Transport of Asian ozone pollution into surface air over the western United States in spring

Abstract

[1] Many prior studies clearly document episodic Asian pollution in the western U.S. free troposphere. Here, we examine the mechanisms involved in the transport of Asian pollution plumes into western U.S. surface air through an integrated analysis of in situ and satellite measurements in May–June 2010 with a new global high-resolution (∼50 × 50 km²) chemistry-climate model (GFDL AM3). We find that AM3 with full stratosphere-troposphere chemistry nudged to reanalysis winds successfully reproduces observed sharp ozone gradients above California, including the interleaving and mixing of Asian pollution and stratospheric air associated with complex interactions of midlatitude cyclone air streams. Asian pollution descends isentropically behind cold fronts; at ∼800 hPa a maximum enhancement to ozone occurs over the southwestern U.S., including the densely populated Los Angeles Basin. During strong episodes, Asian emissions can contribute 8–15 ppbv ozone in the model on days when observed daily maximum 8-h average ozone (MDA8 O₃) exceeds 60 ppbv. We find that in the absence of Asian anthropogenic emissions, 20% of MDA8 O₃exceedances of 60 ppbv in the model would not have occurred in the southwestern USA. For a 75 ppbv threshold, that statistic increases to 53%. Our analysis indicates the potential for Asian emissions to contribute to high-O₃episodes over the high-elevation western USA, with implications for attaining more stringent ozone standards in this region. We further demonstrate a proof-of-concept approach using satellite CO column measurements as a qualitative early warning indicator to forecast Asian ozone pollution events in the western U.S. with lead times of 1–3 days.

Key Points

• Using a high-resolution global chemistry-climate model and observations
• Asian pollution contributes to high-O3 events in western U.S. surface air
• Develop a space-based indicator to inform Asian influence on U.S. surface O3

1. Introduction

[2] Analyses of available observations indicate an increase of springtime ozone (O₃) mixing ratios in the free troposphere and at surface sites of the western United States since the 1980s [Jaffe et al., 2003a; Oltmans et al., 2008; Parrish et al., 2009; Cooper et al., 2010], coincident with rising Asian anthropogenic emissions of O₃ precursors as inferred from satellite measurements of column NO₂and bottom-up inventories [e.g. Richter et al., 2005; Ohara et al., 2007; Q. Zhang et al., 2009]. Global model simulations further support the role of rising Asian emissions in contributing to the springtime O₃ trend in the western U.S. [Jacob et al., 1999; Zhang et al., 2008]. Many prior studies clearly document episodic Asian pollution in the free troposphere over this region [e.g. Jaffe et al., 1999; Yienger et al., 2000; Heald et al., 2003; Jaeglé et al., 2003; Cooper et al., 2004a; Goldstein et al., 2004; Hudman et al., 2004; Zhang et al., 2008]. The extent to which this pollution is entrained into the boundary layer is unclear [Brown-Steiner and Hess, 2011] yet has implications for attainment of U.S. air quality standards. Here, we seek greater understanding of the mechanisms that transport Asian pollution into western U.S. surface air through an integrated analysis of in situ and satellite measurements with a new high-resolution (∼50 × 50 km²) global chemistry-climate model (GFDL AM3) [Donner et al., 2011]. In particular, we quantify the contribution of Asian pollution to surface O₃ levels on highly polluted days in both densely populated regions such as the Los Angeles (LA) Basin and in rural areas such as national parks.

[3] The threshold value for the U.S. primary National Ambient Air Quality Standard (NAAQS) for ground-level O₃, based on the 4th highest daily maximum 8-h concentration (MDA8) averaged over three years, has been lowered twice since the 1980s, with the current 75 ppbv threshold implemented in May 2008. In 2010, the U.S. EPA proposed to further revise the NAAQS O₃ threshold to one in the range of 60–70 ppbv [Environmental Protection Agency, 2010]. If a 60 ppb threshold were to be adopted, the number of violating counties across the USA would double (based on the 2006–2008 monitoring data, http://www.epa.gov/air/ozonepollution/pdfs/20100104maps.pdf). The rising Asian contribution to U.S. surface O₃ levels poses an additional challenge to meeting more stringent NAAQS for O₃ [Lefohn et al., 1998; Fiore et al., 2002; Keating et al., 2004]. Earlier work indicates that the contribution of Asian emissions to surface O₃ in the U.S. reflects mostly an enhancement in background levels [e.g., Goldstein et al., 2004; Zhang et al., 2008], and generally decreases on highly polluted days in summer [e.g., Fiore et al., 2002]. However, a few recent studies suggest a correlation between O₃ entering the U.S. west coast and local pollution episodes in California [e.g., Parrish et al., 2010; Huang et al., 2010; Pfister et al., 2011]. At present, quantitative estimates of the specific sources of O₃flowing into the western USA (e.g., recirculation of North American emissions, intercontinental transport, stratosphere-to-troposphere exchange, and other natural sources) are lacking. Advancing our understanding of the contribution from these sources to total surface O₃over the western U.S. is crucial to inform the NAAQS-setting process and to develop effective state implementation plans (SIPs; the process by which states demonstrate their path to achieving compliance with the NAAQS;http://www.epa.gov/air/urbanair/sipstatus/).

[4] Pollution transported from Asia to North America originates in the boundary layer of East Asia, where primary air pollutants are directly emitted from combustion and industrial sources. The polluted air masses are exported from the Asian boundary layer to the free troposphere via large-scale lifting such as deep convection or warm conveyor belts embedded in midlatitude cyclones [e.g.,Carmichael et al., 2003; Liu et al., 2003; Cooper et al., 2004a; Ding et al., 2009; Lin et al., 2010] as well as via mesoscale processes such as orographic forcing and mountain-chimney effects [e.g.,Chen et al., 2009]. The exported pollutants traverse the North Pacific Ocean in the mid- and upper troposphere, and eventually descend toward the surface [e.g.,Task Force on Hemispheric Transport of Air Pollution (TFHTAP), 2011; Brown-Steiner and Hess, 2011, and references therein]. Observational studies document the frequent transport of Asian dust, sulfate aerosol and mercury to the surface of western North America, and on rare occasions these events can be relatively strong [Husar et al., 2001; Jaffe et al., 2003b; McKendry et al. 2008; Weiss-Penzias et al. 2007; Ewing et al. 2010]. While these events are common, modeling studies suggest that only a small portion (<10%) of the Asian pollutants reaching the North American free troposphere actually descends into surface air [Yienger et al., 2000; Cooper et al., 2004a; L. Zhang et al., 2009]. However, these prior estimates, based on global-scale chemical transport models (CTMs) or Lagrangian trajectory models, both driven with meteorological fields typically at 2° × 2° horizontal resolution, may be conservative as these coarse-resolution models have limitations in capturing meso-scale processes contributing to the exchange of pollutants between the boundary layer and the free troposphere. The dilution of intense plumes due to numerical diffusion in global-scale models further diminishes the modeled impacts of vigorous episodic transport events [e.g.,Lin et al., 2010; Rastigejev et al., 2010, and references therein].

[5] An additional goal of this study is to extend earlier efforts that developed meteorological indices to characterize the daily to interannual strength of Asian pollution outflow and trans-Pacific transport [Liu et al., 2005, Liang et al., 2005]. We hypothesize that space-based instruments are useful for developing these indices given their vast spatial coverage at near-daily intervals and their ability to capture variability in chemical composition associated with synoptic-scale weather patterns. Over the last decade, satellite remote sensing has been widely used to advance our understanding of large-scale outflow of carbon monoxide (CO) from major biomass-burning regions [e.g.,Edwards et al., 2003; McMillan et al., 2005, 2008], pollution transport to the Arctic [e.g., Stohl et al., 2007], intercontinental transport of air pollution [e.g., Heald et al., 2003; Zhang et al., 2008; Pfister et al., 2011], and natural dust [e.g., Chin et al., 2007; Uno et al., 2009; Li et al., 2010]. Specifically, we explore the potential for near real-time space-based measurements of CO columns to help inform regional air quality management decisions regarding the potential for Asian pollution to enhance surface O₃ concentrations over the western USA.

