Episodic Dust Events along Utah’s Wasatch Front W. JAMES STEENBURGH AND JEFFREY D. MASSEY 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 Saturday, February 06, 2016 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: jim.steenburgh@utah.edu Abstract Episodic dust events contribute to hazardous air quality along Utah’s Wasatch Front urban corridor and, through despoition onto the snowpack of the adjacent Wasatch Mountains, regional hydroclimatic change. This study creates a climatology of these events using surfaceweather observations, GOES visible satellite imagery, and the North American Regional Analysis. In hourly weather observations from the Salt Lake International Airport (KSLC), a dust storm, blowing dust, and/or dust in suspension (i.e., dust haze) with a visibility 10 km (6 mi) or less occurs an average of 4.3 days per water year (Oct–Sep), with considerable interannual variability evident during the 1930–2010 study period. The monthly frequency of days with at least one dust report is strongly bimodal with primary and secondary maxima in Apr and Sep, respectively. Dust reports exhibit a strong diurnal modulation and are most common in the late afternoon and evening. Of the 33 recent (2001–2010) events, 16 were produced by transient cold-frontal troughs, 11 were produced by airmass convection and related outflow, 4 were produced by persistent southwest flow ahead of a stationary trough, and 2 were generated by other mechanisms. During the transient trough events, 25% of the dust reports occur with strong south–southwest winds in the prefrontal environment, 100% during and/or within 3 h of a cold-frontal passage, and 13% more than 3 h after cold-frontal passage. Synoptic composites and case studies further illustrate the important parameters behind these mechanisms. GOES satellite imagery and backtrajectories from the 33 recent events at KSLC, as well as 61 additional events identified in the surrounding region, indicate that the primary dust emission sources are clustered in the deserts and dry lake beds of southern Utah, the burn area of the 2007 Milford Flat Fire, and the Carson Sink of Nevada. Efforts to reduce dust emissions in 2 these regions, especially those in southwest Utah, may help mitigate the frequency of severity of hazardous air quality episodes along the Wasatch Front and dust loading in the adjacent Wasatch Mountains. 1. Introduction Despite the numerous studies on dust transport and origin in the Asian and African Deserts, relatively few studies occurred in the western United States, and even fewer are specific to Utah. 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 for geographic and topographic locations). The most productive sources in these regions are the modern and ancient depositional environments, such as dried lakebeds and runoff regions (Gillette, 1999). However, the land surfaces in these regions are naturally resistant to wind erosion due to the presence of physical, biological, and other vulnerable soil crusts (Gillette, 1980), but are easily disturbed. Based on alpine lake sediments collected over the western United States, Neff et al. (2008) found that dust loading increased by 500% following western expansion during the 19th century, a likely consequence of livestock grazing and other activities that disturbed the natural crusts. This post-western-settlement dust loading greatly exceeds the dust loading of the severe decadalscale mega-droughts that affected the southwest over the past 2000 years (Neff et al. 2008, p. 192). Episodic dust events over the Wasatch Front produce hazardous air quality and contribute to dust loading in the snowpack of the nearby Wasatch Mountains (Fig. 2), which provides water for approximately 400,000 people (Salt Lake City Department of Public Utilities 1999). The Wasatch Mountain snowpack is also vital for Utah's $1 billion winter sports industry, internationally known for its “Greatest Snow on Earth” (Steenburgh and Alcott 2009). The 3 accumulation of dust in the seasonal mountain snowpack of western North America and other regions of the world can change the regional climate and hydrologic cycle of those areas (Hansen and Nazarenko 2004; Painter et al. 2007; Painter et al. 2010). Dust loading causes a decreased snow albedo, which shortens the duration of snowcover by several weeks as seen in recent studies of Colorado's San Juan Mountains (Painter et al. 2007). Dust is a multiple scatterer of sunlight (in particular the visible wavelengths), which reduces snow’s albedo from ~0.9 down to as low as ~0.33 (Skiles et al, in preparation) – [insert spectral albedo figure here – will send this for Wasatch sites]. The impact in the visible wavelengths is powerful because half of the solar radiation reaching the Earth’s surface lies in this wavelength range. Distributed hydrologic modeling studies further suggest that the earlier snowmelt results in an advancement of peak runoff and a decrease in annual runoff volume in the Upper Colorado River Basin (Painter et al. 2010). Add something on air quality????? Dust is transported in mesoscale and synoptic scale disturbances. Almost half of the dust emission in the Sahara, the world’s largest Aeolian dust source (Swap et al., 1996), is due to the winds associated with large mesoscale convective systems that propagate west into the North Atlantic (Goudie, Middleton, 2001). These systems have associated dust plumes that travel in the easterly flow for several days at a time (Carlson, 1979). In comparison, intense northeast Asian dust events are typical in the spring and early summer in the postfrontal environment of cold fronts associated with Mongolian Cyclones (Shoa, Wang, 2003; Qian et al., 2001). Here we develop a climatology of dust events for the Wasatch Front and adjacent valleys using hourly weather observations from the Salt Lake City International Airport (KSLC), hourly weather observations from nearby stations, Geostationary Operational Environmental Satellite 4 (GOES) imagery, and the North American Regional Reanalysis (NARR). Dust-events occurred throughout the study period (1930-2010 water years) and recent events are shown to be associated with four different meteorological emission mechanisms. Satellite imagery and backtrajectories show that the Sevier Desert, Carson Sink, Escalente Dester, West Desert, and Milford Flat area are the primary emission sources for many dust events, suggesting that mitigation efforts in this region 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 Dust reports are identified in the