OClO as observed by TROPOMI: a comparison with meteorological parameters and PSC observations

Abstract. Chlorine dioxide (OClO) is a by-product of the ozone depleting halogen chemistry in the stratosphere. Although being rapidly photolysed at low solar zenith angles (SZAs) it plays an important role as an indicator of the chlorine activation in polar regions during polar winter and spring at twilight conditions because of the nearly linear dependence of its formation to chlorine oxide (ClO). Here we compare slant column densities (SCDs) of chlorine dioxide (OClO) retrieved by means of differential optical absorption spectroscopy (DOAS) from spectra measured by the TROPOspheric Monitoring Instrument (TROPOMI) with meteorological data for both Antarctic and Arctic regions for the first three winters in each of the hemispheres (November 2017–October 2020). TROPOMI, a UV-VIS-NIR-SWIR instrument on board of the Sentinel-5P satellite monitors the Earth’s atmosphere in a near polar orbit at an unprecedented spatial resolution and signal to noise ratio and provides daily global coverage at the equator and thus even more frequent observations at polar regions. The observed OClO SCDs are generally well correlated with the meteorological conditions in the polar winter stratosphere: e.g. the chlorine activation signal appears as a sharp gradient in the time series of the OClO SCDs once the temperature drops to values well below the Nitric Acid Trihydrate (NAT) existence temperature TNAT. Also a relation of enhanced OClO values at lee sides of mountains can be observed at the beginning of the winters indicating a possible effect of occurring lee waves on chlorine activation. The dataset is also compared with CALIPSO Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) polar stratospheric cloud (PSC) observations. In general, OClO SCDs coincide well with CALIOP measurements for which PSCs are detected. Very high OClO levels are observed for the northern hemispheric winter 2019/2020 with an extraordinarly long period with a stable polar vortex being even close to the values found for Southern Hemispheric winters. Also the extraordinary winter in 2019 in the Southern Hemisphere with a minor sudden stratospheric warming at the beginning of September was observed. In this winter similar OClO values were measured in comparison to the previous (usual) winter till that event but with a 1–2 week earlier OClO deactivation.


Given that also here the systematic error component is mainly dominating, the detection limit thus is expected to be below ∼2.5×10 13 cm −2 with systematic error as the dominating source of the uncertainty.

