• Barron, R., , and Yates V. , 2004: Overview of the Juneau terrain-induced turbulence and wind shear project. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., CD-ROM, 4.1.

  • Bond, N. A., , and Walter B. A. , 2002: Research aircraft observations of the mean and turbulent structure of a low-level jet accompanying a strong storm. J. Appl. Meteor., 41 , 12101224.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Browning, K. A., 1971: Structure of the atmosphere in the vicinity of large-amplitude Kelvin–Helmholtz billows. Quart. J. Roy. Meteor. Soc., 97 , 283297.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cohn, S. A., , Braid J. T. , , Dierking C. , , Politovich M. K. , , and Wade C. G. , 2004: Weather patterns of Juneau Alaska and their relationship to aircraft hazards. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., CD-ROM, 4.2.

  • Colle, B. A., , and Mass C. F. , 1998: Windstorms along the western side of the Washington Cascade Mountains. Part II: Characteristics of past events and three-dimensional idealized simulations. Mon. Wea. Rev., 126 , 5371.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colman, B. R., 1986: The winter climate of Juneau: A mean of contrasting regimes. Natl. Wea. Dig., 11 , 2. 2934.

  • Colman, B. R., , and Dierking C. F. , 1992: The Taku wind of southeast Alaska: Its identification and prediction. Wea. Forecasting, 7 , 4964.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dawson, P. J., , and Marwitz J. D. , 1982: Wave structures and turbulent features of the winter airflow in southern Wyoming. Geol. Soc. Amer., 192 , 5563.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and Bond N. A. , 2001: Research aircraft observations and numerical simulations of a warm front approaching Vancouver Island. Mon. Wea. Rev., 129 , 978998.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and Durran D. R. , 2004: Recent developments in the theory of atmospheric rotors. Bull. Amer. Meteor. Soc., 85 , 337342.

  • Doyle, J. D., , Shapiro M. A. , , Jiang Q. , , and Bartells D. , 2005: Large-amplitude mountain wave breaking over Greenland. J. Atmos. Sci., 62 , 31063126.

  • Hodur, R. M., 1997: The Naval Research Laboratory’s Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS). Mon. Wea. Rev., 125 , 14141430.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hogan, T. F., , and Rosmond T. E. , 1991: The description of the U.S. Navy Operational Global Atmospheric Prediction System’s spectral forecast model. Mon. Wea. Rev., 119 , 17861815.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lester, P. F., , and Fingerhut W. A. , 1974: Lower turbulent zones associated with mountain lee waves. J. Appl. Meteor., 13 , 5461.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lilly, D. K., 1978: A severe downslope windstorm and aircraft turbulence event induced by a mountain wave. J. Atmos. Sci., 35 , 5977.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mass, C. F., , and Albright M. D. , 1985: A severe windstorm in the lee of the Cascade Mountains of Washington State. Mon. Wea. Rev., 113 , 12611281.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mellor, G. L., , and Yamada T. , 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys., 20 , 851875.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nance, L. B., , and Colman B. R. , 2000: Evaluating the use of a nonlinear two-dimensional model in downslope windstorm forecasts. Wea. Forecasting, 15 , 715729.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sharp, J. M., 2002: Columbia Gorge gap flow: Insights from observational analysis and ultra-high-resolution simulation. Bull. Amer. Meteor. Soc., 83 , 17571762.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, R. B., 1987: Aerial observations of the Yugoslavian bora. J. Atmos. Sci., 44 , 269297.

  • Winstead, N. S., and Coauthors, 2006: Using SAR remote sensing, field observations, and models to better understand coastal flows in the Gulf of Alaska. Bull. Amer. Meteor. Soc., 87 , 787800.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yamada, T., 1983: Simulations of nocturnal drainage flows by a q2 l turbulence closure model. J. Atmos. Sci., 40 , 91106.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Topography and station locations in the vicinity of Juneau, AK. Dark shading indicates elevations greater than 1000 m. The King Air flight track for the period 1720:00–1730:00 UTC is indicated with a thick, solid line.

  • View in gallery

    An aerial view of Juneau airport and vicinity. The photograph is from west of the airport looking east; Gastineau Channel is in the top third of the picture.

  • View in gallery

    Operational upper-air analysis from NCEP valid at 1200 UTC 18 Oct 2004 for 850 hPa. The large asterisk indicates the location of Juneau.

  • View in gallery

    As in Fig. 2 but for 500 hPa.

  • View in gallery

    Time series of surface winds at (a) Lemon Creek (LMN), (b) the Federal Building (JFB), (c) Mayflower Island (MAY), (d) South Douglas (SDI), (e) Point Bishop (PBP), and (f) the Mount Roberts tram (TRM) using standard convention. See Fig. 1 for the station locations. Wind directions unavailable for the Federal Building and Point Bishop (plotted from the north as default with the shafts of the wind barbs indicated by dotted lines). Time increases from right to left.

  • View in gallery

    (a) Time–height section of the winds (every full barb = 10 m s−1) from the 915-MHz wind profiler at South Douglas. Time increases from right to left. (b) As in (a) but for Lemon Creek.

  • View in gallery

    Time series of flight-level observations from the King Air of (a) altitude (solid line, m) and θ (dashed line, K), and of (b) u (solid line, m s−1) and υ (dashed line, m s−1) for the period 1720:00–1730:00 UTC.

