PUBLICATIONS Journal of Geophysical Research: Atmospheres RESEARCH ARTICLE 10.1002/2014JD021694 Key Points: • A 10 year Asian dust record is presented on the U.S. West Coast • Seasonal variability of transported Asian dust can be explained by wet removal
Correspondence to: J. M. Creamean,
[email protected]
Citation: Creamean, J. M., J. R. Spackman, S. M. Davis, and A. B. White (2014), Climatology of long-range transported Asian dust along the West Coast of the United States, J. Geophys. Res. Atmos., 119, 12,171–12,185, doi:10.1002/ 2014JD021694. Received 27 FEB 2014 Accepted 4 OCT 2014 Accepted article online 9 OCT 2014 Published online 4 NOV 2014
Climatology of long-range transported Asian dust along the West Coast of the United States Jessie M. Creamean1,2, J. Ryan Spackman1,3, Sean M. Davis1,2, and Allen B. White1 1
Earth System Research Laboratory, National Oceanic and Atmospheric Administration, Boulder, Colorado, USA, Cooperative Institute for Research in Environmental Sciences, University of Colorado at Boulder, Boulder, Colorado, USA, 3 Science and Technology Corporation, Boulder, Colorado, USA 2
Abstract The contribution of trans-Pacific dust estimated from satellite observations has been shown to be 3 times greater than domestic dust in North America throughout the year. Thus, a quantitative understanding of the frequency and locations where Asian dust is transported is necessary to improve global dust modeling for weather and climate predictions. This work presents a 10 year record (2002–2011) of dust along the U.S. West Coast estimated from the Interagency Monitoring of Protected Visual Environments network in an effort to characterize the seasonal cycle and interannual variability of Asian dust transport. In addition, observations of dust exported from East Asia were analyzed along with air mass trajectories and satellite and ground-based precipitation data to investigate seasonal variability of Asian dust transport. On average, Asian dust concentrations (0.08–0.60 μg m 3) from ground-based observations were 1.7 times those of local dust (0.00–0.53 μg m 3) and 23% (up to 44%) of fine particulate matter (particles with diameters ≤ 2.5 micrometers, or PM2.5) mass concentrations at high elevations in the spring. The maximum in springtime Asian dust on the U.S. West Coast was attributed to higher source concentrations (10.98–36.27 μg m 3) and reduced potential for wet removal over the Pacific Ocean and U.S. West Coast. Although trans-Pacific transport was more favorable during the winter, minimum concentrations of Asian dust were observed on the U.S. West Coast (0.11 μg m 3) due to a lower source influence and higher potential for wet removal during transport. Multiobservational approaches such as these should be taken into account when modeling transport of Asian dust to the western U.S. 1. Introduction Mineral dust aerosols affect weather and climate by influencing Earth’s radiative balance and altering cloud properties, among other environmental and ecological effects [Goudie, 2009]. The global and annual mean radiative forcing of dust can range from 0.56 to +0.27 W m 2 [Forster et al., 2007; Yue et al., 2010]. Dust can affect Earth’s temperature profiles, influencing atmospheric dynamics and near-surface weather conditions [Perez et al., 2006]. Further, dust can modify cloud optical properties and precipitation efficiency [Ou et al., 2009] by serving as ice nuclei (IN) [Creamean et al., 2013; DeMott et al., 2003] or as giant cloud condensation nuclei (CCN) [Kumar et al., 2009; Sullivan et al., 2010]. Dust as IN has been shown to enhance precipitation in mixed phase clouds [Ault et al., 2011; Creamean et al., 2013], whereas dust as giant CCN has been shown to both enhance warm rain formation and suppress precipitation [Rosenfeld et al., 2001]. Dust indirect effects, although potentially more significant than the direct effects, harbor large uncertainties. On top of having large uncertainties associated with radiative and climate forcings, dust is one of the most abundant aerosols in Earth’s atmosphere, comprising up to 75% of the global aerosol mass load [Mona et al., 2012]. Most of the global dust burden is caused by natural sources in North Africa, the Arabian Peninsula, and East Asia [Ginoux et al., 2012; Prospero et al., 2002]. Asian sources, including the Taklimakan, Gobi, and the Chinese loess plateau represent ~25% of global dust emissions [Ginoux et al., 2004]. North America receives twice as much dust from other continents than it emits per year; a large contribution is from Asian dust transported in the midlatitudes (30–60°N) to the western U.S. from February to June [Ginoux et al., 2004; Merrill et al., 1989; VanCuren and Cahill, 2002]. Uematsu et al. [1983] estimated that 6–12 million tons of Asian dust is transported annually to the central North Pacific, while Zhao et al. [2006] concluded that 3% of the Asian dust reaches the U.S. West Coast, equating to 180,000–360,000 t. Thus, large quantities of dust are transported thousands of kilometers to the U.S. West Coast and impact radiative and cloud microphysical properties far from its source regions. The large burden of dust and the uncertainty associated with its weather and climate effects demonstrate the need for a better understanding of the transport of dust, particularly since global warming has the potential
CREAMEAN ET AL.
