, movement, temperature, dewpoint, compass heading and tilting, and pressure every 2–3 s. At typical car speeds, a spatial resolution of 50 m could be achieved.

Table 1 lists the details of the sensors. A fully adjustable roof mount makes it possible to use any car and thus to rely on rental cars. The only necessary purchase in the field is a fit kit for the load bars. The position of the car was determined by GPS with Wide Area Augmentation System (WAAS) correction. Since horizontal pressure changes of less than a few hectopascals and often less than 1 hPa are the footprint of the hydraulic flow overhead, particular effort was spent on the measurement of pressure and its reduction to a common altitude. Computational fluid dynamics simulations of flow around vehicles performed by car manufacturers showed a minimal flow disturbance by the car just above the front axle at a height that exceeds the roof height by about 1 m. This is where the boom with the quadplate static pressure port (Nishiyama and Bedard 1991), the sonic anemometer, and the electronic compass were placed. A hose connected the pressure port to the pressure sensor, which sat inside the car cabin to minimize temperature effects on the pressure measurements. Calibration drives along a level stretch of a road with speeds alternating between approximately 10 mph (≈4.5 m s−1) and values increased in 5–10-mph steps up to 75 mph (≈33.5 m s−1) found no speed dependence of temperature and dewpoint but a speed dependence of static pressure as a linear function of the square of driving speed. All pressure measurements were corrected with this calibration equation. The pseudovertical pressure reduction method developed by Mayr et al. (2002) was used to reference pressure to a common altitude. An altitude error of 1 m introduces an error of reduced pressure of about 0.1 hPa. To achieve this and even better accuracies of altitude, differential processing of GPS measurements is necessary. Because of shading by the complex topography, a sufficient number of satellites for accurate differential processing of vertical position is available only for a few hours each day. Horizontal positions from each meteorological measurement drive were therefore used to look up the corresponding altitude obtained in surveys during periods of optimal satellite visibility. The horizontal mean absolute position difference between the differential GPS (DGPS) and GPS was 0.95 m.

The vertical temperature profile, which we only know slanted along the slope, is the largest error source given a maximum elevation difference along the drive of 1.6 km. Its magnitude depends on the particular flow situation. In general, a 1-K difference of the mean temperature of a 1-km-thick layer yields a pressure difference of ⅓ hPa, that is, 0.5 hPa over 1.6-km elevation difference on the most frequently driven stretch of the road. The vertical temperature profile during foehn breakthrough is most affected, because it is warmer above the valley than the profile along the slope because of the descending isentropes. Using the slanted temperature profile will therefore dampen the pressure contrast along the slope.

The relatively large horizontal extent of the drive of almost 20 km requires consideration of the superimposed synoptic-scale pressure gradient. The synoptic scale [(O(500 km)] pressure increased by between 0.2 and 0.4 hPa from the western to the eastern end of the road in all four cases. This synoptic-scale impact is opposite in sign to the effect from the slanted pressure reduction.

Vector addition of the raw wind rotated into the common geographical reference system and the displacement vector of the car obtained from the GPS measurements yielded the horizontal wind vector υ.

All sensors other than the temperature sensor have response times of 1 s or less. A response time of more than 20 s of the temperature sensor in its radiation shield introduced a lag, which became particularly noticeable on the steep parts of the road (e.g., Onion Valley). The corresponding parts in the figures are therefore whitened and should be interpreted with care. All figures from car measurements will be shown against the windstorm direction for better comparability. The gap in the cross section figures between x = 17 km and x = 18 km comes from data in the town of Independence, which are left out because they are not representative. Unfortunately changes in the flow regime happened frequently near that region, maybe because of the terrain levels out there.

To identify the airmass footprints of the windstorm from radiosonde profiles and car measurements, conserved variables were used—mixing ratio m and virtual potential temperature θυ:
i1558-8432-47-10-2581-e1
where T is temperature, p0 is a reference pressure, Rd is the gas constant of dry air, ϵ is the ratio of molecular weight of water vapor to the mean molecular weight of dry air, and cpd is the heat capacity of dry air at constant pressure.

As θυ is conserved for frictionless, unsaturated airflow, it is possible to distinguish the origin of the flow (pass or crest) by comparing windstorm virtual potential temperatures computed from surface measurements with the vertical profile on the upstream side of the barrier.

Bernoulli’s equation,
i1558-8432-47-10-2581-e2
with pr being pressure reduced to a common altitude, can be used for a zeroth-order interpretation of the flow. Of the underlying assumptions, the absence of (turbulent) friction is clearly not fulfilled for flow close to the surface or passing through hydraulic jumps. The computation of wind speeds from Eq. (2) with measured pressure changes will therefore yield higher values than were measured. Equation (2) states that a region of decreasing reduced pressure will be accompanied by an increase in wind speed. Therefore, in the region of the accelerating downslope flow, a pressure drop is expected. A relative minimum of (reduced) pressure and a relative maximum of wind speed will be found immediately upstream of a hydraulic jump. After the hydraulic jump, reduced pressure should increase and wind speed decrease.

Stationarity of the flow during the time it takes to complete a measurement drive [O(15–30 min)] varied from case to case. Data from the DRI AWS network suggest that stationarity can be assumed. Maximum pressure differences at DRI stations were 0.1–0.2 hPa during the drives, wind direction was very constant, and the maximum wind speed change was 4 m s−1.

Car measurements were checked against the DRI AWS, but the wind profiler had a lot of data gaps and quality issues on most drives. Data were consecutively checked against direct cloud observations, photographs, and a time-lapse video system from Yale University mounted at the MAPR site.

