Arctic on the verge of an ozone hole?

Severe vortex-wide ozone loss in the Arctic would expose nearly 650 million people and ecosystem to unhealthy ultra-violet radiation levels. Adding to these worries, and extreme events as the harbingers of climate change, clear signature of an ozone hole (ozone column values below 220 DU) appeared over the Arctic in March and April 2020. Sporadic 15 occurrences of ozone hole values at different regions of vortex for almost three weeks were found for the first time in the observed history in the Arctic. Furthermore, a record-breaking ozone loss of about 2.0–3.4 ppmv triggered by an unprecedented chlorine activation (1.5–2.2 ppbv) matching to the levels of Antarctic ozone hole conditions was also observed. The polar processing situation led to the first-ever appearance of loss saturation in the Arctic. Apart from these, there were also ozone-mini holes in December 2019 and January 2020 driven by atmospheric dynamics. The large loss in 20 ozone in the colder Arctic winters is intriguing and that demands rigorous monitoring of the region. Our study suggests that the very colder Arctic winters in near future would also very likely to experience even more ozone loss and encounter ozone hole situations, provided the stratospheric chlorine levels still stay high there.

Three satellite-based total column ozone (TCO) data are also employed (level 3) to analyses the ozone hole: These total column measurements have an uncertainty of 2-5%. The ozone and other trace gas profiles are provided in pressure co-ordinates, which are converted to isentropic coordinates using the temperature data from the same satellite. We 75 use the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalyses ERA5 potential vorticity (PV) on a 1°x1° grid to determine the vortex edge. The PV data are also converted to isentropic coordinates using the ER5 temperature data. We computed the equivalent latitude at each isentropic level at 5 K intervals from 350 to 800 K, and are then used to compute the vortex edge using the Nash et al. (1996) criteria. We use measurements inside the polar vortex for ozone loss analysis. The missing values in satellite measurements were filled with linear interpolation (poison_grid_fill). 80 We have taken ozone, ClO, HNO3 and N2O from the Aura MLS measurements. The ozone measurements at 240 GHz have a vertical resolution of 2-3 km, vertical range of 261-0.02 hPa and an accuracy of 0.1-0.4 ppmv. The vertical range of HNO3 measurements is 215-1.5 hPa, vertical resolution is 2-4 km, with an accuracy of 0.1-2.4 ppbv, depending on altitude. The N2O measurements are available for the 68-0.46 hPa vertical range, and 68 hPa roughly equivalent to 400 K isentropic level. 85 The data were extrapolated up to 350 K by performing exponential fitting to N2O vertical distribution at 400-600 K by considering the exponential change of N2O with altitude. The accuracy of retrievals at 190 GHz is about 2-55 ppbv at this altitude range and the vertical resolution is about 2.5-3 km. The vertical resolution of ClO measurements at 640 GHz is about 3-3.5 km over 147-1 hPa, and the accuracy of measurements is about 0.2-0.4 ppbv. The measurements also have latitude-dependent bias of about 0.2-0.4 ppbv, depending on altitude (Livesey et al., 2013;Santee et al., 2008;Froidevaux et 90 al., 2008).
The OMPS consists of three sensors that measure scattered solar radiances in overlapping spectral ranges and scan the same air masses within 10 min. The nadir measurements are used to retrieve ozone total column and vertical profiles (NP). The Limb Profiler (LP) measures profiles with high vertical resolution (∼ 2-3 km) and the LP retrievals are in good agreement 95 with other satellite measurements and the differences are mostly within 10% (Kramarova et al., 2018). The OMPS TCO shows 0.6-1.0% differences with Brewer and Dobson ground-based TCO measurements across the latitudes, and are also biased https://doi.org/10.5194/acp-2020-1313 Preprint. Discussion started: 24 February 2021 c Author(s) 2021. CC BY 4.0 License.
in temperature on 5 February 2020 (i.e. a minor warming) and a corresponding change in zonal winds.
The temperatures were consistently lower than the nitric acid trihydrate (NAT) threshold of 195 K and therefore, large areas of Polar Stratospheric Clouds (PSCs) are observed from December to mid-February. The PSC area (APSC) was about 4 million km 2 in December at 460 K, but it doubled in January through mid-March. The APSC from mid-February to late March is also largest on the observational record ( Figure 1). The low temperatures (i.e. lower than 188 K) also produced very high amount of ice PSCs at the end of January and early February (up to 4 million km 2 ) when the lowest temperatures in 40 years were recorded in the Arctic. This is the largest ice PSC ever observed in terms of its area, volume and number of days of appearance (i.e. frequency) in the Arctic and the area is twice that of the winter 2011. The PSC area shrunk to half of its area in late January and February, as the lower stratospheric temperature increased during the period. This was the only 135 occasion that the temperature increased and PSC areas limited to below 4 million km 2 in the winter 2020. Note that the PSC area and volume were largest in 2016, not in 2020 ( Figure S1) (Kirner et al., 2015).
The potential vorticity (PV) analyses at ~17 km (about 460 K potential temperature level) show that the polar vortex was very strong in the lower stratosphere in 2020. The PV values were consistently higher than the previous cold (i.e. 1995, 140 3. 2 Strong air mass descent and associated ozone distribution Figure 3 shows the distribution of ozone, ClO, N2O, HNO3 and the ozone loss estimated for the winter 2020 using satellite observations. We use the measurements from MLS on the Aura satellite (Livesey et al., 2015). The MLS measurements are 165 one of the best currently available data for polar ozone loss analyses, as the instrument provides measurements of some key ozone-related chemistry trace gases such as ClO, N2O and HNO3 to delineate the features of chlorine activation, vortex descent and denitrification, respectively (Manney et al., 2020). The ozone distributions in the vortex show < 1.0 ppmv in and therefore, we have quantified the ozone loss for the winter. We use the profile descent method using the trace of air motions N2O and is a widely used method for ozone loss estimation (Bremer et al., 2002;Rex et al., 2002;Jin et al., 2006). distributions show below 50 ppbv at all altitudes from early February onwards; suggesting substantial dynamic descent in the stratosphere. When a particular altitude is considered, e.g. the 450 K potential temperature level, the N2O values show 160 ppbv in early December, 100 ppbv in early January, 50 ppbv in early February and smaller than 50 ppbv thereafter. On the other hand, the N2O distributions show 50 ppbv in early December and below that value afterwards at 500 K. The severe air 185 mass descent in this winter is further depicted in Figure S2, where monthly correlation between ozone an N2O are presented.