Global Integrated Surface Hourly Database (DS-3505) for KSLC, which was obtained from the National Climatic Data Center (NCDC). The analysis covers the 1930–2010 water years (Oct–Sep) when the availability of hourly observations exceeds XX% annually. Throughout this period the observing site was located at what is presently known as the Salt Lake City International Airport, just east of downtown Salt Lake City and the Wasatch Mountains (Fig. 1). DS-3505 includes data collected and stored from multiple sources and it is decoded and processed either at the operational weather centers or the Federal Climate Complex in Asheville, NC (NCDC 2001, 2008). Similarly, Brazel (1989) used MF-10A surface weather observation forms from the National Weather Service and the Local Climatological Data (LCD) publications, which are also visual weather observations from an observer, to develop a dust climatology for the southwestern United States. The identification and classification of dust using observer notes is somewhat subjective and inconsistencies arise from observer bias, as well as changes in instrumentation, reporting guidelines, and processing algorithms. These inconsistencies 5 complicate the interpretation of long-term trends and variability and likely result in the misreporting of some events (e.g., dust erroneously reported as haze). Belnap et al. (2009) took advantage of deposition traps in their analysis of sediment flux across the Colorado Plateau, which offers a much more objective approach to dust detection. Nevertheless, our data offers a sufficiently high reporting frequency to identify the general climatological characteristics and synoptic environment of dust events. Consistent with WMO guidelines (WMO 2009), the present weather record in DS-3505 includes 11 dust categories (Table 1). At KSLC, dust is most commonly reported as blowing dust (category 7, 905 reports), followed by dust in suspension (category 6, 178 reports), several categories of dust storm (categories 9, 30-32, and 98; 7 total reports), and dust or sand whirl(s) (category 8, 2 reports). There were no severe dust storm reports (categories 33-35). Analysis of the meteorological conditions in each category revealed 68 reports with a visibility exceeding 6 statute miles (10 km), which is near the thresholds used today by the WMO and national weather agencies for reporting blowing dust or dust-in-suspension (Shao et al. 2003; FMH 2005). These 68 reports include all but one of the 7 total reports in the duststorm categories. Since these events are weak, or may be erroneous, they were removed from the analysis. Reports in category 8 were also removed since we are interested in widespread events rather than localized dust whirl(s) (a.k.a. dust devils). Thus, our analysis of dust events is based on the remaining 1023 reports of blowing dust, dust in suspension (i.e., dust haze) or dust storm with a visibility of 10 km or less. A dust day is any day (MST) with at least one such dust report. b. Recent Events (2001-2010) 6 A recent climatology was compiled to take advantage of modern satellite and reanalysis data. The recent ten-year climatology of dust events for KSLC was examined and each event was given a synoptic type. Synoptic typing utilized GOES imagery, NARR data, and surface observations and comments from the DS-3505 dataset. The NARR is a 32-km, 45-layer dataset based on the eta model three-dimensional variational data assimilation (3DVAR) system covering North America (Mesinger et al. 2006). This data was obtained from the National Oceanic and Atmospheric Administration Operational Model Archive Distribution System (NOMADS) at the National (http://nomads.ncdc.noaa.gov/#narr_datasets). Climatic Data Center web site 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 (outlined in Jeglum et al. 2010). A synoptic classification scheme divided dust events into groups based off of the primary synoptic mechanism for dust emission and transport. The three main mechanisms are airmass convection, transient baroclinic troughs, and stationary baroclinic troughs. The airmass convection events usually had a thunderstorm, thunderstorm in the vicinity, or squall comment within an hour of the dust observation(s), but events with visible convective towers on satellite and a synoptic setup conducive to convection in the NARR reanalysis (i.e. back side of a warm season upper level ridge) are also placed in this group. The transient baroclinic trough group had a surface cold front passage within 24 hours of the dust observation, a distinct frontal band on visible satellite, and evidence of a baroclinic trough or intermountain cyclone on NARR 700 hPa temperature, and 850 hPa height fields (850 hPa heights used instead of MSLP by reasoning in Jeglum et al. 2010) that is mobile through the research area. Events within this group are further 7 broken into prefrontal, frontal, and postfrontal. Prefrontal events have dust at least three hours before frontal passage, frontal have dust within three hours of frontal passage, and postfrontal have dust at least three hours after frontal passage. Events can belong to more than one of these categories. The stationary baroclinic troughs are identified in the same manner as the transient baroclinic troughs, but their axis never crosses KSLC in an organized manner within 24 hours of when dust is reported. Several events had stationary axis initially as flow around the Sierra- Nevada created the boundary before up level support pushed the boundary through, but these were still considered transient events as long as the boundary made it through KSLC within 24 hours. The last category is other, which is a catch all for the synoptically forced events that do not fit the previous two categories. Other classifications schemes in the literature are Brazel and Nickling’s (1986) five synoptic types for Arizona dust storm generation: (1) pre-frontal, (2) postfrontal, (3) thunderstorm/convective, (4) tropical disturbance, (5) upper level/cut-off low. Henx and Woiceshyn (1980) created a hierarchy of weather-duststorm systems for the southern and central Great Plains and their classifications were (1) dust devils, (2) Haboob, (3) Severe Mountain Dust Storm, (4) Frontal, (5) Cyclogenic. The cyclogenic category was further broken down into (a) low level jet, (b) upper level jet, (c) surface storm circulation, and (d) Severe mountain downslope windstorm. Our categories are only specific to recent KSLC dust days. Weather conditions during dust events are compared to a local ten-year (2001-2010) monthly NARR climatology to describe the anomalous weather associated with dust events. The climatology is specific to each of the 8 synoptic time steps (00Z, 03Z, 06Z, etc.) and is computed over an area bounded by 37N, 42N, 112W, and 116W since many of the recent dust events originate and are transported through this area (See Results section). This climatology was compared to the area-averaged variables found at the onset of each dust event. 