You are here:
DEVELOPMENT AND EVALUATION OF PM 2.5 SOURCE APPORTIONMENT METHODOLOGIES
Citation:
REFF, A. H. DEVELOPMENT AND EVALUATION OF PM 2.5 SOURCE APPORTIONMENT METHODOLOGIES. Presented at 2006 EPA Science Forum, Washington, DC, May 16 - 18, 2006.
Impact/Purpose:
The objectives of this task are to continue development and improvement of EPA's mesoscale (regional through urban scale) air quality modeling systems, such as the Community Multiscale Air Quality (CMAQ) model, as air quality management and NAAQS implementation tools. This task focuses on needed research and development of air quality models targeted for a major CMAQ model release in FY08. Model development for a broad scope of application is envisioned. For example, CMAQ will need to be able to simulate air quality feedbacks to meteorology and climate as well as intercontinental transport. The 2008 release of CMAQ is timed to coincide with EPA/OAR's and the states' needs for an improved model for assessments of progress (mid-course corrections) in the post-SIP submittal timeframe.
Description:
The receptor model called Positive Matrix Factorization (PMF) has been extensively used to apportion sources of ambient fine particulate matter (PM2.5), but the accuracy of source apportionment results currently remains unknown. In addition, air quality forecast models such as the Community Multi-Scale Air Quality (CMAQ) model do not currently posses the ability to adequately apportion sources of PM2.5 on a regional or national scale. The objectives of this project are to 1) develop a synthetic dataset of ambient PM species concentrations with known source contributions, 2) analyze this dataset with PMF to evaluate the accuracy of current PMF methods, 3) use this dataset as a basis explore new and more accurate procedures for the application of PMF to ambient PM data, and 4) compare source apportionment results obtained by "tagging" PM2.5 emissions in CMAQ to the results of applying the newly developed PMF methodologies to ambient PM2.5 data. Currently the creation of the synthetic dataset is underway, and preliminary exploration of PMF methods is ongoing.