  • View in gallery

    Time series of flight-level observations of w (solid line; m s−1; scale on left axis) and horizontal wind speed (dashed line; m s−1; scale on right axis) for 30-s periods beginning at (a) 1722:00, (b) 1722:30, (c) 1723:00, (d) 1723:30, and (e) 1724:00 UTC. The King Air was at an altitude of approximately 450 m during this time interval. The aircraft speed was about 70 m s−1 and so each 30-s segment represents a horizontal distance of about 2.1 km.

  • View in gallery

    Model-simulated winds at 35 m for 1730 UTC. Color scale indicated on right; vector scale indicated at lower right. The model terrain is contoured every 250 m with the zero contour suppressed. The locations of cross sections for Figs. 10a,b are also indicated.

  • View in gallery

    Vertical cross section of θ (contours, interval = 1 K) and cross-channel component of the wind (shaded, color scale at right) for (a) line AA′ and (b) line BB′ in Fig. 9.

  • View in gallery

    Flight-level observations (dark solid lines) and COAMPS model output from the 3- (dotted lines), 1- (gray lines), and 0.33-km (dashed lines) domains for θ (top; K), u (middle; m s−1), and υ (bottom; m s−1) along the King Air flight track in Gastineau Channel. The x axis refers to distance down the channel (toward the southeast).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 43 43 5
PDF Downloads 27 27 3

Research Aircraft and Wind Profiler Observations in Gastineau Channel during a Taku Wind Event

View More View Less
  • 1 University of Washington, and Joint Institute for the Study of the Atmosphere and Ocean, Seattle, Washington
  • | 2 NWS Forecast Office, Juneau, Alaska
  • | 3 Naval Research Laboratory, Monterey, California
© Get Permissions
Full access

Abstract

The flow in Gastineau Channel near Juneau, Alaska, during the moderate Taku wind event of 18 October 2004 is examined using observations from the University of Wyoming’s King Air research aircraft, two wind profilers, and surface weather stations. These data sources reveal low-level winds directed down the central portion of Gastineau Channel, that is, gap flow. Farther down the channel, and above this gap flow, the winds were strongly cross channel in association with the downslope flow that characterizes Taku events. The transition region between these two flows included strong vertical wind shear and severe turbulence; measurements from the King Air indicate turbulent kinetic energy locally exceeding 50 m2 s−2. A high-resolution simulation of this case using the Naval Research Laboratory’s Coupled Ocean–Atmosphere Mesoscale Prediction System reproduced the observed character of the mean flow. This case illustrates the hazard to aviation posed by even a moderate Taku wind event and shows the value of a wind profiler for monitoring the vertical wind shear responsible for the hazard.

Corresponding author address: Nicholas Bond, 7600 Sand Point Way NE, Seattle, WA 98115. Email: nicholas.bond@noaa.gov

Abstract

The flow in Gastineau Channel near Juneau, Alaska, during the moderate Taku wind event of 18 October 2004 is examined using observations from the University of Wyoming’s King Air research aircraft, two wind profilers, and surface weather stations. These data sources reveal low-level winds directed down the central portion of Gastineau Channel, that is, gap flow. Farther down the channel, and above this gap flow, the winds were strongly cross channel in association with the downslope flow that characterizes Taku events. The transition region between these two flows included strong vertical wind shear and severe turbulence; measurements from the King Air indicate turbulent kinetic energy locally exceeding 50 m2 s−2. A high-resolution simulation of this case using the Naval Research Laboratory’s Coupled Ocean–Atmosphere Mesoscale Prediction System reproduced the observed character of the mean flow. This case illustrates the hazard to aviation posed by even a moderate Taku wind event and shows the value of a wind profiler for monitoring the vertical wind shear responsible for the hazard.

Corresponding author address: Nicholas Bond, 7600 Sand Point Way NE, Seattle, WA 98115. Email: nicholas.bond@noaa.gov

1. Introduction

The term “Taku wind” refers to a bora-type windstorm that occurs in the vicinity of Juneau, Alaska. Its characteristics have been reviewed by Colman (1986) and the criteria for its occurrence have been outlined in Colman and Dierking (1992). In brief, Taku winds resemble many other severe downslope windstorms in that they are associated with high-amplitude stationary waves induced by prominent quasi-2D terrain. The synoptic conditions conducive for Taku winds are well known by local forecasters, and the three ingredients identified by Colman and Dierking (1992), namely cross-barrier flow at ridgetop, a critical level aloft, and an inversion at or just above ridgetop, are used in a weighting system by the Juneau National Weather Service Forecast Office for operational forecasts. Also, a simple two-dimensional numerical model, implemented by Nance and Colman (2000), has proven useful in evaluating their potential intensity. While the development of Taku winds can generally be anticipated, details on the manifestations of these events remain of concern and importance.

Taku wind situations are accompanied by substantial horizontal and vertical wind shears and turbulence and hence represent a particular hazard for aircraft. The hazard is exacerbated by the terrain in the immediate vicinity of Juneau airport, which constrains the maneuvers possible on departure and approach. In recognition of this and other wind shear problems for a variety of synoptic settings, a series of field campaigns were conducted and an array of monitoring stations were deployed, culminating in the Juneau Airport Wind System (JAWS). An overview of JAWS is provided by Barron and Yates (2004); measurements on the distributions of turbulence are presented by Cohn et al. (2004). The work carried out in the development of JAWS has helped inspire the present contribution.

