« Evaluation of Surface Ozone over Europe and USA: Present-day observations & hindcasts, future projections from chemistry-climate models. »
By Michael Prather, University of California, Irvine, Visiting Scientist at Cerfacs
Most large-scale atmospheric chemistry models do not know how to deal with surface measurements of ozone and fail to recognize the inherent mistake in comparing a single-point measurement with a modeled value that represents the average of a grid cell. Many papers argue that a ‘remote' or ‘clean air' site is representative of what the model predicts, regardless of the model resolution. Here we present a totally new, testable and verifiable approach for generating grid-cell average values from the USA and European surface air quality network of about 2000 stations in each domain. This mapping of many points to an area-average is shown to be robust and even shows that cells with higher ‘quality' indices are better forecast in the re-simulation (hindcast) with the UC Irvine chemistry-transport model (CTM). Our method allows non-coincident but overlapping networks to be compared (EMEP vs AirBase). This new dataset allows for rigorous and fair testing of the diurnal and seasonal cycle of the global Chemistry-Climate Models (CCMs) with resolutions from 1-degree to 3-degrees. Air quality extreme (AQX) events are identified locally as statistical extremes of the ozone climatology and not as air quality exceedances. With the UCI CTM there is skill in hindcasting these AQX episodes, and thus identify an additional diagnostic that we use to study how AQX episodes in a warming climate are projected to change based on the CCM simulations from 2000 to 2100 in a rapidly warming (4C) climate.