©2014. American Geophysical Union. All Rights Reserved.
12,171
Journal of Geophysical Research: Atmospheres
10.1002/2014JD021694
to cause major changes in dust emissions [Gong et al., 2006; Goudie, 2009]. However, large discrepancies exist between global aerosol models that simulate dust in the atmosphere and its impact on climate as shown by the AeroCom phase I project [Huneeus et al., 2011]. Few long-term studies of dust transport exist; most previous studies involving continuous Asian dust observations are on the time scale of months to a year [Shimizu et al., 2004; Yumimoto et al., 2008]. VanCuren [2003] and Yu et al. [2012] presented long-term observations of Asian dust transported to the U.S. using an empirical assessment and remote sensing techniques, respectively, and found that Asian dust is regularly transported to high-elevation sites and even dominates over local dust emissions on the West Coast of North America. We present a 10 year (2002–2011) evaluation of Asian dust transport to the U.S. West Coast by looking at 25 Interagency Monitoring of Protected Visual Environments (IMPROVE) sites and expand upon the work of VanCuren [2003] by using more recent and supplemental observations. Seasonal dust concentration patterns are analyzed in the context of transport patterns and lidar measurements from the National Institute for Environmental Studies (NIES), Japan network. Dust wet removal during transport is investigated by using measurements of precipitation accumulation over the U.S. West Coast and Pacific Ocean from the Tropical Rainfall Measuring Mission (TRMM) satellite and calcium deposition measurements in rain and snow at multiple locations from the National Atmospheric Deposition Program/National Trends Network (NADP/NTN). Combined, these measurements provide a long-term record of dust, which we analyze in an effort to characterize the relative seasonal impact of wet removal processes on Asian dust burdens along the U.S. West Coast.
