Air quality model application and evaluation with aircraft data in Houston, Texas

Realistic simulation of air pollutant concentrations requires accurate 3-D air quality model treatment of three primary features of the real atmosphere: faithful numerical simulation of the relevant photochemistry, accurate description of meteorological transport and mixing, and the use of appropriate emissions inventories as inputs to the model. Prior to 2000, model simulations of ozone formation in Houston, Texas had not been able to reproduce the very high levels of ozone characteristic of, and unique to, that urban center. This marked discrepancy raised serious doubts about the photochemistry incorporated into the current generation of 3-D models, and the ability of these models to reproduce the meteorology on relevant spatial scales.

The Texas Air Quality Study of 2000 provided sufficient data from ground and aircraft platforms to evaluate this discrepancy. Analysis of the available data showed the highest ozone exceedances, with 1-hour averages in excess of 150 ppbv, were always associated with spatially narrow plumes originating from the numerous petrochemical industrial facilities in the Houston area. Further analyses strongly suggests these exceedances arise from very rapid and highly efficient ozone production from the OH-initiated oxidation of light alkenes, primarily ethene and propene, emitted in abundance from many of the industrial facilities in the area. The aircraft data also suggested that tabulated emissions of these light alkenes were consistently and substantially under-reported by many of the industrial facilities, by factors of 20 to nearly 100.

Most important, 3-D models were able to better simulate the observed enhancements in petrochemical plumes after the input inventories were adjusted to reflect the greatly enhanced industrial light alkene emissions inferred from aircraft observations. Most current models simulate the light alkene oxidation pathways explicitly, so the model photochemical representations are sufficient. While model grid size is important in simulating the observed ozone, the original lack of accurate petrochemical emission inventories was the primary handicap in successfully modeling the Houston urban airshed. The Texas 2000 field study illustrates the importance of reconciling top-down with bottom-up emissions inventories, and of evaluating existing inventories against ambient measurements, to build confidence in 3-D model representations of the atmosphere.