[6] Section 2 briefly describes the model simulations and observational data. We focus our analysis on May–June 2010, leveraging intensive in situ vertical profiling of O₃ and related species during the NOAA CalNex (Research at the Nexus of Air Quality and Climate Change) field campaign in California (http://www.esrl.noaa.gov/csd/calnex). Section 3 examines the transport pathways of Asian pollution over the western U.S., including the influence of midlatitude cyclones, mixing of Asian polluted air masses with stratospheric air, and isentropic transport mechanisms. We then summarize the model evaluation (section 4) and the average impacts of Asian anthropogenic emissions on U.S. surface O₃, including the contribution to high-O₃ events during the CalNex period (section 5). Section 6illustrates a proof-of-concept approach developing a space-based index to identify situations under which episodic Asian pollution plumes may affect surface O₃ air quality over the western USA.

2. Measurements and Model

2.1. Meteorological and Chemical Environment in May–June 2010

[7] The Asian contribution to O₃ in the western U.S. is determined by the amount of O₃ produced in the Asian boundary layer, by the strength and position of the North Pacific storm track, and by additional O₃ produced during transport. May–June is the beginning of the O₃ pollution season at northern midlatitudes, with the East Asian coast experiencing maximum O₃ concentrations prior to the onset of the East Asian summer monsoon [e.g., Tanimoto et al., 2005; Lin et al., 2009]. The ridge-trough pattern over the eastern North Pacific and western U.S. was amplified during May–June 2010, resulting in stronger surface anticyclones and northwesterly surface winds relative to climatology [Cooper et al., 2011]. According to the NOAA Climate Prediction Center (http://www.cpc.ncep.noaa.gov), the 2009–2010 winter was influenced by strong El Niño conditions. Koumoutsaris et al. [2008] show that springtime Asian pollution outflow may be enhanced following an El Niño winter due to changes in convective activity in East Asia as well as the strengthening and southeastward extension of the North Pacific storm track. We therefore expect that meteorological conditions during May–June 2010 led to a particularly active season for export of Asian pollution and its subsequent transport to the southwestern U.S. These conditions, in conjunction with the suite of observations collected during the CalNex campaign (section 2.2), offer an invaluable opportunity to improve our mechanistic understanding of Asian pollution transport.

2.2. In Situ Measurements

[8] During the CalNex campaign, near-daily ozonesondes were launched at six sites in California (Figure S1 in theauxiliary material) between May 10 and June 19, 2010, to measure baseline O₃ at the U.S. west coast [Cooper et al., 2011]. For comparison with these measurements, we sample AM3 at the location and times of sonde launches, interpolate to the sonde pressure, and then average both the measured and model values over 0.5 km altitude bins. Three-hourly model fields are used, linearly interpolated to the sonde launch times (14:00–16:59 Pacific Daylight Time). We also use collocated measurements of O₃, NO_x, NO_y, CO, sulfate and organic aerosols aboard the NOAA WP-3D aircraft during CalNex to characterize Asian pollution layers above California (section 3.3) and to evaluate the model (section 4). Measurement techniques are described in prior publications [Ryerson et al., 1998; Holloway et al., 2000; Drewnick et al., 2005; Bahreini et al., 2009]. For ground-level O₃, we compare daily maximum 8-h average O₃ (hereafter MDA8 O₃) sampled at the model surface layer with observations from the U.S. EPA's Clean Air Status and Trends Network (CASTNet) and Air Quality System (AQS) database. For all model comparisons to in situ measurements, we apply a bilinear interpolation technique, obtaining the value at the sample location as a quadratic function of model results at the nearest four grid cells.

2.3. Satellite Data

[9] We use observations of CO and O₃total column amounts from the Atmospheric Infrared Sounder (AIRS) onboard the NASA Aqua satellite to identify trans-Pacific Asian pollution events and stratospheric O₃intrusions, respectively. The AIRS cross-track scanning grating spectrometer coupled with the Aqua cross-track scanning Advanced Microwave Sounding Unit (AMSU) provides vertical profiles of the atmosphere with a nadir 45 km field of regard (FOR) across a 1650 km swath [Aumann et al., 2003]. The broad swath, infrared spectral coverage, and reconstructed cloudy pixels [Susskind et al., 2003] of AIRS enable retrievals over nearly 70% of the planet every day with substantial portions of the globe observed twice daily (ascending and descending orbits), thus providing a unique opportunity for process studies of synoptic-scale chemical transport in the global atmosphere [McMillan et al., 2005, 2008]. In this study we use version 5.2 Level 3 daily 1° × 1° gridded standard products (AIRX3STD) of AIRS CO [McMillan et al., 2011] and O₃ total column retrievals [Susskind et al., 2003], downloaded from the NASA Goddard Earth Sciences Data and Information Services Center (http://disc.sci.gsfc.nasa.gov/). The ascending and descending retrievals are averaged to minimize the day-night difference and to provide better spatial coverage.

[10] The main sensitivity of AIRS to tropospheric CO occurs between approximately 300 and 600 hPa with a typical Degrees Of Freedom for Signals (DOFS) at northern midlatitudes of approximately 1 [Warner et al., 2007, 2010]. AIRS V5 CO retrievals are biased high by 6–10% between 900 and 300 hPa in the northern midlatitudes with a root-mean square error of 8–12% [McMillan et al., 2011]. A number of validation exercises reveal that AIRS O₃profile retrievals capture the large-scale dynamic variability and gradients of O₃ in the extratropical UT/LS region [e.g., Pan et al., 2007; Pittman et al., 2009; Wei et al., 2010].

2.4. GFDL AM3 Model Simulations

[11] AM3 is the atmospheric component of the GFDL global coupled atmosphere-oceans-land-sea ice model (CM3) [Donner et al., 2011; Griffies et al., 2011]. The AM3 cubed-sphere grid greatly improves polar representation and computational efficiency [Putman and Lin, 2007]; the model's horizontal resolution is denoted as Cn, where n is an integer number indicating total number of cells (finite volumes) along each edge of the cube. We apply C180 (where the size of the grid cell varies from 43 km at the 6 corners of the cubed sphere to 62 km near the center of each face) and C48 (163–231 km) horizontal resolution. Model simulations at the C48 resolution starting in January 2009 were used to initialize the C180 simulations spanning January–June 2010. Both coarse- and high-resolution simulations employ the same physical parameterizations and include 48 vertical levels, ranging in thickness from 70 m near the Earth's surface to 1–1.5 km near the tropopause and 2–3 km in much of the stratosphere.

[12] AM3 includes a fully coupled stratosphere-troposphere-aerosol chemistry within a general circulation model. The chemistry models ofHorowitz et al. [2003] for the troposphere and Austin and Wilson [2006] for the stratosphere are merged, with updates in isoprene nitrate chemistry [Horowitz et al., 2007], chlorine and bromine chemistry [Austin and Wilson, 2010], gas-phase reaction rates, dry deposition velocities and heterogeneous reactions as described byDonner et al. [2011]. The unique feature of full stratospheric and tropospheric chemistry distinguishes the GFDL AM3 model from most current generation global and regional tropospheric CTMs, and enables a process-oriented analysis for the intermingling of Asian pollution with stratospheric air over the U.S. west coast (section 3.1). We implement a stratospheric O₃ tracer by defining O₃ above the WMO thermal tropopause each model time step as “stratospheric” (O₃-strat), with its transport into the troposphere driven by dynamic processes (such as tropopause folds and cut-off lows). Once mixed into tropospheric air, O₃-strat is subject to transport and loss (chemical and depositional) in the same manner as O₃ of tropospheric origin. This technique likely overestimates the stratospheric contribution since any O₃ above the thermal tropopause is instantly labeled as “stratospheric” regardless of its actual origin.

