In this study, only one event for each of the following categories of sky conditions: the clear-sky, strong dust, strong pollution, and mixed pollution with dust categories is presented. Backward trajectories are used to track the dust events to the eastern Pacific where these events were observed by the DC-8 DIAL and nephelometer measurements in addition to satellite retrievals. Backward trajectories are also used to obtain information on the possible origins of the dust plumes as they travel from the Asian mainland to the remote Pacific Ocean.
3.1. Clear Sky (Clean, No Dust, Cloud-Free)
 To determine the baseline and provide the background values for studying Asian dust at Xianghe, China, a clear-sky event on 10 May 2006 is analyzed first with surface and satellite data. As shown in Figure 3a, the maximum total, direct, and diffuse downwelling SW fluxes are 1003 W m−2, 849 W m−2, and 160 W m−2, respectively. These values represent the typical meteorological conditions of a cloud-free day, when aerosol loading is low and SW scattering/absorbing is mainly due to atmospheric molecules. The surface AOD and α values are nearly constant throughout the whole day with the averages of 0.17 (at λ = 500 nm) and 1.26, respectively. Comparing the Xianghe α values with the annual mean of Beijing (the nearest city) indicates that fine mode particles may have dominated in this event (Figure 3b).
Figure 3. The total, direct, and diffuse downwelling shortwave (SW) fluxes measured at the Xianghe site. Aerosol optical depth (AOD) was derived from the Cimel-318 Sun photometer measurements. (a, b) Clear-sky event and (c, d) strong dust event.
Download figure to PowerPoint
 As illustrated in Figures 4a and 4b, the AOD values (∼0.16–0.2) retrieved from MODIS over the surface site are very close to the surface retrieved AOD, and the TOA albedo is also low (∼0.2). The MODIS visible channel image (Figure 4c) further proves that this is a clear-sky event with little aerosol loading right over the surface site, but has some aerosol loading about 200 km south of the surface site (Figure 4c) and clouds in the southeast region. In Figure 4d, the NCEP daily averaged sea level pressure (SLP) showed a high-pressure system extending from the northwest over the Gobi Desert to Xianghe during this event. The subsidence associated with the high-pressure system as well as a weak pressure gradient kept wind speeds low at the surface and suppressed any dust or aerosols from lifting high into the atmosphere. Therefore, the integrated observations provide accurate dust-free background information for us to investigate the dust events.
Figure 4. (a, e) Retrieved AOD from MODIS, (b, f) observed TOA albedo from CERES, (c, g) MODIS visible image, and (d, h) pressure pattern, for clear-sky event on 10 May 2006 (Figures 4a–4d) and for strong dust event on 17 April 2006 (Figures 4e–4h). Note that the red color in Figures 4a and 4e represents the AOD values ≥0.55.
Download figure to PowerPoint
3.3. Asian Dust and Aerosol Transport Pathways as Observed by Satellite and DC-8
 Figure 5 shows the averaged AOD values retrieved from MODIS observations on Terra and Aqua during the INTEX-B period (from 7 April 2006 to 15 May 2006). The major pathways of aerosol plumes as observed from both Terra and Aqua are quite consistent, but the absolute values are slightly different over the land regions. As illustrated in Figure 5a and 5b, there are two dust plumes: the pollution plume (bottom arrow) and the dust plume (top arrow) originated from two different source regions. Note that there is a large gap in both Terra and Aqua (white, no retrievals from MODIS) over Mongolia, which may represent very high AODs which cannot be retrieved by the current MODIS aerosol retrieval algorithm. Both dust plumes originated from two different regions which have two different paths that finally meet over the remote Pacific. In this study, we intend to qualitatively demonstrate the plume pathway trend using the Terra/Aqua observations, but not quantitatively address the absolute difference of their AOD retrievals, which is beyond the scope of this study. Transpacific transport during INTEX-B appears to have a steady northeasterly direction as barely any aerosols are found below 35°N latitude once they reach the eastern Remote Pacific. Though there are indications of meridional flow in certain individual cases, the general flow across the Pacific for the aerosols is zonal.
Figure 5. Averaged MODIS AOD values with a 0.5° × 0.5° grid box during the period 17 April to 15 May 2006 on (a) Aqua and (b) Terra, and (c) their transpacific transport overall zonal trend.
Download figure to PowerPoint
 Plume characteristics are given by the AOD values from both the Terra and Aqua retrievals. As the aerosols are transported across the Pacific Ocean, they are prone to both wet and dry deposition as well as dispersion over time. Figure 5c shows a steady decline in aerosol optical properties over time along their transport pathway. The concentration of fine mode aerosols generally decreases in the absence of dust but when mixed with dust, heterogeneous chemical reactions can occur which may cause a slow degradation of the total aerosol layer [Leaitch et al., 2009]. For example, the AOD values peak over the east coast of China, then drop significantly over the East China Sea, and eventually level off to the values close to 0.2 as they approach the remote Pacific. The major plume transport pathway can be seen clearly by following the retrieved AOD values which suggests: plumes exiting the Asian mainland containing very high AOD values transport slightly northeast, then turn slightly eastward after passing through the middle of the Pacific until reaching close to the 145° meridian, and ultimately turn northeast toward the west coast of Canada.
 The time periods and altitudes of dust plumes for all flight tracks over the remote Pacific can be easily identified by using DIAL measurements of aerosol scattering ratios at 1064 nm (Figure 2). However, it is difficult to quantitatively analyze the dust plumes by using the DIAL measurements only, especially for discerning the fine and coarse mode aerosols in the dust plumes. Therefore, it is necessary to use the nephelometer data to do further study. Nephelometer data in conjunction with the technique developed by Gobbi et al.  are utilized to verify which mode (fine or coarse) aerosols dominated in the selected four dust plumes. Scattering ratios and Angström exponent values are used as proxies for both Asian dust and pollution (Figures 6 and 7).