8 Surface, 700 hPa, and dynamic tropopause composites of recent events also used NARR data, which extends from 1000 hPa to 100 hPa. Unfortunately our analysis of the 2 potential vorticity units (PVU) dynamic tropopause is limited by the 100 hPa contraint so all higher levels of the dynamic tropopause are forced to 100 hPa. c. Dust Emission Sources and Transport More stations across the northeastern intermountain west were added to the analysis in an effort to locate all dust plumes over the recent ten year period. All stations with at least 5 years of hourly data from 2000 to 2010 were examined within the bounds of 38N, 43N, 116W, and 109W. The dust days from these stations were added to our sample using the same criteria as the KSLC dust days. Personal notes from email correspondence, blog posts, and Utah Avalanche Center annual reports also supplemented our dust days climatology for the Wasatch Front. A dust retrieval algorithm is applied to GOES data approximately every 15 minutes to the daylight hours (14Z to 2Z) of each dust day going back to 2001 in an effort to track and discover the origin of dust. GOES imager data is downloaded from the Comprehensive Large Array-data Stewardship System (CLASS) electronic library through NOAA (website: http://www.class.ncdc.noaa.gov). Researchers have used other methods to track dust such as visibility isoplething (Morales, 1979), taking photographs from satellite (Nakata et al., 1976), aircraft measurements (Reid et al., 2000, Tanre et al., 2003) and, more recently, using satellite data (Christopher et al., 2010). Christopher et al (2010) lists the Moderate Resolution Imaging SpectroRadiometer (MODIS), Multi-angle Imaging SpectroRadiometer (MISR), Clouds and the 9 Earth's Radiant Energy System (CERES) scanner, Ozone Monitoring Instrument (OMI), Polarization and Directionality of the Earth's Radiances (POLDER), Cloud Aerosol Lidar with Orthogonal Polarization (CALIOP), Spinning Enhanced Visible and Infrared Imager (SEVIRI), and Atmospheric Infrared Sounder (AIRS) as some of the satellite instruments able to observe dust aerosols on regional to global scales. Currently, National Weather Service forecasters use a GOES dust detection algorithm for their blowing dust product, which was created by Cooperative Institute for Research in the Atmosphere (CIRA). GOES data has the advantage of a large spatial and temporal scale. The basic characteristics of dust that make dust detection possible are that dust becomes increasingly absorptive with decreasing visible wavelengths, dust has a higher spectral absorption at 11 microns than 12 microns, and at 3.9 microns than 11 microns. Therefore the following algorithm is applied: if a pixel has a low reflectance (i.e. cloud free), has a positive 12 and 11 micrometer brightness temperature difference, and has the 11 micrometer brightness temperature within 10 degrees Celsius of the 3.9 micrometer brightness temperature then there is a high probability of dust in the pixel (modification of a MODIS algorithm in Zhoa et al., 2010). Unfortunately the presence of clouds, a low sun angle, weak dust concentrations, and low dust near the desert surface prevent dust retrieval and/or can lead to false positives. GOES images for all the recent events were examined and if any dust plumes were visible over the intermountain west then the image with the best depiction of the plume was further examined to locate the starting location and orientation of the plume. The starting positions of the visible plumes are all located based off of the location of the first dusty pixel of the plume and the orientation is the line from the starting positions to the center of the end of the plume. Plumes are only recorded if the starting location is visible and the plume is visible for 10 more than one frame to protect against false positive contamination. Since the GOES data only has 4km resolution, and the dust retrieval algorithm is contaminated with false positives, these locations and orientations are very approximate. In order to avoid a bias towards cloud free events and daylight events a back trajectory analysis offered an easy way to approximate dust sources. We used the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler and Rolph, 2011; Rolph, 2011) to compute 6 hour backjectories ending at 1000 meters above KLSC. Gebhart et al. (2001) used the ARL-ATAD model, which is an older lagrangian parcel model, to compute back trajectories of fine particulates at Big Bend National Park. Our trajectories were computed using the 40km Eta Data Assimilation System (EDAS), which only has data going back to 2004. The only dataset available with a complete record for our recent 10 year recent climatology is the 2.5 degree NCEP/NCAR Reanalysis, so we decided to sacrifice a full temporal extend for a higher spatial resolution. 1000 meters is used as the ending height above the surface because it is below the typical boundary layer height for this region, but high enough to avoid substantial interference with the ground. The six hour time frame mimics the time for dust to reach KSLC after it is emitted into the atmosphere, which is based off satellite animation of events with visible plumes. Backtajectory analysis cannot pinpoint source regions, but it does provide a comprehensive approach to determine the path dust takes. d. Case Studies NARR data and satellite imagery were also used for the three selected case studies, along with archived Nexrad level 2 KMTX http://www.ncdc.noaa.gov/nexradinv/). 