The purpose of the present paper is to use measurements collected on 18 October 2004 to show that pronounced vertical wind shear and turbulence can occur at low levels in Gastineau Channel during Taku winds. The case study of this event was made possible through observations collected by the University of Wyoming’s King Air research aircraft, as part of the Southeast Alaska Regional Jet Experiment (SARJET; Winstead et al. 2006), and data from a pair of 915-MHz boundary layer wind profilers operating as part of JAWS. These measurements are used to detail characteristics of the mean and turbulent flow in Gastineau Channel (see Fig. 1) near Juneau. Our observational analysis is supplemented with results from a high-resolution numerical weather prediction (NWP) model, which is shown capable of simulating important aspects of the flow in this region on scales of a few kilometers. This case is likely to be of particular interest from an aviation forecasting point of view in that severe turbulence is documented for a Taku wind event of only moderate intensity.

This paper is organized as follows. We begin with a brief review of the synoptic setting, focusing on how this case compares with the canonical Taku windstorm. The bulk of the paper consists of analysis of the flow in Gastineau Channel based on observations from selected surface weather stations, the wind profilers, and the King Air research aircraft. We then present results from a simulation of the case using a Coupled Ocean– Atmosphere Mesoscale Prediction System (COAMPS1) simulation of the case. We conclude with a summary and some final remarks.

2. Synoptic setting

A map of the region of interest is provided in Fig. 1. Our analysis focuses on the flow over Gastineau Channel, a 1–2-km-wide sea level channel that stretches about 20 km from near Juneau airport to Stephens Passage and the mouth of Taku Inlet. Gastineau Channel is bounded by steep terrain, especially on its northeast side by Salisbury Ridge. The high terrain north and east of Juneau and Gastineau Channel features permanent snow and ice fields and numerous narrow valleys. This region, hence, represents a potential supply of cold air that may, under northeasterly flow, enhance the low-level static stability, a characteristic often associated with bora-type winds such as the Taku (e.g., Colman and Dierking 1992). Also, the Taku River valley cuts deeply through the coastal mountains and hence provides an outlet for near-surface arctic air, that is, gap flow, from interior northwest Canada. Figure 1 shows these geographic features, the locations of the fixed stations used in the analysis, and the flight track of the King Air. An aerial photograph of the region is provided in Fig. 2.

The period of interest for this case study is the morning of 18 October 2004. This is when the strongest winds were being reported in the Juneau area, and when the relevant measurements were collected by the King Air (1720–1730 UTC or 0920–0930 local time). The large-scale conditions for this period are summarized using the operational analyses from the National Centers for Environmental Prediction (NCEP) at 1200 UTC 18 October for 850 hPa (Fig. 3) and for 500 hPa (Fig. 4). The 850-hPa geopotential height analysis shows a trough oriented from northwest to southeast along the coast of southeast Alaska. A height gradient supporting moderately strong easterly geostrophic winds was present over the southeast Alaska panhandle. As mentioned earlier, cross-barrier flow at the ridgetop level constitutes one of the ingredients for a Taku windstorm (Colman and Dierking 1992). It is expected that the magnitude of Taku winds increases with the strength of the cross-barrier flow, as found in a sensitivity analysis of a similar type of storm in Washington State by Colle and Mass (1998). While a comprehensive sensitivity analysis has not been carried out for Taku winds, we do note that the cross-barrier flow (and cross-barrier pressure gradient) was roughly one-half to two-thirds as strong as for the intense case of 8 January 1975 described in Colman and Dierking (1992).

The 500-hPa analysis (Fig. 4) shows a low center just ENE of the Juneau area. The local 500-hPa height gradients imply weak north-northwesterly flow over Juneau, indicating the presence of a critical level (i.e., zero flow from the northeast) slightly above the 500-hPa level. This is another element of Taku wind events that was identified by Colman and Dierking (1992). The overall configuration of the large-scale background flow resembles that for the anticyclonic-type cases (the more common variety) from Colman and Dierking (1992).

The third criterion for Taku winds according to Colman and Dierking (1992) is an inversion at or just above ridgetop. The measurements collected during the height of the event during an ascent of the King Air at the east end of Gastineau Channel and Douglas Island indicate an average increase of potential temperature θ with height of ∼3.5 K km−1 (N ∼1.1 × 10−2 s−1) from 1.2 to 3 km (illustrated later in Fig. 7). This profile included a 200-m-deep layer of enhanced static stability near 2 km, but the robustness of this feature is uncertain. A descent by the aircraft west of Juneau airport about 3 h later also found a layer of enhanced static stability at the 2-km level. It bears noting that each of these aircraft soundings is downstream of the terrain forcing the Taku winds. The most representative upstream sounding was at Whitehorse, Yukon Territory (YXY). The profile at 1200 UTC 18 October (not shown) indicates moderate static stability, specifically, /dz ∼ 4 K km−1 or N ∼ 1.2 × 10−2 s−1, between the 878-hPa (1151 m) and 784-hPa (2012 m) levels. Taku wind events occur most frequently in winter, and because it was early in the cool season, the air from the interior was relatively warm because of the lack of snow cover below about 1500 m. It is an open question whether the magnitude of the static stability near the ridgetop, and in particular the lack of an inversion, played an important role in determining the intensity of this particular event. From a forecasting point of view, it is important to note that moderately strong downslope Taku winds did develop in the absence of a ridgetop inversion.

