RAND Statistics Seminar Series

Data Assimilation for Epidemiologic Air Exposure Assessment

Presented by Ana G. Rappold, Ph.D., USEPA
Tuesday, October 14, 2008
10:30 a.m. – 12:00 p.m. PST
Conference Room 5312 & 1228
RAND Corporation, Santa Monica, CA
Please contact Denise Miller if you would like to attend this seminar.

Abstract

A substantial focus of air pollution epidemiologic research is the nature of adverse health effects associated with human exposure to particulate matter (PM). Currently, the assessment of population based daily exposure is limited by the spatial and temporal availability of the monitoring networks. However, due to recent advances in physical models such as EPA's Community Multiscale Air Quality model (CMAQ), potentially valuable air quality maps may become available. CMAQ is a deterministic model of atmospheric pollutant transport and provides volume averaged predictions on the grid. Although such model predictions are biased they contribute important spatial and temporal features of systems such as weather patterns, land use, emissions etc. The process of data assimilation involves several important statistical problems such as integration of areal and point data, modeling bias surfaces, and treatment of spatial and temporal similarities in data.

This work illustrates a fully Bayesian multivariate hierarchical state-space time series model for the joint distribution of the two data types. The model handles a “change of support” problem and provides insight on bias surfaces associated with CMAQ predictions. Time permitting, I will present results from the extension of the model which studies bias surfaces through time on the multi-resolution scale. Also available are estimates of missing daily values of particulate matter, complete spatial surfaces, and the associated uncertainty estimates, within a fully probabilistic framework. Posterior estimation was performed via a nested Gibbs algorithm where partial closed form solutions were obtained by an Information Filter. The posterior information can be readily incorporated with human health data in studies of temporal lag effects and spatial variation of adverse health effects.

Speaker Bio

Ana Rappold received her PhD from the Institute of Statistics and Decision Sciences, Duke University, in 2005. She spent several years as a Post Doc at EPA's National Exposure Research Laboratory and has recently joined the Human Studies Division of the EPA as a research statistician. Her research is focused on the health effects of air pollution.



Attending a Seminar

Other Locations/Times:
Washington, D.C. Conf. Rm. 4128: 1:30 p.m. ET
Pittsburgh Conf. Rm. 6206: 1:30 p.m. ET

RAND visitors are welcome to attend and must RSVP at least one day prior to the seminar. To ensure your attendance please contact Denise Miller at dmiller@rand.org with your name, company (or university) affiliation, and national citizenship (for security purposes).

For parking and directions to RAND's Santa Monica office, please see: http://www.rand.org/about/locations/santa-monica.html.

For parking and directions to RAND's Pittsburgh office, please see: http://www.rand.org/about/locations/pittsburgh.html.

For further information and to be added to the mailing list contact Denise Miller at dmiller@rand.org.