Secondary Logo

Journal Logo

ISEE 2007 CONFERENCE ABSTRACTS SUPPLEMENT: Abstracts

Melding Measurements and Models to Enrich the Study of Climate, Air Quality, and Health

Kinney, P*; Knowlton, K*; Hogrefe, C; Chiou, C*; Goldberg, R; Rosenzweig, C

Author Information
doi: 10.1097/01.ede.0000276720.73668.53
  • Free

ISEE-566

Objective:

Ambient measurements of air pollution and temperature tend to be relatively rich in time but poor in space. Air quality models provide detail in both time and space, but often lack accuracy. The objective of this study was to combine observed and modeled data for PM, ozone, and meteorology to develop highly resolved space-time data fields over a large region, and to evaluate whether these exposure fields offer information gains over simple observations in the context of an epidemiologic analysis of acute effects on daily mortality.

Materials and Methods:

Air quality and weather observations were taken from all sites in operation between 1988 and 2002 in New York, New Jersey, and Connecticut. Observations were first interpolated on a daily basis using kriging to 92 county centroids over the region. Next, we used the MM5 regional climate model and the CMAQ regional air quality model to simulate hourly meteorology and air quality at each county centroid, nudged at 6-hour time steps by observed weather data. The observed, kriged, and modeled exposure data sets are then used in time series analyses to estimate the independent and joint effects of air pollution and heat events on acute mortality risk, and to evaluate whether the kriged or fully-modeled exposure field change the magnitude or precision of the risk estimates.

Results:

Only 19 of 92 counties had sufficient monitoring data for inclusion in initial time series analyses based on observed data. Analyses limited to the observed data demonstrated significant effects of both ozone and heat waves on acute mortality. Analyses using the daily kriged and modeled data for 92 counties are ongoing.

Conclusions:

Melding of monitored and modeled data offers a potentially valuable tool for exposure assessment in epidemiology studies.

© 2007 Lippincott Williams & Wilkins, Inc.