Article pubs.acs.org/est
Quantifying the Contribution of Long-Range Saharan Dust Transport on Particulate Matter Concentrations in Houston, Texas, Using Detailed Elemental Analysis Ayse Bozlaker,† Joseph M. Prospero,‡ Matthew P. Fraser,§ and Shankararaman Chellam*,†,∥ †
Department of Civil and Environmental Engineering, University of Houston, Houston, Texas 77204, United States Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida 33149, United States § School of Sustainable Engineering and the Built Environment, Arizona State University, Tempe, Arizona 85287, United States ∥ Department of Chemical and Biomolecular Engineering, University of Houston, Houston, Texas 77204, United States ‡
S Supporting Information *
ABSTRACT: The trans-Atlantic transport of North African dust by summertime trade winds occasionally increases ambient particulate matter (PM) concentrations in Texas above air quality standards. Exemptions from such exceedences can be sought for episodic events that are beyond regulatory control by providing qualitative supportive information such as satellite images and back-trajectories. Herein we demonstrate that chemical mass balancing can successfully isolate, differentiate, and quantify the relative contributions from local and global mineral dust sources through detailed measurements of a wide suite of elements in ambient PM. We identified a major dust storm originating in Northwest Africa in mid-July 2008 which eventually impacted air quality in Houston during July 25, 26, and 27, 2008. Daily PM2.5 and PM10 samples were collected at two sites in Houston over a 2-week period encompassing the Saharan dust episode to quantify the transported mineral dust concentrations during this peak event. Average PM concentrations more than doubled during the Saharan intrusion compared with non-Saharan. Relative concentrations of several elements often associated with anthropogenic sources were significantly diluted by crustal minerals coincident with the large-scale Saharan dust intrusion. During non-Saharan days, local mineral dust sources including cement manufacturing and soil and road dust contributed in total 26% to PM2.5 mass and 50% to PM10 mass; during the three-day Saharan episode the total dust contribution increased to 64% for PM2.5 and 85% for PM10. Importantly, this approach was also able to determine that local emissions of crustal minerals dominated the period immediately following the Saharan dust episode: simple quantification of bulk crustal materials may have misappropriated this elevated PM to trans-Atlantic transport of Saharan dust.
■
INTRODUCTION The arid regions of North Africa are estimated to emit about 800 Tgy−1 of soil dust each year, 70% of the global total.1,2 Summertime trade winds carry a portion of this North African particulate matter (PM) across the Atlantic Ocean to the Caribbean and the southern continental United States3,4 increasing ambient PM10 and PM2.5 concentrations5,6 in Texas typically a few times during June, July, and August. The ability to quantitatively distinguish between locally emitted versus long-range transported PM is vitally important because it enables us to exclude exceptional events that impact local air quality but which are beyond regulatory control especially when responding to more stringent PM standards. To date, source regions of long-range transported dust have only been qualitatively identified using concentration ratios of selected major elements4,7,8 (e.g., Al, Fe, Ti), backward air trajectories,9−12 and satellite data.13,14 Additionally, Saharan dust contributions to peak PM events have been quantitatively estimated for regions close to North Africa (e.g., in the Mediterranean8,9,12,15−17 and Canary islands10) but not in the © 2013 American Chemical Society
United States. Calculations of transported dust contributions have been based on comparisons of the composition of samples from remote locations not impacted by local emission sources16 with that of locally generated material. This strategy is difficult to use with soil dusts because of the similarities between the elemental compositions of different mineral dusts. However, such discrimination is vital to determine the impact of longrange aerosol transport on urban air quality where regulatory standards are designed to protect human health. The primary objective of this research is to show that by accurately measuring a suite of major and trace elements in different hypothesized mineral dust sources and in ambient PM in receptor locations, we can distinguish and quantitatively estimate contributions of long-range dust and locally entrained crustal material and local anthropogenic sources. For accurate Received: Revised: Accepted: Published: 10179
April 10, 2013 July 25, 2013 August 19, 2013 August 19, 2013 dx.doi.org/10.1021/es4015663 | Environ. Sci. Technol. 2013, 47, 10179−10187
Environmental Science & Technology
Article
Figure 1. Track and vertical profile of the Saharan dust cloud as observed with CALIPSO, July 27, 2008. CALIPSO data (V3-02) were obtained from the NASA image browse site [http://www-calipso.larc.nasa.gov/]. a) Track of the satellite across the midwestern US and the Caribbean. b) Total attenuated backscatter (532 nm) along the path; arrow indicates the approximate location of Houston along the track. c) Rawinsonde profile from Corpus Christi, 00:00UTC showing the elevated hot-dry Saharan Air Layer23 which tops out at 3,200 m, matching the CALIPSO dust-layer top altitude in part b. d) Aerosol type as determined in the CALIPSO vertical feature mask: yellow is “dust”; brown is “polluted dust”. e) Satellite track in the vicinity of Houston. Track figures are based on Google Earth images.