3. Regional surface and upper-air observations

The spatial and temporal variations in the surface winds associated with this case are illustrated in the form of time series (Figs. 5a–f) from selected weather stations in the Juneau area (see Fig. 1 for locations). The first station shown, Lemon Creek (Fig. 5a), is situated at the mouth of a small valley and relatively distant from the higher terrain; it experienced modest (10–15 kt; 1 kt = 0.5144 m s−1) sustained winds from the NE and no strong gusts. Conditions were much different only about 5 km to the southeast at the Federal Building in downtown Juneau where the winds reached a sustained value of 31 kt with gusts to 45 kt (Fig. 5b). This location is near the base of the steep terrain at the northwest end of Salisbury Ridge, and its relatively strong winds are probably a result of a combination of gap winds out of the small valley extending east into the higher terrain and downslope-enhanced flow. There are two stations across Gastineau Channel from the northwestern portion of Salisbury Ridge, Mayflower Island (Fig. 5c) and South Douglas (Fig. 5d). The winds at Mayflower Island were relatively weak in the mean but were gusty, while the winds at South Douglas were predominantly from the northwest in the mean and included gusts (directions unknown) as strong as 49 kt.2 The presence of northwesterly winds along the south shore of Gastineau Channel and the windward flank of Douglas Island suggests that gap flow down Gastineau Channel (sea level pressure was about 2 hPa higher at the airport than at South Douglas) intersected the northeasterly downslope flow off of Salisbury Ridge. Visual observations of wind waves from the King Air aircraft indicated the penetration of this northeasterly flow to the surface occurred in the southeastern portion of Gastineau Channel.

The wind data from three other stations help fill in the regional picture. Point Bishop experienced winds exceeding 40 kt (Fig. 5e); its location is subject more to the gap flow out of Taku Inlet than downslope winds. The co-occurrence of gap flow with nearby downslope flow has been found for other locations [e.g., the Cascade Mountains of Washington State; Mass and Albright (1985)]. The top of the Mount Roberts tram (TRM) is situated at an altitude of 536 m on the flank of Salisbury Ridge above downtown Juneau; it reported north winds with sustained speeds up to 26 kt and a peak speed of 49 kt (Fig. 5f). Another higher-altitude station, Sheep Mountain at an altitude of 1078 m near the midpoint of Salisbury Ridge, reported a peak wind of 68 kt at 1700 UTC (not shown).

Additional information on the flow along the flank of Douglas Island during the Taku wind event and the flight of the King Air is provided by the South Douglas wind profiler. A time series of the vertical profile of the winds (Fig. 6a) shows northeasterly winds prevailed above about 700 m through the period from 1500–1900 UTC 18 October. These northeasterly winds were particularly strong (typically 25 and up to 35 m s−1) in the 800–1500-m level. The winds in the lowest 500 m were more variable, especially in terms of direction. The vertical wind shear that was present in the 300–400- to 900-m layer (a vector difference of ∼20 m s−1) was of particular relevance to the conditions encountered by the aircraft at 1720–1730 UTC at an altitude of 450 m, as will be elaborated upon below.

The marked regional variations in the flow in the vicinity of Juneau during these kinds of events can be appreciated by comparing the winds at the South Douglas profiler with those at the Lemon Creek profiler (Fig. 6b). Above ∼1 km, the winds were similar at the two sites, with the Lemon Creek site experiencing steadier and generally slower winds. Not surprisingly, the differences were much more profound near the surface; while South Douglas featured large vector wind shear in this layer and periods of near-surface northwesterlies, at Lemon Creek this layer exhibited little turning and moderate speed shear in northeasterly winds with height. We attribute the differences in the flow at the two sites to local topographic effects. The Lemon Creek site appears to have been influenced largely by the more distant, and less steep, terrain upstream, in general, while the South Douglas site experienced the interaction between the flows forced by the gap of Gastineau Channel and by the steep leeward slope of Salisbury Ridge.

The King Air flight of 18 October included a track down Gastineau Channel at an altitude of 450 m during the period 1720–1730 UTC, which, considering the mean conditions summarized above, represented a propitious place and time to collect turbulence measurements. First, the mean conditions on this track are summarized in Fig. 7 using 1-Hz data along the flight track. The flight profile itself for this period consisted of a climb to ∼450 m, an approximately level leg at 450 m from west of downtown Juneau to the mouth of Taku Inlet, and then a climb to 2500 m. The level portion of this segment at 450 m is of special interest. The first part, from about 1721:30 to 1723:00 UTC, included mean zonal u and meridional υ winds of about 3 and 4 m s−1, respectively, and substantial fluctuations in these winds on 5–10-s time scales. The second part, from about 1723:00 to 1724:30 UTC featured a rapid increase of the mean u and υ winds to about 20 m s−1 each from the east and north, respectively. The variations in each component were 5–10 m s−1. The last part, from about 1724:30 UTC to the beginning of the climb at about 1726:30 UTC, included a modest decrease in the easterly component, but continued substantial high-frequency variability in the winds, and relatively large turbulent fluctuations in θ. Note that the mean vertical gradient in θ was minimal below about 1200 m.