Ozone loss and mini-holes in December and January
There were vortex-wide PSC occurrences in the first week of December, about 2-4 million km 2 in area (APSC) and about 70 million km 3 in volume (VPSC). The APSC and VPSC dropped significantly afterwards and then gradually increased again by mid-December to 10 million km 2 and 120 million km 3 , respectively. An unusual increase in activated chlorine is observed 190 during the first week of December in conjunction with the appearance of PSCs. The temperatures began to decrease from 198 K in mid-December to 187 K by the end of January. The chlorine activation peaked and showed record levels of ClO, about 1.5-2.0 ppbv at 400-600 K, during this period. The chemical ozone loss began in early January with about 0.5 ppmv https://doi.org/10.5194/acp-2020-1313 Preprint. Discussion started: 24 February 2021 c Author(s) 2021. CC BY 4.0 License. and increased to 1.5 ppmv by the end of January below 500 K. The loss above that altitude is always lower than 0.5 ppmv, which shows that the ozone loss is restricted to the altitudes below 21 km (i.e. 550 K). 195 In general, the ozone loss starts in December in the middle stratosphere and then gradually progresses towards the lower stratosphere by January. The loss would be below 0. In addition to the ozone loss inside the vortex, there is another interesting phenomenon in December and January. The analyses of TCO show that there were ozone holes of about 300-700 km 2 size during the first week of December (1-6 December 2019) and on 26 January 2020 ( Figure 4). The lowest TCO measured of the winter was also on the latter date.
However, the ozone holes were not inside the vortex, but in the mid-latitudes. A detailed analysis with TCO, PV, 210 temperature and ClO reveal that those ozone holes were dynamically driven, as there was rapid air mass transport to the southern Arctic in early December and late January. These ozone hole occurrences due to rapid changes in weather patterns are generally called "ozone mini holes" and they return to normal levels of ozone in few days (e.g. Rieder et al., 2013). We used the HYSPLIT trajectory model to find the air mass transport at three different altitudes (17, 18 and 19 km) in the lower stratosphere, where the mini-holes are found ( Figure 4, right panels). The air mass exported from mid-and low latitudes has 215 very low PV values, lower temperature and high ClO. It suggests that the TCO transported from mid-latitudes triggered the It should be mentioned that there was already large chemical loss of ozone inside the Arctic vortex in early December and late January owing to the conventional polar ozone loss chemistry (as shown in Figure 3). However, the ozone mini-holes that appeared outside the vortex were primarily triggered by the dynamics. We cross-checked TCO from OMI (Bias et al.,

