Improved Correlations Between Satellite Aerosol Measurements and PM2.5 Air Quality Measurements in the San Joaquin Valley
Anthony W. Strawa (1), Robert B. Chatfield (1), Marion J. Legg (2), Barbara V. Scarnato (3), PATRICK W. HILLYARD (2)
(1)NASA Ames Research Center, (2)Bay Area Environmental Research Institute, (3)Oak Ridge Associated Universities
Abstract Number: 194
Preference: Poster Presentation
Last modified: April 28, 2010
Working Group: Urban Aerosols
Surface air quality monitoring is traditionally done by point measurements at localized ground sites. Satellite measurements of aerosol properties can potentially augment surface measurements, given their better spatial coverage. This spatial coverage provides the advantages of seeing pollutant levels in remote and non-monitored areas, tracking pollution transport, validating and guiding models prediction, suggesting the placement of future surface sensors, and providing an additional metric in epidemiological studies.
The major problem in augmenting traditional air quality measurements with satellite data is the variability in correlation between surface PM2.5 measurements and satellite aerosol optical depth. The variability is highly location specific and dependent on the seasons. Correlations between 0.2 and 0.98 have been obtained. In some cases, notably in the western US, there is little or no correlation using simple linear regression to match AOD to PM (1,2). Our group at NASA Ames has sought alternative methods by employing generalized additive models (GAMs) aimed at improving the correlation between these two quantities. Our methodology is also unique because it uses only remote observations from several satellites to improve the correlation. The method has resulted in improvements in correlation coefficients between surface PM2.5 and satellite-derived PM2.5 in the Fresno area from 0.4 for a linear model to 0.73 using the GAM. Results for the Bakersfield site were even more dramatic, with correlation coefficients increasing from 0.1 with the linear model to 0.8 with the GAM (3). This presentation will discuss the methodology, results, limitations and potential future applications.
1. J. Al-Saadi et al., Bull. Amer, Met. Soc., 1249 (2005).
2. J. A. Engle-Cox, C. H. Holloman, B. W. Coutant, R. M. Hoff, Atmos. En. 38, 2495 (2004).
3. A.W. Strawa et al., "Toward Obtaining Reliable Particulate Air Quality from Satellites", Int'l Aerosol Modeling Algorithms Conf., Davis, CA, (2009).