11 data from NCDC (website: 3. Results a. Long-term climatology Dust events at KSLC occur throughout the historical record, with an average of 4.3 per year (Fig. 3). Considerable interannual variability exists, with no events reported in seven years (1941, 1957, 1981, 1999, 2000, 2001, 2007) and a maximum of 15 reported in 1934. No effort was made to quantify or assess long-term trends or interdecadal variability given 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 during 95.40%, 2.59%, and 2.01% dust days meets the criteria for blowing dust (>5/8 statute miles), a dust storm (5/16-5/8 statute miles), or a severe dust storm (<5/16 statute miles), respectively (Fig. 4). By observation, 98.04%, 1.20%, and 0.76% fall into the aforementioned categories. Thus, only a small fraction of the dust events and observations meet dust storm or severe dust storm criteria. Given the variability in severity, frequency, and duration, we calculated the total annual near-surface dust-mass transport using eq. 9.81 of Shao (2008, p. 334): ****EQUATION HERE**** where C is the dust concentration in g/m2, delta-t is the reporting interval, V is the mean horizontal wind speed in m/s of the reporting interval, VIS is the mean visibility in meters of the reporting interval, and N is the number of dust observations each year. For all dust flux calculations dust is assumed to be present until an observation does not report dust. For example if a 21Z observation had dust reported with a 5 km visibility, 5 m/s wind and a sequential 22Z 12 observation had no dust reported with a 10 km visibility and 10 m/s wind then V=7.5m/s, VIS=7.5 miles, and N=1. Using this approach the average annual near-surface dust-mass transport is 399.4 g/m2, with a maximum of 2810.2 g/m2 in 1935 (Figure 5). The annual dust flux (Figure 5) provides a somewhat different perspective from the annual dust day frequency (Figure 3) through its integration of event frequency, duration, and severity. For example, 1934 featured the most dust events, but the greatest near-surface dust-mass transport occurs in 1935. In 2010, there were only 2 events, but this year features a pronounced decadal-scale maximum in near-surface dust-mass transport. Although observing system limitations preclude a confident assessment of long-term trends and interdecadal variability, these results indicate dust emissions, transport, and deposition occurred over northern Utah throughout the historical record, consistent with sediment cores collected from alpine lakes in Colorado and Utah, which suggest an increase in dust accumulation beginning in the mid-1800s (e.g., Neffs et al. 2008; Reynolds et al. 2010). The monthly frequencies of dust days (figure 6) and estimated cumulative dust flux (Figure 7) show a bimodal distribution, with primary and secondary peaks in April and September, respectively. The two distributions are distinctly different during the summer and winter months. The annual dust flux is lower in the summer compared to the annual frequency suggesting summer dust events are characterized by shorter or weaker events than the rest of the year. The local maximum in dust flux during January was 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 January monthly mean. For the spring season of March, April, and May an average of 237 g/m2 of dust is transported through KSLC, which roughly corresponds to peak snowpack in the nearby Wasatch Range. Interestingly, Yasunori and Masao (2002) found a similar bimodal monthly distribution for dust events originating in the Taklimakan desert of 13 China, which is on a similar latitude circle as Utah. The bimodal distribution at KSLC is also very similar to the monthly frequency of strong Intermountain cold fronts Schafer et al (2008) found, and to the intermountain cyclone frequency Jeglum et al. produced (Figure 8). Strong intermountain cold fronts are associated with intermountain cyclones and produce persistently strong winds that can lead to dust emission so the correlation between dust, cold front, and cyclone frequency is expected. Also, 12 out of the top 25 strongest cold fronts identified by Schafer et al (2008) to impact KSLC had dust reported within 24 hours of its passage. Strong winds are needed to entrain and transport dust and the mean wind speed during dust events at KSLC is 11.6 m/s with a sigma of 4.0 m/s. A threshold velocity for dust emission was set to 10 m/s based on the mean and Holcombe et al. (1996) distribution of mean hourly wind speeds for 1190 dust episodes over the Sonoran Desert being between roughly 5 and 15 m/s. Threshold velocities are contingent upon many factors such as particle size, vegetation, biological crusts, and soil moisture (Neff et al., 2008, Belnap et al., 2009, Gillette, 1999), but the Sonoran Desert has a similar basin and range characteristic as Utah and Nevada, and similar alluvial and evaporite deposits that emit most of the dust. The 10 m/s threshold is obtained or surpassed most frequently in the spring months of March and April (Figure 9) and the peak in March is 50% higher than the annual mean occurrences and are fairly constant for the remainder of the year with only a weak minimum in July, which is 23% below the mean. Winds alone cannot explain the bimodality of the monthly dust distribution. Land surface processes appear to play a major role since threshold velocities are rather high in the winter months, but dust event frequency is at its minimum. Dust reports also exhibit a strong diurnal cycle and are most common in the late afternoon and evening hours (Figure 10). Mbouro et al. (1997) found a similar diurnal cycle for 14 North African dust days with visibilities less than 5 km, except the diurnal cycle was not as pronounced. Strong surface winds are needed to entrain dust and the maximum number of occurrences with winds greater than our threshold velocity of 10 m/s occurs in the afternoon (Figure 11), which is consistent with the timing of a well mixed boundary layer. Winds are three times more likely to reach our threshold velocity in the afternoon than the early morning. The peak in threshold wind occurrence (14 MST) precedes the peak in dust event start times (18 MST) by four hours and this discrepancy is likely due to the travel time needed for the dust to travel from its source to KSLC after it becomes entrained. The wind direction at the onset of each event has a bimodal distribution (Figure 12). About 50 percent of the wind directions at the onset of when dust is reported were from the south-southwest, south-southeast, or south and another 28 percent were from the northwest, north-northwest, and north. These two modes probably correspond to prefrontal and postfrontal conditions, respectively. b. Recent (2001-2010) events 22 (67%) of recent dust events at KSLC are synoptically forced and 11 (33%) are forced by mesoscale airmass convection. 