3. Case studies

A pressure gradient forces air to flow over or around a mountain range. We discern two major factors, namely the synoptic-scale pressure field, hence called the dynamic forcing because it causes the synoptic wind field, and the contribution from differences in temperature between the upstream and downstream side of the barrier, hence called the hydrostatic forcing. To distinguish the magnitude of the two forcings for the four cases, we compare the vertical structure of the atmosphere both upstream and downstream of the Sierra Nevada. Analyses from the European Centre for Medium-Range Weather Forecasts (ECMWF) global model are used instead of directly comparing observations from, for example, radiosoundings. This method is more consistent, because suitable measurements were not available for all days, or not at representative locations. One drawback of this approach is that the model topography is much smoother and Owens Valley is not resolved (cf. Fig. 3).

All cases showed a more or less similar synoptic setting with a trough over the Pacific coast directing relatively colder air toward the Sierra Nevada. A closer look at the vertical structure of the upstream and downstream air masses at or shortly before the breakthrough of downslope windstorms to Owens Valley floor is given in Fig. 4. Shown is a comparison of model soundings at representative upstream and downstream locations for each day. Therefore the upstream profile locations were chosen depending on the synoptic-scale flow direction and the downstream profile was taken at a place where the model resembles a saddle behind the main barrier. This is marked by dots in Fig. 3; the connecting lines are aligned with the direction of the synoptic-scale flow.

A prerequisite for the impinging flow to reach the valley floor is that its mean virtual potential temperature is lower than the air it replaces. Therefore, Δθυ = θυupθυdown is used to describe the extent of the hydrostatic forcing. Here, θυup and θυdown are the mean virtual potential temperatures of the layer between z = 2.5 and 3.5 km on the upstream side of the model’s representation of the Sierra Nevada and between z = 1.775 and 2.5 km on the downstream side. The upstream layer was chosen, because the crest height of the Sierra Nevada in the model is 2.5 km and the downstream layer represents the layer between valley surface and crest height. The effects of stability and stratification will be discussed in the following sections on the basis of the real soundings, where possible.

On 18 April 2004 (cf. Fig. 4a), flow was very weak and Δθυ was −0.3 K along section B (cf. Fig. 3). No downslope windstorm developed. The flow system in Owens Valley and along the eastern slope of the Sierra Nevada was mainly thermally driven.

The 15 April 2004 case shows a stronger cross-barrier flow and a Δθυ of −3.1 K along section B (Fig. 4b). A downslope windstorm developed. Signatures suggest a superposition of gap flow through the Kearsarge Pass and flow across the crest with a hydraulic jump near Independence. There were indications of a flow separation with a horizontal vortex over the eastern slope of the sierra.

The 20 April 2004 case combines a more moderate airmass difference of −1.4 K between the upstream and downstream “reservoirs” with a strong impinging cross-barrier flow component (Fig. 4c). A strong downslope windstorm developed, again with flow separation, gap flow, and hydraulic jump. Cross section A in Fig. 3 is used because of the prevailing northwesterly flow.

The last case of 17 April 2004 shows a downslope windstorm triggered by the passage of a cold front. With a weak impinging westerly flow component but airmass differences associated with the cold front (Δθυ = −1.4 K along section B, Fig. 4d) extending vertically up to 8 km MSL, the downslope winds reached the valley floor and confined the passage of the cold front to the eastern parts of the valley. The cold front had gone around the northern part of the Sierra Nevada and marched down Owens Valley.

We split observations from interpretation to better distinguish between (limited) measurements and the footprint as deduced from hydraulic theory, leaving some room for other interpretations as well. Local standard time (LST) will be used throughout the paper, since diurnal changes are important for the flow system in Owens Valley. LST is 8 h behind UTC.

a. 18 April 2004, no intensive observation period (no-IOP): Weak dynamic and hydrostatic forcing

This day showed the evolution of the thermally driven valley and slope wind system in Owens Valley despite a weak southwesterly flow across the region and a slightly colder air mass (Δθυ = −0.3 K) west of the sierra (cf. Fig. 4a). No gravity wave structures were visible in the cirrus clouds, which appeared at midafternoon and covered most of the sky by the end of the day.

1) Observations

The diurnal evolution of θυ computed from Eq. (1) along the western slope of Owens Valley is shown in Fig. 5. The car drove from the approximate valley center at Independence up to Onion Valley, about 600 m below the Kearsarge Pass.

Since no dedicated radiosondes were launched on that day, a comparison of the model soundings in Fig. 4a is used to estimate the state of the atmosphere upstream and downstream of the Sierra Nevada at 1300 LST. The upstream side shows a 1.8-km-deep stagnant neutral layer topped by a much deeper stable layer. The relatively colder air mass below z = 3.5 km exerts a weak hydrostatic forcing on the flow. The dynamic forcing was very weak, expressed by the 5 m s−1 south-southwesterly flow in this layer, which did not mix down into the valley. Instead the upvalley and upslope wind regime persisted.

According to measurements at the fixed DRI AWSs (not shown), flow up the western valley slope had commenced about a half-hour before the first measurement drive at 0907 LST. Figure 6a shows that the upslope winds observed by the car were fairly uniform from the valley center at Independence to the beginning of the complex Onion Valley. As the road twists its way up the canyon-like valley, wind direction changed with the orientation of the canyon axis. The mixing ratio remained constant at 2 g kg−1 along the western slope of Owens Valley.

Figure 6b depicts the drive starting at 1636 LST. In the afternoon the car encountered a valley atmosphere that was closer to neutral but not completely mixed yet, with the exception of humidity, which had already been uniform in the morning. The reduced pressure still decreased toward the sierra crest, albeit weaker. Consequently, upslope winds were still present. Upvalley wind speed had increased to 8–10 m s−1.

