Seventeen years of ozone sounding at L’Aquila, Italy: evidence of mid-latitude stratospheric ozone recovery and tropospheric profile changes

Ozone profile measurements collected at L’Aquila (Italy, 42.4°N) during seventeen years of radio-sounding 30 (2000-2016) are presented here, with an analysis of derived trends. Model results from the SPARC-CCMI exercise are used in parallel to highlight the physical and chemical mechanisms regulating mid-latitude ozone trends. The statistically significant trends highlighted in time series at L’Aquila are those in the mid-upper stratosphere (+5.9±4.2), mid troposphere https://doi.org/10.5194/acp-2020-525 Preprint. Discussion started: 1 July 2020 c © Author(s) 2020. CC BY 4.0 License.

(+5.9±2.4) and upper troposphere (+2.5±0.9),all in percent/decade.The upper stratospheric positive trend was already well documented in recent WMO assessments and attributed to the starting decline of stratospheric Cly and Bry and to the stratospheric cooling induced by increasing well mixed greenhouse gases, thus slowing down gas-phase reactions that destroy ozone in the upper stratosphere.The ozone increase in the mid-upper troposphere is largely regulated by the increasing strength of the Brewer-Dobson circulation, which moves more ozone from the tropics to the extratropics and enhances the tropospheric influx from the lowermost stratosphere.This climate feedback mechanism on tropospheric ozone is only partially compensated by the increasing chemical ozone loss associated to higher H2O values in response to the tropospheric warming.We also note that ozone trends obtained in the lower stratosphere are negative (-2.2 percent/decade), but do not result to be statistically significant in our analyses.

Introduction
Recent ozone assessments (WMO, 2011;WMO, 2014) have proved that the declining emission of ozone-depleting substances (ODS) in compliance of the limitations imposed by the Montreal protocol has significantly decreased the chemical destruction of stratospheric ozone.This has resulted in a strong reduction in the negative ozone trends observed from 1970 to 1997 or even in the change of their sign, with emerging evidences of positive trends, at least in the upper stratosphere (globally, Chipperfield et al., 2017, and with particular regard to the Antarctic, Strahan and Douglass, 2018).
Coordinated model inter-comparison initiatives, as SPARC-CCMVal (SPARC-WCRP, 2010) and SPARC-CCMI (Eyring et al., 2013), made use of coupled chemistry-climate coupled models (CCMs) to better understand the complex interconnections between ozone photochemistry and radiative-dynamical processes in the middle atmosphere, as well as their evolution in time.These model inter-comparison projects have highlighted a robust scientific understanding of the driving mechanisms of ozone depletion.
CCMs demonstrated a significant impact of ozone changes on stratospheric heating rates and circulation changes and also showed how increasing greenhouse gases (GHGs) cool the stratosphere (Rind et al., 1990) and slow down homogeneous chemical reactions that deplete ozone in the upper stratosphere (Bekki et al., 2013;Marsh et al., 2016).Finally, the model simulations give a robust indication of a time-increasing strength of the Brewer-Dobson circulation (BDC) (Garcia and Randel, 2008;McLandress et al., 2010;Chipperfield et al., 2013;Polvani et al., 2018), producing an ozone decrease in the tropical lower stratosphere and an increase out of the tropics with respect to a control case with fixed GHGs.
The impact of an enhanced mid-latitude tropospheric influx from the lowermost stratosphere was discussed by Stevenson et al. (2006).In that case, model simulations were designed with fixed surface emissions of tropospheric ozone precursors and climate conditions projected for year 2030, in comparison with the standard case of year 2000 climate.They concluded that two climate feedbacks, i.e., tropospheric water vapor increase and enhanced stratospheric input, appear to be the dominant mechanisms operating on the budget of tropospheric ozone.The impact of stratospheric ozone on tropospheric ozone variability has been previously studied by Ordonez et al. (2007), Terao et al. (2009) and Hess and Zbinden (2011).
In this work we analyze the results of a time-series of ozone soundings taken in L'Aquila, Italy (42.3°N, 13.4° E) from the year 2000 to the year 2016 that have never been published before in the scientific literature.We first compare our results with available data from other ozonosondes in the same latitudinal band previously described and validated (Considine et al., 2008;Tilmes et al., 2012).We then derive linear trends for different partial columns from our collected data spanning from the Boundary Layer to the mid-upper stratosphere and compare them with available simulations from the CCMI project, trying to determine the causes of the upper tropospheric and lower stratospheric recovery by analyzing the differences between the historical simulations (refC2) and sensitivity simulations where GHGs are kept fixed at 1960 levels (fGHG).

