Fine particulate matter modeling in Central California, Part I: Application of the Weather Research and Forecasting model

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 26 January 2011: 4:30 PM
Fine particulate matter modeling in Central California, Part I: Application of the Weather Research and Forecasting model
3A (Washington State Convention Center)
Raphael Rogers, Penn State University, University Park, PA; and A. Deng, D. Stauffer, Y. Jia, S. T. Soong, S. Tanrikulu, S. Beaver, and C. Tran
Manuscript (2.0 MB)

Fine particulate matter (PM2.5) is deleterious to human health. Its public health impacts may well exceed the combined impacts of all other currently regulated pollutants in the United States. In 2008, the United States Environmental Protection Agency (US EPA) proposed the attainment designations for PM2.5 under the National Ambient Air Quality Standards (NAAQS) of the Clean Air Act. Attainment status has two components: daily (24-hr) and annual average ambient PM2.5 levels.

In Central California, five different non-attainment areas for the 24-hr PM2.5 standard have recently been designated by the EPA, including the San Francisco Bay Area (SFBA). As a result, the Bay Area Air Quality Management District (BAAQMD) is required to develop a State Implementation Plan (SIP) tentatively scheduled for 2013. The SIP development process will largely be guided by results obtained from current scientific studies. They include the analysis of ambient measurements, emissions inventory development and air quality modeling.

Nearly all SFBA exceedances of the daily NAAQS occur during the winter season (November-February). Episodes mostly develop under: stable atmospheric conditions inhibiting vertical dispersion; clear and sunny skies favoring enhanced secondary PM2.5 formation; and pronounced overnight drainage (downslope) flows off the Central Valley rims, causing low-level air in the Central Valley to empty through the Delta and into the SFBA along its eastern boundary. Atmospheric transitions of aloft weather systems profoundly influence the surface winds that determine PM2.5 levels. Surface conditions stagnate when an aloft high pressure system moves over Central California. Persisting high pressure conditions allow PM2.5 to accumulate to the exceedance level in the SFBA, typically after 2-4 days.

This abstract is Part I of a two-part series presenting preliminary modeling efforts of BAAQMD. It describes meteorological modeling using the Weather Research and Forecasting (WRF) model. Part II (Tanrikulu et al., 2010) describes the emissions inventory development and air quality modeling using the Community Multiscale Air Quality (CMAQ) model. Central California is a large mountain-valley area with complex terrain, air flow features and air quality patterns. The air quality modeling domain is likewise large. In addition to the SFBA, the domain includes the entire Central Valley (CV) comprising the Sacramento Valley (SV) to the north and the San Joaquin Valley (SJV) to the south. Outlying areas within the modeling domain include the areas over Pacific Ocean, coastal locations along the Coast Range, and the inland Sierra Nevada.

The BAAQMD has been using MM5 for meteorological modeling and is currently developing its capabilities in using the WRF model. To assist the District in the transition from MM5 to WRF, the advanced research version of WRF (WRF-ARW) was applied over Central California. WRF simulations were conducted to reproduce the patterns of low-level air flows in the SFBA and the CV. The coupling between these basins is relevant to the transport and dispersion of the air pollutants in the region. Evaluation of the WRF results was mostly focused over the SFBA for selected cases to determine the best model configuration for the region.

Three nested modeling domains were used in the meteorological model. They had 36-, 12- and 4-km horizontal grid resolutions, respectively. WRF simulated 50 vertical layers. The bottom layer was approximately 20 m thick.

Several iterations of model application, evaluation and improvement were conducted to obtain the best performance from the WRF model. For this purpose, multiple simulations were conducted for the same five-day simulation period. The period used for model testing exhibited conditions that were representative of SFBA winter-season PM2.5 episodes. First a baseline model was configured using the authors' experience with the WRF model and the Central California domain. Then the background Eta analysis was used as input. Subsequent simulations additionally incorporated meteorological observations into the initial and lateral boundary conditions. Next, a land surface model (LSM) was used to improve WRF performance. A benchmark was established using the tested and optimized model setup. Further improvements were made with the use of four dimensional data assimilation (FDDA). Performance improvements from FDDA were assessed by comparison of WRF outputs against the benchmark simulation and observations.

The results showed that the best simulation for reproducing the meteorological patterns that impact air quality in the region used the multiscale FDDA, combining analysis and observational nudging in the nested grid framework. A mesoscale analysis using special surface and upper-air data was used to further examine the details in the observed versus simulated air flows. WRF model results, including both statistical verification and subjective analysis, will be presented with special attention given to the mesoscale features in the SFBA and CV. The core winter season dates 1 December through 31 January were simulated for 2000-01 using the optimum model configuration. The model was ultimately evaluated against the rich database of the 1999-2001 California Regional Particulate Air Quality Study (also known as CRPAQS).

References: Tanrikulu, S., Beaver, S., Tran, C., Soong, S.-T., Jia, Y., Rogers, R., Deng, A., Stauffer, D., 2010. Fine Particulate Matter Modeling in Central California, Part II: Application of the Community Multiscale Air Quality Model. Abstract submitted to AMS 2011 Annual Meeting, 13th Conference on Atmospheric Chemistry.