Prolonged chlorine activation and chemical ozone loss
When the Arctic winters are very cold, chlorine activation occurs in the lower stratosphere at 400-500 K in January and February there. In 2011, the chlorine activation was observed up to the end of February and was intermittent with a peak 230 value of about 1.6 ppbv, and was mostly at 400-500 K (e.g. Manney et al., 2011;Kuttippurath et al., 2012;Livesey et al., 2015;Griffin et al., 2018). Conversely, in the Arctic winter 2020, there was continuous and sustained chlorine activation  (Pommereau et al., 2018;Lindenmaier et al., 2012). The gravest denitrification was in December, with values of about 0-2 ppbv below 400 K and 4-6 ppbv at 400-450 K. Therefore, high chlorine activation and strong denitrification (as deduced from the HNO3 analyses shown in Figure 3) provided an unprecedented situation for large ozone loss of about 2-3.4 ppmv in the lower stratosphere in March-April. 250 Since the ozone loss in 2020 is exceptionally large, we have employed another set of measurements to estimate ozone loss to reconfirm that the derived results are robust. The loss estimated from OMPS measurements together with other analyses are shown in Figure 5 (left). The maximum ozone loss profile extracted from the OMPS data show very good agreement with that from the MLS measurements for the Arctic winter 2020. The peak ozone loss values show about 2-2.8 ppmv in the 255 lower stratosphere below 550 K. Since the maximum ozone loss profiles are averaged for few days, the loss values are slightly lower than that of MLS measurements. The lower stratosphere shows similar ozone loss values, but the loss above 500 K shows slightly smaller values (0.1-0.5 ppmv) due to the low bias of OMPS measurements at these altitudes as compared to the MLS measurements (Kramarova et al., 2018). The comparison with OMPS confirms that the method https://doi.org/10.5194/acp-2020-1313 Preprint. Discussion started: 24 February 2021 c Author(s) 2021. CC BY 4.0 License. adopted for ozone loss is robust. Our estimates are in good agreement with that of Manney et al. (2020) and Wohltmann et 260 al. (2020), as they also derive a loss of about 2.3-2.8 ppmv below 450 K from the MLS measurements inside the vortex.

5 The Arctic ozone loss in the context of other Arctic winters
Arctic winters are normally warmer and occurrences of PSCs are sparse and infrequent. Therefore, chlorine activation and ozone loss is limited to the winters with very low temperatures in December-February (Tilmes et al., 2014;Goutail et al., 2015;WMO, 2018;Newman et al., 2008;Kuttippurath et al., 2012). The ozone loss observed in warm winters (e.g. the Arctic winter 2020 is about 0.7 ppmv higher than that in 2011, about 2.8 ppmv. The difference in ozone loss between the winters is negligible above 480 K. Therefore, it is evident that the ozone loss in the Arctic winter 2020 is undoubtedly the largest on the record and is significantly higher than that of any previous Arctic winter. Furthermore, we applied another loss estimation method to test robustness of the extreme ozone loss values; the passive 280 method that uses a passive tracer (i.e. no chemistry) simulation. We have used the well-known and widely used . This ozone loss is slightly higher than that of the Arctic winter 2011, about 0.2 ppmv. It is also observed that the ozone loss in 2020 is higher than that of 2011 below 475 K, but the loss 285 estimated in latter winter exceeds about 0.3-0.5 ppmv above 475 K up to 700 K (e.g. Manney et al.,2020;Wholtmann et al., 2020). However, these ozone loss estimates are lower than that estimated with the descent method, about 0.5-0.7 ppmv depending on altitude. The analysis with ozone and N2O from the model indicates that modelled ozone is higher than (about 1-1.5 ppmv) the measurements at these altitudes, which could be due to the slower dynamical descent in the model.

290
It is clear that the ozone loss in 2020 is the largest among Arctic winters. Therefore, we also examined the evolution of chlorine activation in terms of the amount of ClO in each Arctic winter, as the total chlorine is decreasing in the stratosphere https://doi.org/10.5194/acp-2020-1313 Preprint. Discussion started: 24 February 2021 c Author(s) 2021. CC BY 4.0 License.
due to the effect of Montreal Protocol (e.g. Strahan et al., 2017;WMO, 2018;Dhomse et al., 2019) and we expect a corresponding response in ozone loss in the polar winters. Figure 6 shows the MLS ClO observations in each winter since 2005. The analyses show that the chlorine activation was very severe and continuous for about four months in 2020. 295 However, the highest ClO was observed in winter of 2016, in February. Many colder winters had the ClO values around 2 ppbv as found in 2020, but the sustained chlorine activation that observed in 2020 was unique. Although the high ClO values in March were also observed in 2011, the chlorine activation was not as severe as in 2020 in early winter (December -January). On the other hand, the unprecedented chlorine activation observed in 2016 was more episodic, such as in mid-December, mid-January to early February and late February. Therefore, the continuous and severe chlorine activation from 300 December through March was the key for the record-breaking ozone loss in 2020. Figure 6(b) and (c) further illustrate that the peak ClO profiles or the time series of average ClO for the entire winter will not reveal the depth of chlorine activation.

