4.2. Transport-Related Variations in Locally Produced Ozone in the Middle East
 The anomalies in locally produced ozone are shown in Figure 3b. We define locally produced ozone as ozone produced within the region between 15°N–35°N and 30°E–60°E. Compared with Asian ozone (Figure 3a), locally produced ozone varies less year-to-year in the mid-troposphere in summer, with a range of variability within ±3 ppbv or ±15% from the mean. Unlike Asian ozone, the local ozone anomalies show no extrema in 1994 and 2002. As discussed by Liu et al. , the confinement of locally produced ozone in the mid-troposphere of the Middle East is a result of the presence of the anticyclone over the region. A comparison of the anomalies in the geopotential heights at 400 hPa in July between years with positive local ozone anomalies and years with negative local ozone anomalies reveals differences in geopotential heights of about ±5 m over the Arabian Peninsula (not shown). The covariance between locally produced ozone anomalies and the geopotential heights over the 20-year period at 400 hPa is shown in Figure 9. The areas with largest covariances (∼10 m) are located over the northeast of the Arabian Peninsula, within the anticyclone (shown in Figure 1b), where the correlation coefficient between the two variables reaches ∼0.8.
Figure 9. Covariance (in m) between the locally produced ozone anomaly (in Figure 3b) and the geopotential heights at 400 hPa. Filled contours are statistically significant at the 90% level using the Student's t-test. Data source for the geopotential heights: NCEP/NCAR reanalysis [Kalnay et al., 1996].
Download figure to PowerPoint
 In the GEOS-Chem simulation, the chemically produced ozone is the same throughout the 20 years. It is the atmospheric circulations that control how the locally produced ozone is distributed. The simulated 20-year mean of this ozone is localized over the Middle East (not shown; see Figure 9a of Liu et al.  for a case in 2005), reflecting the confinement by the anticyclone in the mid-troposphere. Nevertheless, the spatial distribution of this ozone varies from year to year. For example, 2003 was a year with a maximum anomaly, indicating that locally produced ozone was mostly confined in the region in the 20-year period. This extreme was associated with a strong Iranian High, which was split into two and resulted in two high ozone centers over the Middle East. In the cases of 1988 and 1996 with negative extrema in ozone anomalies, the extreme in 1996 was due to the weakest Iranian High in the 20-year period and the high ozone center was shifted into the southwestern part of the Middle East, while in 1988, a large amount of locally produced ozone was transported out to the northeast of the anticyclone by the westerlies and out to the southwest by the easterlies.
4.3. Variability in Long-Range Transport of Ozone From Other Regions
 Although the ozone produced locally and the ozone transported from Asia predominantly contribute to the total ozone mixing ratios over the Middle East in summer, we find that the interannual variations in the ozone mixing ratios in the Middle East are only partially explained by the variations of these two large sources. For example, in Figure 10, the correlation coefficient is 0.54 between the ozone anomalies and the sum of Asian and locally produced ozone anomalies, suggesting that the interannual variations in ozone in the Middle East is also likely to be influenced by ozone transported from other regions.
Figure 10. Correlation between the total ozone anomaly and the sum of Asian and locally produced ozone anomalies in the Middle East in July at 400 hPa. The correlation coefficient and the p-value in the Student's t-test are indicated by r and p, respectively. The 1:1 line is indicated by a dashed line.
Download figure to PowerPoint
 As shown by Liu et al. , and in Table 1, in the middle troposphere of the Middle East there is a seasonal trade-off between ozone produced in Asia and the Middle East and ozone produced outside these regions, with the total contribution of ozone from outside of the Middle East and Asia at a minimum in summer. We find that there is a similar trade-off in the year-to-year variations between ozone from Asia and the Middle East and ozone from other regions. For example, the year-to-year variations in ozone transported from North America are strongly anti-correlated with the Asian ozone anomalies over the Middle East, with a correlation coefficient of r = −0.75 in July at 400 hPa. Similarly, the correlation between the ROW anomalies and the Asian anomalies in July is r = −0.87. The contributions from both of these regions were a maximum in 2002 and a minimum in 1994, when the Asian source was at a minimum and a maximum, respectively. Therefore, although the locally produced ozone and ozone transported from Asia are the dominant contributions to ozone abundances over the Middle East, transport of ozone from other regions is important in the context of driving the interannual variations in ozone in the Middle East.
