Journal of Geophysical Research: Atmospheres

Dust storm over the Black Rock Desert: Larger-scale dynamic signatures

Authors


Abstract

[1] A dust storm that originated over the Black Rock Desert (BRD) of northwestern Nevada is investigated. Our primary goal is to more clearly understand the sequence of dynamical processes that generate surface winds responsible for entraining dust from this desert. In addition to reliance on conventional surface and upper-air observations, we make use of reanalysis data sets (NCAR/NCEP and NARR)—blends of primitive equation model forecasts and observations. From these data sets, we obtain the evolution of vertical motion patterns and ageostrophic motions associated with the event. In contrast to earlier studies that have emphasized the importance of indirect transverse circulations about an upper-level jet streak, our results indicate that in this case the transition from an indirect to a direct circulation pattern across the exit region of upper-level jet streak is central to creation of low-level winds that ablate dust from the desert. It is further argued that the transition of vertical circulation patterns is in response to adjustments to geostrophic imbalance—an adjustment time scale of 6–9 h. Although unproven, we suggest that antecedent rainfall over the alkali desert 2 weeks prior to the event was instrumental in lowering the bulk density of sediments and thereby improved the chances for dust ablation by the atmospheric disturbance. We comprehensively compare/contrast our results with those of earlier investigators, and we present an alternative view of key dynamical signatures in atmospheric flow that portend the likelihood of dust storms over the western United States.

1. Introduction

[2] The impact of dust storms on human activity is often dramatic as in the infamous San Joaquin Valley dust storms of 1977 [Wilshire et al., 1981] and 1991 (the “Interstate-5 Storm” [Pauley et al., 1996]) that caused a multitude of accidents and associated loss of life, and in the erosion of agricultural land so evident during the 1930s over the Midwestern USA—the “Dust Bowl” years [Worster, 1979; Schubert et al., 2004]. Further, the long-term and widespread suspension of dust in the atmosphere that stems from disturbances over the world's gigantic deserts (such as the Gobi and Sahara) certainly impact climate [Idso, 1976; Washington et al., 2003; Goudie and Middleton, 2006; Goudie, 2009]. Yet, significant uncertainty attends parameterization of this widespread dust in the governing equations for global circulation models. This is due in part to the delicate interplay between reflection of sunlight by the dust and the trapping of upwelling radiation by this same dust [Idso and Brazel, 1977; Miller and Tegan, 1998]. Despite these uncertainties, the recent successful application of regional dynamical simulation to dust transport over the Gobi provides some encouragement [Liu et al., 2003].

[3] Prediction or analysis of a dust storm requires knowledge of the surface soil characteristics and the atmospheric processes that give rise to the wind. Determination of soil characteristics—crustal roughness, bulk density, moisture content, etc.—and the relationship of these characteristics to the likelihood of dust storm generation is one of the most challenging aspects of prediction. In part, this difficulty stems from reliance on remote observations from space and the incompleteness and uncertainty of these observations. But we are also ignorant of the micrometeorological processes that lead to lift of the sediments—for example, sand/sediment entrainment and saltation [see Bagnold, 1973; Shao, 2000]. Laboratory work indicates that strong shear in the presence of unstable stratification, in essence a small Richardson number, is conducive to lift [Richardson, 1920; Bagnold, 1973; Stull, 1988]. Unpublished laboratory work by coauthor J. Hallett also indicates that a vortex ring whose axis is perpendicular to a dusty surface causes the dust to expand outward and upward while remaining undisturbed at the stagnation point directly beneath the ring. And although we know that drought conditions favor dust storms, it has now become apparent that antecedent rainfall or run-off, occurring on the order of several weeks before a dust storm event, is a positive contributor. This influence stems from salt efflorescence (production of powdery salts) and swelling of certain clays that serve to reduce the bulk density of the sediment [Gillies et al., 1999; Bullard et al., 2008; J. A. Gilles, personal communication, 2009].

[4] The initial source of dust for our study comes from the Black Rock Desert (BRD) located in northwestern Nevada. The BRD and its neighboring desert, the Smoke Creek Desert (SCD), are dry lakebeds (playa) that lie within the late Pleistocene Lake Lahontan. A topographical map including the area covered by these deserts is found in Figure 1. A panoramic view of the BRD from a mountain location west of the desert, and an aerial view of the BRD and the SCD are found in Figure 2. These deserts are often referred to as “alkali deserts” where the dominant elements are those associated with alumino-silicate minerals (e.g., Si, Al, K, and Ti) [Gillies et al., 1999]. In line with statements above regarding the influence of antecedent rainfall and run-off from neighboring mountains, we examine precedent conditions over these deserts; but our primary goal focuses on understanding the atmospheric processes that produce the low-level winds responsible for raising dust from these deserts.

Figure 1.

(left) A map showing the locations of the upper-air stations and (right) topographic map of Nevada (source: 1 km resolution United States Geological Survey data) and the surface stations, used in the entire study (Black Rock Desert (BRD; elevation = 1193 m)); the Smoke Creek Desert (SCD) lies between the two stations GEL (Gerlach) and Smoke Creek (SMC).

Figure 2.

(a) A view of BRD from Calico Mountain (northwest of BRD) (courtesy of Stan White), and (b) an aerial view of BRD and SCD (source: Google Earth; http://earth.google.com). Alkali deserts appear white in Figure 2b.