The Arctic ozone loss equals the levels of the Antarctic ozone loss
The peak ozone loss in the Antarctic happens at around 500 K and the loss is severe from 400 to 600 K for five months continuously from August to November (Tilmes et al., 2006;Huck et al., 2005;Sonkaew et al., 2013;Kuttippurath et al., 305 2015). In contrast, the Arctic winters are normally short and maximum ozone loss occurs at around 425-475 K for a period of about two months, from mid-January to mid-March (e.g. Kuttippurath et al., 2010;Manney et al., 2004). The ozone loss in the Arctic is restricted to the altitudes below 500 K. The ozone loss in the Arctic winter 2020 is very high and therefore, we 1.0 ppmv at 370 K, and 2.6 ppmv at 460 K, 1.5 ppmv at 550 K, 0.5 ppmv at 650 K and it terminates at 700 K in the Antarctic winter 2015. In the Arctic winter 2020, the ozone loss shows about 0.3 ppmv at 370 K, 2.0 ppmv at 430 K and 480 K, 1.5 ppmv at 550 K and loss terminates above that altitude. The peak ozone loss is about 2.3 ppmv and is at 460-470 K. On the other hand, the loss in the Antarctic winters above 470 K is very large and reaching up to 700 K. The peak ozone loss in the Arctic winter 2020 is about 2.8 (2.3) ppmv and is at 460-470 K. This is also the main difference between the Arctic and 320 Antarctic ozone loss, as the broader and larger ozone loss above the 470 K in the Antarctic. The difference is almost 1.0 ppmv above the peak ozone loss altitude. Therefore, the ozone loss in the Arctic winter 2020 is either equal or larger than that of the Antarctic winter 2019 below 470 K, but the loss is smaller than that of the Antarctic winters above 525 K.
We applied the passive method also to further examine the estimated loss in the Arctic and Antarctic winters ( Figure 5, 325 second panel from the left). The ozone loss estimated with the passive method exhibits smaller values in the lower stratosphere in comparison with that derived from the descent method. The loss is about 0.2 ppmv at 350 K, 1.6 ppmv at 400 K and 2.3 ppmv at 450 K in the Arctic winter 2020. The peak loss is recorded at 450-460 K and the loss decreases with altitude, about 1.5 ppmv at 500 K and 0.1 ppmv at 530 K. In the Antarctic winter 2019, the ozone loss shows similar values as that of the Arctic winter 2020 at 370-420 K, but slightly smaller than that of the Arctic winter at 420-470 K. The 330 maximum ozone loss in Antarctic winter 2019 is estimated at 470 K, about 2.3 ppmv, and about 0.5-1.5 ppmv above that altitude, which is higher than that of the Arctic winter 2020. Furthermore, the Arctic ozone loss halts at about 550 K, whereas the Antarctic ozone loss at this altitude is as high as 1.5 ppmv. In the Antarctic winter 2015, the ozone loss is about 1.0 ppmv at 370 K, 2.0 ppmv at 400 K and the peak loss of about 2.8 ppmv at 475 K. The loss gradually decreases with altitude, such as 2.1 ppmv at 500 K, 1.5 ppmv at 550 K, 1.0 ppmv at 600 K and 0.5 ppmv at 650 K. The ozone loss in the 335 Antarctic winter 2015 is thus, higher than that of the Antarctic winter 2019 and the Arctic winter 2020, about 0.5-1.5 ppmv, depending on the altitude. The assessment further gives strong evidence that the peak ozone loss in the Arctic winter 2020 equals to that of the warm winters of Antarctic (e.g. 2019). The loss estimation method can have uncertainty in the range of 3-5%, depending on the winter months. For instance, the monthly mean ozone loss and its standard deviation for each winter month of 2020 are shown in Figure S3. A complete error analyses of the ozone loss method is already presented in 340 Kuttippurath et al. (2010).

