A Holistic Evaluation of the Houston, Texas Ozone Attainment Episode

In early 2006, the Texas Commission on Environmental Quality (TCEQ) released a simulation episode to form the basis for its 8-Hour Ozone State Implementation Plan (SIP). The SIP details steps to be taken to bring the Houston-Galveston-Brazoria region into attainment of the National Ambient Air Quality Standard for ozone (O3). The modeling episode will have significant impacts on regulatory decision-making, thus affecting the regional economy and human health. At present model performance evaluations (MPEs) are based on guidelines set by the Environmental Protection Agency (EPA) that rely heavily on statistical measures. While the episode falls within the recommended range for these EPA statistical metrics, examining the level of agreement between model predictions and observations for O3 is not a sufficient method to assess model accuracy. We have discovered severe inaccuracies in the 8-Hour modeling episode with an evaluation methodology that recognizes that (1) the formation of tropospheric O3 system is non-linear with respect to its precursors and (2) is the result of a multitude of chemical, emission, and meteorological processes. This paper presents the results of this improved evaluation method using new tools and software developed at the University of North Carolina at Chapel Hill. These tools were designed specifically to compensate for inadequacies in the existing statistical methods currently used for MPE. These tools enable investigators to review O3 production and concentrations in the context of meteorological and chemical conditions, allowing more holistic analytical techniques, not captured by traditional statistical metrics. In our analysis all modeling results are compared with high-resolution data from multiple sources, including an extensive ground-monitoring network, automatic gas-chromatographs (auto-GCs), and aircraft. First in our investigation, as a measure of progress, the latest 8-Hour episode is compared to TCEQ"s previous 1-Hour O3 SIP modeling episode (2003). We then apply 8-hour metrics in the analysis of the new TCEQ episode to investigate what impacts EPA's new 8-hour guidelines (2005) have on model evaluation. We contrast the EPA statistical measures of model performance with the day-by-day, site-by-site approach developed at UNC- CH. Results of the analysis indicate that, while meteorological modeling has improved somewhat, especially with predicted daytime wind-speeds, significant problems still exist with the formation of O3. Other problems include: wholesale overprediction of nitrogen oxides (NOx) and volatile organic compounds (VOCs); underprediction of formaldehyde; strange behavior of modeled CO levels; and excessive nighttime wind-speeds. Moreover, these problems are not adequately captured through the use of standard statistical metrics. Questions still remain about the adequacy of the model to predict winds aloft and our analysis would be better facilitated by more high-resolution carbon monoxide (CO) data at various vertical heights. The lack of formaldehyde, a source of radicals needed for O3 formation, could be the cause for suppressed ozone formation in the model and warrants further investigation. Current available observational data of this radical source, however, is sparse.