CICERO - Center for International Climate Research

Technical note: Quality assessment of ozone reanalysis products and gap-filling over subarctic Europe for vegetation risk mapping

Stefanie Falk, Ane Victoria Vollsnes, Aud Else Berglen Eriksen, Frode Stordal, Terje K. Berntsen

We assess the quality of regional and global ozone reanalysis data for vegetation modeling and ozone (O3) risk mapping over subarctic Europe where monitoring is sparse. Reanalysis data can be subject to systematic errors originating from, for example, quality of assimilated data, distribution and strength of precursor sources, incomprehensive atmospheric chemistry or land–atmosphere exchange, and spatiotemporal resolution. Here, we evaluate two selected global products and one regional ozone reanalysis product. Our analysis suggests that global reanalysis products do not reproduce observed ground-level ozone well in the subarctic region. Only the Copernicus Atmosphere Monitoring Service Regional Air Quality (CAMSRAQ) reanalysis ensemble sufficiently captures the observed seasonal cycle. We also compute the root mean square error (RMSE) by season. The RMSE variation between (2.6–6.6) ppb suggests inherent challenges even for the best reanalysis product (CAMSRAQ). O3 concentrations in the subarctic region are systematically underestimated by (2–6) ppb compared to the ground-level background ozone concentrations derived from observations. Spatial patterns indicate a systematical underestimation of ozone abundance by the global reanalysis products on the west coast of northern Fennoscandia. Furthermore, we explore the suitability of CAMSRAQ for gap-filling at one site in northern Norway with a long-term record but not belonging to the observational network. We devise a reconstruction method based on Reynolds decomposition and adhere to recommendations by the United Nations Economic Commission for Europe (UNECE) Long-Range Transboundary Air Pollution (LRTAP) convention. The thus reconstructed data for 2 weeks in July 2018 are compared with CAMSRAQ evaluated at the nearest-neighbor grid point. Our reconstruction method's performance (76 % accuracy) is comparable with CAMSRAQ (80 % accuracy), but diurnal extremes are underestimated by both.

More details