Figure 6. (a, b) The Case I (17 April 2006) and (c, d) Case II (24 April 2006) used in this study. The scattering coefficients at three wavelengths (450, 550, and 700 nm) (Figures 6a and 6c), and Angström exponent α versus spectral curvature as observed by the TSI Model 3563 nephelometer onboard the DC-8 aircraft during the INTEX-B field campaign (Figures 6b and 6d).
Download figure to PowerPoint
 For the 17 April 2006 dust event (Case I), the spectral data show a weak wavelength dependence of scattering coefficient values at three different wavelengths (Figure 6a) until a peak at 2100 UTC appears where the wavelength dependence increases suddenly. Immediately after the peak, the dependence becomes negligible. Figure 6b shows that the mean (and standard deviation) α is 0.46 ± 0.63, suggests that an abundance of coarse mode particles existed in the plume as given by values in Table 1. The scattering coefficients and α values in Figures 6a and 6b indicate that coarse model aerosols are dominant in the Case I.
 The 24 April 2006 dust event (Case II, Figure 6c) shows a strong wavelength dependence of the scattering coefficients at three wavelengths. The scattering coefficients at the shorter wavelength (λ = 450 nm) are consistently larger than those at the higher wavelength (λ = 700 nm) because the backward scattering coefficients decrease with increasing the wavelength with the presence of fine mode aerosols in the plume. The averaged α (Figure 6d) is 1.53 ± 0.23 and the δα values lie mostly in the negative range with a mean value of −0.31 ± 0.63, which indicates an influence of Asian pollution in this case (Table 1).
 The 29 April 2006 dust events (Cases III and IV, Figure 7) were observed at different time periods (Figure 2c). The α values were 0.40 ± 0.36 for Case III and 0.60 ± 0.29 for Case IV with both being less than the Case II values. The α values for Case III and IV, however, are less and greater, respectively, than those in Case I, indicating a mixture of coarse and fine mode aerosols within both dust events. For Case III, the scattering coefficients at the three wavelengths are nearly the same (as those in Case I) during the period 1.3–1.54 UTC (Figure 7a); subsequently, sharp peaks and strong wavelength dependence occurred for the scattering coefficients after that (as those in Case II). The spectral dependences also strongly increased during the period of 1.54–1.77 UTC (not shown here). Case IV had a wavelength dependence throughout the time interval though not as strong in magnitude as Case II. Case IV had a larger α mean value as compared to Case III which suggests even more fine aerosols or pollution but again, not as much as Case II.
 To further track these four cases back to their regions of origin, we used the NOAA backward trajectory model at three different altitudes: 3 km, 5 km, and 7 km AGL (R. R. Draxler and G. D. Rolph, HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) Model, 2003, http://www.arl.noaa.gov/ready/hysplit4.html). According to Qiu and Sun , aerosol dust particles can be transported eastward from China to the Pacific Ocean in the 2 to 7 km height range, and this is the reason to choose these three altitudes in this study. It should be noted that the altitudes used in some of the HYSPLIT analyses may differ from the chosen levels as dust plume altitudes will change during transport.
 Figure 8 shows the back-trajectory analysis of the selected four cases (using the 5 km (blue) trajectory line as a proxy for plume altitude). Key points describing the similarities and differences from the model output include: (1) for Cases I and III, the dust plumes originated from the Gobi Desert, and (2) for Cases II and IV, the dust plumes originated from the Taklimakan desert. The transport time of dust plumes for the four cases are around 7.5, 8, 5, and 7 days, to reach the remote Pacific regions where the dust plumes were observed by the DC-8 aircraft. However, the transport time was only approximated because the measurements were not taken at the same location. From previous studies, the dust plumes normally take, on average, 7 to 10 days to move out of the Gobi Desert area to the United States. Therefore, the model outputs are 240 h in duration in order to explore all possible sources of the air masses.
Figure 8. The NOAA HYSPLIT Backward Trajectories of dust plumes intercepted by DC-8 aircraft (denoted by black circles) over the remote Pacific Ocean for (a) Case I (2100 UTC 17 April 2006), (b) Case II (0300 UTC 24 April 2006), (c) Case III (0200 UTC 29 April 2006), and (d) Case IV (0400 UTC 29 April 2006). The heights of the dust plumes used in the analysis, 3000 m (red line), 5000 m (blue line), and 7000 m (green line), represent the range of heights of the dust plume intercepted by the DC-8 aircraft.
Download figure to PowerPoint
 For Case I, the center height of dust plumes is around 5 km, ranging from 3 to 7 km (Figure 2a). From the back trajectory analysis, the blue and red lines (Figure 8a) originated from the Gobi Desert area and did not pass over the highly polluted area, whereas for Case II, the dust and pollution plume was located around 5–7 km and sampled by DC-8 at roughly 7 km (Figure 2b). The back trajectory showed that the green and blue lines originated from the Taklimakan desert and did pass through highly polluted areas. Therefore, the outputs of the back trajectory model did support the findings from DC-8 measurements in Cases I (coarse mode) and II (fine mode).
 For Cases III and IV, the green and blue lines originated from central China, and the data suggested a possible scenario of air masses containing dust plumes that were lifted above the boundary layer and passed through many urban areas (Figures 8c and 8d). Trajectories that come from different regions but cross paths over either the Gobi or Taklimakan Deserts are also scrutinized in order to see how pollution and dust can comingle at various points during their transpacific transport processes.