Ozone sounding results and trend analysis
Since 1994, CETEMPS/DSFC at the of University of L'Aquila has operated an ozonesonde station (42.38N, 13.31E, site elevation 683m a.s.l.).In the period 2000-2016 the quality-checked balloon-borne ozone measurements conducted in troposphere and stratosphere were 295, with a pace of about 2 ozone-soundings per month; in some periods the observational activity was stopped or reduced for lack of funding (2002 and 2003), or because of a major earthquake (middle of 2009).The ozone profile measurements are carried out with an ElectroChemical Cell (ECC, 6A Science Pump Corporation®) coupled to pressure, temperature, relative humidity sensors and GPS receiver (Vaisala® RS80, RS90, RS91, and RS41-SG) with the Vaisala® MW11, DIGICORA III and MW41 sounding systems along the different periods; the flying platform is a rubber balloon (Totex® 1200 g) able to reach altitudes above 10 hPa.
In order to homogenize the data (ozone profiles), the ECC sensors preparation's procedure accurately follows the prescriptions about the sensor solution concentrations, the pump flow rate and the ECC background current measurements (Deshler et al., 2017).The vertical resolution of the raw ozone profile is better than 30 m, and the typical accuracies of pressure, temperature and relative humidity sensors are below 0.5 hPa, 0.3 K, and 2%, respectively.Concerning the ECC measurement uncertainties, following the procedure outlined in Witte et al. (2018), it is possible to state that the ozone partial pressure profile has an overall (uncorrelated) error between 4 and 12% in the troposphere, and about 4% in the altitude range between 100 and 10 hPa levels.The various sources of systematic errors, that are also altitude dependent, increase the total error to about ±12% (Komhyr et al., 1995;Witte et al., 2018).Uncertainties of the total column of ozone https://doi.org/10.5194/acp-2020-525Preprint.Discussion started: 1 July 2020 c Author(s) 2020.CC BY 4.0 License.
(TCO) are roughly less than 20 DU, that is about 5-8% of the TCO.The TCO is estimated using the ozone profile, plus the residual ozone (quantity from the maximum altitude of the flight (usually above 30 km or 10 hPa level) to the top of the stratosphere).This quantity is estimated assuming that the ozone has a constant mixing ratio equal to the measured value at the top of the measured profile, up to a pressure of 0 hPa (as detailed in the Quality Assurance and Quality Control for Ozonesonde Measurements in GAW GAW Report-No. 201, WMO 2014b).
In the recent past the L'Aquila ozone-sounding data were used in the validation of stratospheric ozone profiles retrieved by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS), onboard ENVISAT (Rizi et al., 2002;Cortesi at al., 2007), and in the validation campaign of tropospheric ozone profiles estimated by the Ozone Monitoring Instrument (OMI) data on the NASA-Aura platform (Di Noia et al., 2013).

Ozone profiles
The results of ozone soundings at L'Aquila from 2000 to 2016 are presented in Figs.1-4, in terms of the time-averaged vertical profiles, in log-pressure height coordinates, of mixing ratio (Fig. 1) and partial columns (Fig. 2-3), as well as time series of the partial columns (Fig. 4).The vertical layers on which the partial columns are calculated are listed in Table 1; the basic statistics on these columns and on the calculated linear trends are presented in Table 2.The different variabilities shown between the L'Aquila measurements (black bars) and the other measurements (red shading) can be explained by considering that for the former, the error bars only represent the time variability while for the latter the shaded area represent the spread of the time-averaged results from the different soundings, i.e. some spatial variations in ozone.The difference can be seen in Fig. 2, where the same results are shown considering the partial ozone columns (in DU) with the altitude limits described in Table 1.In this case we show in Fig. 2a the variability defined as 1 standard deviation of the time average of the 9 stations, while in Fig. 2b the variability is defined as the standard deviation of the ensemble of measurements of all 9 stations in that altitude range: this is much more comparable with the errorbars for the L'Aquila soundings calculated the same way.We further show in Fig. 3 the results for the single sondes location used in this paper.
The larger errorbars in the uppermost stratosphere are a consequence of the decrease in data availability, since not all ozonosondes reach this altitude, and the values are not always as reliable as those at lower latitudes.
Table 1.Layers where partial O3 columns are calculated.Z is the pressure altitude, defined as Hlog(po/p), where H is the atmospheric mean scale height (7 km), p is the pressure and po the reference globally averaged surface pressure (1013 hPa).