[13] AM3 is typically run in the climate mode, forced with prescribed sea surface temperatures (SSTs) and sea ice [e.g., Austin and Wilson 2010; Golaz et al., 2011; Fang et al., 2011; Rasmussen et al., 2011] or as a component in the GFDL coupled model CM3 [Griffies et al., 2011]. To enable direct comparisons with CalNex observations, AM3 simulations at both C48 and C180 horizontal resolution in this study are forced with observed SSTs as in the standard climate simulations, but horizontal winds are nudged to those from the NCEP Global Forecasting System (GFS) at T85 horizontal resolution (approximately 1.4° × 1.4°, 64 sigma levels, archived 3 hourly) [Kanamitsu et al., 1991]. Our goal is to preserve the large-scale features of the observed airflow while allowing AM3 to simulate meso-scale patterns, similar to the approach taken for high-resolution regional weather forecasting [e.g.,Skamarock et al., 2008; Lin et al., 2009, 2010] and hurricane modeling [Knutson et al., 2008]. In order to minimize the impacts of noise introduced via nudging, which has been shown to lead to excessive stratosphere-to-troposphere O₃ transport [van Noije et al., 2004, and references therein], we implemented a pressure-dependent nudging technique, relaxing the model to GFS U and V with a time scale of 6 h in the surface level, but weakening the nudging strength with decreasing pressure (e.g., relaxing with a time scale of ∼60 h by 100 hPa and ∼600 h by 10 hPa). No nudging is implemented above ∼10 hPa since the GFS model top (∼0.32 hPa) is lower than that of AM3 (∼0.01 hPa).

[14] The impact of Asian emissions is determined as the difference between a base simulation with all emissions included, and a sensitivity simulation with the East Asian (15°–50°N, 95°–160°E) anthropogenic emissions (including aerosol) shut off in the model. Methane is set to observed values (1767 ppbv) as a lower boundary condition in both simulations. The sensitivity simulation begins in January 2009 with the C48 horizontal resolution, allowing for a one-year spin-up period sufficient to capture changes in the background atmospheric composition due to Asian emissions. Horizontal winds in the sensitivity simulation are also nudged to the GFS winds, thus we expect that the large-scale transport should be similar between the two simulations although completely removing Asian emissions may induce some feedbacks on hydrological processes (particularly through aerosol-cloud interactions which are included in AM3). Further, the O₃ response to emission perturbations is nonlinear due to chemistry as discussed by Wu et al. [2009], who found that the perturbation from a 100% NO_x emission reduction yields O₃ responses greater than 5 times the 20% NO_x emission reductions. Differencing O₃ mixing ratios with sharp gradients in the vicinity of tropopause folds may also introduce noise. Therefore, for each event discussed, we carefully checked for consistent patterns in Asian enhancements to O₃, CO, and NO_y in the model, constrained wherever possible by independent observations (e.g., satellite images and ozonesondes).

2.5. Emissions

[15] We implement emission inventories to reflect 2010 conditions. All emission inventories described here are regridded to 1° × 1° for C48 and 0.5° × 0.5° for C180. The global emissions from anthropogenic sources and international shipping for the year 2000 (0.5° × 0.5°) were taken from Lamarque et al. [2010] as a base inventory. We then replaced anthropogenic emissions in the East Asian domain (13°S–53°N, 70°–150°E) with the 2006 Asian emission inventory (0.5° × 0.5°) from Q. Zhang et al. [2009], scaled to 2010 as described below, and the North American (24°–60°N, 135°–40°W) emissions with the U.S. National Emission Inventory (NEI) for 2005 (4 × 4 km²) (ftp://aftp.fsl.noaa.gov/divisions/taq/emissions_data_2005).

[16] For our 2009 and 2010 simulations, North American anthropogenic emissions (including shipping) are held at the 2005 level since changes are expected to be small over this period. We increase the Chinese NO_x emissions from Q. Zhang et al. [2009] by 32.6% for 2010 and 27.9% for 2009, based on changes in the 2006–2010 satellite measurements of NO₂ columns over central eastern China (30°–40°N, 110°–123°E) from the SCIAMACHY sensor (available on www.tenis.nl, with retrieval technique previously described by Boersma et al. [2004]). Our estimate reflects the relatively lower rate of increase of Chinese NO_x emissions during the economic downturn of 2009–2010 [Lin and McElroy, 2011] as compared to earlier years [Q. Zhang et al., 2009]. Using NO₂ columns directly will tend to overestimate changes in emissions over polluted regions since an increase in NO_x emissions consumes OH and increases the NO_x lifetime, but underestimate changes in emissions over remote regions where an increase in NO_x emissions decreases the NO_x lifetime through feedbacks on O₃ and OH [Lamsal et al., 2011]. By considering the influence of NO_x lifetime when deriving emission changes from SCIAMACHY NO₂ columns, Lamsal et al. [2011] reported a ∼18% increase in Chinese emissions in 2009 relative to 2006. This change is ∼9% lower than our estimate using NO₂ columns over polluted central eastern China without adjustments to NO_x lifetime.

[17] We also increase NMVOC emissions in China by 23.2% for 2010 and 17.4% for 2009, relative to 2006, assuming the same annual growth rate as 2001–2006 previously reported by Q. Zhang et al. [2009], due to lack of better information. Emissions of SO₂, CO, black carbon and organic carbon remain at the 2006 level, based on recent evidence from satellites [Lu et al., 2010; Lei et al., 2011; D. G. Streets, personal communication, 2010].

[18] Daily biomass burning emissions for 2009–2010 are adopted from Wiedinmyer et al. [2011] (FINN v1, 1 × 1 km²) and distributed over six ecosystem-dependent altitude regimes between the surface and 6 km, following the recommendations ofDentener et al. [2006] as support from recent satellite observations [Val Martin et al., 2010]. Isoprene emissions from vegetation are calculated online based on the Model of Emissions of Gases and Aerosols in Nature (MEGAN) [Guenther et al., 2006], as implemented in Emmons et al. [2010] with modifications described by Rasmussen et al. [2011]. Other natural emissions–including acetone, ethane, propane, ethylene, propene, methanol, ethanol and monoterpenes from vegetation, CO from vegetation and ocean, volcanic SO₂, soil and lightning NO_x, dimethyl sulfide (DMS), sea salt and dust–are included as in the standard AM3 simulation [Donner et al., 2011].

3. Transport Pathways of Asian Pollution

[19] We identify three major trans-Pacific Asian pollution events from AIRS measurements of daily CO columns for May–June 2010.Figure 1illustrates the daily progression of these large-scale CO plumes from the East Asian coast to the North American west coast during May 4–8, May 16–20, and June 15–19. These events are typical cases of rapid trans-Pacific warm conveyor belt (WCB) transport [Cooper et al., 2004a]. The June plume (Figure 1c) took a more northerly route to the Gulf of Alaska followed by southeastward advection toward southern California, spending ∼1 day longer over the North Pacific Ocean than the two plumes in May. Model CO columns exhibit a similar large-scale structure as the AIRS data, and reveal that these CO plumes are dominated by Asian pollution which was lofted from the continental boundary layer of eastern China approximately 1 week prior to their arrival to the U.S. west coast (figure not shown). Our analysis described below mainly revolves around the three Asian pollution events displayed inFigure 1, the mechanisms governing their transport into western U.S. surface air, and the ability of AM3 to capture these processes.