[5] As shown by Danielsen [1968, 1974a, 1974b] and Pauley et al. [1996], large-scale ascent/descent couplets in the vicinity of the jet stream or jet streaks (maxima in wind speed along the axis of the jet stream) were central to atmospheric processes that led to dust storms. Reviews of these vertical circulations around the jet stream are provided by Keyser and Shapiro [1986] and Carlson [1991]. Eliassen's several-decadal studies of vertical circulations—secondary circulations—were more general and not restricted to couplets near the jet stream. He viewed these circulations as responses to imbalance in a variety of dynamical systems— in the context of quasi-static vortex motion [Eliassen, 1952], quasi-geostrophic dynamics in the vicinity of frontal systems [Eliassen, 1962] (Sawyer-Eliassen circulations – primarily based on Sawyer [1956] and Eliassen [1962]), and in the more general primitive equation dynamics [Eliassen, 1983; T. Iversen, personal communication, 2010]. And it is fair to say that stimulation for this line of research came from the pioneering geostrophic adjustment studies of Rossby [1937, 1938] and Obukhov [1949] (English translation of Obukhov [1949] is available from the authors). Although these pioneers relied on simplified dynamics (“shallow water” dynamics), they provided insight into the temporal scales of adjustment for their system—a fast gravity wave mode and the slower inertial mode with periods of several hours to the half pendulum day (∼17 h at 45° latitude), respectively. A masterful pedagogical review of these works is provided by Blumen [1972].

[6] To set the stage for our study, we review the dynamical processes that have been linked to dust storms and present a brief summary of dust storm climatology in the USA. The particular dust storm event in February 2002 is then viewed macroscopically with the aid of satellite imagery and surface-based measurements of particulate matter. Upper-air analysis follows with support from the 6 h updates of NCEP (National Center for Environmental Prediction)/NCAR (National Center for Atmospheric Research and 3 h updates of NARR (North American Regional Reanalysis). The NCEP/NCAR data set is a synoptic-scale analysis with 2.5° grid resolution [Kalnay et al., 1996], and the NARR data set is a subsynoptic-scale analysis with a 32 km grid resolution [Mesinger et al., 2006]. These archived data sets permit analysis of dynamically consistent vertical motion fields. The secondary vertical circulations are viewed in the context of dynamic imbalance and adjustments to imbalance. We end with section 7, where we present: (1) a comparison/contrast of our results with those from earlier investigations, and (2) a summary and schematic diagram of the physical processes germane to the dust storm over the BRD appended by a list of key signatures that portend the likelihood of dust storms over the western United States.

2. Background Information on Dust Storms: General and Case Specific

[7] Throughout this research paper we make frequent reference to various upper air stations in the U.S. West Coast and surface weather stations in Nevada. These stations along with topographical and geographical features in northwestern Nevada are shown in Figures 1 and 3.

Figure 3.

Geomorphology of Nevada. This map follows the style of Erwin Raisz, artist/illustrator of U.S. 40 Cross Section of the United States [Stewart, 1953, p. 234].

2.1. Dynamics Linked to Dust Storms

[8] Two categories of dynamical processes have been linked to dust storms: (1) storms associated with cyclogenesis where Danielsen's paradigm of large-scale descending trajectories is the central theme [Danielsen, 1968, 1974a, 1974b; Pauley et al., 1996] (also http://marrella.meteor.wisc.edu/Martin_2008.pdf, and J. E. Martin, personal communication, 2010), and (2) secondary vertical circulations where geostrophic adjustments around jet streaks are central to the generation of surface winds [Karyampudi et al., 1995]. In both cases, deep mixed layers adjoining the surface are often present. The dust storms that originate over the world's gigantic deserts are typically associated with cyclogenesis in springtime [Liu et al., 2003]. In these storms, the transport is often on a continental/hemispheric space scale as found for the April 2001 storm that formed over the Gobi Desert. This voluminous dust plume was tracked for 9 days (6–14 April) as it moved from the Gobi to the western USA [Szykman et al., 2003].

2.2. Climatology of Dust Storms in North America

[9] Compared to the southwestern plains of the United States and the northeastern plains of Mexico—notably that area on the eastern edge of the Rocky Mountains and the Sierra Madre Oriental Mountains that runs from Monterrey (Mexico) up through western Texas and along the border of Kansas and Colorado—the state of Nevada has relatively few dust storms [Changery, 1983]. Based on the annual average number of hours with visibility less than 3 miles (about 5 km) due to dust storms, the southwestern plains of the U. S. have a maximum of 45 h (centered near Lubbock, TX). Nevada, on the other hand, has a maximum of 20 h (near Lovelock (LOL), NV). The frequency of dust storms and associated visibility for several stations in Nevada are displayed in Table 1 (locations shown in Figure 1). In every category of dust duration and associated poor visibility, LOL ranks first. Since the primary dust emission sources are northwest of LOL, an area devoid of reporting stations in the Changery [1983] study, the maximum frequency of dust storms in Nevada likely lies northwest of this station. A brief review of the literature relevant to dust storms in the contiguous U. S. is provided by Orgill and Sehrnel [1976].

Table 1. Dust Climatology for Cities in Nevadaa
CityVisibility <11.2 km (hours/year)Visibility <1 km, (hours/year)Visibility <11.2 km (number of episodes/year)Visibility <1 km (number of episodes/year)Visibility <11.2 km (duration (hours)/episode)Visibility <1 km (duration (hours)/episode)
  • a

    Station abbreviations are given in parentheses, and their locations are shown in Figure 1.

Lovelock (LOL)43.412.3019.78.52.21.5
Fallon (NFL)27.22.7019.65.01.40.5
Las Vegas (LSV)18.91.2013.23.61.40.3
Battle Mountain (BAM)10.60.905.60.71.90.6
Winnemucca (WMC)10.91.35.80.91.91.4
Elko (LKN)6.10.103.10.32.00.3
Ely (ELY)3.30.012.90.11.10.1
Tonopah (TPH)2.30.12.00.21.20.7
Yucca Flats (UCC)2.10.011.30.11.60.2
Reno (REV)1.10.010.60.031.60.2

2.3. Precipitation Antecedent to the Dust Storm

[10] The dust storm over the BRD was initiated at approximately 2100 UTC on 28 February 2002. It occurred in conjunction with an upper-level jet streak that passed northeast of the BRD. We follow the evolution of this jet streak and associated larger-scale features over a 48 h period, from 0000 UTC 27 February 2002 until 0000 UTC 1 March 2002. Two weeks prior to the dust storm, widespread precipitation occurred over NW Nevada. This precipitation was associated with a surge of midtropospheric to lower-tropospheric moisture that traveled from the mid-Pacific to northern California and Nevada. The rainfall totals (including the liquid equivalent of snow) for stations in NW Nevada are listed in Table 2 (station locations are shown in Figure 1). Although this was the only significant precipitation event in NW Nevada during the month of February 2002, the amount of precipitation exceeded the monthly average at five of the seven stations. Gerlach (GEL on Figure 1), at the southern end of the BRD, was left with more than an inch of precipitation over a 3–4 day period.