 According to Liu et al. , much of the ozone transported from North America to the Middle East originates in the North American boundary layer and is exported to the upper troposphere over the North Atlantic, where it is transported along the subtropical westerly jet and descends into the Middle East, near the Mediterranean. Shown in Figure 11 is a latitude-altitude cross section of the zonal wind along at 40°E (across the Middle East). In 1994, when the North American ozone contribution was at a minimum, the subtropical westerly jet was stronger and shifted further north, to between 40°N–50°N, compared to 2002, when the North American ozone contribution was at a maximum and the jet was weaker and located between 30°N–45°N. We find that when the westerly jet was further south like in 2002, the Arabian anticyclone circulated the North American ozone, descending from the upper troposphere, around the eastern flank of the anticyclone and into the Middle East region. In contrast, when the westerly jet was further north like in 1994, the North American ozone contribution was confined north of the Middle East region.
Figure 11. Zonal winds (in m s−1) in a latitude-altitude cross section in July at 40°E in (a) 1994 and (b) 2002. The latitude of 35°N is marked with a line. Brown areas denote topography.
Download figure to PowerPoint
 In the Northern Hemisphere, the ROW region includes the Northern Hemispheric Pacific and Atlantic Oceans. It is ozone produced in these regions (downwind of Asia and North America) that is transported to the Middle East through the subtropical westerlies. Little ozone in the Southern Hemisphere of the ROW is transported to the Northern Hemisphere in the boreal summer (similarly South American and Australian ozone abundances are small in the Middle East in the boreal summer as shown in Table 1). Similar to the case for North America, we find that when the jet shifts to the north in 1994, less ozone from the ROW is transported through the westerly jet to the Middle East. Conversely, when the jet shifts to the south in 2002, more ozone from the ROW is transported into the Middle East.
 The subtropical westerly jet over Asia exhibits seasonal variations in strength and its meridional shift. In summer, the core of the jet over Asia is generally located near 200 hPa with wind speeds around 20–40 m s−1, compared with 50–70 m s−1 in winter [Zhang et al., 2006; Schiemann et al., 2009]. From winter to summer, the jet transits northward [Zhang et al., 2006; Schiemann et al., 2009]. Schiemann et al.  found that during the northward transition, the jet intensity and its latitudinal location vary greatly from year to year. As shown in Figure 4a, in 1994 the jet was characterized with strong cyclonic flow over Europe and the equatorward component of the flow contributed to the stronger westerly jet in central Asia. In 2002, the cyclonic flow over Europe was weaker (see Figure 4b) and extended into Siberia, which contributed to a weaker westerly jet that was shifted further south over the Middle East.
 Our analysis suggests a strong connection between the trade-off in transport from the different source regions and the meridional location and strength of the subtropical westerly jet. The Asian ozone anomalies are negatively correlated with zonal winds at 200 hPa over a region of 20°N–40°N and 30°E–90°E from South Asia to the Middle East at the 90% significance level (Figure 12a). In contrast, over almost the same latitudes, positive correlations of 0.5–0.8 are found between zonal winds and the North American and ROW ozone anomalies (Figures 12b and 12c), although the details are slightly different. The American and ROW ozone anomalies are more sensitive to zonal wind over the Middle East (30°E–60°E), while the Asian ozone anomalies are more sensitive to zonal wind from Asia to the Middle East (60°E–80°E). A more southward shift of the westerly jet is linked to stronger westerly flow over the Middle East, which favors transport of ozone from North America and ROW to the region. The transport of ozone from these regions into the Middle East is through strong descent over North Africa and the eastern Mediterranean. When the westerly jet is displaced north, the transport of this ozone down into the Middle East is restricted.
Figure 12. Correlation between zonal winds at 200 hPa and averaged ozone anomalies over the Middle East from (a) Asia, (b) North America, and (c) ROW. Filled areas are statistically significant at the 90% level using the Student's t-test. All values are for July.
Download figure to PowerPoint
 In their analysis, Liu et al.  found that transport from the stratosphere does not contribute significantly to the ozone maximum over the Middle East. Similarly, we find that the interannual variations in transport of ozone from the stratosphere is not significantly correlated with variations in ozone in the Middle Eastern mid-troposphere (r = 0.26). In addition, the stratospheric ozone contribution is only weakly anti-correlated with the Asian contribution in the Middle East (r = −0.30).
