Gridded global surface ozone metrics for atmospheric chemistry model evaluation

Abstract

Abstract. The concentration of ozone at the Earth's surface is measured at many locations across the globe for the purposes of air quality monitoring and atmospheric chemistry research. We have brought together all publicly available surface ozone observations from online databases from the modern era to build a consistent dataset for the evaluation of chemical transport and chemistry-climate (Earth System) models for projects such as the Chemistry-Climate Model Initiative and Aer-Chem-MIP. From a total dataset of approximately 6600 sites and 500 million hourly observations from 1971–2015, approximately 2200 sites and 200 million hourly observations pass screening as high-quality sites in regional background locations that are appropriate for use in global model evaluation. There is generally good data volume since the start of air quality monitoring networks in 1990 through 2013. Ozone observations are biased heavily toward North America and Europe with sparse coverage over the rest of the globe. This dataset is made available for the purposes of model evaluation as a set of gridded metrics intended to describe the distribution of ozone concentrations on monthly and annual timescales. Metrics include the moments of the distribution, percentiles, maximum daily eight-hour average (MDA8), SOMO35, AOT40, and metrics related to air quality regulatory thresholds. Gridded datasets are stored as netCDF-4 files and are available to download from the British Atmospheric Data Centre (doi:10.5285/08fbe63d-fa6d-4a7a-b952-5932e3ab0452). We provide recommendations to the ozone measurement community regarding improving metadata reporting to simplify ongoing and future efforts in working with ozone data from disparate networks in a consistent manner.

Comment on gmd-2021-39

“…The error may be reduced by averaging measurements made at different stations within a model grid box, although atmospheric measurements may be too sparse to permit this (Schultz, 2016). The representation error is sometimes taken as the mean of the spatial standard deviation of different measurements within a grid-box (Sofen et al 2016). However, this measure quantifies the spatial variability of measured O3 within a grid-box and may not match the representation error To test the effect on parameter estimates of varying this representation error, we use synthetic data from the control run of the model using parameters set to their nominal default https://doi.org/10.5194/gmd-2021-39 Preprint.…”