The transition from the modest northwesterlies at 1722:00 UTC to strong northeasterlies at 1724:30 UTC was accompanied by severe turbulence. The properties of this turbulence are described here using the high-rate (25 Hz) data available from the King Air, for a sequence of 30-s blocks. Time series of vertical velocity w and wind speed are plotted for each block (Figs. 8a–e); averages of selected turbulence parameters for each block are indicated in Table 1. The block for 1722:00–1722:30 UTC (Fig. 8a) included significant fluctuations (∼2–3 m s−1) in w, and in wind speed. As stronger northeasterlies were encountered from 1722:30 to 1724:00 UTC (Figs. 8b–d), the typical magnitude of w fluctuations increased to ±6 m s−1. There is a broad spectral peak in the wind fluctuations at a period of 10 s (not shown), which translates to an along-track scale of 700 m given the speed of the aircraft. Note that there were multiple instances of exceedingly rapid changes in w, specifically, vertical accelerations significantly greater than the acceleration due to gravity g. The maximum downward acceleration was about −3g. As might be imagined, the flight was extremely rough. The plots of w and wind speed show a strong negative correlation; that is, downdrafts (updrafts) systematically transported strong northeasterly (weak northwesterly) momentum at flight level. Beginning at about 1724:20 UTC (Fig. 8e) the mean w remained strongly positive (∼6 m s−1) and the horizontal wind also steadied, although it was still quite choppy by usual standards.

The magnitudes of the mean w and selected turbulent properties for each of these 30-s blocks of Figs. 8a–e (and one block before and after) are itemized in Table 1. Especially noteworthy are the tremendous magnitudes of the turbulent kinetic energy (TKE), including a peak value of greater than 50 m2 s−2 for the 1723:00–1723:30 UTC segment. We are unaware of any larger values of boundary layer TKE that have been documented by a suitably equipped research aircraft. By way of comparison, the research aircraft observations in the immediate lee of steep terrain of which we are aware (Dawson and Marwitz 1982; Smith 1987) found TKE on the order of 5–10 m2 s−2.3 The turbulent momentum fluxes listed in Table 1 are also impressive. The maximum values of ∼20 N m−2 are almost an order of magnitude greater than their counterparts found by Bond and Walter (2002) in the vicinity of a 45 m s−1 low-level jet with a storm off the coast of Oregon. It is a tribute to the airworthiness of the King Air, and even more so, the skill of the pilot, that these conditions were encountered without incident.

There is self-consistency in the turbulence parameters summarized in Table 1 in terms of the production of TKE. An in-depth evaluation of the TKE budget is outside the scope of this paper, and the data are insufficient for a robust estimation of the various terms. Our results do show that the momentum fluxes were down the gradient of vertical wind shear; their magnitudes in concert with the strength of the mean vertical shear yield an estimated peak rate of shear production of TKE of roughly 1 m2 s−3. This is a very high rate of production, but it is also realistic. Not only were there also losses of TKE due to dissipation, but the mean cross-channel and upward motions imply a relatively short residence time scale (∼100–200 s) for the air within the turbulent/shear zone. Rapid production must have been occurring given that the TKE was so large.

Even though the turbulence was severe, the turbulent heat fluxes between 1721:30 and 1724:30 UTC were moderate in intensity and of varying sign. This is compatible with the turbulence having been driven by Kelvin–Helmholtz (KH) instability (e.g., Browning 1971). The conditions were certainly conducive for KH instability because of the combination of low static stability and strong vertical wind shear, and hence a Richardson number of very nearly zero (∼0.1). The KH waves that were breaking within the layer of shear caused vigorous enough vertical circulations to apparently cause air parcels of higher-θ air to roll under lower-θ air and vice versa, essentially resulting in little in the way of systematic vertical transports in heat, at least up until 1724:30 UTC. It is unknown why the heat fluxes became so much more strongly negative (downward) after this point.

4. Numerical results

The NWP simulations for this case have been carried out using COAMPS (Hodur 1997). The model configuration used is summarized in Doyle and Bond (2001) and Doyle et al. (2005). The parameterization of the planetary boundary layer and free-atmosphere turbulent mixing was likely to be of special relevance for the present application, and so we incorporated the relatively sophisticated scheme based on the level-2.5 formulation of Mellor and Yamada (1982) and Yamada (1983) as modified for implementation within COAMPS (e.g., Doyle et al. 2005; Hodur 1997). The simulations were made using a series of four nests with 60 vertical levels, with the inner nest having a horizontal grid spacing of 0.33 km. The terrain is specified using the U.S. Defense Mapping Agency’s dataset, which has a native 100-m resolution. Lateral boundary conditions were constructed from the Navy Operational Global Analysis and Prediction System (Hogan and Rosmond 1991); the run commenced at 0000 UTC 18 October to ensure sufficient time for the full development of small-scale structures in the flow.

The objective of the modeling component of this study is to supplement the available observations, rather than carry out a detailed sensitivity study. This kind of study does depend on proper specification of the flow upstream of the area of interest, and while there is rather limited data available to check the model, it does appear accurate enough to make its predictions meaningful. More specifically, simulated temperature and wind profiles 12 h into the run were compared with the observed sounding at YXY for 1200 UTC 18 October. This comparison indicated that the model included a slightly weaker stable layer in the 2.5–3.5-km layer than was observed, but a similar wind profile. The results from the run of a series of nested simulations that were performed indicate that proper specification of the terrain, and high horizontal resolution, that is, a grid spacing of 0.33 km, was required for realistic representation of the flow in Gastineau Channel.