Figure 1

(a–c) Major trans-Pacific Asian pollution events in May–June 2010 as seen from AIRS retrievals of CO total columns (10¹⁸ molecules cm⁻²; level 3 daily 1° × 1° gridded products). Day and nighttime retrievals are averaged, with white areas indicating no retrievals. Data bins are reordered so that neighboring gridded cells of data at the international dateline are no more than a swath of time apart (about 90 min).

3.1. Cyclone Passages and Merging Air Streams

[20] Prior work has shown that Asian pollution is often intermingled with stratospheric O₃ intrusions above the North American west coast [e.g., Jaeglé et al., 2003; Cooper et al., 2004b; Liang et al., 2007; Ambrose et al., 2011]. Indeed, we find that deep tropopause folds occurred one day prior to the arrival of Asian pollution above the California coast on May 18 and June 17, while the May 4–8 pollution event was followed by two consecutive stratospheric intrusions on May 9–11 [Langford et al., 2011; M. Lin et al., Springtime high surface ozone events over the western United States: Quantifying the role of stratospheric intrusions, manuscript in preparation, 2012]. Taking the May 16–20 event as an example, we examine here the dynamic processes leading to the interleaving and mixing of Asian and stratospheric originating air masses and quantify their contributions to O₃ entering western North American.

[21] Observed and simulated O₃ vertical profiles at two coastal sites from May 17–19 are shown in Figure 2. The GFDL AM3 model (section 2.4) successfully reproduces the layered structure, sharp vertical gradients, and day-to-day variability in the observed O₃ profiles, a major improvement upon previous models [Jaeglé et al., 2003; Liang et al., 2007; Jonson et al., 2010].

Figure 2

(a–d) Observed vertical profiles of relative humidity (gray) and ozone (black) at Point Reyes (RY) and Point Sur (PS) sondes on May 17–19, together with model total ozone (red), stratospheric ozone tracer (blue), and model ozone from a sensitivity simulation with Asian anthropogenic emissions shut off (green). Model results have been interpolated to the sonde pressure and averaged over 0.5-km altitude bins.

[22] Figure 3depicts the evolution of two consecutive midlatitude cyclones sweeping over the Gulf of Alaska and the associated upper level structure during May 17–19. On May 17, AIRS-retrieved total O₃ shows a sharp spatial gradient, with a narrow band of elevated O₃ columns in excess of 450 DU (Figure 2d) coinciding with an elongated dry air stream in the GFDL AM3 model (Figure 2a), both extending to the California coast. This typical feature of a stratospheric intrusion [Wimmers et al., 2003; Felker et al., 2011] is also represented in the GFS FNL analysis by the potential vorticity (PV) contours on the 300 hPa surface. The ozonesonde launched at Point Reyes, located immediately to the right of the tropopause fold off the California coast, recorded high O₃ mixing ratios in excess of 100 ppbv penetrating down as low as 6 km a.s.l. (Figure 2a), coincident with low humidity, a marker for air of stratospheric origin.

Figure 3

Evolution of two midlatitude cyclones and associated flow structures over the Pacific North America on May 17–19, 2010. Shown are (a–c) AM3 6-h mean sea level pressure (contours, hPa) with 300–400 hPa layer average specific humidity (shading, mg/kg) at 21:00 UTC and (d–f) AIRS retrievals of daily mean total column amounts of ozone (shading, DU) with 300 hPa potential vorticity (contours, PVU) computed from GFS Final analysis at 18:00 UTC. Only PV values greater than 2 PVU are contoured and signify air of essentially stratospheric origin. The thick black line along the California coast in Figure 3e indicates the location of vertical cross-section shown inFigure 4.

[23] Between May 17 and 18, this dry air stream of stratospheric origin sheared into three components in a similar manner previously described by Cooper et al. [2004b]. Depicted in the AIRS retrievals of total column O₃ and model 300–400 hPa humidity on May 18 (Figures 3b and 3e), the first component travels across the western U.S. with the upper level flow, while the second component representing the bulk of the dry air stream moves northward. A three dimensional view of AM3 stratospheric O₃ tracer reveals that the third component had descended into the lower troposphere by 21:00 UTC, May 18, leading to the formation of a distinct O₃ layer just 2 km above the California coast (Figure 4a). Four ozonesondes launched along the California coast on May 18 confirm the existence of this dry layer of enhanced O₃ (Figure S3, middle panels). Transported stratospheric remnants contribute 40–80 ppbv to the simulated 60–100 ppbv range of total O₃ mixing ratios in this approximately 2 km thick layer of enhanced O_3, with collocated CO enhancements (Figure 4b). AIRS observations of CO columns indicate transport of an Asian pollution plume into the eastern North Pacific on May 18 by the WCB of the next midlatitude cyclone entering this region (Figure 1a). We find in the model that the bulk of Asian polluted air masses were located in the upper troposphere (6–10 km) (Figure 4b), immediately adjacent to stratospheric air on the left (Figure 4a). Asian anthropogenic emissions contribute 20–30% of total O₃ mixing ratios in the upper troposphere when the air was sampled by the Point Reyes ozonesonde on May 18 (Figure 2b). Similar to the findings of Stohl and Trickl [1999], these processes reveal two interleaving O₃ layers with dominating stratospheric and Asian influence in the lower and upper troposphere, respectively.

Figure 4

Vertical cross-section along California coast (thick black line inFigure 3e), Showing (a) simulated stratospheric ozone tracer (shading, section 2.3) and (b) total CO (shading) with Asian CO (contoured every 5 ppb starting at 20 ppb) as determined by the difference between the standard simulation and a sensitivity simulation with Asian anthropogenic emissions shut off. Letters in red denote locations of ozonesondes (Figure S1).

[24] By 21:00 UTC, May 19, the first cyclone decayed and a newly formed cyclone was located off the coast of Washington (Figure 3c). A portion of stratospheric air that had rolled back from the deep intrusion on May 17 was now entrained into the newly formed tropopause fold of the second cyclone (Figure 3f). A distinct high-O₃ layer with lower humidity between 4 and 8 km a.s.l. was observed by the Point Reyes and Point Sur ozonesondes and successfully captured by the model (Figures 2c and 2d). In the ∼60–95 ppbv of simulated total O₃ in this layer, remnants of stratospheric intrusions and Asian anthropogenic emissions contribute ∼40–55 ppbv and 10–25 ppbv, respectively, suggesting the intermingling of polluted and stratospheric air.

[25] Between May 19 and 20, stratospheric and Asian polluted air masses became further intermingled and descended into the mid- and lower troposphere above southern California (Figure S2). These air masses reached the western U.S. surface on May 20–21, contributing to elevated surface O₃ episodes in the national parks (section 5.2).

[26] Other stratospheric intrusions identified during the CalNex campaign were also mixed, to varying degrees, with Asian pollution, suggesting that the dispersion of O₃-rich stratospheric air into polluted air masses is a common feature in the vicinity of active storm tracks in the North Pacific as well as in the Atlantic [e.g.,Stohl and Trickl, 1999; Parrish et al., 2000; Esler et al., 2003]. Our analysis further supports prior publications suggesting that a positive O₃-CO correlation does not necessarily indicate anthropogenic influence on O₃ enhancements in the continental outflow (such as Asian Pacific Rim) or inflow regions (such as North American west coast) [Cooper et al., 2004b; Voulgarakis et al., 2011]. Thus the O₃-CO correlations derived from collocated in situ or satellite measurements (such as TES) reported in earlier studies [e.g.,Zhang et al., 2006, 2008; Voulgarakis et al., 2011; Pfister et al., 2011] should be interpreted with caution.