Table 2. Precipitation in Northwestern Nevada During February 2002a
StationTotal Precipitation Feb 2002 (inches)Percent Above (+) or Below (−) Normal, Feb 2002Precipitation 15–20 Feb (inches)
  • a

    Numbers in parentheses in Precipitation 15–20 Feb refer to the days in February when precipitation was recorded. Locations of the stations are shown in Figure 1.

Denio Jct (DJC)0.91+180.91 (18–20)
Gerlach (GEL)1.17+601.06 (17–20)
Lovelock (LOL)0.39−190.39 (19–20)
Reno (REV)0.24−770.12 (15–17)
Rye Patch (RYP)0.71+30.70 (17–20)
Smoke Creek (SMC)0.82+30.78 (16–20)
Winnemucca (WMC)0.89+440.87 (17–20)

2.4. Satellite Imagery of the Dust Storm

[11] Visible imagery from National Oceanic and Atmospheric Administration's (NOAA's) Geostationary Operational Environmental Satellite (GOES-8) indicated that dust was raised from the playa (NE corner of the BRD) at 2045 UTC 28 February 2002 (1300 LST 28 February 2002). The visible image at this time and those at subsequent times are shown in Figure 4. Based on distinct features at the leading edge of the dust storm, the southward-directed propagation speed of the storm was estimated to be 10 m s−1. At 0102 UTC 1 March 2002 (1700 LST 28 February 2002) (Figure 4d), the leading edge of the storm has just passed through Fallon and is slightly north of Reno (REV in Figure 1). Dust striae at the front edge of the dust extend backward toward the NNE—an alignment close to the basin-range orientation shown in Figure 1.

Figure 4.

GOES-8 satellite imagery of cloud and dust valid at (a) 2045 UTC, (b) 2215 UTC, (c) 2315 UTC (28 February), and (d) 0102 UTC (1 March) (1700 LST 28 February 2002). Locations of BRD, SCD, Reno, and Fallon are indicated. Arrows point to the area covered by dust storm.

2.5. PM10 Observations

[12] The particulate matter from the northern Nevada playa is generally classified as fine silt or clay with diameters in the range of 2–8 μm or less (see Udden-Wentworth particle scale of Shao [2000, Table 5.2]). Evidence of the storm's passage through Reno is found from observations of PM10 (particulate matter of aerodynamic diameter ≤ 10 μm). Observations of PM10 concentration are shown in Figure 5. Based on this time series, the dust storm lasted about 8 h in Reno.

Figure 5.

Observed PM10 (particulate matter of aerodynamic diameter ≤ 10 μm) concentration (units in μg m−3) at Reno, Nevada.

3. Salient Features of Upper-Air Circulation

[13] In future reference to time designation, we numerically abbreviate as follows: hours and minutes UTC (month/day) i.e., 1200 UTC 27 February 2002 will be denoted as 1200 UTC (27 February).

3.1. Evolution of the Vorticity Pattern

[14] The 500 hPa vorticity patterns displayed in this section are based on Bellamy's graphical method [Bellamy, 1949]. This method calculates relative vorticity at the centroids of adjoining triangles where triangle vertices are the locations of the upper-air wind observations. This approach obviates the need to interpolate wind observations to grid points. Such interpolation generally introduces interpolation noise—that is, it alters the observations and introduces uncertainty into kinematic fields such as convergence/divergence and vorticity. These kinematic fields are especially sensitive to incremental changes in the wind vectors. By choosing adjoining triangles of smallest area we analyze vorticity on a spatial scale intrinsically tied to station separation. The structure of the field follows from scalar analysis of the point values of vorticity at centroids of the triangles [Saucier, 1955].

[15] At 1200 UTC (27 February), 36 h prior to the onset of the dust storm, an upper-air disturbance was evident over southwestern Yukon Territory (Figure 6a). The vorticity center at this time is located just south of Whitehorse (YXY in Figures 6a and 6b). The progression of this disturbance is shown in Figures 6c and 6d. The positive relative vorticity center intensified and moved southward by 0000 UTC (28 February). At this time there is evidence of a jet streak that runs from Annette Island, AK (PANT) down through Quillayute, WA (UIL). Some wavering in subjective analyses of this vorticity center is evident over the 36 h period from 0000 UTC (28 February) to 1200 UTC (1 March), but this fluctuation is due in part to the limited resolution of the rawinsonde network (∼400 km station spacing and 12-hourly balloon launch cycle). What is beyond doubt is the large-amplitude positive vorticity center in the triangular region between Boise (BOI), Elko (LKN), and Reno (REV) at 0000 UTC (1 March). The rapid amplification of this vorticity center between 1200 UTC (27 February) and 0000 UTC (1 March) was followed by rapid decay. It is appropriate to mention that the evolution of vorticity as well as amplitude that came from the Bellamy approach was in close agreement to the NARR reanalysis.

Figure 6.

