Wednesday 15 January 2014, 1-2pm
Speaker: Prof. Michael Prather, Earth System Science, University of California, Irvine
Location: Training rooms 1&2, Gordon Manley Building
One would not choose global atmospheric chemistry models to study air quality on a metropolitan scale, yet we need them to assess change in air quality at these scales when global changes are occurring in composition and climate. This study seeks metrics that allow direct comparison of surface air quality monitoring stations with global chemistry-transport models (CTMs) on an equal footing. Comparison of point-based station measurements to modeled grid-cell values, while done for most CTMs, is fundamentally unjustified and described in the literature as incommensurability. It has been addressed in part through interpolation techniques such as Kriging or inverse-distance weighting, but is still regarded as an unresolved problem. We use these methods in a new approach that seeks to calculate a model grid-cell averaged value based on station data for surface O3 from all US and EU air quality monitoring sites, thus allowing direct comparison with CTM simulations and providing clear statistical information on model errors under different conditions. These diagnostics provide insight into the source of error as well as clear metrics of any model improvements. Over the 21st century, we anticipate that (i) local/regional emissions will change in response to air quality mitigation efforts; (ii) atmospheric composition will change globally; and (iii) meteorological conditions contributing to the worst pollution episodes will shift with climate. In addressing point (iii), we need to identify and characterize extreme air pollution episodes, which we define in a climatological manner as the 10 worst days in a year. A clustering algorithm is used to construct multi-day, multi-cell air pollution episodes. The model has considerable skill in hindcasting these episodes for 2004-2005. About 75% of the individual cell events occur in coherent, multi-day, connected episodes covering areas greater than 1000 x 1000 square km, and model skill is greatest with these large episodes. These tests are specific to exact-day matches and cannot be used for free-running chemistry-climate models. Even the traditional approach of using the probability distribution is interpreted incorrectly when used as climatology. The use of a fixed return time in a future climate has problems given changing baseline O3 levels. We investigate the climate statistics of air pollution episode size from a decade of observations and show that his may provide a clear measure of how climate change may alter the meteorology of extreme pollution episodes.
Michael Prather has been a Professor of Earth System Science at UCI since 1992 (Fred Kavli Endowed Chair 2002-2013), is a Fellow of the AGU and AAAS, Member of the Norwegian Academy of Science and Letters, and was a Jefferson Science Fellow at the U.S. Department of State in 2005-2006. He has been a lead author on UNEP/WMO Ozone Assessments 1985-1994, was Editor-in-Chief of GRL from 1997-2001, and he has been a Lead Author on the last four IPCC Assessment Reports from 1994-2013.