The winds at 35 m AGL for the inner domain valid at 1730 UTC are shown in Fig. 9. The modeled winds are strongest (>20 m s−1) over the higher terrain of the Juneau Icefield and in selected locations over Salisbury Ridge. As observed, the simulation produced stronger northeasterly winds in the southeastern portion of Gastineau Channel than in its northwestern portion.

The spatial variations in the modeled flow are further illustrated using a pair of vertical cross sections oriented normal to Gastineau Channel. In the southeast cross section (Fig. 10a), the strongest northeasterly (cross channel) winds extend from just above the top of Salisbury Ridge to across Gastineau Channel and then to just above the windward slope of Douglas Island. The modeled wind speed at the location of the aircraft track was 20–25 m s−1 as compared with a measured value of about 25 m s−1.

The northwest cross section (Fig. 10b) indicates some significant differences in the cross-channel winds, namely, separation of the core of the strongest winds about halfway down the steeply sloping terrain behind Juneau. The consequences include weaker winds near the surface over Gastineau Channel, and an elevated wind maximum (about 400–500 m higher) on the upstream side of Douglas Island, relative to that modeled for the southeast cross section. The northwest cross section also indicates a 200-m-deep layer of near-zero or even reverse cross-channel flow on the upstream side of Douglas Island. This feature is consistent with the observations, albeit somewhat shallower than that indicated by the South Douglas wind profiler and suggested by the aircraft. Perhaps most importantly, the model simulated much greater vertical wind shear at the 450-m flight level in the northwest section than in the southeast section. Presumably this difference in the vertical shear is responsible for the model’s greater TKE at the aircraft’s location in the northwest section than in the southeast section (not shown).

We conclude this section with direct comparisons between the model forecasts along the flight track down Gastineau Channel and the corresponding aircraft observations. The results for θ (Fig. 11, top) indicate a much better match between observations and the model for the inner (0.33 km) nest than the next two coarser nests. The latter do not appear to properly characterize the sources of cool air at low levels in Gastineau Channel, presumably due to inadequacies in resolving gaps in the terrain. With respect to the u (Fig. 11, middle) and υ (Fig. 11, bottom) components of the wind, the simulation on the 3-km grid indicates little systematic change along the flight track, while the inner two nests indicate trends toward more negative values of u and υ down the track roughly similar to those observed. The wind simulation from the inner domain along the flight track is not clearly superior to that from the 1-km nest. In making this comparison, it should be recognized that the observed winds include the signatures of vigorous eddies that are lacking in the model simulations. These eddies are too small to be resolved by even the 0.33-km grid, and while they may have systematic relationships to small-scale variations in the terrain, they will also tend to be transient. The best that can be hoped for is that the effects of these eddies on the mean flow are handled in the boundary layer parameterizations of NWP models such as COAMPS. There are crucial differences between the 0.33-km- and 1-km-domain winds in terms of their vertical structure within the northwest portion of Gastineau Channel. Specifically, the finer grid indicates much greater vertical wind shear in the cross-channel component over the shore of Douglas Island (Fig. 10b) than does the 1-km grid (not shown), with implications for the generation of turbulence.

5. Final remarks

A case study was carried out on the Taku wind (a bora-type windstorm in the vicinity of Juneau, Alaska) event of 18 October 2004. Moderately strong northeast winds occurred in the lee of Salisbury Ridge over the southeast portion of Gastineau Channel. A wind profiler situated on Douglas Island along the northwestern flank of the strongly downsloping flow indicated north to northwest winds 300–400 m deep below northeast winds aloft. The resulting layer of strong vertical wind shear (∼20 m s−1 over 500 m) was probed by the University of Wyoming’s King Air research aircraft at an altitude of 450 m; flight-level measurements indicated extremely large values of TKE (up to ∼50 m2 s−2) and turbulent momentum fluxes. The turbulence was consistent with KH instability as a source. A COAMPS simulation at very high-resolution (0.33-km inner-grid spacing) reproduced the salient aspects of the mean flow. Notably, the simulation suggests that the region of strong vertical wind shear at flight level can be attributed to separation of the downslope flow off the very steep terrain in the immediate vicinity of Juneau, thereby allowing undercutting by a low-level north-northwesterly gap flow.

The aforementioned transition zone between low-level northwesterlies and strong northeasterlies aloft bears some resemblance to the leading edge of a rotor (Doyle and Durran 2004). Both the observations and numerical results show that this zone was in a region of ascending motion, and the presence of weak or even reverse flow in the cross-terrain component of the flow near the surface. The aircraft’s encounter of severe turbulence is consistent with previous experience in the updraft region of a rotor (e.g., Lester and Fingerhut 1974). It should be noted that rotors are typically considered to be quasi-two-dimensional structures, and in the present case, there was a significant change in the character of the flow along Gastineau Channel (the primary axis of the terrain), which complicates the interpretation of the observations and model results in the context of the prototypical rotor. Moreover, the transition zone was only about 2 km downstream of the base of the leeward terrain as compared with a typical value of 10 km for rotors. Presumably this difference in scale, and perhaps the existence of the transition zone itself, was because of the terrain of Douglas Island. It bears noting that stronger ascent was measured and modeled farther southeast along Gastineau Channel, but in that location the turbulence was weaker because there was not as much vertical shear. The factors controlling the position and nature of the transition zone are unknown, but presumably are related to interactions between the gaplike flow and the downslope flow in the central portion of Gastineau Channel. The presence of this transition zone is associated with the separation of the downslope flow off the lee slope of Salisbury Ridge, and we speculate that its existence and location are related to the strength of the mountain-induced stationary wave associated with Taku winds.4 A point worth repeating here is that a relatively modest Taku wind event by the usual standards brought about conditions leading to severe turbulence.