The 500 hPa relative vorticity (solid line; contour interval = 2 × 10−5 s−1) and geopotential heights (dashed line; contour interval = 6 dm), air temperature (T; °C), air temperature minus dew point temperature (T - Td; °C) and winds (kt; 10 kt = 5 m s−1) from the upper air soundings valid at (a) 1200 UTC (27 February), (b) 0000 UTC (28 February), (c) 1200 UTC (28 February), and (d) 0000 UTC (1 March). Note that the analysis area shifts further southward in Figures 6c and 6d. X in Figure 6d marks the location of the Black Rock Desert.

3.2. Transverse Circulation

[16] We rely on the reanalysis to provide information on the evolution of the synoptic-scale vertical motion. The analysis of vertical motion greatly benefits from these dynamically consistent analyses that blend forecasts with observations. The 500 hPa NCEP/NCAR reanalyzed vertical motion between 0000 UTC (28 February) and 0000 UTC (1 March) is shown in Figure 7. At 0000 UTC (28 February), the ascending branch of the circulation pattern resides over the left side (NE side) of the exit region with weaker descent to the right (SW side) of the exit region. This pattern is similar to the four-quadrant model of a jet streak as discussed by Carlson [1991, section 14.1]—dynamically forced ascent (descent) in the left (right) exit region and right (left) entrance region (entrance regions not shown in Figure 7).

Figure 7.

NCEP/NCAR reanalyzed vertical motion (contour interval = 1 cm s−1; solid line, upward; dashed line, downward) at 500 hPa overlain on the 500 hPa jet position (isotachs; m s−1; gray) valid at (a) 0000 UTC (28 February), (b) 1200 UTC (28 February), and (c) 0000 UTC (1 March), relative to the jet streak position at 500 hPa (isotachs plotted in m s−1). Location of BOI is shown.

[17] In this case, the vertical motion couplet is consistent with an indirect circulation pattern, i.e., a pattern where the cold air northeast of the exit region is rising and the warm air to the southwest of the exit region is sinking. Over the next 24 h period, the jet streak moves southeastward and the ascending branch migrates southwestward across the jet's exit region until it covers most of northern and central Nevada at 0000 UTC (1 March). At this time, the ascent/descent couplet is principally oriented along the jet streak as opposed to across the streak. The development of curvature in the geopotential field—creation of a smaller-scale trough from BOI toward REV shown at 0000 UTC (3/1)—is an important component in analyzing the vertical motion [Eliassen, 1984]. That is, the propagation of this trough along the stream contributes to lift ahead of the trough and sinking behind it.

3.3. Cooling

[18] As shown in Figure 6, dramatic cooling takes place over Boise, ID (BOI) between 1200 UTC (28 February) and 0000 UTC (1 March). At 500 hPa, the temperature drops from −24°C to −38°C over this 12 h period —slightly greater than 1°C h−1. There is minimal cold air advection in the vicinity of BOI at 1200 UTC (28 February). By 0000 UTC (1 March), however, the 500 hPa vorticity pattern has intensified at a location just south of BOI, and the 500 hPa wind shifts from NW to NE at this location. Based on this NE wind and the temperature gradient, cold air advection amounts to 0.58°C h−1. One can assume that the other primary contribution to cooling in the absence of high humidity/precipitation stems from adiabatic cooling, i.e., adiabatic ascent in a stably stratified atmosphere. Judging from the vertical motion pattern displayed in Figure 7, there is weak lifting at 500 hPa over BOI at 1200 UTC (28 February). This gives way to descending motion by 0000 UTC (1 March). To achieve adiabatic cooling the order of 0.5°C h−1 at BOI, ascent of 2–3 cm s−1 is required. The reanalysis failed in this context. Thus, the vertical motion fields from the NCEP reanalysis are inconsistent with the observed cooling at BOI. This inconsistency is discussed further in section 6.

[19] Cooling occurred at REV, delayed relative to the cooling at BOI. In Figure 8, we display the Reno soundings at 2324 UTC (28 February) and 1116 UTC (1 March). The soundings were taken by NCAR personnel as part of the special field program for graduate students at the University of Nevada-Reno [Cohn et al., 2004, 2006]. The tropopause is close to the 250 hPa level based on the isothermal structure of the atmosphere above this level. Comparison of soundings in Figure 8b indicates a 12 h cooling of 10°–20°C below 700 hPa and 5°C atop the inversion.

Figure 8.

Rawinsonde soundings at Reno (REV), Nevada valid at (a) 2324 UTC (28 February) and (b) 1116 UTC (1 March) plotted on a Tephigram. Wind speeds (kt; 10 kt = 5 m s−1). Point values of temperature at 2324 UTC (28 February) and 0516 UTC (1 March) are overlain on the temperature profile at 1116 UTC (1 March).

4. Surface Analysis

4.1. Evolution of the Gust Front

[20] We have amassed reports from first-order stations (manned National Weather Service stations at airports) and stations in the Regional Automated Weather Station (RAWS) network. The locations of these stations are displayed in Figures 1 and 3, and the observations are shown in Figure 9. Visibilities are not recorded at the RAWS stations and we only show visibilities less than 10 miles (16 km) at the first-order stations.

Figure 9.

Time series of surface observations at stations in northwestern Nevada, Doyle (California), and Boise (Idaho) from 1200 UTC (28 February) to 0600 UTC (1 March). Station identifiers are shown in Figure 1 (temperatures (°C), winds (full barb = 10 kt; 10 kt = 5 m s−1), wind gust (m s−1), and visibility (km; indicated in bold)).