3.2. Isentropic Transport Mechanisms

[27] We explore here the transport of Asian pollution into western U.S. surface air and its impacts on the regional variability in surface O₃, focusing on the Asian pollution event in June (Figure 1c). Figure 5 shows observed and simulated O₃ profiles at Trinidad Head and Point Reyes on June 17–19, illustrating a layered structure of O₃ with stratospheric and Asian origin, similar to those discussed in section 3.1. The large Asian influence on O₃ mixing ratios (up to 30 ppbv) mainly occurred between ∼3 and 9 km above the California coast during June 17–19. Less stratospheric O₃ was entrained into the polluted WCB during this event than during May 19–20. This Asian pollution plume was transported into the California coast by the WCB of a midlatitude cyclone, which intensified and moved into the Gulf of Alaska during June 19–21 (Figure S4). A southeastward shift of the Pacific anticyclone and a weak surface low extending northward from Mexico produced onshore flow, containing Asian pollution, across central and southern California. These conditions led to a strong latitudinal variation of Asian CO which descends along isentropic surfaces (S4), a typical feature of Asian plumes entering the western U.S. [Brown-Steiner and Hess, 2011]. By 21:00 UTC, June 20, the plume has progressed to southern California, with the vertical distribution of Asian O₃ up to 10–30 ppbv exhibiting a continuous band sloping from 5 to 8 km a.s.l. all the way to the surface of the Sierra Nevada and southern California (Figure 5d).

Figure 5

(a–c) As in Figure 2, but for the June 17–19 event at Trinidad Head (TH) and Point Reyes (RY). (d) Subsidence of Asian ozone (shading) along isentropic surfaces (contours, K) above the California coast on June 20, 2010 (also see Figure S4).

[28] The subsiding Asian pollution along isentropic surfaces episodically enhances surface O₃ levels above the surrounding “background” over the intermountain regions. Figure 6 shows that the regions experiencing high Asian influence in excess of 8–15 ppbv are where the observed and simulated total O₃ pattern is enhanced. The model estimates Asian enhancements of 8–15 ppbv in locations where observed MDA8 exceeds 60 ppb on June 20–22 over the intermountain west, and of 5–8 ppbv where observed MDA8 exceeds 75 ppb over parts of southern California on June 22. While these estimates are likely an upper limit for current conditions (see section 2.4), they nevertheless indicates the potential for Asian emissions to contribute to high-O₃episodes over the high-altitude western USA.

Figure 6

Spatial patterns of daily maximum 8-h average ozone in western U.S. surface air on June 20–22, 2010, showing (a) CASTNet (circles) and AQS (squares) observations, (b) results from the surface layer in the GFDL AM3 model, and (c) the enhancements due to Asian anthropogenic emissions, calculated using the same 8-h interval identified for the MDA8 in the base simulation.

[29] Asian influence appears to be minimal where O₃ exceeds 75 ppbv in the Central Valley of California on June 21–22 (Figure 6), indicating that this O₃ pollution mostly reflects photochemical production from local emissions and suppressed ventilation under the influence of a surface anticyclone. Langford et al. [2010] suggested that O₃ produced in the LA Basin can be lifted by topographic venting, followed by westerly transport to Utah and western Colorado. An additional sensitivity simulation with North American anthropogenic emissions switched off in the model suggests that North American influence is relatively smaller over the intermountain regions where the model suggests Asian enhancements contribute to enhancing total surface O₃ than at the local pollution hot spots around each city (Figure S6). We note that the ∼50 × 50 km²horizontal resolution of AM3 may be insufficient to capture the meso-scale mountain-valley flows described byLangford et al. [2010].

[30] Another notable feature in Figure 6 is the modeled O₃ minimum above the ocean that parallels the coast and varies strongly with distance from shore. This O₃minimum largely reflects the influence of the low-level jet along the California coast which is typically at a maximum in summer [Burk and Thompson, 1996; Parish, 2000]. The lower observed O₃ at the AQS sites near the coast supports the existence of this transport pattern. Asian enhancements to O₃ in the coastal marine boundary layer do not exhibit strong episodic variations, consistent with prior publications regarding background influence at the Trinidad Head surface site [Goldstein et al., 2004; L. Zhang et al., 2009; Parrish et al., 2010; Cooper et al., 2011]. Our results further reveal that the majority of the Asian pollution plumes arriving to the U.S. west coast are located in the free troposphere where a portion of the pollution descends isentropically behind cold fronts, which is directly intercepted by the elevated terrain and eventually becomes entrained into the daytime boundary layer over land. This transport mechanism is the key driver for the stronger signal of episodic Asian pollution in surface air over the U.S. intermountain regions than over the North Pacific Ocean.

3.3. Dispersion of Asian Pollutants Into Southwestern U.S. Surface Air

[31] We discuss in this section the potential impacts of Asian pollution in a densely populated region like the LA Basin, where Asian pollution can mix with high levels of locally produced O₃pollution. The NOAA WP-3D intercepted a pollution layer approximately 2 km above the LA Basin on May 8, with 3–8μg/m³ sulfate aerosol, < 1 μg/m³ organic mass, 150–175 ppbv CO, and 80–90 ppbv O₃ (Figure 7). This pollution layer was encountered three times (denoted as red arrows in Figure 7c) by the aircraft, and shows distinct sulfate enhancements, in contrast to the boundary layer pollution with dominant signatures of primary emissions (such as organic matter and CO). The observed high sulfate concentration is unlikely to be from oxidation of SO₂ emissions from diesel ships off California's coast. GOES 1 km visible images for this day show that the weather was calm and the skies completely cloud free above southern California and the nearby offshore regions, indicating no vertical transport of moist marine boundary layer air up to 2 km a.s.l. Under these types of conditions aircraft in situ observations of ship plumes off the California coast show that the plumes remain within the very stable marine boundary and are less than 200 m above the surface [Chen et al., 2005]. The presence of an aerosol layer dominated by sulfate is a distinguishing feature of Asian pollution layers that have been transported to the eastern North Pacific [Brock et al., 2004; Peltier et al., 2008; Dunlea et al., 2009]. When a polluted air mass rises out of the Asian boundary layer, the existing aerosol is washed out if precipitation occurs during lifting, while the less-soluble gas phase compounds such as SO₂ are not entirely removed. SO₂that “escaped” wet removal during lifting is converted into sulfate during subsequent trans-Pacific transport in the free troposphere, enhancing the sulfate-to-organic aerosol ratio in eastern North Pacific air masses influenced by Asian pollution.

Figure 7

Transport of Asian pollutants to the lower troposphere of southern California on May 8, 2010. (a) Model Asian enhancements to the tropospheric column O₃in dobson unit. The dashed black line denotes the location of the ozone vertical cross-section shown in Figure 7b. The white line indicates the NOAA WP-3D flight path. (b) Model vertical distribution of Asian ozone (shading) with isentropic surfaces (contours, K) along the California coast. (c) Observed distributions of CO (green), O3 (blue), sulfate (red) and organic aerosols (orange) along the WP-3D flight path, with altitude shown as gray shading.

[32] Indeed, AIRS CO retrievals confirm the occurrence of large-scale trans-Pacific transport of Asian pollution during May 4–8 (Figure 1a). The plume arrived at the coast of California on May 8, evident both in AIRS CO and in the AM3 model Asian enhancements to tropospheric column O₃ (Figure 7a). A vertical cross-section of model Asian O₃ in Figure 7bdepicts a plume-like feature of Asian O₃enhancements in excess of 10–25 ppbv sloping from 3 to 6 km above northern California to 2–3 km above the LA Basin where the air masses were sampled by the WP-3D. Other WP-3D flights during CalNex also intercepted sulfate-dominated aerosol layers above California, to varying degrees, on May 7, May 24, and June 20. The model indicates Asian enhancements to O₃mixing ratios but fails to reproduce the magnitude of sulfate concentrations, likely due to excessive wash out during trans-Pacific transport consistent with an overestimate in precipitation [Donner et al., 2011].