Partial column height (km)
Pressure (hPa) Name of vertical layers

Ozone trend analyses
In order to maintain consistency with the results presented in the WMO (2014) Scientific Assessment of Ozone Depletion, a multiple linear regression (MLR) method has been applied to the ozonesonde measurements taken in L'Aquila, for deriving the trends shown in Table 2 and the following figures.This is done in order to remove natural variability due to the following factors: QBO, ENSO, solar cycle and seasonal variability.
In Fig. 4 we show the time distribution of the soundings in L'Aquila for the partial columns described above.While data is sparser in the first 5 years of retrieval, we chose to include it anyway since results for the linear trend derived do not change significantly by excluding them.Superimposed to the single data points, we show the linear trend (after applying a multivalue linear regression described above) for all partial columns, separating between a positive slope (red lines) and a negative one (blue line).2).The blue line is for other northern mid-latitude sounding stations (redrawn from WMO, 2014).The green line shows the ozone trend derived from multiple observational datasets (i.e., satellites, ozone soundings, lidar, Umkehr, etc.) (redrawn from WMO, 2014; whiskers here highlight ±2σ uncertainty).
In Fig. 5 we compare the results derived in Fig. 4 with those described in the WMO report, both for mid-latitudinal soundings and for all observations.The results seem to agree in terms of positive/negative trend from the surface to the troposphere, but to diverge in the lower stratosphere, while the mid-stratosphere decrease and the uppermost stratosphere increase seem to be consistent with the average of all observations.Large uncertainties are however present both in the reported Mid-Lat soundings and in the trend as calculated for the L'Aquila soundings.
The UTLS positive trend (+1.7±0.9 percent/decade) is driven dynamically by the same mechanisms that regulate the upper tropospheric trend (as explained below in this section), but from our data analysis it does not appear to be statistically significant, contrary to those in the mid and upper troposphere.
The upper stratospheric positive trend was already well documented in the WMO (2014) ozone assessment (+3.9±1.3 percent/decade, here with ±2σ uncertainty).It may be attributed to the following concomitant causes: (a) decrease of total organic chlorine and bromine (Fig. 6ab), with associated reductions of Cly and Bry (Fig. 6cd); (b) increases of CO2, CH4, N2O (Fig. 7abc) and H2O (Fig. 8a), with an associated decrease of stratospheric temperatures due to more cooling to space (Fig. 7d and Fig. 8); the increasing amount of stratospheric H2O results from the TTL warming and increasing CH4 oxidation.These radiative-chemical mechanisms affecting upper stratospheric ozone are pretty well known and were widely discussed in past WMO assessments (WMO, 2011;WMO, 2014;WMO, 2018).They can be summarized as a slowing down of the catalytic chemical cycles for ozone destruction involving gas phase homogeneous reactions, because of the negative temperature trend (i.e., more cooling to space produced by the increasing CO2 and other GHGs).Cly and Bry catalytic cycle in the upper stratosphere, in turn, are slowed down due to the decreasing amount of organic chlorine and bromine.proxy for these VSLS (see Morgenstern et al., 2017).The corresponding shading represents 1  of the ensemble of models.
The profiles for the single models are reported in Fig. S1.For changes in the troposphere, the multi-model study of Stevenson et al. (2006) pointed out the importance of climate change for the tropospheric ozone burden.In that study, the ozone chemical precursors were kept fixed in a perturbed experiment operated with year 2030 climate conditions and in a baseline reference case with year 2000 climate.They found a negative climate feedback dominating at low altitudes, which they attributed to higher humidity and hence increased ozone destruction via the reaction O(1D) + H2O.At the same time, they found upper tropospheric ozone to increase, especially in the Northern Hemisphere, and this was attributed to an increased influx from the stratosphere.They concluded that these two climate feedbacks (water vapor and injection from the stratosphere) appear to be the dominant mechanisms operating.
These mechanisms apply to our results as well, together with the changes in ozone chemical precursors during the seventeen years.The time evolution of mean zonal tropospheric NOx and HO2 at northern mid-latitudes from CCMI simulations does not suggest a potential significant impact on in-situ production of tropospheric O3, with the exception of the increasing amount of CH3O2 from CH4 (Fig. 7b).The first of the two climate feedbacks discussed before can be easily observed in the tropospheric H2O increase of Fig. 7d; however, to precisely determine the contribution of the fluxes from the stratosphere at  Morgenstern et al., 2017).The use of the monthly averaged fields only allows an evaluation of the large scale and not of the eddy-driven fluxes: however, considering our goal of evaluating the trends in these quantities between the two time periods, whatever error might be introduced by this calculation is not influent to our overall conclusions.In Fig. 9 we compare both the vertical and horizontal O3 fluxes averaged over all available CCMI models between 10 and 100 hPa.In this figure we also show results for the CCMI simulations with fixed Ozone Depleting Substances (fODS): this is done to check if, for the period we are considering, changes driven by ODS are significant.Fig. 9a  predict different changes in value, all models show consistently a strengthening of the BDC due to the increase of GHGs in the refC2 simulation that result, for the analyzed region, in an increased (i.e., more negative) downwelling and an increased poleward transport.This is further confirmed by the absence, in the fGHG simulations, of significant changes in the fluxes.
Lastly, in Fig. 9e and Fig. 9f, we show these changes in terms of the base fluxes, that result in an average 8.25.1% increase in downwelling and an average 4.12.5% increase in poleward transport.The similarities in the response between the refC2 and fODS show that, unlike the analyses presented by Polvani et al. (2018) (where the authors looked at changes at the end of the 21st century), for the period under our consideration changes due to ozone depleting substances are negligible compared to those due to greenhouse gases.Finally, the enhanced stratospheric influx is well captured in Fig. 10, in the analyses of the trends of the ozone STE from the ULAQ-CCM.This ozone flux is calculated following the specific recommendations available at the SPARC-CCMI website for simple tracer experiments (https://blogs.reading.ac.uk/ccmi/data-requests-and-formats/).The guidelines for the 'o3ste' tracer experiments are the following: ozone tracer intended to map out the stratosphere-troposphere exchange of ozone: (a) o3ste is set to O3 at the model top layer; (b) standard O3 production-loss terms are used to change o3ste every chemistry time step for 'stratospheric' air grid boxes; (c) o3ste is reset to O3 in the lowermost layers (fix these layers to be up to sigma=0.9, ~1 km altitude): this tropospheric reset is recorded as a 2D (latitude-longitude) monthly mean tendency (Tg per month per grid cell); (d) 'stratospheric' air is determined in any way that is most convenient (e90, O3, PV-theta, etc).The ULAQ-CCM was the only one CCMI contributing model running this specific tracer experiment.
At mid-latitudes, the STE has been connected with deep convection (Hegglin et al., 2004;Tang et al., 2011) and is an important term in the budget of tropospheric ozone (Voulgarakis et al., 2010).An increase in downward O3 fluxes is observed in the northern mid-latitudes (Fig. 10a) especially over the continents where this flux is, on average, maximal (Fig. S7).This positive trend is shown to be especially significant over the months of April, May and June (where the flux is also stronger), as shown in Fig. 10b.These results confirm previous projection of STE changes due to the increase in GHG concentrations (Sudo et al., 2003;Eyring et al., 2007;Oman et al., 2010).The decrease of the same STE fluxes for the fGHG simulation shown in Fig. 10c and Fig. 10d is on the other hand simply due to the decrease in ozone-depleting substances, that is not balanced and compensated by the transport changes discussed for the refC2 case.In Fig. 11 we show a comparison of the trend observed by the soundings in L'Aquila and the overall trend predicted by the CCMI models in the same period, as in Fig. 8c, calculated in the same partial column described in Table 2.While the soundings usually overestimate both the tropospheric increase and the upper stratospheric increase compared to the models, 305 a general agreement is present in both with regards to the general sign of the trend: positive in the uppermost stratosphere due to the cooling produced by the increase in greenhouse gases and positive near the surface.