2) Interpretation

During the previous night, nocturnal cooling had stabilized the valley atmosphere. By midmorning (0907 LST), the air between the valley floor at 1200 m MSL and the Sierra Nevada slopes at 2200 m MSL was still stably stratified, as the increase of θυ along the drive in Fig. 5 shows. The whole valley atmosphere continued to warm rapidly. Since Owens Valley is semiarid, a large portion of the net radiation balance goes into sensible heat flux. By the time of the next drive (1048 LST), the lowest 700 m were already well mixed. By 1507 LST the maximum of the layer-averaged θυ was reached and the profile was nearly mixed all the way up. Increasing high cloud cover stopped a further warming and made the profile starting at 1636 LST (which is approximately 1 h before sunset in the valley center) cooler.

The car measurements illuminate the mechanism driving the upslope winds. The driving force for the upslope wind was a decrease of pressure (reduced to a common altitude) between the valley floor and the eastern slope of the sierra, which is directly exposed to the morning sun. The insolation had not yet eroded the nocturnal cold pool in the valley at the time of the midmorning drive. Low values of mixing ratio in Owens Valley reflect its semiarid nature. In the afternoon, the smaller pressure difference between valley and slope, presumably due to increased shading, explains the overall weaker upslope winds compared to the morning. The larger-scale pressure gradient distribution, which led to southerly winds of about 10 m s−1 at 700 hPa (cf. Fig. 4a), reinforced the thermal upvalley winds. However, it did not dominate the valley wind system, since winds at the surface turned downvalley at around 2300 LST. Later that night, upstream winds became less southerly and the downslope windstorm reached the valley for a couple of hours around 0100 LST of the following day (not shown).

b. 15 April 2004, IOP 12: Hydrostatic and dynamic forcing

This day showed impressive wave clouds and strong downslope winds with gusts up to 15 m s−1. A reservoir of colder air (Δθυ = −3.1 K) lay on the upstream side of the Sierra Nevada. Warm air was advected into Owens Valley from the south as the boundary between the two air masses lay SW–NE over the Sierra Nevada. The temperature difference between the upstream and downstream reservoir led to a hydrostatically caused pressure gradient across the Sierra Nevada. Dynamic forcing from the impinging larger-scale flow was strong with 10–15 m s−1 cross-barrier flow at crest height. The downslope windstorm reached the bottom of Owens Valley in the afternoon.

1) Observations

By late morning the θυ profile measured by the car showed a stably stratified valley atmosphere (Fig. 7) similar to the no-show case of 18 April (section 3a). After a breakthrough of downslope windstorms to the valley floor (1640 LST), θυ over the uppermost slopes decreased by about 5 K.

The midmorning soundings in Fig. 8a show an overall colder profile upstream of the Sierra Nevada. A capping inversion was present between 5 and 6 km MSL. The afternoon sounding, Fig. 8b, shows that the upstream profile had become even colder and the inversion even more pronounced. Air up to 2.5 km MSL was blocked. At the MAPR site (downstream of the sierra), the air was neutral up to 5 km MSL at 1600 LST. Winds in this layer were weak southeasterly. Near the top of the descending layer, the downdraft must have overpowered the rise rate of the radiosonde since it was dragged downward. The measurements in the layer between approximately 5.5 and 6 km MSL therefore undulated along the inversion rather than penetrating vertically through it. The sonde only rose again once it had gotten close to the next downstream mountain range.

Three hours later a strong westerly flow was present with its maximum a few hundred meters above Owens Valley. Photographs show torn cumuli fracti just above Independence and again over the western slope of the Inyos. Both were topped by smooth, approximately valley-parallel lenticular banks.

Figure 9a illustrates the details of the footprint of the downslope windstorm as captured in the car drive from 1640 LST. Beginning at the foot of the Sierra Nevada (x = 9 km), reduced pressure fell and dynamic pressure rose. Note the smoother decrease of pressure, which is an integral measure, as compared with dynamic pressure, which is computed from the wind speed along the surface and as such strongly influenced by turbulent friction and topographical details. Winds along the eastern slope of the Sierra Nevada, in the complex topographic part of the Onion Valley canyon, were overall downslope. However, Fig. 9b demonstrates that there was return flow in the upslope direction between the sierra slopes at x = 6 km and the exit of Onion Valley (crossed by the road at x = 8 km).

Past x = 9 km θυ increased to a maximum at x = 16 km, where mixing ratio dropped abruptly. The pressure leveled out (and even increased about 0.4 hPa) after approximately x = 16.5 km, θυ stayed constant, and the mixing ratio rose to its former level. At approximately x = 20 km, air had rebounded high enough in the hydraulic jump region to reach its condensation level and become visible as roll clouds.

Figure 10 portrays the next car drive more than an hour later, beginning at 1803 LST. Here, θυ had cooled by 1–3 K and remained nearly constant throughout the drive as compared with the increase in the previous drive. At the westernmost stretch of the road, dynamic pressure (i.e., kinetic energy) was higher than on the previous drive. Reduced pressure decreased and kinetic energy increased from x = 9 km on. After x = 16 km, reduced pressure leveled out and rose past x = 22 km where kinetic energy dropped sharply. The radiosonde at 1900 LST (Fig. 8b, dotted lines) showed a cooling of the whole valley atmosphere. The downslope windstorm extended to the sounding location, contrary to the previous sounding from 1600 LST.

2) Interpretation

At noon, solar heating of the ground had weakened the stability and brought the virtual potential temperatures at the valley bottom closer to the ones of approximately 300 K at the western end of the road. However, later in the afternoon they never reached these values of around 309 K. Instead, strong westerly downslope winds reached the valley floor at around 1500 LST and extended all the way across the valley, because the impinging air had become colder. With the breakthrough of the downslope windstorm θυ in the valley dropped, which can be seen in the difference between the drives at 1127 and 1640 LST. The density difference may have become the driving force for the foehn wind. The increase of θυ by about 3 K across the valley at 1640 LST is discussed further below.