2. Measurements and Methodology Figure 1 shows all sites and domains used for analysis, including sites from the IMPROVE network (Figure 1a; discussed in section 2.1), the NADP/NTN (Figure 1a; discussed in section 2.3), the NIES network (Figure 1b; discussed in section 2.1), and domains used for TRMM (Figure 1c; discussed in section 2.3). The U.S. West Coast domain for TRMM is approximately equivalent to the domain used for Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) analysis (discussed in section 2.2). 2.1. Dust Observations Soil concentrations were acquired from 25 IMPROVE sites on the U.S. West Coast in Washington, Oregon, and California (http://vista.cira.colostate.edu/improve/Default.htm). Site codes, elevations, latitudes, and longitudes are provided in Table 1. IMPROVE sites in the California Central Valley, Death Valley, and Southern California are excluded from analysis due to their interference from dust sourced from local agriculture and desert soils [Baker et al., 2005; Clausnitzer and Singer, 1996; VanCuren and Cahill, 2002]. Sites are separated into specific regions as discussed below. Twenty-four hour aerosol filter samples were collected every 3 days from 2002 to 2011. Each site corresponded to 9 ± 2 samples per month, on average, from 2002 to 2011. Of those sites, 2% had ≤5 samples per month. IMPROVE also provides reports of uncertainties and minimum detection limits (MDLs) per sample for fine particle mass (≤2.5 μm) and each element analyzed. On average, based on 33,249 samples from all 25 sites from 2002 to 2011, fine particle mass concentrations were measured down to 0.31 μg m 3 with an uncertainty of ±0.20 μg m 3, which was also previously reported by VanCuren and Cahill [2002]. Select “soil elements” from IMPROVE including aluminum (Al), silicon (Si), calcium (Ca), iron (Fe), and titanium (Ti) were measured by X-ray fluorescence. The fine soil concentrations were calculated by the following IMPROVE convention using concentrations of specific metals: SOIL = 2.2[Al] +2.49[Si] +1.63[Ca] +2.42[Fe] +1.94[Ti] [Hand et al., 2011; Malm et al., 1994]. Asian soil (called “Asian dust”) was calculated from IMPROVE soil concentrations based on the method of VanCuren et al. [2005]; they used a Bayesian mixing model for Asian and local dust, using the ratio of Fe/Ca concentrations. Fe and Ca had uncertainties of, on average, 10% and 8%, respectively. Only 3% of the samples had values of Fe and/or Ca with uncertainties that fell below their respective MDLs; these samples were excluded from analysis, leaving a total of 32,237 samples with Fe and Ca concentrations well above the MDLs for analysis. Hyslop et al. [2012] also observed concentrations of Fe and Ca consistently above the MDLs, even considering the changes in IMPROVE analytical methods in 2005. Ratios of elements, as opposed to their absolute concentrations, minimize the effect of nondust contamination, e.g., Ca from industrial emissions [Arimoto, 2001; VanCuren and Cahill, 2002]. This ratio method has also been used on both ambient and snow water samples to investigate Asian influences on snowpack in the Californian Sierra Nevada Mountains [Hadley et al., 2010]. Based on the results from VanCuren et al. [2005], the probability of a sample containing 100% Asian dust corresponds to Fe/Ca ≤1, while 100% CREAMEAN ET AL.
©2014. American Geophysical Union. All Rights Reserved.
12,172
Journal of Geophysical Research: Atmospheres
10.1002/2014JD021694
Figure 1. Maps showing locations of (a) IMPROVE and NADP network sites on the U.S. West Coast and (b) NIES network lidar sites. Note that STAR and LAVO have both a colored and black star due to the colocation of both IMPROVE and NADP sites. The sites that correspond to regions with similar seasonal variability in Asian dust are highlighted in similar colors, including the mountain, coastal, and Columbia River sites. The elevation profile is the same for Figures 1a and 1b. (c) The domains used for TRMM analysis. The U.S. West Coast domain is approximately equivalent to the domain for HYSPLIT.
local dust corresponds to Fe/Ca >2. For the current analysis, the frequency distributions of Fe/Ca from a highelevation mountain site (LAVO), which is frequently exposed to Asian dust, and a local-influenced site (CORI) shown in Figure 2 align with the criteria defined by the Asian and local dust frequency distributions of VanCuren et al. [2005]. We believe that the Columbia River site is influenced predominantly by local sources since it is blocked from westerly flow, as previously shown [Fischer et al., 2009] and discussed in more detail below. A coastal site (SAGA) is shown for comparison, which exhibits a similar probability distribution as the mountain site. For all intents and purposes, 1 < Fe/Ca 1).