[21] As one macroscopically views the surface reports in Figure 9, the feature that is most intimately tied to the dust storm is the gust-front line, the line that defines the shift in wind direction at the leading edge of the cold front. The wind gusts vary from 20 to 40 mph. The gust front passed Winnemucca (WMC) and Catnip Mountain (CTP) at 1800 UTC (28 February). It then proceeded past Bluewing Mountain (BLM) and Lovelock (LOL) 2–4 h later (see also Figure 1 for station locations). Based on these observations, the gust front is oriented NW–SE and propagates toward the SW. A temperature drop of approximately 10°C follows the gust front passage (delayed by approximately 6 h). This delay was in response to significant surface heating during the late morning and afternoon hours—especially evident at LOL, DYL, NFL, and REV as displayed in Figure 9. The lowest visibility was reported at Fallon (NFL): 1/8 mile (0.2 km) in blowing dust (symbol: $) at 0200 UTC (1 March). As a complement to the time series at surface stations, we also include the 850 hPa temperature and winds valid at 0000 UTC (1 March). As can be seen, there is no organized extratropical cyclone development over the area of interest. Rather, there is a frontal push of cold air (Figure 10).

Figure 10.

NARR temperature (contour interval = 2.5°C; dashed line), geopotential height (contour interval = 30 m; solid line), and wind vectors at 850 hPa valid at 0000 UTC (1 March).

4.2. Mixed Layer

[22] In Richardson's classic paper [Richardson, 1920] that extended Osborne Reynolds' criterion for turbulence [Reynolds, 1893], he showed that turbulence would be generated in the atmosphere when the supply of energy to eddies exceeds the losses [Richardson, 1920, section 4]. Using the conservation of energy principle, he found the criterion to be

equation image

where U is the horizontal velocity, z is the geometric height, θ is the potential temperature and g is the acceleration of gravity. A nondimensional number indicating the dynamic instability of the flow known as the Richardson number (Ri) stemmed from this work and is given by

equation image

where u (v) is the east–west (north–south) component of the mean wind. Ri is a ratio of buoyant production of turbulent kinetic energy (which is the energy required to vertically displace the eddies) to the mechanical production (which is the supply of energy derived from the mean wind). Thus, turbulent energy increases when Ri < 1. In practice, it has been found that Ri < 0.25 is a more meaningful criterion for the onset of turbulence [Wallace and Hobbs, 1977; Stull, 1988, Galperin et al., 2007; Zilitinkevich et al., 2008]. This might be expected in light of Richardson's assumption that turbulence developed about a state of rest which is rarely the case in practice.

[23] From upper-air observations near the BRD (REV, LKN, and BOI), we have calculated Ri at 0000 UTC (1 March) (1600 LST (28 February)) and the results are shown in Table 3. Under essentially clear skies at these stations, a superadiabatic layer of about 50 hPa adjoined the ground. Above this unstable layer, a slightly stable layer extended to 700 hPa. Using boundary values of potential temperature (at the surface and 700 hPa) and the corresponding shears, this criterion indicates that turbulent energy in this layer increased at REV and BOI whereas it decreased at LKN.

Table 3. Richardson Numbers (Ri) Valid at 0000 UTC 1 March 2002a
StationLayer (mbar)equation image (°K km−1)equation image (10−5 s−2)Ri = equation image
  • a

    U is the horizontal velocity vector and θ is the potential temperature.

REV842/700−1.205.65−0.69
LKN835/7001.472.242.25
BOI916/700−0.241.43−0.58

5. Adjustment to Thermal Wind Imbalance

[24] At 1200 UTC (28 February), the momentum associated with the jet streak is out of balance with the mass field (pressure gradient force). Analyses of wind and geopotential height at 300 hPa indicate a subgeostrophic flow regime at 1200 UTC (28 February) (Figure 11). This is also evident by examination of the geopotential thickness pattern for the 700/300 hPa layer shown in Figure 12. The thermal wind is greater than the observed shear at most of the upper-air stations. The thickness pattern in Figure 12a exhibits only slight curvature. Under this condition, the difference between the thermal wind and the observed shear is a measure of the geostrophic imbalance. A substantial change takes place in the 700/300 hPa thickness pattern and associated wind shears by 0000 UTC (1 March) as shown in Figure 12b. Most notable is the presence of cyclonic curvature in the vicinity of the jet's exit region. Further, the geopotential thicknesses have precipitously dropped—most noticeable along the line of stations BOI–LKN–REV (cutting across the exit region).

Figure 11.

Geostrophic wind (kt; denoted by the shorter-shafted vectors), observed wind (kt; denoted by the long-shafted vectors) and geopotential heights (contour interval = 6 dm) on 300 hPa valid at 1200 UTC (28 February).

Figure 12.

Observed wind shear (kt; denoted by the long-shafted vectors), thermal wind (kt; denoted by the shorter-shafted vectors) and layer thickness (contour interval = 6 dm) in 700–300 hPa layer valid at (a) 1200 UTC (28 February) and (b) 0000 UTC (1 March).

[25] In this case of curved flow, the difference between the thermal wind and the observed shear is not an accurate measure of imbalance. A more appropriate measure of balance/imbalance is the vector difference between gradient wind shear and observed shear [Forsythe, 1945]. With cyclonic curvature, the centrifugal and Coriolis forces jointly act to oppose the pressure gradient force. In short, the dynamic imbalance in the nearly rectilinear flow at 1200 UTC (28 February) can be ameliorated through development of cyclonic curvature. The previous work by Moore and VanKnowe [1992] lends support to the importance of curvature as a mechanism for restoration of dynamic balance in synoptic-scale flow regimes.

[26] To calculate the gradient winds at 700 and 300 hPa (and the associated shear), we assume that the trajectory curvature (denoted by K) is approximated by streamline curvature on level surfaces. Realizing this assumption is subject to error in the presence of vertical motion and transient flow [Saucier, 1955], we include an uncertainty of ±20% in the radius of curvature (R) estimate

equation image

where α (in radians) denotes wind direction and s denotes distance along the streamline. K is positive (negative) for cyclonic (anticyclonic) curvature associated with the upper-level troughs (ridges). Under these conditions, Table 4 displays the gradient wind shear in the 700/300 hPa layer at BOI, LKN, and REV alongside the associated thermal wind and observed shear. Even in the presence of curvature uncertainty, the results shown in Table 4 indicate near balance between momentum and mass when curvature effects are included. Based on the 3 h data sets from NARR discussed in section 6, we know that the adjustment period is 6–9 h, considerably smaller than the half pendulum day at 45° latitude (∼17 h).