[33] Our analysis suggests that the signature of Asian pollution enhancements in the lower troposphere of the densely populated LA Basin is sufficiently strong to be observed, such as through analysis of chemical composition demonstrated here. Daytime mixed layers in this region range from 1 to 4 km a.s.l. [Langford et al., 2010; Cooper et al., 2011; Neuman et al., 2012; D. Seidel et al., Climatological variations in planetary boundary layer mixing heights over the Continental United States and Europe, submitted to Journal of Geophysical Research, 2011]. Asian pollution ∼2 km aloft is thus available to be entrained into the boundary layer and contributes to the local pollution burden. It should be noted that strong dilution may occur when Asian polluted air masses are dispersed into the boundary layer. For example, Figure 8 shows that much of Nevada experiences a pronounced maximum of Asian enhancements (10–15 ppbv) to surface O₃ on May 8, primarily reflecting the free tropospheric signal intersected by the elevated terrain. By May 9, Asian influence covers a larger spatial extent (including southern California), but the magnitude is weakened by a factor of two. Observed O₃at the CASTNet and AQS ground stations exhibits a similar day-to-day variability in the spatial pattern as model surface O₃. Despite dilution, Asian emissions contribute at least 50% of the surface O₃ increases from May 8 to May 9 over Arizona and western Colorado.

Figure 8

As in Figure 6, but for May 8–9, 2010.

4. Model Evaluation

4.1. Variability and Bias in O₃ Flowing Into the Western U.S.

[34] Generalizing from the process-oriented analysis insections 3.1–3.3, we conclude that the GFDL AM3 model captures much of the observed dynamic variability in O₃ flowing into the western U.S. resulting from complex air stream interactions. A systematic evaluation of model O₃ profiles at seven ozonesonde sites during CalNex shows that AM3 captures the latitudinal variation of surface to near tropopause O₃along the California coast, with correlation coefficients for day-to-day variability ranging from 0.50 to 0.96 (Lin et al., manuscript in preparation, 2011). For most cases, AM3 at C48 horizontal resolution (∼163–231 km) also captures the general features of the observed O₃ profiles (Figures S3 and S5). The model bias in the upper troposphere tends to increase when the ozonesondes are primarily influenced by stratospheric air, likely due to limitations to resolve adequately (at both C48 and C180) the filamentary structure of tropopause folds and the associated sharp vertical and horizontal gradients in O₃ mixing ratios.

4.2. Variability and Bias in Ground-Level O₃

[35] Table 1 summarizes the model mean bias in MDA8 O₃and correlation coefficients at 18 CASTNet sites over the western USA during May–June 2010. Despite the large bias (up to 10–25 ppbv) at the coastal sites (e.g., at MOR, PIN, LAV), the model reproduces much of the day-to-day variability in surface O₃ with correlation coefficients ranging from 0.4 to 0.8 and standard deviations within 15% of those observed for most CASTNet sites. Comparison with AQS observations (Figure S7) further shows that the model captures the regional variability of mean observed surface O₃ over the western U.S., including a minimum in the coastal areas and a maximum in the southern California and intermountain regions. The surface O₃bias is typically smaller at the high-elevation sites over the intermountain regions, suggesting that downward transport from the free troposphere is not likely to be the primary cause of the surface O₃ bias. Supporting this conclusion, we find that the model bias in surface O₃at both the low-elevation AQS sites and high-elevation CASTNet sites does not have a direct relationship with the Asian enhancement (Figure S8).

Table 1. May–June 2010 Mean Statistics for Observed and Model MDA8 O₃ in Surface Air at the CASTNet Ground Stations in the Western USA

State	Site ID	Site Name	Elevation (m)	Observed^a (ppbv)	Model^a (ppbv)	R
AZ	GRC474	Grand Canyon NP	2073	59 ± 7	64 ± 8	0.56
AZ	PET427	Petrified Forest	1723	56 ± 7	62 ± 9	0.46
AZ	CHA467	Chiricahua NM	1570	57 ± 7	60 ± 11	0.67
CA	PIN414	Pinnacles NM	335	46 ± 7	62 ± 10	0.83
CA	YOS404	Yosemite NP	1605	53 ± 10	61 ± 10	0.73
CA	LAV410	Lassen Volcanic NP	1756	48 ± 8	60 ± 8	0.80
CA	JOT403	Joshua Tree NP	1244	74 ± 12	72 ± 8	0.42
CA	SEK430	Sequoia NP - Ash Mountain	457	59 ± 13	70 ± 10	0.80
CO	MEV405	Mesa Verde NP	2165	55 ± 8	64 ± 8	0.60
CO	ROM406	Rocky Mtn NP	2743	59 ± 10	61 ± 10	0.72
CO	ROM206	Rocky Mtn NP Collocated	2743	55 ± 11	61 ± 11	0.75
NV	GRB411	Great Basin NP	2060	54 ± 9	61 ± 10	0.71
TX	PAL190	Palo Duro	1050	49 ± 7	54 ± 7	0.46
TX	BBE401	Big Bend NP	1052	49 ± 8	50 ± 7	0.54
UT	CAN407	Canyonlands NP	1809	58 ± 7	65 ± 9	0.57
WA	MOR409	Mount Rainier NP	415	29 ± 8	55 ± 5	0.71
WY	PND165	Pinedale	2388	53 ± 8	58 ± 8	0.69
WY	CNT169	Centennial	3178	56 ± 8	58 ± 8	0.55

• a Mean ± standard deviation.

4.3. Ozone Production Efficiency

[36] Next we evaluate the O₃ production efficiency (OPE), O_x versus NO_z derived from collocated measurements of O₃, NO_x, and NO_yaboard the WP-3D aircraft during CalNex, over the LA Basin and the Central Valley, respectively (Figure 9). The observations suggest a slightly smaller OPE in the LA Basin than Central Valley. The model captures this urban versus rural difference in OPE [Trainer et al., 1993; Kleinman et al., 1994], with the slope of O_x versus NO_z linear least squares correlation within ∼10% of those observed for both regions. The intercept of the correlation can be interpreted as a background O₃ level [Kleinman et al., 1994]. We find that the intercept of O_x versus NO_z over the LA Basin in the model is close to that in the observations (Figure S9a). For the Central Valley (Figure S9b), however, the intercept is ∼10 ppbv higher in the model, suggesting a possible problem in the boundary layer “background” O₃ level over this region. Figure S10 confirms that the bias gets worse in the low tail of the cumulative probability distribution of O₃ in the Central Valley_.

Figure 9

Asian pollution enhancements to (a) mean ozone mixing ratios at ∼800 hPa and (b) daily maximum 8-h ozone in surface air for May–June 2010, estimated with ∼50 km AM3. (c) Same as Figure 9b, but for ∼200 km AM3. The dashed rectangle in Figure 9a indicates the surface region analyzed inFigure 10. The symbols denote locations of 12 CASTNet high-elevation sites shown inFigure S8b.

[37] We conclude that the model has systematic positive biases in the northern California and coastal regions. While the source of this problem is unclear, it may reflect some combined influence from model limitation in resolving coastal gradients and sharp topography, an underestimate in coastal marine boundary layer cloudiness [Donner et al., 2011] which affects oxidant production and loss, local NO_x emissions or lifetime, and planetary boundary layer mixing heights (MH). A systematic evaluation of continental U.S. boundary layer MH in GFDL AM3 indicates that the model reproduces general features of observed spatial, seasonal, and diurnal MH variability but has difficulty simulating shallow, stable boundary layers at night (Seidel et al., manuscript in preparation, 2011).

