Episodic Dust Events along Utah’s Wasatch Front
W. James SteenburghAND Jeffrey D. Massey[JS1]
Department of Atmospheric Sciences, University of Utah, Salt Lake City, UT
Thomas H. Painter
Jet Propulsion Laboratory, Pasadena, CA
In preparation for submittal to
Journal of Applied Meteorology and Climatology
Draft of Tuesday, October 11, 2011
Corresponding author address: Dr. W. James Steenburgh, Department of Atmospheric Sciences, University of Utah, 135 South 1460 East Room 819, Salt Lake City, UT, 84112.
E-mail:
Abstract
Episodic dustevents cause hazardous air quality along Utah’s Wasatch Front anddust loading of the snowpack in the adjacent Wasatch Mountains. This paper presents a climatology of episodic Wasatch Front dust events based on surface-weather observations from the Salt Lake City International Airport (KSLC), GOES satellite imagery, and the North American Regional Reanalysis. Dust events at KSLC, defined as any day with at least one report of a dust storm, blowing dust, and/or dust in suspension (i.e., dust haze) with a visibility of 10 km (6 mi) or less, occur an average of 4.3 days per water year (WY, Oct–Sep), with considerable interannual variabilityfrom 1930–2010. The monthly frequency of dust-events is bimodal with primary and secondary maxima in Apr and Sep, respectively. Dust reports are most common in the late afternoon and evening.
An analysis of the 33 most recent (2001–2010 WY) events at KSLC indicates that 16were associated with a cold front or baroclinic trough, 11withairmass convection and related outflow, 4with persistent southwest flow ahead of a stationary trough or cyclone over Nevada, and 2with other synoptic patterns. GOES satellite imagery and backtrajectories from these 33 events, as well as 61 additional events from the surrounding region, illustrate that emissions sources are mostly concentrated in the deserts of southern Utah and western Nevada, including the Sevier dry lake bed, Escalente Desert, and Carson Sink. Efforts to reduce dust emissions in these regionsmay help mitigate the frequency and severity of hazardous air-quality episodes along the Wasatch Front and dust loading of the snowpack in the adjacent Wasatch Mountains.
1. Introduction
Dust storms impact air quality (Gebhart et al. 2001; Pope et al. 1996), precipitation distribution (Goudie & Middleton, 2001), soil erosion (Gillette 1988;Zobeck 1989), the global radiation budget (Ramanathan 2001), and regional climate (Nicholson 2000, Goudie & Middleton 2001). Recent research regarding regional hydrologic and climatic change produced by dust-radiative forcing of the mountain snowpack of western North America and other regions of the world has initiated a newfound interest in dust research. (Hansen and Nazarenko 2004; Painter et al. 2007; Painter et al. 2010). For example, observations from Colorado’s San Juan Mountainsindicate that dust loading increases the snowpack’s absorption of solar radiation,thus decreasing the duration of snow cover by several weeks (Painter et al. 2007). Modeling studies further suggest that dust-radiative forcing results in an earlier runoff with less annual volume in the upper Colorado River Basin (Painter et al. 2010).
Synoptic and mesoscale weather systems are the primary drivers of global dust emissions and transport. Mesoscale convective systems that propagate eastward from Africa over the Atlantic Ocean produce half of the dust emissions from the Sahara Desert, the world’s largest Aeolian dust source (Swap et al. 1996; Goudie Middleton 2001). Dust plumes generated by these systems travel for several days in the large-scale easterly flow (Carlson 1979), with human health and ecological impacts acrossthe tropical Atlantic and Caribbean Sea (Goudie & Middleton 2001; Prospero & Lamb 2003). In northeast Asia, strong winds in the post-cold-frontal environment of Mongolian Cyclones drive much of the dust emissions (Yasunori and Masao 2002; Shoa, Wang, 2003; Qian et al., 2001). The highest frequency of Asian dust storms occurs over the Taklimakan and Gobi Deserts of northern China where dust is observed 200 d yr-1 (Qian et al., 2001). Fine dust from these regions can be transported to the United States, producing aerosol concentrations above National Ambient Air Quality Standards (Husar et al. 2001; Jaffe et al., 1999; Fairlie et al., 2007)
The Great Basin, Colorado Plateau, and Mojave and Sonoran Deserts produce most of the dust emissions in North America (Tanaka and Chiba 2006;see Fig. 1[JS2] for geographic and topographic locations). Most land surfaces in these deserts are naturally resistant to wind erosion due to the presence of physical, biological, and other crusts (Gillette1980). However, these crusts are easily disturbedleading to increased dust emissions, in some cases long after the initial disturbance (e.g., Belnap et al. 2009). Based on alpine lake sediments collected over the interior western United States, Neff et al. (2008) found that dust loading increased 500% during the 19th century, a likely consequence of land-surface disturbance by livestock grazing, plowing of dryland argricultural soils, and other activities.