The morning soundings (Fig. 8a) are puzzling. Why had the flow not already reached the valley by that time since the upstream sounding had colder virtual potential temperatures than the one downstream? A comparison with the soundings after breakthrough (Fig. 8b) provides the necessary clues to solve the puzzle. We hypothesize that by the time upstream air from the late morning sounding (Fig. 8a) reached the crest, turbulent mixing across the interface (between 5 and 6 km MSL far upstream) had brought the virtual potential temperature to a value higher than the ones of the valley air. The low Richardson number of 0.3 across the interface supports the turbulent mixing argument. Similar mixing across the upper interface of a flowing foehn layer was documented by Armi and Mayr (2007) in the European Alps and by Pawlak and Armi (2000) and Baines (2001, 2005) in laboratory experiments of stratified flows down a slope. In the afternoon, the upstream air above 3 km MSL became approximately 2 K colder. This was enough for air from the lower end of the interface passing through the lower terrain formed by the gap of Kearsarge Pass to be colder than air on the upper leeward slopes of the sierra. As the impinging flow plunged downslope, further turbulent mixing along the lower interface with the terrain and along the upper interface took place, ultimately mixing down air with a virtual potential temperature that was the average of the upstream interface between 5 and 6 km MSL (Fig. 8b). The increase of θυ showed clearly in the car drive begun at 1640 LST (Fig. 7).

As we expected from Eq. (2) and in accordance with the assumed supercritical flow from hydraulic theory (cf. section 2), pressure minima and wind speed maxima were found over the eastern Sierra Nevada slopes after the foehn breakthrough. The westernmost and uppermost part of the road (cf. Fig. 1c) is nearly perpendicular to the local sierra crest and consequently the car captured westerly flow between x = 0 and x = 3 km. The subsequent terrain is highly complex and strongly modifies the flow. We interpret it as follows: at x = 3 km, the air from Kearsarge Pass was deflected by a protruding ridge. Only wind direction (southerly) but neither θυ nor reduced pressure changed. However, as the road continues farther north, it veers from the direct swath of the canyon. Here the car found upslope flow during the 1640 LST drive, indicating the presence of a horizontal vortex and flow separation in this topographically highly complex area. As winds were turning westerly on the moraine bordering the Onion Valley canyon to the north at x = 7.5 km, it can be assumed that the vortex was terrain induced and confined between the sierra slopes at around x = 5 km to the exit of Onion Valley, corroborated by the fact that at x = 8 km the road crosses the Onion Valley canyon, and again found upslope winds. At x = 9 km, the road is back in the Onion Valley swath. Consequently, some westerly winds were captured again (cf. Fig. 9b). The air coming down the uppermost part of Onion Valley had most likely passed through the gap of Kearsarge Pass. This is corroborated by lower virtual potential temperatures there after breakthrough of the downslope windstorm (Fig. 7, between 1127 and 1640 LST) and by comparison with the sounding at Lemoore (Fig. 8b), which showed comparable values at the altitude of Kearsarge Pass. Past x = 9 km, θυ increased until matching the mean θυ over the inversion in Fig. 8b, indicating that air from heights above the sierra crest mixed down. The short drop in mixing ratio centered around x = 16 km could indicate that air originated from the mixing ratio minimum at 5 km MSL found in the Lemoore sounding. The nearly constant θυ and the weak pressure increase together with visual observations suggest a transition of the flow from supercritical to subcritical over Independence. The continuing strong westerlies in this jump region are still puzzling.

Later on, the two airstreams from Kearsarge Pass and over the Sierra Nevada crest were well mixed. Overall, the downslope flowing air was already colder since the sun had set in the meantime. Additionally, the upstream air mass might have become colder still. The assumed jump was about in the same location at x = 17 km. Most likely, stronger turbulence had mixed the only weakly stratified air coming through the Kearsarge gap and across the crest (Fig. 8b) more quickly. The upstream influence of the Inyos slowed down the flow, deflected it downvalley, and increased the pressure from x = 22 km on.

The 1600 LST sounding at Independence had mostly weak easterly flow while the car measured already strong westerly foehn flow. Since the downstream radiosonde location is outside of the swath of air coming down from the Kearsarge Pass gap, it is possible that the radiosonde location was in a deep eddy with vertical axis at the southern side of the gap flow jet, as simulated by Grubišić and Billings (2007) for the IOP 8 case. Another possibility though is that the flow separated from the surface farther upstream and that the radiosonde location was in a hydraulic jump region. The evidence from the radiosonde at 1600 LST (Fig. 8b) itself is not conclusive. The wind profile fits both the hypothesis of an eddy with a vertical axis and a hydraulic jump.

c. 20 April 2004, IOP 14: Mainly dynamic forcing

Similar to the previous case of 15 April, a colder (Δθυ = −1.4 K) air mass lay upstream of the sierra (Fig. 4c). The impinging cross-barrier component was even stronger (20–25 m s−1) and the flow direction more perpendicular to the crest than in Fig. 4b. A thermally driven upvalley and upslope wind system in the morning was replaced by a strong downslope windstorm, which progressed down into the valley and reached the bottom at around 1300 LST.

1) Observations

Figure 11 shows θυ profiles for all drives in the Independence area. The drive starting at 1247 LST showed a cold pool near the valley floor. Subsequent drives at 1329 and 1457 documented the progression of the downslope windstorm into the valley. The eastern half of the valley at this time was shaded by lenticular clouds. Later, at 1546 and 1703, the downslope windstorm was fully established and virtual potential temperatures increased across the valley. At 1758 LST a more uniform profile was found across the valley. By 1912 LST a cold pool was established again.

