Assessing multi-year changes in modeled and observed daily maximum 8-hour ozone with a dynamic evaluation approach

Since photochemical air quality models are being applied in regulatory settings to determine how particular emission control strategies impact ozone (O3) air quality, it is critical that a model's response to emission changes is investigated and better understood. This dynamic evaluation approach, an emerging technique being employed as part of a comprehensive model evaluation effort, is specifically designed to assess a model's ability to reproduce observed changes in pollutant concentrations that could be attributed to changes in emissions and variability in meteorological conditions.

In this study, changes in the maximum daily average 8-hour (MDA8) ozone concentrations from 3 summer (June, July, August) months spanning 5 consecutive years (2002-2006) from CMAQ (Community Multiscale Air Quality) model results and measurements at CASTNet (Clean Air Status and Trends Network) monitoring sites are investigated. Substantial NOx emissions reductions (close to 40%) occurred during 2003-2004 in the major point source sector, primarily power plants, in the eastern US due to the implementation of the U.S. Environmental Agency's NOx SIP (State Implementation Plan) Call program. Additionally, estimated mobile NOx emissions exhibited a steady decline of about 5% per year. Consequently, this multi-year period serves as a valuable case study for applying dynamic evaluation to test how well the model's O3 response compares to observed changes due to these notable NOx precursor emission reductions as well as to meteorological variability that took place over these periods.

The CMAQv4.7 modeling system using the CB-05 (Carbon Bond 2005) chemical mechanism was applied with a 12-km grid cell size and 24 vertical layers on a regional domain covering all states and Canadian provinces east of the Rocky Mountains. Hourly meteorological fields were generated by the MM5 meteorological model and hourly gridded emissions inputs were developed by the SMOKE emissions processing system based on the 2002 NEI (National Emissions Inventory) with updates including county-specific MOBILE6 on-road NOx emissions and hourly CEMS (Continuous Emissions Monitoring Systems) data at major NOx point sources from each summer period. Graphical results of modeled gridded fields of the MDA8 O3 at the 95th percentile reveal broad areas displaying decreases of up to 20% between 2002 and 2006 across the Ohio River Valley region and various eastern states with much smaller changes in the southern and southwestern portions of the domain. Using summer 2002 MDA8 O3 results as a reference, comparisons of relative (%) change reveal the model results closely track the pattern of the overall observed decreases in MDA8 O3 from summer-to-summer. Modeled and observed percentage changes of MDA8 O3 concentrations in groups of individual locations from different sections of the domain are also analyzed to examine how well the model captures spatial variations across the region.