Conclusions
We have reported in this work the results from over 200 ozone soundings-derived vertical profiles measured in the town of L'Aquila (42.3°N, 13.4° E) between 2000 and 2016, and never published before.The results available from other locations in the northern mid-latitudes have been compared with our results showing a good agreement both in terms of the overall time averaged O3 partial column and with the trends already reported in the WMO report (WMO, 2014).These trends show a similar increase in O3 in the layers ranging from the surface to the uppermost stratosphere, while a decrease is reported in our trends between 16 and 30 km that, while in disagreement with the other reported ozone soundings, is more similar to the overall trends reported from a wider array of instruments.
To further understand the observed trends in the measurements we used the modeling results from the Chemistry-Climate Models Intercomparison (CCMI) (Morgenstern et al., 2017).In particular, we focused on the reference simulation refC2 (with greenhouse gases that follow the RCP6.0IPCC scenario (Meinshausen et al., 2011)  Changes in mid-latitudinal high tropospheric ozone are mainly due to two different factors: changes in the concentration of ozone-depleting substances and modifications of the stratospheric transport and of the stratospheric-tropospheric connections that result in different influxes of O3 from the stratosphere.For the former, we show that, since the global surface mixing ratio in the CCMI models of chlorine and bromine substances is prescribed to decrease in the considered time period (Eyring et al., 2013;Chipperfield et al., 2013), the vertical profiles also show a substantial decrease in their concentration: this would, in turn, result in less ozone depletion at mid-latitudes.Together with this, however, we show in the CCMI models an overall increase in mid-latitudinal humidity due to temperature changes in the stratosphere that would enhance ozone destruction via the reaction O(1D) + H2O.
For the latter, we prove that all models tend to agree over the sign of the changes in mid-latitudinal transport for the considered period: all models detect (albeit with differences in the magnitude of the predicted changes) a strengthening in the mid-latitudinal downwelling, coupled to an increase in poleward transport from the tropical area, where most of the ozone is produced.For one of the models (ULAQ-CCM) we show that these two factors are then coupled to an increase in Stratospheric-Tropospheric Exchange fluxes, further confirming the positive trend of tropospheric ozone that is detected by most instruments, and in particular by the L'Aquila ozonosondes, over the same time period at those latitudes.latitude deep convective system, Atmos.Chem.Phys., 4, 741-756, https://doi.org/10.5194/acp-4-741-2004, 2004. Hegglin, M. I. and https://doi.org/10.5194/acp-2020-525Preprint.Discussion started: 1 July 2020 c Author(s) 2020.CC BY 4.0 License.

