Operational Evaluation of Houston-Galveston Area Air Quality Forecasts with two Different Meteorological Inputs

To provide air quality guidance to the public routinely and to investigate causes of high air pollution events, the Institute for Multidimensional Air Quality Studies (IMAQS), University of Houston, is running an air quality forecasting (AQF) system for the Houston-Galveston area (HGA). The AQF results can be further used to build a long-term data base for the regional air quality to help state and local agencies for the regulatory applications. Accurate meteorological and photochemical modeling is essential to achieve these objectives. The IMAQS AQF system performs downscale meteorological modeling of the National Centers for Environmental Prediction (NCEP)'s ETA forecasts with the Mesoscale Meteorological Model version 5 (MM5), processes emissions input with the Sparse Matrix Operator Kernel Emissions (SMOKE), and simulates air quality with the U.S. EPA's Community Multiscale Air Quality (CMAQ) model. We analyzed the ozone forecasting results for Summer 2005. To understand the effects of different meteorological inputs on air quality forecasting, we utilized two different MM5 simulations, one with the 5-layer slab model and MM5-default land use and land cover (LULC) data and the other with modified NOAH LSM(land-surface model) and MRF PBL (Planetary Boundary Layer) scheme and the high resolution LULC obtained from LANDSAT multi-spectral data. Although there were day- and location-specific differences, the former in general predicted higher PBL heights and temperatures during daytime. But at nighttime, the latter showed higher temperatures. Also, the latter resulted in stronger wind speed than the former during both day and night times. The latter seemed to represent the temperature and wind conditions a little better than the former. Comparison of the ozone forecasts with the surface measurements in the HGA showed that the former results were better than the latter for the non-violation days. The latter predicted ozone peaks and locations better during the ozone violation days but showed high bias of the low to middle ozone concentrations. Some ot these problems may be due to the too high background ozone concentrations assumed for the Gulf of Mexico.