16 (73%, 48% of total) of the synoptically forced events are forced by a transient cold-frontal trough, 4 (18%, 12% of total) are stationary troughs upstream of KSLC, and 2 (9%, 6% of total) fall into the other synoptic category. Out of the 16 transient cold-frontal trough events, 4 (25%) reported dust in the prefrontal environment, 16 (100%) reported dust as the surface front passed, and 2 (13%) reported dust in the postfrontal environment. A list of all the recent events and their respective typing is presented in Table 2. The September 16, 2003 event is placed in the other category because a trough and surface front 15 formed downstream of KSLC, placing KSLC in the postfrontal environment. A distinct boundary never moved through KSLC. The March 13, 2005 event is in the other category because an arctic front coming from the north forced the dust event instead of the more typical Intermountain Cold front that comes from the northwest. The August 30, 2009 event is questionable since all the other observations with low visibility after the one dust observation had smoke comments, and smoke plumes are visible on satellite images, but this event is included in all analyses since the dust observation cannot be disproven. Three of the airmass convection events did not have a thunderstorm reported within an hour of the dust observation, but satellite imagery showed convective towers in the vicinity of KSLC during the time dust was reported and NARR analysis showed a distinct monsoonal setup. The recent event climatology resembles the long term climatology except for a disproportionate number of summer events. 2001 and 2007 water years do not have any reported dust days and 2009 has a maximum of 7 days, but there is no observable recent trend in the annual number of dust days (Figure 13). The monthly frequency (Figure 14) shows the primary springtime peak and secondary fall peak, much like the long term climatology, but July and August also have a local maximum. Since airmass convection dust events typically occur in the summer months, the recent percentage of these events may be unrepresentative of the long term mean. NARR analysis of these events suggests upper level troughing upstream of KSLC coupled with a low level pressure minimum is the most common setup for dust emission. The dynamic tropopause composite (Figure 15) shows a mean trough to the west and an embedded southwesterly jet over Utah. The 700 hPa temperature and wind analysis (Figure 16) further shows a tight temperature gradient over Nevada being advected towards Utah and strong 16 southwest flow out ahead of the baroclinic zone. 700 hPa wind speeds associated with dust events are considerably higher than climatology (Figure 17a) and skewed to more southwesterly compared to climatology (Figure 18a). These features are coupled with a surface trough over northern Utah and Eastern Nevada, which is evident on the 850 hPa height analysis (Figure 19). The height of the planetary boundary layer is also much higher than climatology for dust events (Figure 20a) so the strong winds above can be better realized at the surface. Interestingly, soil moisture during dust events does not vary significantly from climatology (Figure 21). Past studies suggest soil moisture increases the friction velocity of a soil making it harder for the soil to become entrained in the atmosphere (Saleh and Fryrear 1995, Bisal and Hsieh 1966, Chepil and Woodruff, 1963). McKenna-Neuman and Nickling (1989) showed how the capillary effect of soil moisture held sand grains together thus decreasing erodibility. However, Gillette (1999) observed wind erosion 10-30 minutes after a soaking rain because the eroding layer needs only to be a millimeter thick and strong winds can dry a layer that thin in a very short time frame. The NARR soil moisture content is calculated from the soil surface down to 200 cm. Transient cold-frontal trough events and airmass convection events were compared against climatology and 700 hPa wind speeds were drastically different comparatively for these two groups. The maximum for the transient troughs is centered around 15m/s (Figure 17b), which occurs only 4% of the time climatologically. The airmass convection events roughly following climatology and are only skewed slightly towards higher values (Figure 17c). Wind directions are only in the southwest quadrant of the compass for transient trough events (Figure 18b), whereas airmass convection events have some more northerly directions (Figure 18c). Planetary boundary layers are larger than climatology for both groups (Figures 20b and 20c), but soil moisture does not appear to be any different from climatology (Figures 21b and 21c). 17 b. Dust emission sources and transport Dust is reported in the hourly observations at four different stations across the IMW since 2001 (33 at KSLC, 30 at Delta, 18 at Pocatello, 6 at Elko for 87 total, but 79 individual events due to overlap). The dust days from these stations were further supplemented with personal observations of dust along the Wasatch Front bringing the total number of dust days to 94 (15 from personal notes). The characteristics of these events are consistent with the long-term climatology of KSLC in terms of annual, monthly, and diurnal distribution. A GOES satellite dust retrieval algorithm is applied to all 94 dust days in an effort to locate all visible dust plumes originating in the IMW. Out of the 94 dust days 47 (50%) had visible plumes. 120 plumes were identified so each dust day with visible dust had an average of 2.55 plumes. The remaining 50% of dust days did not have any observable plumes due to at least one of the following reasons: (1) clouds blocked the dust from the satellite, (2) the dust occurred at night or during a low sun angle, or (3) the dust concentration was too weak for the detection algorithm to pick it up. Airmass convection events and frontal passages with only a couple of dust observations were the most common types of dust events without any visible plumes. GOES data indicates the ancient depositional environments (e.g., ancient lake beds) in southwest Utah and Western Nevada are the primary dust plume emission sources for the IMW. Figure 22 shows the approximate plume origins are mostly clustered in certain lowland regions, most notably the Sevier Desert, Milford Flat area, Escalente Desert, and West Desert in southwest Utah (Figure 22a) and the Carson Sink in Nevada (Figure 22b). Gillette (1999) calls small areas of frequent dust production “hot spots”, which are depositional environments in transitional arid regions that have had their biological and physical crust disturbed. 