■
MATERIALS AND METHODS Sampling. Daily PM2.5 (n = 28) and PM10 (n = 27) samples were collected on 47-mm diameter Teflon filters between July 20 and August 2, 2008 at Clinton Drive (Texas Commission on Environmental Quality (TCEQ) CAMS 403; latitude +29.734, longitude −95.258) and Channelview (TCEQ CAMS 15; +29.803, −95.126) using Rupprecht & Patachnick 2025 samplers. A major outbreak of desert dust emerged off the west coast of Africa on July 14−16, 2008 (Figure S1 of Supporting Information (SI) and the graphical abstract). The air mass moved across the tropical Atlantic and entered the Western Gulf of Mexico, impacting air quality over the Houston area on July 25−27, 2008 (see satellite images and NOAA HYSPLIT21 back-trajectories in SI Figures S1, S2, and S3). The dust cloud covered a large area of the southwestern U.S. as shown by CALIPSO22 (Cloud - Aerosol Lidar Infrared Pathfinder Satellite Observations) which on July 27, toward the end of the event, moved along a track (Figure 1) that passed a short distance west of Houston. In Figure 1a and 1e, the position marked (29.84, −96.10) is 70 km west of Houston center, 80 km west of Clinton Drive, and 94 km west of Channelview. Figure 1b shows the aerosol distributions along the track as classified by the CALIPSO Vertical Feature Mask (VFM) which uses an algorithm to classify seven aerosol types, two of which are “dust” and “polluted dust” as shown in the color scale in Figure 1d. The northern edge of the dust cloud extends to ∼36N (near Tulsa, OK) where it is lost in a cloud; the southern extreme is located over Central America. Over Texas, the dust layer extends to ∼2.5−3.0 km altitude. The
source apportionment, we developed a representative source profile of aerosols originating in North Africa by collecting samples in a remote location in the Caribbean (Barbados) during dust episodes before they entered the continental United States. Sampling transported aerosols from remote locations to characterize the transported aerosol source profile reduces potential biases from atmospheric transformation or selective deposition of larger particles during trans-Atlantic transport.18,19 Importantly, equivalent analytical methods based on inductively coupled plasma − mass spectrometry (ICP-MS) were used to measure the elemental composition of source samples and ambient aerosols at receptor sites.20 As isolating air pollutant contributions is difficult in complex urban environments, the ability of detailed chemical speciation to differentiate sources is challenged because of the existence of myriad local sources. During a PM2.5 and PM10 sampling campaign at two locations in the highly industrialized Houston Ship Channel area, a large-scale intrusion of Saharan dust was first identified through back-trajectories and satellite imagery. Source profiles for Saharan dust, local soil and road dust and several others developed as part of this research or obtained from the literature were input to the chemical mass balance (CMB) receptor model. This approach isolated multiple sources of mineral material to ambient aerosol loadings, including long-range transport, industrial emissions, and entrainment of local soil and road dust. To our knowledge, this is the first use of the CMB model to quantify the relative importance of desert dust to PM2.5 and PM10 in the United States and complements other approaches put forth in Europe.16 10180
dx.doi.org/10.1021/es4015663 | Environ. Sci. Technol. 2013, 47, 10179−10187
Environmental Science & Technology
Article
Figure 2. Time series of PM2.5 and PM10 mass concentrations at the two monitoring sites. Green circles are hourly TEOM PM2.5 data, whereas red and blue squares are daily filter based PM2.5 and PM10 data, respectively.
by dynamic reaction cell − quadrupole − ICP-MS (ELAN DRCII, PerkinElmer). High purity NH3 (99.999%) was used as the cell gas with 0.2−1.0 mL min−1 flow rates and 0.25−0.8 RPq values to better measure27Al, 51V, 52Cr, 57Fe, 60Ni, 63Cu, 64 Zn, 75As, and 112Cd using the reaction cell.20,28 74Ge, 115In, and 209Bi were employed as internal standards: quantitative recoveries (90−110%) of elements from National Institute of Standards and Technology Standard Reference Materials (NIST SRMs); 1633b coal fly ash and 1648a urban particulate matter demonstrated the accuracy of our protocols (SI Figure S3). Source Apportionment. Source profile abundances (as mass fraction of each element) and receptor ambient concentrations, with appropriate uncertainty estimates, were input to a chemical mass balance model (EPA-CMB v8.2) to estimate source contributions along with their uncertainties for individual samples. Species used for model fitting included Na, Mg, Si, K, Ca, Ti, Co, Cu, As, Se, Sr, Ba, Pb, Al, V, Cr, Fe, Ni, Zn, La, Ce, Pr, Nd, and Sm; either because they represent major constituents of crustal material or because they represent unique marker species of Houston-area emission sources.29 Other elements were employed as floating species with the exception of Tm and Lu, which were excluded as >75% of their measurements was below their method detection limit (MDL, 0.005 and 0.004 ngm−3, respectively) and signal-to-noise ratios were generally