 Previous studies have related long-range transport of pollution with the state of the climate system through the use of climate indices, such as El Niño-Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO). It was found that ENSO is linked to year-to-year variability of Asian outflow toward North America and Europe [Liu et al., 2003, 2005; Koumoutsaris et al., 2008]. Liu et al.  found that the Asian outflow in spring is strong in the upper troposphere under La Niña conditions while Koumoutsaris et al.  found that El Niño conditions in winter favor the export of Asia pollutants toward Europe in the spring of the subsequent year. Moreover, Liu et al.  suggested that the correlation between ENSO and Asian outflow toward North America is seasonally dependent: it is strong in winter but weak in summer with spring and fall as intermediate cases. In our analysis we found no significant correlation between ENSO (using the Southern Oscillation Index (SOI)) and the westward transport of Asian ozone to the Middle East in summer. The SOI index in July is also insignificantly correlated with the ozone contributions from North America and the ROW. Lin and Lu  suggested that the subtropical westerly jet stream in summer may be correlated with ENSO in the previous winter. We explored possible links between the SOI index in the previous winter and Asian or American ozone anomaly over the Middle East in summer and found only weak correlations (r ≈ ±0.3) that are not significant (p > 0.1).
 The NAO index is found to be closely related to interannual variations in pollution transport from North America to Europe [Li et al., 2002] and from Europe to other continents [Duncan and Bey, 2004]. A positive NAO phase means a strong north-south pressure gradient over the Atlantic Ocean and this condition usually enhances the transport of North American boundary layer pollutants to Europe [Li et al., 2002]. However, it seems that after American ozone reaches Europe, its transport to the Middle Eastern mid-troposphere is not closely linked to NAO as we found only a weak correlation between the NAO and North American ozone over the Middle East.
4.4. Sensitivity to Linearization of the Ozone Chemistry
 To assess the potential impact of our use of chemical fields for 2005 to linearize the ozone chemistry, we repeated the analysis using chemical fields for 1990 and 1995. We found that the differences in the chemical fields had a small impact on the estimated anomalies. For example, the magnitude of the Asian ozone anomalies in 1994 were 26.7% over the Middle East using the 2005 chemistry fields, whereas they were 29.2% and 29.5% using the 1990 and 1995 chemistry fields, respectively. In 2002, the Asian ozone anomalies were −26.6% with the 2005 chemical rates, compared with −27.6% and −27.0% using the 1990 and 1995 chemical rates, respectively. Similarly, the differences in ozone produced over the Middle East were small. Similarly, the differences in the anomalies of locally produced ozone with different chemistry inputs were small. For example, the largest difference between the anomalies in locally produced ozone using 2005 and 1990 chemical inputs was 1.9%.
 We also compared the tagged ozone simulation with a full chemistry run over the 20-year period to assess the potential contribution of nonlinearity in the ozone chemistry to the year-to-year variations in ozone. We found that the largest differences in the ozone abundances in the tagged ozone run, relative to the full chemistry simulation, was 3.8% in July 1992. For 1994 and 2002, the years with the extrema in the Asian ozone contribution to the Middle East, the tagged ozone run produced ozone abundances that were lower than the full chemistry run by −0.7% and −0.3%, respectively. Figure 13 shows the correlation between the ozone anomalies over the Middle East obtained with the tagged ozone simulation (using the 2005 chemistry fields) and the full chemistry run. A significant correlation of r = 0.82 is found between the two sets of anomalies. We also obtained correlations of r = 0.79 and r = 0.83 for the ozone anomalies based on the 1990 and 1995 chemistry fields, respectively (not shown). This suggests that in these simulations, transport can explain over 60% of the interannual variability in the ozone mixing ratios in the Middle Eastern mid-troposphere in summer.
Figure 13. Comparison of ozone anomalies in July at 400 hPa between a full chemistry simulation and a tagged ozone simulation using 2005 chemistry input. The 1:1 line is indicated by a dashed line. The correlation coefficient and the p-value in the Student's t-test are indicated by r and p, respectively.
Download figure to PowerPoint
 Although our focus in this paper is on the impact of interannual variations in transport on ozone abundances over the Middle East, our analysis suggests that the nonlinear interplay between chemistry and transport could also provide a significant contribution to interannual variations in tropospheric ozone over the Middle East, and should be further explored to better understand the underlying mechanisms driving these variations.