Influence of Smoke on O3 and PM in Houston, Texas

The Houston metropolitan area has made significant strides in reducing O₃ concentrations, but the region still has many days above the current O₃ standard. Fires emit O₃ precursors and usually result in enhanced O₃ downwind. In this work, I used a variety of tools to quantify fire impacts on O₃ in the Houston area. First we use Generalized Additive Models (GAMs) to explore the relationship between the maximum daily 8-hour average O₃ (MDA8) for Houston for typical, non-smoke, conditions using a variety of meteorological parameters (temp, RH, transport pattern, CAPE, etc). The statistical model can explain up to 80% of the variance in MDA8 for O₃ for Houston. The model residual can then be used to tell us about unusual O₃ sources.

We use surface PM_2.5 data and the NOAA Hazard Mapping System-Fire and Smoke Product (HMS-FSP) to identify days with a high probability of smoke. We define “high confidence smoke days” as those with both overhead HMS smoke and enhanced PM_2.5 at the surface (>15.1 ug/m³), defined by the mean of non-smoke days (10.6 ug/m³) plus one S.D. (4.5 ug/m³). For these days, the O₃ GAM residuals are elevated by an average of 7 ppb compared to the non-smoke days, but with a high degree of variability. We also find that at similar levels of PM2.5, smoke days show higher O₃ MDA8 values compared to non-smoke days. For the high confidence smoke days, the monthly mean GAM residual (extra O₃) is between 0-4 ppb (in April through July) and 10-14 ppb (in August through October). This indicates that the impact on O₃ from fires in Houston is greatest in late summer-fall. The model residual is correlated with PM_2.5 with a slope of 1.2 ppb per μg/m3 for August-October. This provide an approximate measure of the extra MDA8 O₃ due to smoke for each μg/m3 of PM_2.5.

In addition to smoke, the GAMs can identify “extra O₃” from other unusual events. For September 1-3, 2017 Hurricane Harvey disrupted industrial operations in Houston and substantially increased emissions, compared to normal operations. MDA8 O₃ exceeded the current air quality standards at numerous sites in Houston during this period. The GAM MDA8 O₃ residual for this day was up to 30 ppb and demonstrates the unusual nature of photochemical O₃ production during the post-Harvey period.

Finally, we apply the GAM method to a recent case study and show that the method gives reasonable results that are directly applicable to the EPA guidance on “exceptional events”.