5. Impacts of Asian Pollution on U.S. Surface Ozone Air Quality

5.1. Mean Distributions

[38] Figure 9a shows the mean Asian anthropogenic enhancement to U.S. O₃ mixing ratios at ∼800 hPa averaged over May–June 2010; a pronounced maximum occurred over the southwestern U.S. Enhanced Asian pollution in the lower troposphere toward the south reflects the combined influence from isentropic subsidence (section 3) and production of O₃ from NO_x released from the thermal decomposition of Asian PAN when the air warms as it descends (Figure S11). Earlier studies suggest that Asian pollution reaching the eastern North Pacific Ocean typically splits into northern and southern branches with O₃ production due to PAN decomposition occurring in the southern branch [e.g., Moxim et al., 1996; Heald et al., 2003; Hudman et al., 2004; Zhang et al., 2008; Fischer et al., 2010, 2011]. We find that this mechanism extends into the western U.S. during spring 2010 and contributes to the total Asian enhancement to surface O_3.

[39] The spatial pattern of Asian enhancements at ∼800 hPa is in general agreement with forward trajectory analysis of surface destinations of baseline air masses by Cooper et al. [2011] for the same study period. Our results suggest that the latitudinal variation of the Asian enhancement contributes to O₃ gradients along the California coast as measured by ozonesondes [Cooper et al., 2011]. Compared with the springtime climatological pattern of Asian O₃ tracer of Brown-Steiner and Hess [2011], our May–June 2010 estimates of Asian influence in southwestern U.S. surface air are relatively larger, likely reflecting the southeastward extension of the Pacific storm tracks following an El Niño (section 2.1).

[40] Figure 9b shows the mean Asian influence on MDA8 O₃in the model surface layer for May–June 2010. Trans-Pacific transport of Asian pollution has the greatest impacts on surface O₃in the high-elevation regions over the western U.S., consistent with prior studies [e.g.,TFHTAP, 2011; Brown-Steiner and Hess, 2011; Zhang et al., 2011], where the elevated terrain has a higher probability of intersecting subsiding air masses behind cold fronts than lower elevation regions such as the Central Valley of California and the eastern U.S. (section 3). The minimum Asian O₃ enhancement in surface air occurs in the Southeast U.S. where the average transport in the lower troposphere advects air masses from the Gulf of Mexico, limiting the influence of westerly transport, similar to summertime transport conditions discussed by Fiore et al. [2002]. The AM3 simulation at C48 horizontal resolution (Figure 9c) produces a consistent large-scale view of Asian influence in surface air over U.S., but the high-resolution model spatially refines the estimates and produces a 1–2 ppbv higher impact at the surface in Utah, Arizona and western Colorado.Zhang et al. [2011]also found a greater background impact in a nested version of the GEOS-Chem model than the coarser output from the 2° × 2.5° version.

5.2. Contribution to Air Quality Exceedances

[41] We next examine in Figure 10, the O₃ enhancement from Asian anthropogenic emissions for the entire distribution of springtime MDA8 O₃ concentrations in the model surface layer over the southwestern U.S. (box in Figure 9a). Asian emissions can contribute up to 8–15 ppbv when springtime total O₃ concentrations exceed 70 ppbv over southern California and Arizona, consistent with case study results in section 3 (gray points in Figure 10). This finding is in contrast to an earlier study indicating that foreign influence typically declines below 8 ppbv in the 70–90 ppbv range of U.S. summertime total O₃ concentrations in surface air [Fiore et al., 2002]. While local emissions likely play a dominant role in producing the high-O₃ events in surface air over this region, here we calculate the number of O₃ events in excess of 60, 70 and 75 ppbv thresholds that would not have occurred in the model in the absence of Asian pollution enhancements. These points are denoted by the green and red trapezoids in Figure 10 for 60 and 70 ppbv thresholds, respectively. We include only points where the Asian enhancement is above the 25th percentile to filter out model noise. Dividing the number of points falling within these trapezoids by the total number of points exceeding the threshold value, we estimate that 20% of MDA8 O₃ exceedances of 60 ppbv in the model would not have occurred over southern California and Arizona in the absence of Asian anthropogenic emissions. For a 70 ppbv threshold, this statistic increases to 49% since both the mean and extreme events of Asian enhancements increase in the 70–80 ppbv range of total O₃ and the number of exceedances for the 70 ppbv threshold decreases as compared with the 60 ppbv threshold. For a 75 ppbv threshold–the current NAAQS for O₃–that statistic is 53%.

Figure 10

Asian enhancements to daily maximum 8-h average ozone in surface air, as estimated with the GFDL AM3 model, plotted as a function of total ozone abundances for all land grids within the rectangle indicated inFigure 9a. The blue line represents the 25th percentile value of Asian enhancements. Points falling within the green and red trapezoids denote values in excess of 60 and 70 ppbv, respectively, that would not have occurred in the absence of Asian emissions in the model. The case study periods in section 3 are identified as gray points.

[42] We further examine any influence of model biases on the relationship between Asian enhancements and total O₃abundances, and thereby our estimate of the relative contribution to high-O₃ events. The results are reported in Figure S8a for all AQS sites within the same region used in Figure 10, suggesting no direct relationship (r² = 0.02) between Asian enhancements and model bias in surface O₃. Reidmiller et al. [2009]found an anti-correlation between foreign influence and model biases for a 15-model average (year 2001), and thus suggested that global models may underestimate foreign contribution. We do not find any evidence for this relationship in our model.

[43] Next we explore the influence of Asian emissions on high-O₃ episodes at three representative CASTNet sites (Figure 11). Episodes of elevated O₃ mixing ratios are observed at the national park ground stations, lagged by 1–2 days after the onset of Asian pollution events over the U.S. west coast discussed in section 3 (gray shaded periods in Figure 11). During these events from May–June 2010, surface O₃anomalies of up to +15 ppbv are observed at the three sites, with the model attributing approximately 2–9 ppbv positive anomalies to Asian anthropogenic emissions. The GFDL AM3 high-resolution model captures 45–65% of observed day-to-day variability of surface O₃ at the ground stations, and the Asian anthropogenic enhancements can explain 25–30% of the variability in the model total O₃. The results indicate that subsiding Asian pollution directly contributes to some high-O₃ episodes at the CASTNet stations in the mountain west, consistent with our findings in section 3 and Figure 10.

Figure 11

Day-to-day variability of observed (black), modeled (red) and Asian enhancements (purple) to daily maximum 8-h average ozone at three CASTNet surface stations (triangles inFigure 9a). Shown are anomalies relative to the May–June mean. The gray shading indicates major Asian pollution events discussed in section 3.

6. Forecasting Asian Pollution Events in the Western U.S. From Space

[44] In this section we explore the potential to develop a space-based indicator for Asian pollution enhancements to surface O₃ over the western USA. The method involves the correlation of daily AIRS CO columns at each 1° × 1° grid box with the Asian contribution to MDA8 O₃ in the model surface layer for western U.S. national parks, for time lags of 0 to 3 days between the data sets. Figure 12 shows an example for Grand Canyon National Park. The region with the greatest correlation over the eastern North Pacific represents the region from which air is preferentially transported to the U.S. west coast lower troposphere within the time lag (transport time to the high elevation receptor site) indicated. In Figure 12 strong correlations emerge with AIRS CO columns over approximately 30°–50°N of the northeast Pacific; the region that gives the best correlation with the surface data shifts from west to east as the time offset decreases, reflecting the progression of Asian pollution toward the U.S. west coast consistent with Figure 1. Correlation results for other ground stations in the western mountains show a similar large-scale feature, with the region experiencing the peak correlation falling within the boxes denoted inFigure 12.

Figure 12

Correlation coefficients between May–June 2010 AIRS daily CO columns at each 1° × 1° grid box and AM3 estimated Asian anthropogenic enhancements to daily maximum 8-h average ozone at Grand Canyon National Park (denoted as a white circle), Arizona, considering time lags of 0–3 days. The rectangles indicate the regions where AIRS CO can be used to derive a space-spaced indicator of Asian influence on surface ozone in the western U.S. (section 6).