Several studies suggest that the synoptic and mesoscale weather systems that generate dust emissions and transport over western North America vary geographically and seasonally. In a dust climatology for the contiguous United States,Orgill and Sehmel (1976)proposed several including convective systems, warm and cold fronts, cyclones, diurnal winds, and specifically for the western United States, downslope (their katabatic) winds generated by flow-mountain interactions. They identified a spring maximum in the frequency of suspended dust for the contiguous United States as a whole, which they attributed to cyclonic and convective storm activity, but found that several locations in the Pacific and Rocky Mountain regions have a fall maximum. However, they made no effort to quantify the importance of the differing synoptic and mesoscale systems. In Arizona,Brazel and Nickling (1986, 1987) found thatfronts, thunderstorms, cutoff lows, and tropical disturbances (i.e., decaying tropical depressions and cyclones originating over the eastern Pacific Ocean) are the primarydriversof dust emissions. The frequency of dust emissions from fronts is highest from late Fall–Spring, thunderstorms in the summer, and cutoff lows from May–June and Sep–Nov. Dust emissions produced by tropical disturbances are infrequent, but are likely confined to Jun–Oct when tropical cyclone remnants move across the southwest United States (Ritchie et al. 2011[JS3]). For dust events in nearby California and southern Nevada, Changery (1983) and Brazel and Nickling (1987) also established linkages with frontal passages and cyclone activity, respectively. In addition to synoptic and mesoscale systems, these studies also cite the importance of land-surface conditions (e.g. soil moisture, vegetation) for the seasonality and spatial distribution of dust events.
None of these studies, however, have specifically examined the Wasatch Front of northern Utah, where episodic dust eventsproduce hazardous air quality in the Salt Lake City metropolitan area and contribute to dust loading of the snowpack in the nearby Wasatch Mountains (Fig. 2[JS4]). From 2002–2010, wind-blown dust events contributed to 13 exceedances of the National Ambient Air Quality Standard for PM2.5 or PM10 in Utah (T. Cruickshank, Utah Division of Air Quality, Personal Communication). Dust loading in the Wasatch Mountains affects a snowpack that serves as the primary water resource for approximately 400,000 people and enables a $1.2 billion winter sports industry, known internationally for the “Greatest Snow on Earth” (Salt Lake City Department of Public Utilities 1999; Steenburgh and Alcott 2008; Salt Lake Tribune 2011).
This paperHere we examines the climatological characteristics and emissions sources duringWasatch Front[JM5]dust[JS6] events. We find that Wasatch Front dust-events occur throughout the historical (1930–2010 water year) record, with considerable interannual variability. Events are driven primarily by strong winds associated with cold fronts or airmass convection, with thedeserts and dry lake beds of southern Utah, as well as the Carson Sink of Nevada,serving as primaryregional emission sources[JS7][JM8]. Dust emission mitigation efforts in these regions may reduce the frequency and severity of related hazardous air quality events along the Wasatch Front and dust loading of the Wasatch Mountain snowpack.
2. Data and methods
a. Long-term climatology
Our long-term dust-event climatology derives from hourly surface weather observations from the Salt Lake City International Airport (KSLC), which we obtained from the Global Integrated Surface Hourly Database (DS-3505) at the National Climatic Data Center (NCDC). KSLC is located in the Salt Lake Valley just west of downtown Salt Lake City and the Wasatch Mountains (Fig. 1) and provides the longest quasi-continuous record of hourly weather observations in northern Utah. The analysis covers the 1930–2010 water years (Oct–Sep) when 97.9% of all possible hourly observations are available.
