Spatial Quantile Evaluation of Air Quality Models Using In Situ Observations

• Activity Number: 91
• Type: Invited
• Date/Time: Sunday, August 9, 2015 : 9:30 PM to 10:15 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Elizabeth Mannshardt* and Montserrat Fuentes and Soumendra Lahiri and Kristen Foley
• Companies: North Carolina State University and North Carolina State University and North Carolina State University and EPA
• Keywords: quantile function ; spatial ; data product validation

Abstract

The Air Quality System (AQS) network of monitoring stations provides a rich source of information about air pollution through in situ observations. However, station observations do not have full spatial and temporal coverage. Air quality models can be used to make inferences at unobserved locations. The Community Multiscale Air Quality System (CMAQ) allows for estimations across time and space, but is not as accurate as network sensors. Of interest is the comparison of modeled ozone output to observed ozone, in particular how the metrics calibrate across their respective quantile functions, across the tails of the distributions, and at points of interest such as values associated with ozone regulatory standards. We construct spatial quantile functions for observed ozone (AQS) versus modeled ozone (CMAQ) to evaluate the differences across the distributions. Findings from such evaluation studies can be directly relevant in regulatory decisions regarding primary and secondary standards for ozone