Seasonal cycle of the total ozone content over Southern high latitudes in the CCM SOCOLv3.

This dataset is the output from simulations conducted using the SOCOLv3 model. We use it to study sensitivity of the TCO over Antarctica to the; (1) efficiency of the stratospheric heterogeneous reactions (SR1), and (2) intensity of the meridional flux into the polar regions due to sub-grid scale mixing processes (SR2) and (3) accuracy of the pho-to-dissociation rate calculations. (SR3). The data includes the following output variables: total ozone, ozone (O3), methane (CH4), nitrous oxide (N2O), hydroxyl (OH), ClOx family, O(1D). The data is stored in NetCDF format. Zip file contains the output for four experiments, described below. 1. Reference run (RR) To compare the CCM SOCOLv3 original calculation results with the IKFS2 measurements we have undertaken a 19-year-long reference numerical experiment for the 2000-2018 period. The first 14 years (2000-2013) of the model calculations were considered as spin up. It is necessary for the adaptation of the chemical composition and dynamics of the model atmosphere to the specific boundary conditions. The last 5 years of the model experiment (2014-2018) were used for analysis and comparison with the correspondent IKFS2 observations. As boundary conditions, we considered the long-term (2000-2018) evolution of the ozone depletion substance (ODS), greenhouse gas concentrations GHG, sea surface temperature and sea ice concentration (SST/SIC), and zonal winds in the equatorial stratosphere (QBO). The mixing ratios of ODS in the lower troposphere evolved according to the World Meteorological Organization (WMO) data (WMO, 2018). The atmospheric mixing ratios of the main GHG (CO2, CH4, and N2O) are taken from (Meinshausen et al., 2017) until 2014 and extended to 2018 following the IPCC SSP2-4.5 scenario (Meinshausen et al., 2020). The SST/SIC fields for the 21st century prescribed as monthly means are adopted from the HadISST1 dataset provided by the UK Met Office Hadley Centre (Rayner et al., 2003). The QBO is produced by a linear relaxation (“nudging”) of the model zonal winds in the equatorial stratosphere to a time series of observed winds (28, Giorgetta et al., 1996). The nudging is used between 20o N and 20o S from 90 hPa up to 3 hPa. Within the QBO core domain (10o N–10o S, 50– 8 hPa) the relaxation time is uniformly set to 7 days; outside this region, the damping depends on latitude and altitude (Giorgetta et al., 2006). Also, the evolutions of the 11-year solar activity (Matthes et al., 2017), stratospheric aerosol contents (Kovilakam et al., 2020), and the surface CO and NOx emissions from CMIP6 input4MIPs databases for the historical period to 2014 and following RCP2-4.5 until 2018 (Hoesly et al., 2018) were applied in the model runs as boundary conditions. 2. Sensitivity runs. The above-mentioned TCO discrepancies between model and satellite data probably can be explained by the poor model representations of the processes that are responsible for the polar ozone state. According to the modern scientific paradigm, the ozone content in the polar stratosphere is controlled mainly by the rates of the heterogeneous reactions and the photodissociation of ozone molecules by solar radiation at the large zenith angles of the Sun (Brasseur and Solomon, 2005). Also, the level of TCO inside the inner part of the SH polar night vortex can depend on the horizontal resolution of the model (Shuhua et al., 2002). So, if the model grid is rather rough it makes sense to consider the transport of the model species into the polar night vortexes by the sub-grid scale motions. Table 1. A short description of the sensitivity runs with CCM SOCOLv3. Name of model run Short description of the model run SR1 Two times reduction of the heterogenies reaction HCl + ClONO2 → Cl2 + HNO3 rate. SR2 Intensification of the horizontal mixing process of all transported model species into the SH polar vortex with diffusion coefficient Kyy = 5·106 m2/s in its maximum. SR3 Using the Cloud-J module installed in the SOCOLv3 model to calculate photolysis rates Therefore, to investigate the causes of the TCO underestimation and find some reasonable refinement of the model accuracy, we performed set of additional numerical runs with CCM SOCOLv3 described in the Table 1. All these model runs were designed exactly as the reference run (boundary conditions, spin-up, using the 2016-2018 years model results for comparison with the satellite data).