The first appearance of ozone loss saturation in the Arctic
Ozone loss saturation is a common feature of Antarctic winters since 1987 (Kuttippurath et al., 2018;Jin et al., 1996).
However, as compared to the Antarctic, the Arctic winters are relatively short (December to March), stratospheric temperatures are about 10 K higher, occurrence of PSCs are infrequent, denitrification is modest and thus, ozone loss is 345 moderate. Therefore, the Arctic never encountered the ozone loss saturation (i.e. the complete loss of ozone at some altitudes in the lower stratosphere between 400 and 550 K) there. Apart from these, the ozone loss normally happens only up to 25-30% in the Arctic winters and henceforth, the loss saturation was unexpected for the Arctic conditions. Figure 5 (right) shows the ozone profile measurements by ozonesondes at two Arctic stations, Alert (82.50° N, 62.33° W) and Eureka (80.05 N, 86.42 W), on selected days. The ozone profiles measured at selected Antarctic stations are also shown for 350 comparisons. In general, the ozone loss saturation in Antarctica occurs at the altitude between 400 and 500 K (e.g. Davis: 68.6°S, 78.0°E and Marambio: 64° S, 56° W), and the altitude range would go up to 550 K for the stations that are always inside the vortex, as shown for Syowa. The ozone loss observed at Davis and Marambio is always smaller than that at Neumayer, South Pole and Syowa. Therefore, ozone loss saturation is also comparatively infrequent and occurs at limited altitudes in the Antarctic. Here, the ozonesonde measurements at Alert (on 08 April 2020) show loss saturation at the 355 available measurements, show that the ozone loss saturation occurred at these station in early April ( Figure S4). The vertical the Antarctic ozone column loss is about twice that of the Arctic, about 150-160 DU, but slightly lower about 100-120 DU in very warm winters (1988 and 2002) and in early years (e.g. 1979-1985) of ozone loss there (Huck et al., 2005;Tilmes et al., 2006;Kuttippurath et al., 2015). The analyses clearly suggest that even the partial column ozone loss in the Arctic winter 2020 is about 115 DU at 350-550 K, which is higher than that of the Arctic winter 2011 and equal to that of the loss found in the Antarctic winters 1979-1985, 2002 and 2019. 395 Since the ozone loss in the Arctic winter 2020 is up to the levels of that found in Antarctic winters, we examined the occurrence of ozone holes using TCO data from OMPS and MERRA-2 and the results are presented in Figure 8 for selected ozone hole days. The first appearance of ozone holes in the Antarctic winters are also shown for comparison. There are clear and identifiable ozone holes in March and April 2020 and were hundreds of kilometres wide to demarcate the regions below 400 220 DU. The ozone maps show that the holes in March and April 2020 were larger than that of the Antarctic ozone holes in October 1979 and 1980. Therefore, ozone loss in the Arctic winter 2020 is roughly comparable to the Antarctic ozone loss and the appearance of ozone hole for several weeks demonstrate that the Arctic winters enter a new era of ozone depletion events, and signal significant changes in the climate of the region. However, as the ozone holes were not very large and were not present for continuously for months as they occur over the Antarctic, the situation could termed only as the appearance 405 of signatures of ozone hole in Arctic winter 2020.

Conclusions
The Antarctic ozone hole is present for the past forty years, and the impact of ozone hole on public health is mostly restricted to the southern high and mid-latitudes. The ozone hole has also influenced the climate of the southern hemisphere by 410 changing the winds, temperature and precipitations in different regions. On the other hand, the biggest concern about the polar ozone loss in the stratosphere has always been an Arctic ozone hole, because such an ozone hole can occur anywhere beyond 45° N in the densely populated northern mid and high latitudes. The changes in associated UV radiation incidence would also affect the flora and fauna of the region. If such a situation arose, that would trigger ecosystem damage and impose serious threat to public health (e.g. Newman et al., 2009). Nevertheless, it is believed that an ozone hole over the 415 Arctic would be unlikely due to relatively higher temperature and shorter wintertime ozone loss period there. Furthermore, the winters are always prone to several minor and frequent major warmings (almost a major warming per winter), which would restrict the lifetime of the polar vortex, PSC occurrence and chlorine activation to limit the extent and severity of ozone loss. However, the Arctic winter 2020 was exceptional as it was characterised by strong vortex from December through the end of April, large and widespread PSCs, and unprecedented and prolonged chlorine activation with peak ClO 420 values of about 2.0 ppbv. The high chlorine activation in early December and early January produced larger loss in ozone (e.g. 1-1.5 ppmv below 430 K in early January) in the Arctic that has never occurred before. The continued high chlorine activation from January to mid-April caused a record-breaking ozone loss of about 2.5-3.4 ppmv at 400-600 K, and triggered the first-ever ozone hole of Arctic in March and April 2020. The unprecedented chlorine activation (e.g. January