The hourly weather observations included in DS-3505 derive from multiple sources, with decoding and processing occurring at either operational weather centers or the Federal Climate Complex in Asheville, NC (NCDC 2001, 2008). Studies of dust eventsfrequently use similar datasets (e.g., Nickling and Brazel 1984; Brazel and Nickling 1986; Brazel and Nickling 1987; Brazel 1989; Hall 1981; Orgill and Sehmel 1976; Changery 1983; Weihong 2002; Kurosaki and Masao 2002; Shao et al. 2003; Song et al. 2007;Shao and Wang 2003). Nevertheless, while hourly weather observations are useful for examining the general climatological and meteorological characteristics of dust events, they do not quantify dust concentrations,making the identification and classification of dust somewhat subjective. Inconsistencies arise from observer biases, changes in instrumentation, reporting guidelines, and processing algorithms. These inconsistencies result in the misreporting of some events (e.g., dust erroneously reported as haze) and preclude confident assessment and interpretation of long-term trends and variability.
Consistent with World Meteorological Organization (WMO) guidelines (WMO 2009), the present weather record in DS-3505 includes 11 dust categories (Table 1[JS9]). During the study period, there were 916 blowing dust (category 7), 178 dust-in-suspension (category 6), 7 dust storm (categories 9, 30–32, and 98) and one dust or sand whirl report (category 8) at KSLC. There were no severe dust storm reports (categories 33–35). Amongst the blowing dust, dust-in-suspension, and dust storm reports, there were 69 with a visibility 6 statute miles (10 km), the threshold currently used by the WMO and national weather agencies for reporting blowing dust or dust-in-suspension (Shao et al. 2003; Federal Meteorological Handbook, 2005). Since these events are weak, or may be erroneous, they were removed from the analysis. This includes all but one of the 7 dust storm reports. The dust or sand whirl report was also removed since we are interested in widespread events rather than localized dust whirl(s) (a.k.a. dust devils). The resultinglong-term dust-event climatology is based on the remaining 1033 reports. A dust day is any day (MST) with at least one such dust report.
b. Characteristics of recent dust events
The analysis of the synoptic, meteorological, and land-surface conditions contributing toWasatch Front dust events concentrates onevents at KSLC during most recent ten-year period (2001–2010). This enables the use of modern satellite and reanalysis data, and limits the number of events, making the synoptic analysis of each event tractable.
Resources used to synoptically classify dust events, composite events, and prepare case studies includes the North American Regional Reanalysis (NARR), GOES satellite imagery, Salt Lake City (KMTX) radar imagery, and hourly KSLC surface weather observations and remarks from DS-3505. The NARR is a 32-km, 45-layer reanalysis for North America based on the National Centers for Environmental Predication (NCEP) Eta model and data assimilation system (Mesinger et al. 2006). Compared to the ERA-Interim and NCEP-NCAR reanalysis, the NARR better resolves the complex terrain of the Intermountain West, but still has a poor representation of the basin and range topography over Nevada (seeJeglum et al. 2010). We obtained the NARR data from the National Oceanic and Atmospheric Administration (NOAA) Operational Model Archive Distribution System (NOMADS) at the National Climatic Data Center web site ( the level-IIKMTX radar data from NCDC (website: and the GOES data from the NOAA Comprehensive Large Array-Data Stewardship System (CLASS,
c. Dust emission sources
We identify dust emission sources during this most recent 10-year period using a dust-retrieval algorithm applied to GOES satellite data. Because the algorithm only works in cloud-free areas and many dust events occur in conjunction with cloud cover, we expand the number of events to include those identified in: (1) DS-3505 reports from stations in the surrounding region with at least 5 years of hourly data (Fig. 1[JS10]), (2) the authors’ personal notes, and (3) Utah Avalanche Center annual reports. This analysis is thus not specific to KSLC, but does identify emissions sources that contribute to dust events in the region.
Ourdust-retrieval algorithm follows that described by Zhoa et al. (2010) for MODIS. First,we substitute the GOES 10.7 μm channel for the MODIS 11.02 μm channel. Then the two reflectance condition thresholds used to identify the presence of clouds from the MODIS .47, .64, and .86 μm channels are replaced by a single threshold (.35) that uses the only visible channel on GOES. Finally, the maximum threshold for the brightness temperature differencebetween the 3.9 and 11 μm (11 and 12 μm) bands was changed from -.5 ºC to 0 ºC (25 ºC to 10 ºC). These adjustments enable the identification of visible dust over Utah using GOES data, although uncertainties arise near cloud edges, when the sun angle is low, or when the dust concentrations are low or near the surface. The algorithm is applied every 15 min during the daylight hours (1400–0200 UTC), with plume origin and orientation identified subjectively.