Table 4. Geostrophic Wind Shear (Thermal Wind), Gradient Wind Shear, and Observed Wind Shear Along the Line of Stations Southwest of the Exit Region of the Jet Stream at 0000 UTC 1 March in the 700/300 hPa Layera
StationThermal WindGradient Wind ShearObserved Wind Shear
  • a

    Values given in format: m s−1/standard meteorological convention for direction. The gradient wind shear is the mean of three calculations with ± 20% uncertainty in the radius of curvature.

BOI36/31520/32518/300
LKN46/29534/28539/290
REV39/32024/31523/330

6. NARR Analysis

[27] As a complement to the large-scale depiction of ascent/descent in the vicinity of the jet (Figure 7), we have accessed vertical motion fields from the NARR. This archive contains a high-resolution (32 km) climate data set for the North American continent (including oceanic areas bordering the landmass). Postprocessed forecasts from the operational NCEP's Eta model, augmented by a variety of observations, are the basis for the reanalysis [Mesinger et al., 1988].

6.1. Vertical Motion

[28] The NARR vertical motion serves to deliver a finer-scale space and time complement to the synoptic-scale lifting/descent from NCEP/NCAR discussed earlier. In particular, we have chosen to examine the changes in the NARR ascent/descent pattern over an 18 h period, 0600 UTC (28 February) to 0000 UTC (1 March), shown in Figure 13. Consistent with the earlier analyses, the exit region of the jet streak moves southeastward over the 18 h period. At 0600 UTC (28 February), there is pronounced ascent to the left (NE) of the exit region and weaker descent to the right (SW) of the exit region. As time proceeds, the ascending motion intensifies and moves across the exit region. We examine the cooling at two locations BOI and BRD between 1200 UTC (28 February) and 0000 UTC (1 March). The examination is made using the NARR where analyses are available at 3 h intervals. Table 5 displays results for BOI at 500 hPa and for BRD at 700 hPa. This choice was dictated by our desire to show changes in midtroposphere at BOI where the cold front aloft moved past this station and in the lower troposphere (at BRD where the cold gust front moved past the station). The cooling at BOI was ∼12°C and ∼9°C at BRD. At the earlier times during the 12 h period, cooling due to adiabatic ascent dominated cooling due to advection. During the later part of the period, the cold air advection dominated. From the data, it is difficult to completely understand the concerted action of adiabatic cooling and cold air advection in explaining the temperature drop. That is, averaging the cooling due to both sources does not account for the analyzed drop in temperature.

Figure 13.

NARR vertical velocity fields (contour interval = 1 cm s−1) at 500 hPa overlain on the 500 hPa jet position (isotachs; m s−1) valid at (a) 0600 UTC, (b) 1200 UTC, (c) 1800 UTC (28 February), and (d) 0000 UTC (1 March). Locations of BOI, LKN, REV, and SLC are shown.

Table 5. The 500 hPa Cooling at BOI and 700 hPa Cooling at BRD Based on NARR Analyses Between 1200 UTC (28 February) and 0000 UTC (1 March)
TimeStationPressure Level (hPa)Air Temperature (°C)w-Vertical Velocity (cm s−1)Cooling Rate due to Horizontal Temperature Advection (°C h−1)Adiabatic Cooling Rate (°C h−1)
(28 Feb) 1200 UTCBOI500−24.0+3.93+0.21−0.64
(28 Feb) 1800 UTCBOI500−29.7+0.01−0.45+0.00
(1 Mar) 0000 UTCBOI500−35.6−3.60−0.60+0.70
(28 Feb) 1200 UTCBRD700−2.5+2.75−0.01−0.36
(28 Feb) 1800 UTCBRD700−5.1−1.51−0.36+0.34
(1 Mar) 0000 UTCBRD700−11.2−2.32−1.12+0.11

[29] Further examination of the vertical motion in a cross section extending from north of BOI down through the BRD (SW–NE line of cross section is shown in Figure 13) adds important information to the examination of the cooling process. These vertical cross sections at 1200 UTC (28 February) and 0000 UTC (1 March) are shown in Figure 14. Between 0600 and 1200 UTC, the upward motion increases just north–northeast of BOI (figure not shown). At 1200 UTC, there is significant ascent (4–6 cm s−1) in the 700–400 hPa layer above BOI. At this time, the jet streak was right above BOI. Thereafter, the upward vertical motion pattern rapidly moves to the southwest and across the exit region (see also Figure 13). Although this transition is consistent with a flow pattern where curvature significantly increases with time [Moore and VanKnowe, 1992], the time scale of the change in the vertical motion at BOI and neighboring locations takes place over a period less than 12 h The isentropic analysis at 0000 UTC (1 March) also captures the cold front aloft that gradually descends to the surface with a pronounced well-mixed boundary layer with a depth of about 2 km AGL (NARR planetary boundary layer depths ∼2500 m AGL at 2100 UTC (28 February) in northwestern Nevada). This is also seen in the Reno sounding at 0516 UTC (see Figure 8).

Figure 14.

NARR X-P cross sections of isentropes (contour interval = 2.5 K), horizontal winds (wind barbs; 10 kt = 5 m s−1), isotachs (contour interval = 5 m s−1), and vertical velocity (shaded; highest and lowest magnitudes are indicated in boxes) along the cross section shown in Figure 13 at (left) 1200 UTC (28 February), and (right) 0000 UTC (1 March). Locations of REV, BRD, and BOI are shown.