Our numerical experiments with COAMPS show that current NWP models are capable of simulating a Taku wind event. Qualitatively correct results required a realistic account of the terrain and hence very high spatial resolution, as in the Columbia River Gorge model experiments by Sharp (2002). These types of model runs are currently more feasible in a research rather than an operational mode, but that may not be the case in the not too distant future. While outside the scope of the present study, it would be interesting to more fully explore the relationships between the mean and turbulent aspects of the flow. This might be most effectively carried out via large-eddy simulations. The present simulation included only subgrid-scale aspects of the turbulence and hence could not account for the effects of the eddies on scales of 100–1000 m, which the aircraft observations show were prominent (Fig. 8).

While Taku winds have long been recognized in the Juneau area, our results do have implications for forecasting. In particular, aviation interests should take notice of our finding that a marked transition in the winds (with a vertical profile resembling that of a rotor) can occur over the northeast shore of Douglas Island; the existence of such a feature is also indicated in Cohn et al. (2004). Current practice during Taku wind events is to avoid departures down Gastineau Channel, if possible. Otherwise, it is recommended that pilots stay away from the Juneau side of the channel, in the region of typically strong downward motion, and instead follow a track over the shoreline of Douglas Island where upward motion prevails. Our results offer vivid evidence that the latter track can experience severe turbulence. The vertical wind shear causing the turbulence was observed by the South Douglas wind profiler. This platform is much more effective than an array of surface wind anemometers in providing real-time information on the potential for hazardous flying conditions in the present type of situation.

Acknowledgments

We extend our special appreciation to all of the University of Wyoming research flight facility personnel involved in the SARJET project, specifically Al Rodi, Glenn Gordon, Larry Oolman (Larry initially suggested a Gastineau Channel departure for the flight of 18 October), Don Lukens, Matt Burkhart, and pilot Don Cooksey. The lead scientist for SARJET is Nathaniel Winstead of Johns Hopkins University’s Applied Physics Laboratory. We also acknowledge helpful conversations with Brad Colman and constructive suggestions from three reviewers. Support for N. A. Bond was provided by National Science Foundation (NSF) Grant ATM-0240784. J. D. Doyle was supported by the Office of Naval Research’s Program Element 0601153N, with computing time supported in part by a grant of HPC time from the Aeronautical Systems Center Department of Defense Major Shared Resource Center at Wright-Patterson Air Force Base, Ohio. COAMPS is a registered trademark of the Naval Research Laboratory. This publication is funded by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement NA17RJ1232.

REFERENCES

  • Barron, R., , and Yates V. , 2004: Overview of the Juneau terrain-induced turbulence and wind shear project. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., CD-ROM, 4.1.

  • Bond, N. A., , and Walter B. A. , 2002: Research aircraft observations of the mean and turbulent structure of a low-level jet accompanying a strong storm. J. Appl. Meteor., 41 , 12101224.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Browning, K. A., 1971: Structure of the atmosphere in the vicinity of large-amplitude Kelvin–Helmholtz billows. Quart. J. Roy. Meteor. Soc., 97 , 283297.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cohn, S. A., , Braid J. T. , , Dierking C. , , Politovich M. K. , , and Wade C. G. , 2004: Weather patterns of Juneau Alaska and their relationship to aircraft hazards. Preprints, 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., CD-ROM, 4.2.

  • Colle, B. A., , and Mass C. F. , 1998: Windstorms along the western side of the Washington Cascade Mountains. Part II: Characteristics of past events and three-dimensional idealized simulations. Mon. Wea. Rev., 126 , 5371.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colman, B. R., 1986: The winter climate of Juneau: A mean of contrasting regimes. Natl. Wea. Dig., 11 , 2. 2934.

  • Colman, B. R., , and Dierking C. F. , 1992: The Taku wind of southeast Alaska: Its identification and prediction. Wea. Forecasting, 7 , 4964.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dawson, P. J., , and Marwitz J. D. , 1982: Wave structures and turbulent features of the winter airflow in southern Wyoming. Geol. Soc. Amer., 192 , 5563.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and Bond N. A. , 2001: Research aircraft observations and numerical simulations of a warm front approaching Vancouver Island. Mon. Wea. Rev., 129 , 978998.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and Durran D. R. , 2004: Recent developments in the theory of atmospheric rotors. Bull. Amer. Meteor. Soc., 85 , 337342.

  • Doyle, J. D., , Shapiro M. A. , , Jiang Q. , , and Bartells D. , 2005: Large-amplitude mountain wave breaking over Greenland. J. Atmos. Sci., 62 , 31063126.