The TRMM online visualization and analysis system (http://disc.sci.gsfc.nasa.gov/ precipitation/tovas/) [Liu et al., 2012] was used to estimate the accumulated rainfall (mm) per month from 2002 to 2011 over the U.S. West Coast and the Pacific Ocean midlatitude region where HYSPLIT indicated frequent transport. The U.S. West Coast domain coordinates were approximately 125°, 49°, 34°, and 119° for the west, north, south, and east, respectively. The domain for the Pacific Ocean consisted of the coordinates 146°, 50°, 31°, and 124°. These domains were chosen based on the approximate areas encompassing the IMPROVE and NADP sites on U.S. West Coast and the transport region from HYSPLIT over the Pacific Ocean.
Wet deposition of dust on the U.S. West Coast was investigated using precipitation chemistry data from the NADP/NTN (http://nadp.sws.uiuc.edu/ntn/). Samples at NADP sites were collected during precipitation only. Seasonal wet deposition data (kg ha 1) were acquired from the NADP website for winter (December-January-February), spring (March-April-May), summer (June-July-August), and fall (September-October-November). In particular, the calcium ion (Ca2+) was used as a tracer for mineral dust in Washington, Oregon, and California [Brahney et al., 2013]. The chloride ion (Cl ) was used to eliminate coastal locations that were more influenced by sea salt, thus the Ca2+ at these locations was likely from a marine source versus dust [Brahney et al., 2013]. Coastal locations had much higher Cl deposition on average (5–35 times the amount of Cl ) than the noncoastal locations. Overall, 1, 1, and 8 noncoastal, highelevation NADP sites were active during 2002–2011 from Washington, Oregon, and California, respectively.
3. Results and Discussion 3.1. Dust Concentration Record on U.S. West Coast The Asian signatures of dust at IMPROVE sites demonstrated unique seasonal variability between the different regions on the U.S. West Coast. Sites that showed similar seasonal variability were separated into the regions shown in Figure 1a and Table 1, including the mountain, coastal, and the Columbia River regions. As an example, Figure 3 shows the Fe/Ca ratio from 2002 from sites representative of each of the individual regions and corresponds to sites in Figure 2. The solid brown markers represent Fe/Ca ≤1, meaning the dust in the IMPROVE sample was likely Asian versus local dust (Fe/Ca >1) [VanCuren et al., 2005]. Asian dust has been shown to contain high calcium content compared to dust from elsewhere [Cao et al., 2005; Formenti et al., 2011]. In contrast, soil samples from arid regions in the southwestern U.S. (i.e., local sources) have been shown to contain higher percentages of iron compared to calcium [Goldstein et al., 2007; Goldstein et al., 2008]. Seasonal variability was observed in Fe/Ca, particularly at the higher-elevation sites: Fe/Ca ≤1 occurred from November to June at sites in the mountains. This suggests the mountain sites were impacted by Asian
CREAMEAN ET AL.
©2014. American Geophysical Union. All Rights Reserved.
12,176
Journal of Geophysical Research: Atmospheres
10.1002/2014JD021694
Figure 4. Average IMPROVE concentrations for dust each month from 2002 to 2011, including (left column) all dust, (center column) local dust, and (right column) Asian dust. Rows show the averages for the mountain, coastal, and Colombia River regions. “Local Dust” is the difference in the monthly averages of “All Dust” and Asian Dust.