Soundings at the time of the fully established downslope windstorm in the valley are depicted in Fig. 12a. A very stable, approximately linearly stratified flowing layer on the upstream side between 3.5 and 4 km MSL is bounded above and below by nearly neutrally stratified layers of about 400-m thickness. The downstream sounding showed a shallow neutral layer of 350-m thickness near the ground. Wind speeds in the shallow layer increased from 8 m s−1 (surface in Fig. 12) to ≈20 m s−1 in the stable layer atop the shallow layer. Figure 12b shows that radiosoundings do not measure a vertical but rather a slanted profile, which exaggerates the difference in θυ over the stable layer in the MAPR sounding.1

At noon the accessible parts of the sierra and Inyo slopes still had a thermal upslope regime. The downslope windstorm broke into the valley around 1300 LST when lee waves became visible through altocumulus lenticular clouds. During the establishment of the downslope windstorm, flow meandered from SW over W to NW. Later on the windstorm grew steadier and increased its strength. The drive beginning at 1546 LST (Fig. 13a) found lower θυ over the upper sierra slopes than over the valley center, decreasing again beyond x = 19 km. At x = 0 km, winds were strong westerly, following the 90° northward bend of the canyon axis at x = 4 km. Between x = 6 and x = 9 km the flow was weak easterly (i.e., up the canyon). Further on winds were again westerly, accelerating strongly between x = 15 and x = 16 km, staying strong until x = 20 km. Maximum westerly wind speeds were found in low pressure regions (i.e., on the uppermost and lowermost slopes and over Owens Valley center). A pressure maximum was located over the sierra slopes, between x = 10 and x = 15 km.

Two hours later (Fig. 13b) wind speeds and dynamic pressure were higher on the upper part of the sierra slope than before. Reversed (i.e., upslope) flow in the complex topographical part of the road was missing; instead a weak turning wind region was found at the same place. Overall, westerly wind speed increased to about 15 m s−1. The accelerating stretch upstream of Independence expanded horizontally between x = 13 and x = 16 km. Mixing ratio was several tenths of g kg−1 higher between x = 0 and x = 10 km than over Owens Valley center, and also more fluctuating. The line of jump clouds, which was at Independence at the time of the previous drive, was unstationary: it had first moved westward, then eastward and again westward. Its shape was close to that of a wide open “v” with a break in clouds north of the road and north of Independence.

2) Interpretation

The soundings (Fig. 12a) show a stability structure similar to Wood’s (1968) solution. We assume that the bounded flowing layer (cf. observations) descended with turbulent mixing along the lower boundary and turbulent detrainment on its upper boundary, thereby creating a shallow foehn jet.

Figure 13a shows a situation similar to the 15 April case for the drive beginning at 1546 LST (cf. Fig. 9a): gap flow in the uppermost part of the Onion Valley and upslope flow between x = 6 and x = 9 km. Here some westerly winds mixed down again. Virtual potential temperatures were noticeably higher presumably due to the merging of the airstream through Kearsarge Pass with the one across the sierra crest. At x = 15 km the flow reattached to the ground again. The fall in reduced pressure combined with an increase of wind speed starting at x = 15 km fits a supercritical flow situation. The mixing ratio also decreased by a few tenths of grams per kilogram in agreement with the notion of drier air from even farther aloft being entrained. The end of the pressure fall indicates a hydraulic jump just upstream of Independence at x = 17 km. Visual observations found cumulus fractus, indicating a jump cloud just west of Independence. Air with warmer virtual potential temperature was mixed down. Westerly winds remained relatively strong farther toward the Inyos. From x = 19 km on, pressure increased slightly and wind speeds decreased as the flow impinged upon the Inyos. Another row of clouds parallel to the valley axis was spotted at about this location, so that the assumed jump may have been of undular type.

At 1758 LST the downslope windstorm had increased and turbulent mixing was merging the two airstreams (through the Kearsarge Pass gap and over the sierra crest) as in the 15 April case (cf. section 3b). This is supported by the mixing ratio fluctuations. The windward edge of the region where reduced pressure fell and kinetic energy increased had moved upwind from x = 15 km in the previous drive to x = 13 km.

d. 17 April 2004, no-IOP: Cold front passage

Cold air behind a front had gone around the northern side of the Sierra Nevada and moved down Owens Valley, where it encountered a downslope flow over the Sierra Nevada, which confined the surface frontal passage to the eastern parts of Owens Valley. The dynamic forcing from the inflow was weak, about 5 m s−1 at crest level, and Δθυ was −1.4 K.

1) Observations

This case was unforecast. Therefore, no dedicated radiosoundings or car drives prior to the establishment of the downslope windstorm are available. The ECMWF model soundings (Fig. 4d) show a 1.8-km-deep neutral stratified layer, presumably from the passage of the cold front on the upstream side. Air farther aloft, which moved toward the Sierra Nevada at approximately 5 m s−1, was stably stratified and colder up to the tropopause than on the downstream side. The downslope windstorm therefore had higher virtual potential temperatures than the surface cold front, which had to pass around the northern side of the Sierra Nevada. The day was overcast with low cumulus fractus moving down Owens Valley and from which eventually some rain drops fell. A foehn wall was visible over the Sierra Nevada and a cloud bank could be seen over the valley center. In between these two was a more or less clear space through which cirrus clouds could be observed.

Most measurement drives were carried out on a subsection of the road between Independence and the beginning of the Onion Valley canyon to sample the interaction zone of front and downslope windstorm at a higher temporal resolution. Prior to the frontal passage at Independence, the car measurements starting at 1714 LST and shown in Fig. 14 found a uniform θυ and mixing ratio distribution as well as westerlies all the way across the valley.

An hour later (cf. Fig. 15a), higher mixing ratio values and decreasing θυ were encountered just west of Independence. Wind changed abruptly from downslope to downvalley with distance downstream of x = 15 km and dynamic pressure dropped sharply.

Similar θυ and pressure profiles were present 1 h later (Fig. 15b). Mixing ratio values of almost 4 g kg−1 were found 4 km farther east than before (cf. Figs. 15a,b).