18 The aforementioned areas receive heavy recreation and agricultural use so the crusts in these areas are likely disturbed. The plumes are mostly directed to the NNE and NE indicting they are transported in SSW and SW flow. The extent of the plume lines only represents the length of the plume at one particular time step and the lines are not related to plume strength or to the distance the plume traveled. Only (8.3%) of the plumes had a southerly component and they all originated in Nevada. These plumes all formed in a postfrontal environment after a frontal band on satellite moved over the region. It is important to note that the postfrontal environment of an intermountain cold front is usually cloudy, which effectively blocks satellite detection of dust plumes. 40 (33%) of the plumes coincide to Delta, UT dust days, 29 (24%) to KSLC, 10 (8%) to personal notes, 9 (8%) to Pocatello, ID, 6 (5%) to Elko, and 26 (22%) coincide to dust days reported at multiple stations. Interestingly, many of the dust days do not have plumes directed towards their respective stations. There are several possible explanations for this, including plumes directed towards the station are blocked by clouds, new plumes form near the station after sunset, or the dust over the station is too diffuse to be detected by satellite. Not all of the dust we observed on satellite started as a point source. There are 11 cases when large areas of dust showed up on the satellite with no clear origin and then were transported with the flow. The majority of these cases occurred over western Nevada and moved southeast during the day, but a couple of these cases also occurred over central Utah, and one of the Snake River Plain of Idaho. Back trajectory analysis for all events indicates the plume starting locations from the GOES imagery are a good proxy for the primary emission sources for all KSLC dust events. Archived six hour back trajectory for the 25 KSLC dust days since 2004 reveal the majority of dust comes from the south-southwest and southwest (Figure 23). Only two out of the 25 back trajectories have a starting point north of KSLC; an airmass convection event on July 26, 2006 19 and the “other mechanism” event on March 13, 2005, which is an arctic frontal passage. The airmass convection event did not have a visible dust plume, but the arctic front did have a diffuse area of dust show up over the Snake River Plain in Idaho and move south towards KSLC. Since this was a diffuse area and not a plume it was not recorded on the plume plot (Figure 22). The rest of the events all point towards source regions identified by GOES imagery. c. Case Studies a. 5/10/2004 May 5, 2004 has a common setup for a transient trough induced dust event. The day starts with a baroclinic zone over western Nevada out ahead of a landfalling Pacific trough. By 15Z a weak surface trough takes shape and 700 hPa winds reach 25 m/s downstream in the prefrontal environment (Figure 24a). The frontal band is seen clearly over northwestern Nevada on visible satellite, but dust is not detected (Figure 25a), and winds are only 4.9 m/s from the south at KSLC. By 18Z the trough has become better organized and has shifted to the east (Figure 24b) and satellite imagery shows a similar shift to the frontal band and the initiation of dust near the frontal boundary in Nevada and over western Utah (Figure 25b). At 21Z the baroclinic zone nears northern Utah and 850 hPa heights continue to fall indicting a deepening intermountain cyclone (Figure 24c). The frontal band is now more organized and further east with pre and post frontal dust visible over Nevada, and a very distinct prefrontal dust plume originating from the Escalente Desert and Milford area and extending to near the Utah/Idaho border (Figure 25c). Winds at KSLC during this time are at 16.5 m/s from the south-southeast, but dust is not yet reported. The surface trough nears KSLC at 00Z on May 11, 2004 with apparent deformation frontogenesis (Figure 24d) and the prefrontal dust plume reached KSLC 20 and extended well into Idaho (Figure 25d). Visibilites at KSLC dropped to 8 km at 23Z with a haze comment and dust was reported at 0Z with winds at 15.6 m/s from the south. By 1Z the wind shifted to the north-northwest, the temperature dropped nearly 13 degrees celcius signaling frontal passage, but visibility remained at 8 km with dust observed. c. 10/17/2004 The event on 0ctober 17, 2004 is a stationary trough example that produced one observation of dust at 05Z and dropped visibilities to 10 km. At 18Z on the 16th, a zonal baroclinic zone set up over the Pacific Northwest (Figure 26a) out ahead of an approaching trough. Six hours later, at 0Z on the 17th, a weak surface trough formed over central Nevada and 700 hPa southwesterly winds intensified downstream (Figure 26b). Unlike the May 10th event, the baroclinic zone remained weak over Nevada and the upper level potential vorticity support is much less (not shown). One hour after the dust observation, at 6Z, a surface trough is well established and strong southwesterly winds are evident over Utah. Isotherms are oriented meridionally over western Nevada, but the gradient is weak (Figure 26c). During this time KSLC has winds at 12.1 m/s from 160 degrees and visibility at 13 km. Unfortunately the sun had already set so satellite imagery could not resolve any dust plumes. Six hours later, at 12Z, the surface trough and baroclinic zone over Nevada remained fairly stationary and weakened, but a new surface trough formed over western Idaho (Figure 26d). This feature eventually strengthens downstream of KSLC so KSLC never experiences a frontal passage. All the while a longwave upper level trough with its PV support sits off the west of coast of the United States preventing transient cold-frontal trough event from happening. 21 b. 5/19/2006 The 11 airmass convection events over the recent period of record are characterized by very similar parameters. All of these events occurred between the middle of May and the middle of September in a monsoonal surge scenario. May 19th, 2006 is the earliest in the year of these events, but still demonstrates the typical setup. Figure 27 shows the surface convective available potential energy (CAPE), 700 hPa winds, and the atmospheric column precipitable water (PW). Both the CAPE and PW have a local maximum over KSLC. Although CAPE values around 400 J/kg, and PW values around 20 mm are not that impressive compared to other regions in the United States, it is substantial enough for convection in Utah. As mentioned earlier, one of the distinguishing factors between airmass convection dust events and synoptic dust events is the 700 hPa wind speed, which is much lower for airmass convection. These low wind speeds mean even with the 700 hPa winds mixing down to the surface they would still be insufficient in entraining dust. Strong winds must originate in smaller mesoscale processes within convection. At 23Z on May 19th, 2006 winds at KSLC are only 4.5 m/s from the south and this is the strongest they have been all day. Then, at 23:07 winds go to 24.1 m/s and gust to 27.7 m/s and dust is reported along with a squall at or within sight of the station comment. Radar imagery just two minutes before reveals a convective cell very close to the station with returns greater than 60 dBZs (Figure 28). By 01Z on the 20th winds are back down to 4 m/s since the storm is fully passed. 