Fig. 1 .
Fig. 1.Vertical profiles of the ozone volume mixing ratio observed at L'Aquila, Italy and averaged over the years 2000-2016 (black lines).Black whiskers represent ±1σ of the time variability.The red lines show the mixing ratio averaged over the same period of time from other northern mid-latitude ozone sounding stations, with the shading representing ±1σ of the variability of mean values among these 9 stations (Boulder, Hohenpeissenberg, Legionowo, Madrid, Payerne, Prague, Sapporo, Tsukuba, Wallop) (Tilmes et al., 2012).Panel (a) shows the complete profile extension over the troposphere and the stratosphere (0-10 ppmv), whereas panel (b) is a zooming over the troposphere (0-1 ppmv).

Fig. 2 .
Fig. 2. Vertical profiles of partial O3 columns calculated for the layers in Tables 1-2 (average on years 2000-2016): black 150 and red lines are for L'Aquila and other northern mid-latitude sounding stations, respectively (as in Fig. 1).The shaded area in panel (a) shows ±1σ of the variability of mean values among the 9 stations listed in Fig. 1, whereas in panel (b) it shows ±1σ of the partial column time variability in these same stations.The black whiskers are for ±1σ of the partial column time variability at L'Aquila.155

Fig. 3 .
Fig. 3.As in panel (b) of Fig. 2 for the shaded area and the black line of the partial columns from L'Aquila soundings.The other lines refer to years 2000-2016 mean values at the other northern mid-latitude sounding stations (see legend) (Tilmes et al., 2012).Colored whiskers show the time variability at the single stations.

Fig. 4 .
Fig. 4. Time series of partial O3 columns from L'Aquila soundings (year and DU on x and y axis, respectively).Black circles represent the data, while the superimposed red/blue lines (positive/negative slope, respectively) show a linear fit of the data 175 with the different factors removed by the MLR fitting (see text).

Fig. 5 .
Fig. 5. Ozone trends in the layers of Tables1-2(percent/decade). The black line is for the linear trends obtained from L'Aquila partial column data; the shaded area shows ±2σ of the linear trend (see Table2).The blue line is for other northern https://doi.org/10.5194/acp-2020-525Preprint.Discussion started: 1 July 2020 c Author(s) 2020.CC BY 4.0 License.