The DRI AWS network showed that downslope windstorm and downvalley flow coexisted for several more hours until the cold air finally also filled the western half of the valley.

2) Interpretation

The drive beginning at 1714 LST (Fig. 14) can be interpreted as going through a supercritical flow. Concomitance of abrupt changes to high wind speeds (cf. dynamic pressure–kinetic energy) with an increase of θυ and a drop of mixing ratio may show stronger downward mixing upstream of the jump or express flow nonstationarity We deduce a hydraulic jump between x = 13 and x = 15 km, from the overall decrease of dynamic pressure and the slight but continuous increase and subsequent leveling out of pressure farther downstream. At the jump and downstream of it, downward mixing continued over a larger stretch.

The drive starting at 1806 LST (Fig. 15a) shows that the cold front had already moved into the eastern half of the valley. The downslope windstorm encountered the surface intersection of the front at approximately x = 15 km. The frontal air mass was 2 g kg−1 moister and slightly colder than the foehn air. Winds were downvalley and weaker than the cross-valley foehn flow.

Surprisingly, though, reduced pressure east of the frontal intersection with the surface did not rise but rather fell, although only slightly. A first expectation would be for the reduced pressure to rise because of the increasing thickness of the overlying colder frontal air. Another mechanism must overcompensate. We offer the explanation schematically sketched in Fig. 16. Above the cold front the downslope windstorm continued to plunge down making the average temperature in the column above the front warmer than farther upstream, thus causing the pressure to decrease slightly. As the downslope windstorm air closest to the ground encountered the cold front, it turbulently mixed and lifted off the surface. The largest part of the descending layer, however, still continued its descent. The adjustment to subcritical flow must have happened farther downstream (east). Observations from both NCAR wind profilers between 1800 and 1900 LST (not shown) found northerly (downvalley) winds up to 500 m AGL and weak westerlies above. Unfortunately there were no measurement drives farther across the valley to corroborate that hypothesis. A similar case was studied by Gohm et al. (2000) during the Mesoscale Alpine Programme in the Inn Valley, Austria, on 6 November 1999.

Figure 15b shows that the location of the frontal intersection was unchanged an hour later. The measurements suggest that turbulent mixing of the frontal air with the downslope windstorm air, however, had smoothed the mixing ratio distribution. Turbulent mixing just at the intersection had also mixed downslope windstorm air from aloft down so that θυ increased. The frontal air mass remained stably stratified. Again, pressure decreased past the frontal intersection in accordance with the conceptual view of Fig. 16.

4. Discussion and conclusions

The pilot phase of the Terrain-Induced Rotor Experiment in 2004 provided the opportunity to study the footprints of downslope windstorms in the lee of the largest terrain drop of the contiguous United States (east of the Sierra Nevada) with a car-mounted measurement platform. We covered four cases spanning the spectrum from just-not-occurring to strong downslope windstorm. All four cases had a colder air mass west of the sierra in common, thus providing different “reservoirs” of air on either side of the Sierra Nevada. Different reservoirs together with the steep and large drop in elevation, which leads to strongly nonlinear flow, let us use the hydraulics framework as a conceptual model and “background” information in an atmospheric setting, which was severely underdetermined through the sparse available measurements. The available measurements are suitable for a hydraulic interpretation, as they provide vertical profiles of both reservoirs and line measurements along the floor of the downstream reservoir of the integral quantity pressure. One alternative approach to obtain the necessary further background would have been the application of the (linearized) gravity wave conceptual model. The decrease of surface pressure along the leeward slope of the sierra can then be interpreted as being under the descending part of a standing gravity wave, and the slight increase in pressure as being past the upstream tilted trough of the wave. Interpreting the flow separation in the highly complex terrain part of the sierra slope as being caused by an adverse pressure gradient from lee waves rather then from edges of the topography itself is more difficult. When the separation occurred it was always at the same location. Additionally, the horizontal extent of O(1 km) is rather small for a lee wave to have sufficient pressure amplitude to trigger the separation. However, given the sparsity of the data, some of our interpretations of the flow field have to remain hypothetical and even speculative, just like an archaeologist imagining the acronym-giving tyrannosaurus rex from some footprints and bones. The much richer dataset of the main field phase in 2006 will be able to shed further light upon our explanations and hypotheses.

No downslope windstorm developed when the potential temperature of the upstream air was warmer than the air in the valley and the cross-barrier flow component was weak (18 April 2004). Under these conditions a thermally (through radiation) driven upvalley and upslope wind system developed. The pressure measurements from the car reduced to a common altitude with the slantwise algorithm of Mayr et al. (2002) documenting the resulting decrease of pressure toward the sierra as the radiation-induced driving force behind the circulation. A larger airmass difference with colder upstream potential temperatures (cf. Figs. 4a,b) and an increase of cross-barrier wind speed let the flow reach Owens Valley floor (15 April 2004). The car measurements of pressure as an integral value of processes in the atmosphere aloft in addition to the airmass tracer variables of θυ and mixing ratio fit the hydraulic concept to explain the downslope windstorm. Reduced pressure decreased along the downward-sloping western half of the valley while kinetic energy (i.e., wind speed) increased, which we interpret as a supercritical flow region. The transition to a subcritical state was not conclusive from the pressure measurements alone. A rise of θυ downstream of the leading edge of the jump indicated that stronger turbulence in the jump might have mixed down air from even farther aloft.

During both the 15 and 20 April 2004 cases, reduced pressure was usually level in the jump region after a decrease of approximately 2 hPa in the supercritical flow upstream. On both occasions the downslope windstorm air was sufficiently moist for a cloud bank with a turbulent windward edge to be present at the location where we infer the existence of a jump from the car-based measurements. The lack of a significant pressure increase together with only a small deceleration of the wind across the jump can occur with weak jumps, where the Froude number is slightly above critical upstream of the jump and slightly below critical downstream of it. Since no soundings immediately upstream and downstream of the presumed jump locations were available, Froude numbers could not be calculated to substantiate this hypothesis.