dust sources during this time period, similar to the previous findings of VanCuren et al. [2005] and Hadley et al. [2010]. Short, episodic Asian dust events occurred throughout the year at the coastal sites and less so at the Columbia River site; these sites were predominantly influenced by what is likely local dust sources. VanCuren et al. [2005] also observed short Asian dust episodes at sea level, but only during strong frontal passages. Further, Asian dust is not as commonly observed at low-elevation coastal sites because of the strong and persistent marine layer [VanCuren and Cahill, 2002]. Overall, the purpose of this analysis is to evaluate the degree of similarity in the records of Asian dust in the different regions along the U.S. West Coast. The Fe/Ca ratios were used to calculate the portion of the soil samples that contained what was probably Asian dust from each of the aforementioned regions on the U.S. West Coast for all 10 years (2002–2011). Figure 4 shows the results from these calculations, averaged from all the sites in each region for all dust, local dust, and Asian dust for each year. The average is also shown from all the years combined. Local dust concentrations were typically highest in July–October as previously observed with IMPROVE data [Wells et al., 2007] and satellite data [Prospero et al., 2002] but also had another smaller peak in April/May at the coast. Asian dust concentrations were within the range of those observed by VanCuren and Cahill [2002] (0.2–1 μg m 3) and were highest in the spring (peaking April/May), in addition to smaller concentrations throughout the summer, which is consistent with previous studies [Husar et al., 2001; Jaffe et al., 2005; VanCuren and Cahill, 2002; Wells et al., 2007]. Further, Asian dust concentrations were, on average, 11% (up to 44%) of the total fine particulate matter (particles with diameters ≤ 2.5 micrometers, or PM2.5) mass concentrations measured by IMPROVE at high-elevation mountain sites, with an average of 9%, 23%, 7%, and 4% in the winter, spring, summer, and fall, respectively. The coast saw much lower concentrations of Asian dust compared to the mountains, which is expected because the sites are typically isolated from the free troposphere, where Asian dust is thought to be transported when entering North America [Eguchi et al., 2009; VanCuren and Cahill, 2002; Zhao et al., 2006]. This is also exhibited in Figure 5, which shows the latitudinal dependence of estimated Asian dust concentrations for the mountain sites: the highest elevation sites (around 38°N in California) had the highest concentrations. It is
CREAMEAN ET AL.
©2014. American Geophysical Union. All Rights Reserved.
12,177
Journal of Geophysical Research: Atmospheres
10.1002/2014JD021694
difficult to elucidate the main inflow latitudes of Asian dust since the highest concentrations are typically associated with the highest elevation sites, which are centered around 38°N. However, Asian dust concentrations were higher on average at sites near 38°N when comparing to sites with similar elevations elsewhere, such as LAVO in California (1732 m; 0.28 μg m 3) versus WHPA in Washington (1827 m; 0.10 μg m 3). Further, California sites HOOV (2560 m) and KAIS (2597 m) are both similar elevations, yet HOOV is closest to 38°N and had almost double the average Asian dust concentration as KAIS (0.83 μg m 3compared to 0.48 μg m 3 at KAIS). In general, Asian dust was predominantly transported into California at a maximum around 38°N, with much lower concentrations transported further north to Oregon and Washington. The Columbia River site exhibited very different dust patterns compared to any other site (Figure 4). The site (178 m) is located in the Columbia River Gorge, a ~1200 m deep canyon of the Columbia River that forms the boundary between Washington and Oregon. In the summer, the winds flow up gorge from Portland, while easterly winds flow down gorge in the winter, both cases introducing haze rich in organic, sulfate, and nitrate aerosols [Green and Xu, 2007] and regional windblown dust [Kavouras et al., 2006]. The site is decoupled from westerly transport by the Cascade Range, as shown by the relative absence of Asian dust (note the scale difference for Figure 5. Latitudinal dependence of IMPROVE Asian dust. the Columbia River site versus the other regions). The latitude of each point along the line from each year Further, Columbia River has the highest influence corresponds to the latitude of each of the mountain sites listed in Table 1. The average from all years at each mountain from local dust, peaking in July/August Thus, the fact that Asian dust peaks in the spring at all sites, in site, site elevation shown as site IDs, and the average ± the standard deviation IMPROVE Asian dust concentration per combination with the low influence of Asian dust at state are also shown. the blocked Columbia River site compared to the mountain sites, corroborates the use of Fe/Ca to differentiate what is likely local from Asian dust on the U.S. West Coast. From here on, the Columbia River and coastal sites are excluded from analysis due to the low influence of Asian dust in the IMPROVE data. Both the Asian and local dust concentrations exhibited large interannual variability, as shown in Figure 6 at high-elevation sites. The larger ratio values correspond to what is likely more local compared to Asian dust. Asian dust was dominant (ratio