6.2. Isallobaric Wind

[30] The isallobaric wind Visal is the ageostrophic wind component that forms in response to local pressure changes—quantitatively proportional to the gradient of pressure tendencies and directed opposite to this gradient [Saucier, 1955; Bluestein, 1992; Martin, 2006; Rochette and Market, 2006]. On a constant pressure surface, it is expressed as follows:

equation image

where f is the Coriolis parameter and Φ is the geopotential. On a constant height surface, pressure tendency replaces height tendency. As expected, a rapidly moving pressure/weather system or a short-wave trough as found in our case will lead to significant pressure rise/fall couplets and associated isallobaric winds. This is the case for the rapid pressure/height changes along the line of stations BOI-LKN-REV in our study. The boundary layer isallobaric winds (winds in the surface to 700 hPa layer for our study) are northeasterly/northerly along this line during the afternoon of dust storm initiation (Figure 15). The isallobaric wind has physical meaning and is especially relevant in our case study since its magnitude and direction (as shown in Figure 15) is approximately equal to the ageostrophic wind vector. For example, from the NARR analyses at 700 hPa over the BRD, the total wind speed and ageostrophic wind speed at 0000 UTC (1 March) are 11.2 and 8.6 m s−1, respectively, while the isallobaric wind speed over the period 1800 UTC (28 February) to 0000 UTV (1 March) is 6.1 m s−1. The directions of the ageostrophic and isallobaric winds are NNE.

Figure 15.

NARR surface pressure tendency (mbar (6 h)−1) valid during 1800 UTC (28 February) to 0000 UTC (1 March) and the isallobaric wind vectors at 700 hPa are shown by the bold arrows. Locations of BOI, LKN, and REV are shown.

[31] There is also some evidence of confluence in the isallobaric winds along this line of stations that contributes to strengthening the cold front and its attendant turbulence kinetic energy (TKE). The NARR analysis indicates TKE's the order of 2–3 J kg−1 at 825 hPa (not shown).

[32] The isallobaric winds have formed in response to the mass adjustments above the boundary layer. The time scale of Visal in the boundary layer is linked to the time scale of mass adjustments above this boundary layer—adjustments associated with vertical motions/divergence-convergence patterns.

6.3. Profiler Winds Versus NARR Wind Profiles

[33] As part of the academic field exercise conducted by NCAR/University of Nevada-Reno, a wind profiler was deployed that made measurements at the time of dust storm passage through Reno [Cohn et al., 2004, 2006]. The upper-air wind structure in the lowest 2–3 km at the time of storm passage (∼0130 UTC (1 March)) as well as several hours before and after passage is shown in Figure 16. Beside the profiler data we have plotted the profiles of wind at REV from the NARR data set. A most notable feature of wind profiler data is the presence of a marked discontinuity in wind speed at about 1.5 km (AGL) at the time of storm passage. This would appear to be indicative of a low-level wedge (a 1.5 km deep wedge) of northerly wind similar to the outflow from a thunderstorm. There is an absence of such a discontinuity in the NARR profile. Another feature of importance is evidence of an abrupt shift in wind direction at the lowest levels of the profiler data just prior to dust storm passage—a shift from weak westerly wind (∼1 m s−1) to a stronger northerly wind (∼14 m s−1) at 0145 UTC. Again, this feature is absent in the NARR profiles. This shift is indicative of a confluence zone near the frontal boundary and is likely correlated with a zone of strong convergence and associated strong lift.

Figure 16.

A comparison of (a) wind profiler measurements and (b) NARR wind profiles at Reno valid from 1200 UTC (28 February) to 1200 UTC (1 March) (in knots; 10 kt = 5 m s−1).

7. Discussion and Conclusions

7.1. Comparison/Contrast With Danielsen [1974b] and Pauley et al. [1996]

[34] In their studies of dust storms over the western United States, Danielsen [1974b] (author's full name is abbreviated in the following discussion as EFD) and Pauley et al. [1996] (abbreviated in the following discussion as PBB) indicated that the high-momentum air impacting the surface layer came from the stratosphere. In this case study, the air that impacts the BRD is not associated with a stratospheric intrusion. This is borne out by: (1) structure of isentropic surfaces along the line of stations from St. George, BC (ZXS) to BOI at 0000 UTC (1 March) (Figure 17), and (2) the absence of ozone intrusions from the stratosphere into the troposphere (satellite imagery not shown). In Figure 17 the isentropes tend to parallel the tropopause as opposed to intersecting the troposphere-stratosphere boundary. The mean tropopause pressures at BOI, LKN, and REV during the period 1200 UTC (27 February) to 0000 UTC (1 March) are 276 hPa (9.7 km), 243 hPa (10.6 km), and 228 hPa (11.0 km), respectively.

Figure 17.

X-Z vertical cross section of potential temperature (°K), geopotential height (dm), and wind barbs (full barb = 10 kt, 10 kt = 5 m s−1) valid at 0000 UTC (1 March).

[35] Another primary difference between our results and those of EFD and PBB is the nature of the transverse or secondary circulation. Although EFD refrained from discussing vertical circulation in baroclinic zones, it is clear that the wide sweep of descending air in his case was consistent with the indirect circulation about the jet streak's exit region— descending air on the anticyclonic side of the jet. PBB's arguments rest firmly on the indirect circulation that straddled the jet streak's exit region. In essence, these authors argue that the descending branch of the indirect circulation is responsible for the vertical transport of high-momentum air to the top of the mixed layer—a mixed layer that was present in their cases as well as ours. In our case, synoptic analysis indicates a transition from indirect to direct circulation about the jet streak's exit region—a circulation that leads to cooling and creation of a cold front that tracked from the midtroposphere to the surface. The cold front moved into and likely enhanced the zone of deep mixed layer over the BRD and ablated dust that mixed to 1–2 km. This sequence stands in contrast to the one described by EFD:

During its [momentum] descent, the jet decelerates in speed but wind speeds from 60–80 kt do reach the adiabatic layer and then rapidly mix to the ground as strong gusts. These strong wind gusts sandblast the arid soil, disaggregating soil particles, creating small soil aerosols which then mix vertically through the adiabatic volume into the stable transition layer above…. Danielsen [1974a, p. 171]

[36] The issue of large-scale dynamic imbalance and restoration of balance was central to our theme. In our case, the development of large-scale cyclonic curvature in the flow field gives rise to centrifugal force, that when paired with Coriolis force, counterbalances the pressure gradient force and serves to force the system back toward equilibrium. Complete understanding of the subsynoptic- and synoptic-scale interactions that result in an adjustment time scale less than the half pendulum day is beyond the scope of this investigation. Nevertheless, realizing that this is a crucial issue, we further explore these interactions in a companion paper (M. L. Kaplan et al., Dust storm over the Black Rock Desert: Subsynoptic analyses of unbalanced circulations across the jet streak, manuscript in preparation, 2011) where the Advanced Weather Research Forecast Model [Skamarock et al., 2008] is used.

7.2. Succinct Summary of Dust Storm Dynamics

[37] Based on the surface and upper-air analyses and arguments related to the adjustment, we develop a schematic view of the dust storm that is shown in Figure 18. It is difficult to place a single time on the schematic. In essence, we try to depict processes at work over the period of adjustment —the 6 h period between 1800 (28 February) and 0000 UTC (1 March).

Figure 18.

Schematic diagram depicting the dynamical processes associated with the dust storm during 1800 UTC (28 February) to 0000 UTC (1 March).

[38] The schematic indicates the movement of lifting from the cold-air side (left side when looking downstream) of the jet streak's exit region to the warm air side (right side) (Refer to NARR vertical motion patterns, Figure 13). A corresponding descending motion moves into the region northwest of BOI. In conjunction with the movement of the upward motion pattern, the vorticity increases in response to tilting and stretching the vortex tube. Consistent with the vorticity pattern, a northeasterly wind develops at BOI and this is associated with cold air advection (a cold front aloft) where cooling is also linked to adiabatic ascent. Below the region of lifting, the surface pressure increases (net mass flux convergence in the column) and leads to an ageostrophic (essentially isallobaric) wind from the northeast, which organizes a cold gust front that moves over the BRD and ablates dust.

[39] Despite some difference in synoptic regimes and in the modes of adjustment for the cases we have compared and contrasted, there is sufficient overlap to identify key signatures in the analyses that portend the likelihood of dust storms in the western U.S. The key features are the following: (1) the presence of a pronounced/well-defined jet streak in the midlevel to upper level of the troposphere, (2) evidence of vertical or secondary circulation in response to dynamic imbalance, (3) development of substantial curved flow aloft (increasing with time), (4) lower tropospheric isallobaric component of the ageostrophic wind signal orthogonal to the jet's exit region, and (5) a cold front, initially aloft, that descends to the surface and impacts a dust emission source overlain by a well-mixed layer where the bulk density of the sediments is low.

[40] Although knowledge of these signatures offers forecasting guidance, critical questions remain unanswered. Among them are the following: what are the consequences of adjustments in the presence of supergeostrophy as opposed to the subgeostrophic state we investigated?, and what is the relative frequency of dust storm genesis that follows the scenario we investigated compared to the Danielsen [1974b]/Pauley et al. [1996] scenario? Beyond these questions, it is important to understand the interplay of physical processes that lead to the planetary boundary layer structures found in the profiler data from Reno. To gain more insight into these processes, high-resolution simulations with a new generation mesoscale model such as WRF will be required [Skamarock et al., 2008]. In particular, we will employ WRF in a successor study to determine how the ascending motion on the right side of the jet becomes stronger (Figure 14) as a result of unbalanced mass/momentum adjustments resulting in strengthening the low-level cold front and PBL turbulence. Preliminary work with WRF is underway.

Acknowledgments

[41] We are especially grateful to the graduate students and National Center for Atmospheric Research (NCAR) technical staff who took part in the NCAR/UNR (University of Nevada-Reno) Field Program in the Washoe Valley during February and March 2002. This field program was a major component of the UNR graduate level course in Atmospheric Instrumentation and Observations that was taught by coauthors John Hallett and Steve Cohn. The enthusiastic attitude and the diligent effort of the students and technical support staff to collect data were essential to the project's success. One of the authors, Ramesh Vellore, now a postdoctoral scholar at Desert Research Institute, was a student in that class, and two others, Claudio Mazzoleni and Lynn Reinhart Mazzoleni, now professors of physics and chemistry at Michigan Technical University, respectively, supplied us with the PM10 data. Trond Iversen, a protégé of Arnt Eliassen, is credited with supplying us with an insightful view of Eliassen's contributions to secondary circulations in response to dynamic imbalance. The two teams of meteorologists and computer scientists who developed the NCEP/NCAR and NARR reanalysis data sets are commended for creation of these invaluable archives. The four formal JGR-Atmospheres reviewers spent considerable time on the manuscript and offered cogent suggestions for improvement that were followed, and this led to tighter arguments and clarity in presentation. We offer hearty thanks to these individuals. Finally, we thank George Platzman for his excellent set of lecture notes from courses in geophysical dynamics at the University of Chicago – notes that included in-depth examination of geostrophic adjustment theory and secondary circulations in response to dynamic imbalance. Jim Howcroft's translation of Obukhov [1949] constituted an important element of these notes. Joan O'Bannon, draftsperson at National Severe Storms Laboratory, assisted us in obtaining electronic versions of the hand-drawn weather maps. J. Lewis and R. Vellore acknowledge support from the Office of Naval Research (ONR) (grant N00014-08-1-0451). Further, Darko Koracin the principal investigator of this ONR project, offered encouragement to pursue this line of research. The National Center for Atmospheric Research (NCAR) is sponsored by the National Science Foundation (NSF).

Ancillary