4. Conclusions Dust events at KSLC occur throughout the historical record, with considerable interannual variability. The vast majority of these events have visibilities above dust storm or severe dust 22 storm criteria and blowing dust is the most common observer comment. Dust events have a bimodal monthly distribution with a primary peak in the spring and a secondary peak in the fall. Climatological winds do not have the same distribution, but do have a local maximum during the spring months. Winds alone cannot explain the dust day monthly frequency. Annual and monthly dust flux calculations offer a different perspective than the frequency distributions, but results are very similar with the exception of a more dramatic local minimum during the summer in the monthly dust flux distribution. This difference is from shorter and less intense airmass convection events primarily occurring in the summer. A third of all recent events at KSLC are forced by mesoscale airmass convection and the rest are synoptically forced. Almost three quarters of synoptically forced events are transient cold-frontal troughs with associated intermountain cyclones and cold fronts. Dust is reported within 3 hours of frontal passage in each of these events and in the prefrontal environment 25% of the time. Only two of these events had dust three hours after frontal passage in the postfrontal environment. The remaining synoptically forced events are stationary surface troughs upstream of KSLC and events that do not fit these guidelines. NARR analysis suggests upper level troughing upstream of KSLC coupled with strong southwest 700 hPa winds and a low level pressure minimum are commonly associated with KSLC dust events. These composites are skewed towards the synoptic events since airmass convection 700 hPa wind speed and direction are closer to climatology. Planetary boundary layer heights are higher than climatology for all dust events. Surprisingly, soil moisture in the NARR analysis does not appear to effect dust emission. GOES data and HYSPLIT back trajectories indicate the ancient depositional environments (e.g., ancient lake beds) in southwest Utah and Western Nevada are the primary 23 dust emission sources for the Intermountain West. Specifically Sevier Desert, Carson Sink, Escalente Desert, and the Milford Flat area are common emitters. These areas experience high agricultural and recreational use, which are dust disturbing practices that lead to increased dust emission. Mitigating crust disturbing practices in these areas will help decrease dust flux over the intermountain west, which will improve air quality and decrease dust loading in the mountain snowpack 5. Acknowledgments 6. References Belnap, J., R. L. Reynolds, M. C. Reheis, S. L. Phillips, F. E. Urban, H. L. Goldstein, 2009: Sediment losses and gains across a gradient of livestock grazing and plant invasion in a cool, semi-arid grassland, Colorado Plateau, USA, Aeolian Research, 1, 27-43 Bisal, F. and J. Hsieh (1966) Influence of soil moisture on erodibility of soil by wind. Soil Sci., 102, 143–14 Carlson, T.N., 1979. Atmospheric turbidity in Saharan dust outbreaks as determined by analyses of satellite brightness data. Monthly Weather Review 107, 322–33 Chepil and Woodruff, N.P. 1963. The physics of Wind Erosion and its Controls. Advances in Agronomy; Vol. 15, pg. 211-302 24 Christopher, A. S., T. A. Jones. 2010. Satellite and surface-based remote sensing of Saharan dust aerosols. Remote Sensing of Environment 114, 1002-1007 Draxler, R.R. and Rolph, G.D., 2011. HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php). NOAA Air Resources Laboratory, Silver Spring, MD. FMH, 2005: http://www.ofcm.gov/fmh-1/fmh1.htm Gebhart, K.A., Kreidenweis, S.M. and Malm, W.C. 200. Back Trajectory Analyses of fine particulate matter measured at Big Bend National Park in the historical database and the 1996 scoping study. The Science of the Total Environment; Vol. 276, no. 1-3, pp. 185204 Gillette, D. A., 1999: A Qualitative geophysical explanation for hot spot dust emitting source regions. Contr. Atmos. Phys., 72,1,67-77 Gillette, D.A., Adams, J., Endo, A., and Smith, D. 1980. Threshold Velocities for Input of Soil Particulates into the Air by Desert Soils. Journal of Geophysical Research; Vol 87, no. C11, pp. 9003-9016 Goudie, A. S., and N. J. Middleton, 2001: Saharan dust storms: nature and consequences. EarthScience Reviews, 56, 197-204 Hansen, J., and L. Nazarenko, 2004: Soot climate forcing via snow and ice albedos. Proc. Natl. Acad. Sci. U. S. A., 101, 423-428. Helgren, D.M., and J.M. Prospero. 1987. Wind velocities associated with dust deflation events in the western Sahara. J. Clim. Appl. Meteorol. 26:1147–115 25 Holcombe, T. L., Ley, T. and Gillette, D.A. 1996. Effects of Prior Precipitation and Source Area Characteristics on Threshold Wind Velocities for Blowing Dust Episodes, Sonoran Desert 1948 – 78. Journal of Applied Meteorology; vol. 36, no. 9, pp. 1160-1175. Mbouro, G., Bertrand J., and S. Nicholson, 1997: The Diurmal and Seasonal Cycles if WindBorne Dust Over Africa North of the Equator. J. Appl. Meteor., 36, 868–882 http://journals.ametsoc.org/doi/full/10.1175/15200450(1997)036%3C0868:TDASCO%3E2.0.CO;2 Mc Kenna-Neuman C. and Nickling W.G., 1989: A theoretical and wind tunnel investigation of the effect of capillary water on the entrainment of sediment by wind. Canadian Journal of Soil Science. Vol 69, 79-96 Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343–360 Morales, C., 1979. The use of Meteorological Observations for Studies of the Mobilization, Transport and Deposition of Saharan Soil Dust. In Saharan Dust. John Wiley and Sons Ltd. Pp. 119-131. Nakata, J.K., Wilshire, H.G. and Barnes, G.G. 1979. Origin of Mojave Desert dust plumes photographed from space. Geology; Vol. 4, pp. 644-648. NCDC, 2001: http://www1.ncdc.noaa.gov/pub/data/inventories/ish-tech-report.pdf NCDC, 2008: Data documentation for data set 3505 (DSI-3505) Integrated Surface Data. [Available from http://www1.ncdc.noaa.gov/pub/data/documentlibrary/tddoc/td3505.pdf]. Neff, J. C., and Coauthors, 2008: Increasing eolian dust deposition in the western United States linked to human activity. Nature Geosci., 1, 189-195. 