3. Results
a.Long-term climatology
Dust events at KSLC occur throughout the historical record, with an average of 4.3 per water year (Fig. 3)[1]. Considerable interannual variability exists, with no events reported in seven years (1941, 1957, 1981, 1999, 2000, 2001, 2007) and a maximum of 15 in1934. No effort was made to quantify or assess long-term trends or interdecadal variability given the subjective nature of the reports and changes in observers, observing methods, and instrumentation during the study period.
Based on current weather observing practices (Shao and Wang 2003; Glossary of Meteorology 2000), the minimum visibility when dust is reported on 95.40%, 2.59%, and 2.01% of the dust days meets the criteria for blowing dust (>5/8 statute miles1 km), a dust storm (0.5/165 km -– 1 km5/8 statute miles), or a severe dust storm (<0.5 km5/16 statute miles)[JM11], respectively (Fig. 4). The observed visibility meets these criteria in 98.04%, 1.20%, and 0.76% of all dust reports. Therefore, only a small fraction of the dust events and observations meet dust storm or severe dust storm criteria.
To integrate the effects of event severity, frequency, and duration into an estimate of the annual near-surface dust flux[JS12], we first estimate the dust concentration, C (µg/m3), for each dust report following eq. 9.81 of Shao (2008, p. 334):
C = 2802.29VIS-0.84) VIS < 3.5 km
C = exp(-0.11VIS + 7.62))VIS ≥ 3.5 km
whereVIS is the visibility. Multiplying by the wind speed and integrating across all observation intervals yields an estimated mean annual near-surface dust flux of 399.4 g/m2, with a maximum of 2810.2 g/m2 in 1935 (Fig. 5). Because it integrates event severity, frequency, and duration, the annual near-surface dust flux provides a somewhat different perspective from the annual number of dustdays (Fig. 3). For example, 1934 featured the most dust days, but the greatest near-surface dust flux occurred in 1935. In 2010, there were only 2 dust days, but also a pronounced decadal-scale maximum in near-surface dust flux.
The monthly distributions of dust days (Fig. 6) and estimated dust flux (Fig. 7) arebimodal, with primary and secondary peaks in Apr and Sep, respectively. The dust flux is distinctly lower in the summer compared to the dust-day frequency, suggesting summer dust events are characterized by shorter or weaker events. The local maximum in dust flux during Jan is rather surprising, but careful examination of the data revealed a particularly strong two-day event in January of 1943 that contributed to 83% of the Jan monthly mean[JS13]. From Mar–May, the climatologicalwhich usually encompasses the maximum snowpack snow water equivalent maximum and beginning of the Spring runoff, the mean three-month dust flux is 237 g/m2, or 59XX% of the annual flux.
Similar bimodal or modal distributions with a spring dust maximum have been identified in the Taklimakan desert of China, southern Great Plains of the United States, Mexico City, and the Canadian Prairies (Yasunori and Masao 2002; Stout 2001; Jauregui 1989; Wheaton and Chakravarti 1990). The spring peak appears to be the result of erodible land-surfaces and enhanced upper level support for surface wind events from the migration of the polar jet stream over these areas (Lee, 1995[JM14])?? In fact, At KSLC, the bimodal distribution at KSLC is very similar to that of Intermountain cold fronts and cyclones, which are strongest and most frequent in the spring (e.g., Shafer and Steenburgh 2008; Jeglum et al. 2010). These features produce persistently strong winds capable of generating dust emissions and transport during favorable land-surface conditions. Interestingly fact, dust was reported at KSLC within 24 h of the passage oduringf12 of the 25 strongest cold fronts identified by Schafer and Steenburgh (2008).
The mean wind speed during dust reports at KSLC is 11.6 ms-1 with a standard deviation of 4.0 ms-1, slightly higher than the 8.5 ms-1 and 9.29 ms-1found by Holcombe et al. (1996) for Yuma, AZ and Blythe, CA, respectively. Therefore, and for convenience, we use 10 ms-1 as an estimated threshold velocity for dust emissions and transport. At KSLC winds ≥10 m s-1are most common in Mar and Apr, with a relatively even distribution the rest of the year (Fig. 9). This Mar and Apr maximum resembles the springtime peak in dust days and flux, but the lack of a fall secondary maximum and winter minimum suggests that the secondary peak inother factorsdust days and flux may be related to seasonal changes to vegetation, soil conditions, and soil moisture (Neff et al., 2008, Belnap et al., 2009, Gillette, 1999) are also important contributors to dust flux quantity and dust day frequency.