ML-TOMCAT: machine-learning-based satellite-corrected global stratospheric ozone profile data set from a chemical transport model

[摘要] High-quality stratospheric ozone profile data sets are a keyrequirement for accurate quantification and attribution of long-termozone changes. Satellite instruments provide stratospheric ozoneprofile measurements over typical mission durations of 5–15 years.Various methodologies have then been applied to merge and homogenisethe different satellite data in order to create long-termobservation-based ozone profile data sets with minimal data gaps.However, individual satellite instruments use different measurementmethods, sampling patterns and retrieval algorithms which complicatethe merging of these different data sets. In contrast, atmosphericchemical models can produce chemically consistent long-term ozonesimulations based on specified changes in external forcings, but theyare subject to the deficiencies associated with incompleteunderstanding of complex atmospheric processes and uncertainphotochemical parameters. Here, we use chemically self-consistent output from the TOMCAT 3-Dchemical transport model (CTM) and a random-forest (RF) ensemblelearning method to create a merged 42-year (1979–2020) stratosphericozone profile data set (ML-TOMCAT V1.0). The underlying CTMsimulation was forced by meteorological reanalyses, specified trendsin long-lived source gases, solar flux and aerosol variations. The RFis trained using the Stratospheric Water and OzOne SatelliteHomogenized (SWOOSH) data set over the time periods of the MicrowaveLimb Sounder (MLS) from the Upper Atmosphere Research Satellite (UARS)(1991–1998) and Aura (2005–2016) missions. We find that ML-TOMCATshows excellent agreement with available independent satellite-baseddata sets which use pressure as a vertical coordinate (e.g. GOZCARDS,SWOOSH for non-MLS periods) but weaker agreement with the data setswhich are altitude-based (e.g. SAGE-CCI-OMPS, SCIAMACHY-OMPS). Wefind that at almost all stratospheric levels ML-TOMCAT ozoneconcentrations are well within uncertainties of the observational datasets. The ML-TOMCAT (V1.0) data set is ideally suited for theevaluation of chemical model ozone profiles from the tropopause to 0.1 hPa and is freely available via https://doi.org/10.5281/zenodo.5651194 ( Dhomse et al. ,  2021 ) .

[学科分类] 眼科学