There is evidence that the ozone layer has begun to recover owing to the ban on the production of halogenated ozone-depleting substances (hODS) accomplished by the Montreal Protocol and its amendments and adjustments (MPA). However, recent studies, while reporting an increase in tropospheric ozone from the anthropogenic NOx and CH4 and confirming the ozone recovery in the upper stratosphere from the effects of hODS, also indicate a continuing decline in the lower tropical and mid-latitudinal stratospheric ozone. While these are indications derived from observations, they are not reproduced by current global chemistry–climate models (CCMs), which show positive or near-zero trends for ozone in the lower stratosphere. This makes it difficult to robustly establish ozone evolution and has sparked debate about the ability of contemporary CCMs to simulate future ozone trends. We applied the new Earth system model (ESM) SOCOLv4 (SOlar Climate Ozone Links, version 4) to calculate long-term ozone trends between 1985–2018 and compare them with trends derived from the BAyeSian Integrated and Consolidated (BASIC) ozone composite and MERRA-2, ERA-5, and MSRv2 reanalyses. We designed the model experiment with a six-member ensemble to account for the uncertainty of the natural variability. The trend analysis is performed separately for the ozone depletion (1985–1997) and ozone recovery (1998–2018) phases of the ozone evolution. Within the 1998–2018 period, SOCOLv4 shows statistically significant positive ozone trends in the mesosphere, upper and middle stratosphere, and a steady increase in the tropospheric ozone. The SOCOLv4 results also suggest slightly negative trends in the extra-polar lower stratosphere, yet they barely agree with the BASIC ozone composite in terms of magnitude and statistical significance. However, in some realizations of the SOCOLv4 experiment, the pattern of ozone trends in the lower stratosphere resembles much of what is observed, suggesting that SOCOLv4 may be able to reproduce the observed trends in this region. Thus, the model results reveal marginally significant negative ozone changes in parts of the low-latitude lower stratosphere, which agrees in general with the negative tendencies extracted from the satellite data composite. Despite the slightly smaller significance and magnitude of the simulated ensemble mean, we confirm that modern CCMs such as SOCOLv4 are generally capable of simulating the observed ozone changes, justifying their use to project the future evolution of the ozone layer.@en

Reliable ozone trends after 2000 are essential to detect early ozone recovery. However, the long-term ground-based and satellite ozone profile trends reported in the literature show a high variability. There are multiple reasons for variability in the reported long-term trends such as the measurement timing and the dataset quality.

The Payerne Switzerland microwave radiometer (MWR) ozone trends are significantly positive at 2 % to 3 % per decade in the upper stratosphere (5–1 hPa, 35–48 km), with a high variation with altitude. This is in accordance with the Northern Hemisphere (NH) trends reported by other ground-based instruments in the SPARC LOTUS project. In order to determine what part of the variability between different datasets comes from measurement timing, Payerne MWR and SOCOL v3.0 chemistry–climate model (CCM) trends were estimated for each hour of the day with a multiple linear regression model. Trends were quantified as a function of local solar time (LST). In the middle and upper stratosphere, differences as a function of LST are reported for both the MWR and simulated trends for the post-2000 period. However, these differences are not significant at the 95 % confidence level. In the lower mesosphere (1–0.1 hPa, 48–65 km), the 2010–2018 day- and nighttime trends have been considered. Here again, the variation in the trend with LST is not significant at the 95 % confidence level. Based on these results we conclude that significant trend differences between instruments cannot be attributed to a systematic temporal sampling effect.

The dataset quality is of primary importance in a reliable trend derivation, and multi-instrument comparison analyses can be used to assess the long-term stability of data records by estimating the drift and bias of instruments. The Payerne MWR dataset has been homogenized to ensure a stable measurement contribution to the ozone profiles and to take into account the effects of three major instrument upgrades. At each instrument upgrade, a correction offset has been calculated using parallel measurements or simultaneous measurements by an independent instrument. At pressure levels smaller than 0.59 hPa (above ∼50 km), the homogenization corrections to be applied to the Payerne MWR ozone profiles are dependent on LST. Due to the lack of reference measurements with a comparable measurement contribution at a high time resolution, a comprehensive homogenization of the sub-daily ozone profiles was possible only for pressure levels larger than 0.59 hPa.

The ozone profile dataset from the Payerne MWR, Switzerland, was compared with profiles from the GROMOS MWR in Bern, Switzerland, satellite instruments (MLS, MIPAS, HALOE, SCHIAMACHY, GOMOS), and profiles simulated by the SOCOL v3.0 CCM. The long-term stability and mean biases of the time series were estimated as a function of the measurement time (day- and nighttime). The homogenized Payerne MWR ozone dataset agrees within ±5 % with the MLS dataset over the 30 to 65 km altitude range and within ±10 % of the HARMonized dataset of OZone profiles (HARMOZ, limb and occultation measurements from ENVISAT) over the 30 to 65 km altitude range. In the upper stratosphere, there is a large nighttime difference between Payerne MWR and other datasets, which is likely a result of the mesospheric signal aliasing with lower levels in the stratosphere due to a lower vertical resolution at that altitude. Hence, the induced bias at 55 km is considered an instrumental artifact and is not further analyzed.@en