  • Hodur, R. M., 1997: The Naval Research Laboratory’s Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS). Mon. Wea. Rev., 125 , 14141430.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hogan, T. F., , and Rosmond T. E. , 1991: The description of the U.S. Navy Operational Global Atmospheric Prediction System’s spectral forecast model. Mon. Wea. Rev., 119 , 17861815.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lester, P. F., , and Fingerhut W. A. , 1974: Lower turbulent zones associated with mountain lee waves. J. Appl. Meteor., 13 , 5461.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lilly, D. K., 1978: A severe downslope windstorm and aircraft turbulence event induced by a mountain wave. J. Atmos. Sci., 35 , 5977.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mass, C. F., , and Albright M. D. , 1985: A severe windstorm in the lee of the Cascade Mountains of Washington State. Mon. Wea. Rev., 113 , 12611281.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mellor, G. L., , and Yamada T. , 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys., 20 , 851875.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nance, L. B., , and Colman B. R. , 2000: Evaluating the use of a nonlinear two-dimensional model in downslope windstorm forecasts. Wea. Forecasting, 15 , 715729.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sharp, J. M., 2002: Columbia Gorge gap flow: Insights from observational analysis and ultra-high-resolution simulation. Bull. Amer. Meteor. Soc., 83 , 17571762.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, R. B., 1987: Aerial observations of the Yugoslavian bora. J. Atmos. Sci., 44 , 269297.

  • Winstead, N. S., and Coauthors, 2006: Using SAR remote sensing, field observations, and models to better understand coastal flows in the Gulf of Alaska. Bull. Amer. Meteor. Soc., 87 , 787800.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yamada, T., 1983: Simulations of nocturnal drainage flows by a q2 l turbulence closure model. J. Atmos. Sci., 40 , 91106.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Topography and station locations in the vicinity of Juneau, AK. Dark shading indicates elevations greater than 1000 m. The King Air flight track for the period 1720:00–1730:00 UTC is indicated with a thick, solid line.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 2.
Fig. 2.

An aerial view of Juneau airport and vicinity. The photograph is from west of the airport looking east; Gastineau Channel is in the top third of the picture.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 3.
Fig. 3.

Operational upper-air analysis from NCEP valid at 1200 UTC 18 Oct 2004 for 850 hPa. The large asterisk indicates the location of Juneau.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 4.
Fig. 4.

As in Fig. 2 but for 500 hPa.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 5.
Fig. 5.

Time series of surface winds at (a) Lemon Creek (LMN), (b) the Federal Building (JFB), (c) Mayflower Island (MAY), (d) South Douglas (SDI), (e) Point Bishop (PBP), and (f) the Mount Roberts tram (TRM) using standard convention. See Fig. 1 for the station locations. Wind directions unavailable for the Federal Building and Point Bishop (plotted from the north as default with the shafts of the wind barbs indicated by dotted lines). Time increases from right to left.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 6.
Fig. 6.

(a) Time–height section of the winds (every full barb = 10 m s−1) from the 915-MHz wind profiler at South Douglas. Time increases from right to left. (b) As in (a) but for Lemon Creek.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 7.
Fig. 7.

Time series of flight-level observations from the King Air of (a) altitude (solid line, m) and θ (dashed line, K), and of (b) u (solid line, m s−1) and υ (dashed line, m s−1) for the period 1720:00–1730:00 UTC.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 8.
Fig. 8.

Time series of flight-level observations of w (solid line; m s−1; scale on left axis) and horizontal wind speed (dashed line; m s−1; scale on right axis) for 30-s periods beginning at (a) 1722:00, (b) 1722:30, (c) 1723:00, (d) 1723:30, and (e) 1724:00 UTC. The King Air was at an altitude of approximately 450 m during this time interval. The aircraft speed was about 70 m s−1 and so each 30-s segment represents a horizontal distance of about 2.1 km.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 9.
Fig. 9.

Model-simulated winds at 35 m for 1730 UTC. Color scale indicated on right; vector scale indicated at lower right. The model terrain is contoured every 250 m with the zero contour suppressed. The locations of cross sections for Figs. 10a,b are also indicated.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 10.
Fig. 10.

Vertical cross section of θ (contours, interval = 1 K) and cross-channel component of the wind (shaded, color scale at right) for (a) line AA′ and (b) line BB′ in Fig. 9.

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Fig. 11.
Fig. 11.

Flight-level observations (dark solid lines) and COAMPS model output from the 3- (dotted lines), 1- (gray lines), and 0.33-km (dashed lines) domains for θ (top; K), u (middle; m s−1), and υ (bottom; m s−1) along the King Air flight track in Gastineau Channel. The x axis refers to distance down the channel (toward the southeast).

Citation: Weather and Forecasting 21, 4; 10.1175/WAF932.1

Table 1.

Turbulence parameters in Gastineau Channel. Here, τx and τy refer to the zonal and meridional components of the momentum fluxes (N m−2), respectively; Fh refers to the sensible heat flux (W m−2). Linear trends in u, υ, w, and temperature were removed in the computation of the turbulence properties for each 30-s segment.

Table 1.

* NOAA/Pacific Marine Environmental Laboratory Contribution Number 2790 and Joint Institute for the Study of the Atmosphere and Ocean Contribution Number 1109.

1

COAMPS is a registered trademark of the Naval Research Laboratory.

2

While this event was not as severe as many Taku wind storms, it did cause some minor damage in Douglas.

3

A very large magnitude of TKE (∼150 m2 s−2) was found at the 6-km level in the breaking region of a mountain wave associated with a severe downslope windstorm (Lilly 1978).

4

This idea is consistent with the evidence that stronger Taku events are accompanied by northeasterly winds at Douglas.

Save