**4. Conclusions**

**Figure 12.** Boxplot for the NOx-to-NOy ratio by day of the week

**Season Average ± Standard Dev.**

**(Weekdays)**

**Table 5.** Results of the ANOVA test for the NOx-to-NOy ratio (*p*-values are at an α of 0.05)

regional rainy season.

302 Current Air Quality Issues

Finally, we conducted an Analysis of Variance (ANOVA) test for the NOx-to-NOy ratio by season (see Table 5). For the hot months (spring and summer), lower ratios were observed during weekends compared to weekdays. In fact, the lowest average ratio was obtained during spring, the season with the highest average and peak O3 concentrations. In contrast, the winter of 2013 showed a marginally higher ratio during the weekends than on the weekdays, but the difference was not statistically significant. This could be the result of an intensive use of fuels during the whole week in response to the low temperatures. For the fall of 2012, this statistical test also indicates no difference between weekdays and weekends. Fall, as the results from previous sections suggest, is influenced by days with high O3 production and days with very low O3 levels as a result of the transition from a period with high SR and temperature to the

**Average ± Standard Dev.**

Summer 2012 0.606 ± 0.139 0.582 ± 0.124 0.033 Yes Fall 2012 0.671 ± 0.141 0.668 ± 0.107 0.657 No Winter 2013 0.545 ± 0.086 0.549 ± 0.077 0.504 No Spring 2013 0.553 ± 0.112 0.496 ± 0.142 < 0.001 Yes Summer 2013 0.586 ± 0.110 0.561 ± 0.112 < 0.001 Yes

**(Weekends)** *p-value*

**Difference between Weekdays and Weekends** Observational-based methods have proven valuable for analysis of the chemical interactions that give rise to high air pollution events in urban areas. In particular, the vast amount of information gathered by networks of air quality stations can provide insight into the produc‐ tion of secondary air pollutants such as O3, as atmospheric conditions change throughout the year. This analysis can be enhanced if complementary observations are also made of species that are not routinely registered, such as NOy.

In this study, we presented a statistical analysis performed on the air quality and meteorolog‐ ical data registered by the routine air quality stations of the MMA. The analysis included descriptive statistics, regression analysis, correlation analysis, PCA, and ANOVA, along with the interpretation of bivariate polar plots, wind roses, and boxplots. In addition, ratios of NOx with NOy and CO with NOy provided additional information on the level of chemical processing of the air masses traversing the MMA. When used together, these techniques prove to be complementary, thus providing more robust results.

In the MMA, O3 registers two distinct annual peaks: one in spring and the other in late summerearly fall. The analysis of meteorological conditions and air pollutant levels indicate that the O3 concentrations during winter would be characteristic of background conditions that get transported to the urban center when wind speeds are sufficiently strong. O3 production in winter is small because typical conditions that foster photochemical activity are not present, such as relatively high temperatures and strong solar radiation. During this season, NOy is composed mainly of locally-emitted NOx, which corroborates the low photochemical activity. That is, the ambient conditions do not favor the catalytic effect of NOx to produce O3. Spring is also heavily influenced by local emissions, but meteorological conditions favor photochem‐ ical activity, leading to production of the highest levels of O3. In this season, winds bring background O3 to the MMA, and precursors traverse the long-axis of the urban core, allowing for chemical processing, injection of fresh emissions, and high photochemical production. Spring registers the highest amount of NOz, indicating that NOx is reacting efficiently with VOCs to produce photochemical oxidants, including O3. Fall would seem to be influenced partly by local emissions and partly by transport. Even though fall has the second highest peak O3 levels, the average O3 in this season is relatively low. This can be explained by the fact that the rainy season occurs during this period. Finally, summer would be characterized by the influence of more regional (and aged) emissions. Deep planetary boundary layers are charac‐ teristic of this season, allowing mixing and dispersion of air pollutants that result in the lowest NOy levels of the year. Thus, even though temperature and solar radiation levels could suggest high O3 production, photochemical activity is limited by transport and mixing effects. Finally, for the hot months (spring and summer), a distinct weekday-weekend effect can be identified, as the NOx-to-NOy ratio tends to be higher during weekdays than during weekends. This would indicate changes in NOx emission rates during the week, which could lead to higher O3 events during the weekend, in line with the VOC-sensitive condition of the MMA atmos‐ phere that has been suggested by others.

Overall, the analysis of NOx and NOy levels with other chemical and meteorological variables as well as the correlation and ratios between NOx and NOy provide indicators of the level of photochemical activity that fosters O3 production.