26 Painter, T. H., J. S. Deems, J. Belnap, A. F. Hamlet, C. C. Landry, and B. Udall, 2010: Response of Colorado River runoff to dust radiative forcing in snow. Proc. Natl. Acad. Sci. U. S. A., NEED FINAL VOLUME AND PAGES. Painter, T. H., and Coauthors, 2007: Impact of disturbed desert soils on duration of mountain snow cover. Geophys. Res. Lett., 34, L12502, doi:10.1029/2007GL030284. Qian, Weihong, Lingshen Quan, Shaoyin Shi, 2002: Variations of the dust storm in china and its climatic control. J. Climate, 15, 1216–1229 Reid, J. S., Kinney, J. E., Westphal, D. L, Holben, B. N., Welton, E. J., Tsay, S. -C., et al. (2003). Analysis of measurements of Saharan dust by airborne and ground-based remote sensing methods during the Puerto Rico Dust Experiment (PRIDE). Journal of Geophysical Research, 108(D19), 8586. doi:10.1029/2002JD002493. Rolph, G.D., 2011. Real-time Environmental Applications and Display sYstem (READY). Website (http://ready.arl.noaa.gov). NOAA Air Resources Laboratory, Silver Spring, MD. Saleh, A. and Fryrear, D.W. (1995) Threshold wind velocities of wet soils as affected by wind blown sand. Soil Sci., 160, 304–309 Salt Lake City Department of Public Utilities, 1999: Salt Lake City Watershed Management Plan. Bear West Consulting Team, 129 pp. [Available from www.slcgov.com/utilities/PDF%20Files/slcwatershedmgtplan.pdf]. Shao, Y., and J. Wang, 2003: A climatology of northeast Asian dust events. Meteor. Zeitschrift, 12, 187-196. Steenburgh, W. J., and T. I. Alcott, 2009: Secrets of the "Greatest Snow on Earth". Bull. Amer. Meteor. Soc., 89, 1285-1293. 27 Swap, R., Ulanski, S., Cobbett, M., Garstang, M., 1996. Temporal and spatial characteristics of Saharan dust outbreaks. Journal of Geophysical Research 101 D2 , 4205–4220 Tanaka, T. Y., and M. Chiba, 2006: A numerical study of the contributions of dust source regions to the global dust budget. Glob. Planet. Change, 52, 88-104. Tanré, D., Haywood, J., Pelon, J., Léon, J. F., Chatenet, B., Formenti, P., et al. (2003). Measurement and modeling of the Saharan dust radiative impact: Overview of the Saharan Dust Experiment (SHADE). Journal of Geophysical Research, 108(D18), 8574. doi:10.1029/2002JD003273 WMO, 2009: Manual on Codes. See WMO (2009b) in my directory. We'll need to do full reference eventually. Yasunori, K., and M. Masao, 2002: Seaonsla and Regional Characteristics of Dust Event in the Taklimakan Desert. Journal of Arid Land Studies, 11-4, 245-252 Zhoa, T., S. Ackerman, W. Guo, 2010: Dust and smoke detection for multi-channel imagers. Remote Sensing, 2, 2347-2368 Figure Captions Fig. 1. Topography and geography of the study region. Fig. 2. Examples of dust layering in the late-season Wasatch Mountain snowpack. (a) Ben Lomond Peak (XXXX m), April 2005. (b) Alta 2009, (b) Alta 2010. 28 Fig. 3. Blowing dust climatology at Delta, UT. (a) Blowing dust reports by year. (b) Blowing dust reports by month. (c) Blowing dust reports by time of day (may need to normalize by report frequency given Delta’s frequent lack of reporting). (d) Wind roses of all blowing dust events. (e) Wind roses of blowing dust events with visby less than 1 mile. Fig. 4. Same as Fig. 2 except for KSLC. Fig. 5. Dust layer climatology derived from UAC reports. (a) Number of events by year. (b) Number of events by month. Fig. 6. Satelite (dust product), MesoWest, and manual surface analysis during selected events. (a) 2002 Tax Day storm, (b-?) other events. Maybe pick four. Fig. 7. Camera images from selected events, especially the nasty 5 April (date?) 2010 event with the haboob. Fig. 8. Meteograms for (a) Delta and (b) KSLC during the 10-12 April 2010 event. Fig. 9. PM2.5 or PM10 for sites in the Salt Lake Valley during the 10-12 April 2010 event. Table 1: DS-3505 dust-related present-weather categories, including full and abbreviated (i.e., used in the text) descriptions and the number of total and used reports at Salt Lake City. Category 06 Full Description Widespread dust in suspension in the air, not raised by wind at or near the station at the time of observation 29 Abbreviated Description Dust in suspension Reports 178 (155) 07 08 09 30 31 32 33 34 35 98 Dust or sand raised by wind at or near the station at the time of observation, but no well-developed dust whirl(s) or sand whirl(s), and no duststorm or sandstorm seen Well developed dust whirl(s) or sand whirl(s) seen at or near the station during the preceding hour or at the time of observation, but no duststorm or sandstorm Duststorm or sandstorm within sight at the time of observation, or at the station during the preceding hour Slight or moderate duststorm or sandstorm has decreased during the preceding hour Slight or moderate duststorm or sandstorm no appreciable change during the preceding hour Slight or moderate duststorm or sandstorm has begun or has increased during the preceding hour Severe duststorm or sandstorm has decreased during the preceding hour Severe duststorm or sandstorm no appreciable change during the preceding hour Severe duststorm or sandstorm has begun or has increased during the preceding hour Thunderstorm combined with duststorm or sandstorm at time of observation, thunderstorm at time of observation Blowing dust 905 (867) Dust whirl(s) 1 (0) Duststorm 2(1) Duststorm 1 (0) Duststorm 1 (0) Duststorm 1 (0) Duststorm 0 (0) Duststorm 0 (0) Duststorm 0 (0) Duststorm 2 (0) Table 2 Date Airmass Convection Persistent SW Flow Transient trough Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal 3/23/2002 4/15/2002 6/1/2002 X 9/16/2002 X 2/1/2003 4/1/2003 X 30 Other X X X X Notes 4/2/2003 Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal 9/16/2003 4/28/2004 5/10/2004 7/9/2004 X 10/17/2004 X 3/13/2005 4/13/2005 5/16/2005 7/22/2005 X 7/30/2005 X 5/19/2006 X 7/19/2006 X 7/26/2006 X 4/29/2008 5/20/2008 7/27/2008 X 31 X Trough forms to the south X X X Arctic front X X X X X X Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal Prefrontal Frontal Postfrontal 8/31/2008 3/4/2009 3/21/2009 X 6/30/2009 X 8/5/2009 X 8/6/2009 8/30/2009 X 9/30/2009 3/30/2010 4/27/2010 32 Stationary trough at first X Stationary trough at first X X X Many Smoke Obs X X X X X X