Analyzing Spikiness in Diurnal Ozone Patterns in Houston

1. Analyzing Spikiness in Diurnal Ozone Patterns in Houston A Case Study Using Region 12 Data to Relate Industrial Emissions to Ambient Air Quality Dave Sullivan OCE/MO and John Jolly OEPPA/TAD Why is this important? Because figuring out what causes high O3 in Houston can save $100s of millions & protect public health.

2. In most other areas in Texas, ozone levels typically increase at a fairly predictable, moderate rate z On high ozone days in Dallas, ozone rises at a steady rate of 15-25 ppb / hr z On high ozone days in Houston, ozone often rises suddenly 50 100+ ppb / hr. We believe this comes from the highly chemically reactive air in industrial areas

3. Given a complete set of monitoring data z Upset reports, z 5 minute resolution ozone, wind speed, wind direction data, z Hourly auto-GC & other PAMS VOC data, z Area-wide data, z Then we may be better able to police air quality

4. Using Delta-max to measure steepness of ozone increase/decrease for a monitor for a day z 1) Delta-max equals greatest hourly increase in ozone concentration z 2) Delta-min is greatest hourly decrease following delta- max z 3) High values for these two variables indicate possibility of upset emissions, and need to be investigated z 4) Charts of delta-max vs delta-min y plotted for 5-year period, 1995-99, for all area monitors with >= 100 ppb hourly ozone

5. Delta-max and Delta-min at Deer Park CAMS 35 (industrial/urban monitor) 1995-1999 High Delta-max values show spikiness of ozone concentrations in Houston Ship Channel area

6. Delta-max and Delta-min at Northwest Harris CAMS 26 (rural/suburban downwind monitor), 1995-1999 Note absence of high values i.e. less spikiness

7. Comparing Delta Max in Houston and Dallas z Dallas emissions largely mobile and area sources much less spikiness z Next two slides show distribution of delta-max values in Houston and Dallas over same 5-year period y all area monitors >= 100 ppb hourly ozone included z note that Houstons mean delta-max (34.9 ppb) is above 90th percentile in Dallas

8. Distribution of Delta-Max Values in Dallas Ft Worth, 1995-99

9. Distribution of Delta-Max Values in Houston-Galveston, 1995-99

10. Case Study: 9/20/99 z Severe early morning ozone spike at Deer Park C35 monitor z Ozone traveled downwind; spike subsequently recorded at 3 downwind monitors z 1,3-butadiene spill reported in morning at a Dynegy plant located on north side of channel, some 6-8 miles NW of the Deer Park monitor z Amount reported was small (50 gals) but estimates of upset releases are notoriously inexact -- plus this is a highly reactive VOC z Map, forward trajectory, and diurnal ozone graphs follow

11. Map of Affected Area -- Ozone spike detected first at Deer Park, then subsequently at Texas City, Galveston, and Clute, respectively. Winds were from NW, shifting to NE later.

12. HYSPLIT 24-Hour Forward Trajectory for 9/20/99 Using Upper Air Data 13z (7am cst) start time Red: 10m starting height Blue: 100m starting height Green: 500m starting height filled icons (near shore): noon cst outlined icons: 6pm cst This trajectory starts near Ship Channel, shows initial NW wind flow, followed by NE shift

13. Local Wind Direction Using CAMS (Deer Park met equipment not functioning; 2 adjacent monitors wind data shown)

19. Next: Looking at 1,3-butadiene at Deer Park z Deer Park Auto-GC measures 1,3-butadiene, but was not functioning on 9/20/99, hampering this investigation z Looked at data for this chemical at Deer Park in 1999 overall, instead z Following charts show wind directions, by frequency, of high 1,3-butadiene measurements y In 1999, very high measurements most frequently came from NNW -- same direction as 9/20/99

21. Summary z O3 spikes in Houston from industrial emissions y From upset emissions or regular releases of reactive species y Results of summer 2000 & 2001 studies shed light on this z Auto-GC data at Deer Park on study date would have been useful in identifying 1,3 butadiene. z Without auto-GC data, FO spent time checking company records to find link between spill & ozone z Bottom line: MonOps & TAD need complete auto-GC and CAMS data to do forensic studies for Field Ops