[45] Based on the correlation analysis, we hypothesize that variability of AIRS CO measurements averaged over the eastern North Pacific within these boxes can serve at least as a qualitative warning index to inform regional air quality management decisions about incoming Asian enhancements to surface O₃ in the western mountains. Figure 13 elaborates on this hypothesis by illustrating a time series of AIRS CO averaged over the eastern North Pacific between 30°–50°N latitudes and 145°–125°W longitudes (box in lag = 1d panel of Figure 12), together with Asian enhancements to surface MDA8 O₃ averaged over Great Basin, Grand Canyon and Canyonlands National Parks. The variability in AIRS CO is confounded by the seasonal increase in OH, which leads to a decreasing trend (−6.1 × 10¹⁵ molecules/cm² per day, r²= 0.74) in CO from May to June. To highlight the CO variability due to trans-Pacific transport, we remove the linear trend from the time series of AIRS CO, which more clearly shows the three major trans-Pacific Asian pollution events discussed insection 3(May 7–12, May 17–22, and June 17–22). Further supporting the utility of satellite-based CO column measurements as an early warning indicator for surface air quality, we find that the timing of peak Asian enhancements on surface O₃ over the Rocky Mountains generally appears 1–3 days later than the peak AIRS CO off the U.S. west coast. Generalizing from the analysis in section 3 and Figures 12 and 13, we find that it takes approximately 3–6 days for the Asian pollution plumes arriving over the eastern North Pacific, which can be detected from space, to be transported into western U.S. surface air during the late spring and early summer of 2010.

Figure 13

Time series illustrating increasing Asian enhancements to surface ozone (purple) at CASTNet sites following 1–2 days after peak CO columns represented by AIRS (gray, detrended in black) over the eastern North Pacific (30°–50°N, 145°–125°W). The yellow shading indicates the range of Asian enhancements at the three sites: Great Basin, Grand Canyon, and Canyonlands National Parks.

[46] We note that the relationship between AIRS CO and Asian O₃pollution events in the western U.S. shown here is qualitatively promising in terms of daily to synoptic-scale variability. The ΔO₃ due to Asian influence may not have a quantitative relationship with ΔCO represented by AIRS since Asian pollution transported to the eastern North Pacific may not consistently reach the same locations given variability in meteorological conditions. Future work should explore the interannual variability in the vertical extent of Asian pollution over the U.S. west coast as well as explore whether retrievals at particular levels (rather than the total column) contain a more quantitative relationship with the Asian surface O₃over the U.S. Mountain West. The correlation between satellite-detected CO enhancements over the eastern North Pacific and the Asian pollution reaching western U.S. surface air should be further examined using multiyear data sets.

7. Conclusions

[47] We have applied a new high-resolution (∼50 × 50 km²) global chemistry-climate model (GFDL AM3) to examine the transport pathways of Asian pollution over the western U.S. in spring 2010, with a primary focus on the influence of Asian pollution on surface O₃air quality. To the best of our knowledge, this is the first time a global high-resolution model with full stratospheric and tropospheric chemistry has been employed to address the problem of trans-Pacific pollution transport.

[48] Extensive observations during the CalNex field campaign (May–June 2010) enable a process-oriented assessment of the model in simulating regional transport and chemistry over the western USA. Evaluation with surface O₃networks shows that the model captures much of the observed day-to-day variability in daily maximum 8-h average O₃ (MDA8) in surface air, with correlation coefficients ranging from 0.4 to 0.8 and simulated standard deviations within 15% of those observed for most sites. The most severe limitation for our study is a high surface O₃ bias of 2–10 ppbv on average in the high elevation regions and up to 10–25 ppbv at the Pacific coastal sites (section 4). The model bias in surface O₃ does not have a direct relationship with the Asian enhancement.

[49] An integrated analysis of satellite measurements of CO and total O₃columns from AIRS, ozonesondes, and model diagnostics reveals two cases of interleaving and mixing of Asian pollution with stratospheric air associated with complex airstream interactions as midlatitude cyclones cross the North American west coast during CalNex (May 17–20 and June 17–19) - a transport mechanism noted by prior studies [e.g.,Stohl and Trickl 1999; Cooper et al., 2004b]. The GFDL AM3 model successfully reproduces the prominent features of these events, and implies that transported stratospheric remnants from a deep tropopause fold may contribute 50–60% of distinct O₃ layers between 2 and 4 km above the California coast. Asian pollution, arriving a day after deep tropopause folding, contributes 20–30% of total O₃in the mid-troposphere when the air was sampled by the ozonesondes (section 3.1). We note our estimates should be viewed as upper limits as discussed in section 2.4.

[50] Using both the model and observations, we further illustrate that Asian pollution over the U.S. west coast descends isentropically behind cold fronts toward the south (sections 3.2 and 3.3), consistent with prior model studies [Cooper et al., 2011; Brown-Steiner and Hess 2011]. Combined with the influence of additional O₃ produced from the thermal decomposition of Asian PAN during subsidence, the Asian enhancement to O₃ at ∼800 hPa exhibits a pronounced maximum in the southwestern U.S. (including parts of California, Nevada, Utah, and Arizona; section 5.1). We find that the Asian enhancement in the model surface layer over this region increases for total MDA8 O₃ concentrations in the 70–90 ppbv range compared with that in the 60–70 ppbv range (Figure 10). We estimate that in the absence of Asian anthropogenic emissions, 49% of springtime MDA8 O₃ exceedances of 70 ppbv in the model would not have occurred over the southwestern U.S. For a 60 ppbv or 75 ppbv threshold, that statistic is 20% or 53%, respectively (section 5.2).

[51] During strong trans-Pacific transport events, the Asian enhancement to MDA8 O₃in western U.S. surface air can reach 8–15 ppbv in high-elevation regions experiencing peak surface O₃ levels in excess of 60 ppbv (Figures 6 and 8). We further demonstrate the potential for using near real-time satellite measurements of CO columns (AIRS) over the eastern North Pacific at daily intervals to develop space-based criteria to identify these strong Asian pollution events (Figures 12 and 13). Such a space-based index could serve as an early warning indicator (lead time of 1–3 days) for the potential influence of Asian pollution on surface O₃ air quality and any associated health advisories for National Park Services.

[52] Our analysis suggests that consideration of enhancements from Asian anthropogenic emissions will be important to develop effective O₃ abatement strategies for the western USA. There is a high degree of uncertainty regarding the trend of Asian emissions over the next few decades. The RCP (Representative Concentration Pathways) scenarios for IPCC Fifth Assessment Report [Moss et al., 2010] indicate that East Asian NO_x emissions peak in 2020–2040 with ∼10–30% increase from 2010 levels. It therefore seems likely that the Asian component of U.S. baseline O₃ will continue to increase at least in the coming decade, although coincident reductions in other northern hemispheric source regions may cancel such increases, with little change in baseline O₃ over North America. Constant baseline O₃ would make O₃ abatement more difficult under more stringent NAAQS, and increasing baseline would further confound O₃ abatement efforts (the “tightening vise” of Keating et al. [2004]).

Acknowledgments

[53] This work is supported by the Cooperative Institute for Climate Science (CICS), a collaboration between Princeton University and NOAA Geophysical Fluid Dynamics Laboratory (GFDL). The satellite data analysis is funded by the NASA Air Quality Applied Science Team (AQAST) (NNH09ZDA001N). Funding for the IONS-2010 field campaign was provided by NOAA ESRL Health of the Atmosphere Program, NASA Tropospheric Chemistry Program, U.S. Navy, Environment Canada, and NOAA's National Air Quality Forecast Capability. We acknowledge the free use of tropospheric NO₂ column data from the SCIAMACHY sensor from www.tenis.nland Bahreini Roya at NOAA ESRL for help collecting the AMS aerosol measurements onboard the WP-3D aircraft. We are grateful to Songmiao Fan and Hiram Levy II at NOAA GFDL and Pat Dolwick at EPA for helpful comments in improving an earlier version of this manuscript.