Since the crest of the Sierra Nevada is not as homogeneous as it appears at first sight, two airstreams could be identified from their respective virtual potential temperatures: a colder one going through the gap of Kearsarge Pass and a warmer one crossing the crest, which is about 400 m higher there. Consequently the breakthrough of the downslope windstorm occurred downstream of the gap prior to the onset downstream of the surrounding crest. The importance of gaps for the development of downslope windstorms was recently demonstrated by Armi and Mayr (2007), Mayr et al. (2007), and Gohm and Mayr (2005), and had already been suggested by Vergeiner (1983).

The peculiarity of the 17 April case was the coexistence of a surface cold front marching down the eastern side of Owens Valley and the downslope windstorm plunging into the western part of the valley. The downslope windstorm air rode over the cold frontal air.

Similar to other downslope windstorm regions, the penetration of the downslope windstorm to the ground is strongly diurnally modulated. Insolation was needed to warm the nocturnal cold pool to θυ values close to the ones observed on the upstream side in the proximity of the crest level.

Acknowledgments

We thank everybody involved in the Sierra Rotors Project, and particularly Vanda Grubišić and her team from DRI for AWS data and NCAR for the radiosoundings. Special thanks are given to Larry Armi for discussions and for being there on 17 April 2004. Harold Klieforth shared his experience from the Sierra Wave project, his enthusiasm, and his vast knowledge of the weather in the area. Thanks! Some financial support from the University of Innsbruck is acknowledged. Author TR was partly supported by Austrian Science Foundation Grant FWFP18940-N10.

REFERENCES

  • Armi, L., , and G. J. Mayr, 2007: Continuously stratified flows across an Alpine crest with a pass: Shallow and deep föhn. Quart. J. Roy. Meteor. Soc., 133 , 459477.

    • Search Google Scholar
    • Export Citation
  • Baines, P. G., 2001: Mixing in flows down gentle slopes into stratified environments. J. Fluid Mech., 443 , 237270.

  • Baines, P. G., 2005: Mixing regimes for the flow of dense fluid down slopes into stratified environments. J. Fluid Mech., 538 , 247267. doi:10.1017/S0022112005005434.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and D. R. Durran, 2002: The dynamics of mountain-wave induced rotors. J. Atmos. Sci., 59 , 186201.

  • Durran, D., , and J. Klemp, 1987: Another look at downslope winds. Part II: Nonlinear amplification beneath wave-overturning layers. J. Atmos. Sci., 44 , 34023412.

    • Search Google Scholar
    • Export Citation
  • Farmer, D. M., , and L. Armi, 1999: Stratified flow over topography: The role of small-scale entrainment and mixing in flow establishment. Proc. Roy. Soc. London, A455 , 32213258.

    • Search Google Scholar
    • Export Citation
  • Gohm, A., , and G. J. Mayr, 2005: Numerical and observational case-study of a deep adriatic bora. Quart. J. Roy. Meteor. Soc., 131 , 13631392.

    • Search Google Scholar
    • Export Citation
  • Gohm, A., , G. J. Mayr, , S. Mobbs, , S. Arnold, , J. Vergeiner, , L. Darby, , R. Banta, , and S. Sandberg, 2000: Foehn flow in the Austrian Alps interrupted by a cold front passage: Part II. Preprints. Ninth Conf. on Mountain Meteorology, Aspen, CO, Amer. Meteor. Soc., 83–86.

    • Search Google Scholar
    • Export Citation
  • Grubišić, V., , and B. J. Billings, 2007: The intense lee-wave rotor event of sierra rotors IOP 8. J. Atmos. Sci., 64 , 41784201.

  • Grubišić, V., and Coauthors, 2008: The terrain-induced rotor experiment: An overview of the field campaign and some highlights of special observations. Bull. Amer. Meteor. Soc., in press.

    • Search Google Scholar
    • Export Citation
  • Holmboe, J., , and H. Klieforth, 1957: Investigation of mountain lee waves and the air flow over the Sierra Nevada. Final Rep. Contract AF 19(604)–728, University of California, Los Angeles, 283 pp.

  • Kuettner, J., 1959: The rotor flow in the lee of mountains. Tech. Rep., Research Notes 6, Air Force Cambridge Research Center G.R.D., 20 pp.

  • Long, R. R., 1953: Some aspects of the flow of stratified fluids. I. A theoretical investigation. Tellus, 5 , 4257.

  • Long, R. R., 1954: Some aspects of the flow of stratified fluids. II. Experiments with a two-fluid system. Tellus, 6 , 97115.

  • Long, R. R., 1955: Some aspects of the flow of stratified fluids. III. Continuous density gradients. Tellus, 7 , 341357.

  • Mayr, G. J., , J. Vergeiner, , and A. Gohm, 2002: An automobile platform for the measurement of Foehn and gap flows. J. Atmos. Oceanic Technol., 19 , 15451556.

    • Search Google Scholar
    • Export Citation
  • Mayr, G. J., and Coauthors, 2007: Gap flows: Results from map and beyond. Quart. J. Roy. Meteor. Soc., 133 , 881896.

  • Nishiyama, R. T., , and A. J. J. Bedard, 1991: A “Quad-Disc” static pressure probe for measurement in adverse atmospheres: With a comparative review of static pressure probe designs. Rev. Sci. Instrum., 62 , 21932204.

    • Search Google Scholar
    • Export Citation
  • Pawlak, G., , and L. Armi, 2000: Mixing and entrainment in developing stratified currents. J. Fluid Mech., 424 , 4573.

  • Prandtl, L., 1944: Zwei Flüssigkeiten verschiedener Dichte. Führer durch die Strömungslehre. 2nd ed., Friedr. Vieweg & Sohn, 326–330.