Fig. 6 .
Fig. 6.Time series (2000-2020) of surface mixing ratios of total organic chlorine (a) and bromine (b); values are those used as prescribed boundary conditions for the CCMI RefC2 numerical experiments (Eyring et al., 2013).Panels (c,d): vertical profiles of northern mid-latitude changes (35N-55N) of total inorganic chlorine (Cly, ppbv) (c) and total inorganic bromine (Bry, pptv) (d) (2016-2000) averaged over all available models for refC2 (black line).Part of the models explicitly include the contribution of CH2Br2 and CHBr3 to the Bry budget, other use a 5 pptv increase of the CH3Br surface mixing ratio as a

Fig. 7 .
Fig. 7. Time series (2000-2020) of surface mixing ratios of CO2 (a), CH4 (b) and N2O (c); values are those used as prescribed boundary conditions for the CCMI RefC2 numerical experiments (Eyring et al., 2013).Panel (d): vertical profiles of northern mid-latitude trends (35N-55N) of temperature (K/decade) averaged over all available models for refC2 (black line) and fGHG (red line).The corresponding shadings represent 1  of the ensemble of models.The profiles for the single models are reported in Fig. S2.
https://doi.org/10.5194/acp-2020-525Preprint.Discussion started: 1 July 2020 c Author(s) 2020.CC BY 4.0 License.mid-latitudes, three distinct elements are necessary: a) the increased mid-latitudinal stratospheric ozone mixing ratio due to increased horizontal isentropic transport from the main photochemical source region; b) the increased mid-latitudinal stratospheric downwelling fluxes and finally c) the increase in Stratosphere-Troposphere Exchange (STE) fluxes.We show these three contributions to the increase of upper-tropospheric ozone in Fig. 9 (for stratospheric horizontal and vertical mass fluxes) and Fig. 10 (for STE fluxes).Both horizontal and vertical fluxes are calculated starting from the velocities, ozone concentration and temperature 3D monthly-averaged (as done already by Visioni et al., 2017) fields made available by each CCMI group, interpolated on a common vertical grid (considering that different models have different vertical coordinates, as explained by

Fig. 8 .
Fig. 8. Vertical profiles of northern mid-latitude trends (35N-55N) of H2O (ppmv/decade) (a) and ozone (ppbv/decade in (b) and percent/decade in (c)) averaged over all available models for refC2 (black line) and fGHG (red line).The corresponding shadings represent 1  of the ensemble of models.The profiles for the single models are reported in Fig. S3 (for panel a) and Fig. S4 (for panels b-c).
and Fig. 9b show the base fluxes averaged over the years 1999 and 2000 for refC2, fGHG and fODS, showing an overall agreement of all models https://doi.org/10.5194/acp-2020-525Preprint.Discussion started: 1 July 2020 c Author(s) 2020.CC BY 4.0 License.regarding the mid-latitudinal stratospheric sign of the fluxes: a downwelling for the vertical flux and a poleward movement for the horizontal fluxes.The changes in the 2000 to 2016 period are shown in Fig. 9c and Fig 9d: whilst different models

Fig. 9 .
Fig. 9. a) Left panels: a) mid-latitude LS vertical (g/m2/yr) fluxes of O3 (1999-2000) for refC2 (grey bar, average over 9 models) and fGHG (red bar, average over 6 models).The error bars represent 1  of the ensemble of values for all models.c) Linear trend (g/m2/yr/decade) of the LS vertical fluxes of O3 between 2000 and 2016 for the same models.e) Same as c) but expressed in % of the fluxes in panel a).The values for all separated models are shown in Fig. S5.Right panels: b) midlatitude LS horizontal (kg/m2/yr) fluxes of O3 (1999-2000) for refC2 (grey bar, average over 15 models) and fGHG (red bar, average over 8 models).d) Linear trend (kg/m2/yr/decade) of the LS horizontal fluxes of O3 between 2000 and 2016 for the Fig. 10.a) STE O3 flux linear trend (g/m2/yr/decade) between the years 2000 and 2016 for the ULAQ-CCM refC2 simulation.b) Same as in a), but for the fGHG simulation.c) Monthly STE O3 flux linear trend (g/m2/yr/decade) averaged 300 between 35N and 55N for the refC2 simulation.d) Same as in c), but for the fGHG simulation.

Fig. 11 .
Fig. 11.Ozone trends in the layers of Tables 1-2 (percent/decade), with blue line and shading as in Fig. 5 (L'Aquila sounding).Black and red lines show the northern mid-latitude trends (35N-55N) derived from CCMI model runs, for cases RefC2 and fGHG, respectively.
to understand the 2000-2016 trend https://doi.org/10.5194/acp-2020-525Preprint.Discussion started: 1 July 2020 c Author(s) 2020.CC BY 4.0 License.and the scenario with greenhouse gases concentration fixed at 1960 levels (fGHG) to understand the role of the increased surface warming (and connected stratospheric cooling) that can be revealed by looking at the difference between refC2 and fGHG.