Spatio-Temporal Modeling and the Operational Evaluation of Air Quality Models
Mon Mar 31 17:57:13 PDT 1997

Paul D. Sampson, Peter Guttorp, Don Percival, Sharon LeDuc

The aim of this proposal is the development and demonstration of methods for modeling and analysis of spatio-temporal data for the primary purpose of assessing air quality model predictions against field monitoring data. However, much of the methodology is directly relevant to other purposes, particularly the analysis of air quality data for the health effects research project to be directed by Lianne Sheppard. The background for the current proposal is summarized in the original NRCSE proposal, sections 4.1.1, 4.1.2, 4.1.3, and 4.4.3 (and the associated Proposed Research aims). The essentials of these sections are reproduced below with modification and elaboration of our specific plans. First we list here a brief summary of our tasks. We intend to:

1. Refine and extend our current methods for modeling the nonstationary spatio-temporal correlation structure of hourly ozone monitoring data for application to the entire SARMAP database for the San Joaquin Valley (only subsets of these data for limited spatial domains have been modeled to date);
2. Develop wavelet-based methods for estimating spatial and temporal trends (trends which are subtracted in order to model the spatio-temporal correlation structure of ``residuals'' and which must be spatially interpolated for the estimation of task 3);
3. Compute (areal) grid cell estimates for comparison with SARMAP Air Quality Model (SAQM) output. This estimation will incorporate geostatistical trend estimates previously computed, but may consider also wavelet-based trend estimates, depending on the timing of progress under task 2;
4. Develop methods for fitting our deformation model for nonstationary spatio-temporal correlation (task 1) to gridded model output data (in contrast to point field monitoring observations); and
5. Apply these methods, in cooperation with EPA scientists, to the evaluation of (``Models-3'') model predictions over larger spatial and temporal domains in the Eastern U.S.

• Operational and Diagnostic Assessment of Air Quality Models
  • Estimation of Spatial and Spatio-temporal Covariance/Correlation
  • Assessing Spatial and Temporal Trends
• Proposed research
• Funding request
• References
• About this document ...