    • Search Google Scholar
    • Export Citation
  • Smith, R. B., 1979: The influence of mountains on the atmosphere. Advances in Geophysics, B. Saltzman, Ed., Academic Press, 87–230.

  • Smith, R. B., 1985: On severe downslope winds. J. Atmos. Sci., 42 , 25972603.

  • Vergeiner, I., 1983: Dynamik alpiner Windsysteme. Tech. Rep., Institut für Meteorologie und Geophysik, Universität Innsbruck, 129 pp.

  • Wood, I. R., 1968: Selective withdrawal from a stably stratified fluid. J. Fluid Mech., 32 , 209223.

Fig. 1.
Fig. 1.

(a) Topographic map (m) of California; the rectangle marks the area of the Sierra Nevada and OV. (b) Detail with surveyed roads. Bishop (BI) was the operations center and IN is a small village in the center of OV; LE, MGLASS, and MAPR are radio sounding sites. (c) The road leg between OV, Seven Pines (7P), IN, and MC.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 2.
Fig. 2.

The measurement platform mounted on a custom roof rack is an aluminum beam carrying the wind, temperature, and humidity sensors and the quadplate pressure port. The GPS receiver was put into the case of a PC mouse. The attachment for the DGPS unit is also visible. The small insert shows a detail of the radiation shield.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 3.
Fig. 3.

ECMWF model topography (shading interval is 200 m). The actual topographic map of this region is shown in Fig. 1. The lines show the sections A and B, which are aligned with the synoptic flow between the chosen sounding locations during the individual cases. Upstream sounding location A is at 38.4°N, 121.2°W; B is at 37.6°N, 120.8°W. Downstream sounding location is at 38.4°N, 118.0°W.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 4.
Fig. 4.

The θυ profiles from ECMWF analysis with 40-km horizontal mesh width, upstream (thick) of the model’s representation of the Sierra Nevada and downstream (thin) and upstream wind (barbs; short barb: 5 kt; long barb: 10 kt; triangle: 50 kt) for 1300 LST (a) 18 Apr, (b) 15 Apr, (c) 20 Apr, and (d) 17 Apr 2004. Upstream sounding location for (a), (b), and (d) is at B and for (c) is at A in Fig. 3.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 5.
Fig. 5.

Virtual potential temperature profiles for across-valley drives between (right) IN and (left) OV (cf. Fig. 1c) annotated with their starting times for 18 Apr 2004. The gray parts of the curves indicate the region where the road passes through highly structured terrain, which induces pronounced small-scale variations not representative for the slope-scale flow.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 6.
Fig. 6.

Across-valley drive between IN (E) and OV (W) at (a) 0907 and (b) 1636 LST 18 Apr 2004. (panel top curves) Virtual potential temperature (thick) and mixing ratio (thin); (middle) pressure reduced to starting point (thick) and dynamic pressure (thin); and (bottom) elevation (thick) and horizontal wind (same as Fig. 4, averaged for improved legibility).

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 7.
Fig. 7.

As in Fig. 5, but for 15 Apr 2004 except that the leg extends now to the eastern part of OV (MC).

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 8.
Fig. 8.

Vertical atmospheric structure at (a) 1116 LST at LE and 1000 LST at IN (MAPR) and (b) 1700 LST at LE and 1600 and 1900 LST 15 Apr 2004 at IN (MAPR). (left) Virtual potential temperature; (middle) mixing ratio and relative humidity (dashed); and (right) cross-barrier component of wind. Line SC marks the height of the Sierra Nevada crest, and KP is the height of Kearsarge Pass. Thick lines represent the upstream, and thin lines represent the downstream sounding. The 1900 LST MAPR sounding is dotted.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 9.
Fig. 9.

Measurement drive MC to OV at 1640 LST 15 Apr 2004. (a) As in Fig. 6, except that the leg extends now to the eastern end of OV (MC), and (b) horizontal wind direction (thick) and speed over the upper part of the leg.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 10.
Fig. 10.

As in Fig. 9a, but at 1803 LST 15 Apr 2004.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 11.
Fig. 11.

As in Fig. 7, but for 20 Apr 2004.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 12.
Fig. 12.

(a) As in Fig. 8, but at 1744 LST 20 Apr 2004 at Fresno (MGLASS) and at 1700 LST at Independence (MAPR). (b) Horizontal displacement of the two radiosondes while ascending. Dots are MSL height markers in 1000-m intervals. Also visible are the ground footprints of the radiosondes (gray line).

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 13.
Fig. 13.

As in Fig. 9a, but at (a) 1546 LST 20 Apr and (b) 1758 LST 20 Apr 2004.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 14.
Fig. 14.

As in Fig. 9a, but for the drive between MC (E) and 7P (W) at 1714 LST 17 Apr 2004 and leg ending at 7P.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 15.
Fig. 15.

Across-valley drive between IN (E) and 7P (W) at (a) 1806 LST and (b) 1907 LST 17 Apr 2004; as in Fig. 14, except that the leg begins at the eastern end of IN.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Fig. 16.
Fig. 16.

(top) Schematic interaction of the downslope windstorm with a cold front in the eastern part of OV. The foehn air closest to the ground rides up the frontal surface while the air aloft still descends. (bottom) Depending on the ratio of the depths of the cold front and descending foehn layer, pressure (reduced to a common altitude) may still decrease past the location where the front intersects the surface and only rise at the hydraulic jump farther downstream.

Citation: Journal of Applied Meteorology and Climatology 47, 10; 10.1175/2008JAMC1675.1

Table 1.

Instrumental specification.

Table 1.

1

After crossing the bottom of the stable layer at around 5900 m MSL, the sonde moved almost horizontally for several kilometers until it ascended again and crossed the crest of the Inyo Mountains at around 6300 m MSL.

Save