CALIOP PSCs observations
In addition, we relate the retrieved OClO SCDs with the Level 2 Polar Stratospheric Cloud provisional version 1.10 product (Pitts et al., 2009) satellite. From the CALIOP PSC product we use the provided PSC cloud mask profiles indicating whether a PSC is detected above a certain location as a function of altitude. The advantage of the use of the PSC mask product in our opinion is that it reduces possibility to misinterpret the aerosol information which would be the case if backscatter data would be used instead.
We neglect the available distinction with respect to different PSC types as the aim of the current study is to check how the 155 general existence of PSCs relates with the OClO SCDs we have measured. We also consider the detection sensitivity which is provided in the PSC product where the horizontal averaging which was necessary to detect PSC is provided. To be able to match an OClO SCD at a given location which is not altitude resolved with a single piece of information about PSCs, we merge the PSC existence profile information as well as the altitude resolved detection sensitivity to a single generic quantity. This quantity, which we call PSC evidence E in the following and which up to our knowledge have not been used in the literature 160 so far, is calculated as a sum of the PSC signals originating from all different altitudes at a given location: where M i is boolean being unity if a PSC is reported in the CALIOP data at an altitude level i more than 4 km above the tropopause. A i is the reported horizontal averaging being either 1, 3, 9 or 27 corresponding to the horizontal averaging of 5, 15, 45 or 135 km, respectively, which was necessary to detect the PSC. 165 For the comparison each CALIOP measurement is collocated with the average of TROPOMI measurements within the range of 89°<SZA<90°on the same day that are less than 100 km away. It is done because of the larger spatial coverage of TROPOMI as well as to largely eliminate random error contribution of individual TROPOMI measurements.
In addition also daily mean and maximum evidences are obtained from PSC evidences calculated beforehand for all CALIOP measurement locations above 60°latitude. While the collocated PSC evidences describe the PSC existence at and near the 170 analysed TROPOMI measurements, these two additional parameters provide additional information about PSC extent in the whole polar region.
4 Interpretation of the TROPOMI OClO measurements with respect to meteorological quantities and CALIOP PSC The first winter (2017/2018) after TROPOMI was launched was a rather cold stratospheric winter especially with cool temperature anomalies in January until the beginning of February over the polar cap (Wang et al., 2019). A sudden stratospheric warming event has been reported for 12 February characterized by a polar vortex split Hall et al., 2021).
For this winter unfortunately many days of measurements are missing due to calibration processes. The time series of OClO 180 SCDs daily averages for SZA between 89 and 90°during this winter are plotted in the top panel of Fig. 1       before which the vortex area seems to have stayed rather constant for a few days (Fig. 1, second plot from top). Nevertheless the OClO values continue to decrease afterwards, the temperature gradient becomes very large within the split vortex which can be deduced by the increased OClO at high temperatures in the temperature resolved time series of OClO SCDs (third panel in Fig. 2). After this short cooling the temperature rises rapidly, the vortex area decreases and the OClO SCDs continue to decay. The breakup of the polar vortex is also evident in the bottom plot of Fig. 2 where still increased OClO SCDs are found 230 towards lower PV values. A second similar event, but not as strong, is observed at the last days in February (26 February).
Here also PSCs are barelly evident at a longitude (120°W) among the longitudes at which largest OClO SCDs are observed.
The vortex eventually strengthens again at the beginning of March when mean zonal winds become westerly again  but it has no relevance for chlorine activation because of the high temperatures.
The following winter 2018/2019 has been reported as being unusual in terms of the polar vortex variability ): with both a major sudden stratospheric warming and a reformation of a strong vortex later. In terms of minimum temperature (see third plots from top in Figs. 4 and 5, for technical explanation of plots please see the description for the previous winter) the beginning of the winter was rather warm, the temperatures dropped below T NAT only in December. However the mean OClO SCDs (Fig. 4, upper plot) appear to be slightly but consistently increased above zero already during the last days 240 of November with enhanced OClO SCDs above Greenland and Northern Asia (upper plot in Fig. 5). This increase however technically is still below the detection limit of 2×10 14 cm −2 . An OClO production in the area covered by the plotted SZA range (89°<SZA<90°) can likely be excluded because no OClO enhancements at the lowermost temperature bins in the temperature resolved time series of OClO SCDs are found (Fig. 5, third panel). This finding does not exclude that such an activation could have taken place in some other area not covered by the SZA range investigated here. Lee and Butler (2020) report a begin of 245 the increase of a vertically propagating wave activity during November and thus local drops of the temperature below T NAT induced by mountain waves could have been a possibility for OClO formation because the minimum temperature at 600K reaches T NAT in that period. The CALIOP data (Fig. 6) however do not show any evidence of PSC formation.
The mean OClO SCDs increase further at the beginning of December a few days after the temperature dropped below T NAT .
This delay probably indicates that the area where this drop occurs is small or that the drop was not sufficient to overcome the . From the other hand, such or even higher OClO SCDs not necessarily correspond to an observation of the PSC evidence above zero. The largest OClO SCDs on these days are clearly limited to the area with temperatures below T NAT which are located eastwards of the Scandinavian mountains and around the Ural mountains: this could be an indication for mountain waves having enhanced the chlorine activation process. The OClO SCDs in the rest of the analysed polar vortex area remain lower but well above the random uncertainty level and at or above the detection limit and looks like remnants of the 265 chlorine activated earlier. After this cooling the polar vortex slowly starts to shrink (Fig. 4, second plot from top), is warmed up at the end of December (Fig. 4, third plot from top) as the prelude for an early sudden stratospheric warming event reported on 2 January . The atmospheric temperatures rise above T NAT on 27 December and stay slightly above T NAT eventually dropping once more below it on 3 and 4 January 2019. However the area with temperatures below T NAT is very small for these days. The appearance of one additional OClO peak at the beginning of January can be attributed to the 270 irregular shape of the polar vortex and to the fact that the earlier activated air masses are moved inside the 89°<SZA<90°r ange. This interpretation is supported by the temperature resolved time series of OClO SCDs (third panel of Fig. 5) where the enhanced OClO SCDs appear at very warm temperatures. These enhanced OClO values especially at the end of December and in January even appear for very high temperatures (> 20K above T NAT ). On these days also an increase of the potential vorticity (above 50 PVU) is observed (bottom panel of the same figure) which indicates that here air masses are seen which 275 were not observed before, because they were located deep in the centre of the polar vortex. Afterwards the OClO SCDs decay until mid of January to values below the detection limit. In February and March the formation of a very strong polar vortex has been reported (Lee and Butler, 2020) but the temperatures never fell again below the threshold of the chlorine activation.

Winter 2019/2020
In the winter 2019/2020 an exceptionally strong and cold stratospheric polar vortex was formed which maintained cold temper-  Fig. 9 illustrates the PSC evidence from CALIOP observations. The hemispheric T min dropped below T NAT as early as on 16 November 2019, but increased OClO SCDs were observed on 21 November when T min was already lower than T' NAT (Fig. 7). In the third panel of Fig. 8 it can be further seen that this increase happened exactly when the local temperature 285 fell below T' NAT . Also nonzero PSC evidences (at longitudes 30°-60°E and few days later 0°-60°E) coincide with some of the increased OClO SCDs (Fig. 9). In the third panel of Fig. 8 it can further be seen that the OClO SCDs show a new enhancement when the temperatures again drop T' NAT at the beginning of December. Also PSCs are reported (Fig. 9, middle panel) as evident at a few longitudes (mainly 60°-90°E). With temperatures staying at these low levels or even dropping below T ICE the OClO SCDs almost linearly increase till the end of the second week of January 2020. More variation can be seen in the polar mean 290 and maximum hemispheric PSC evidences which increase by an order of magnitude whenever T min drops below T ICE . This increase in the PSC evidence however seems not to have a clear relation with the observed OClO SCDs. Since mid January, with temperatures still being low, the OClO SCDs remain nearly constant at about 2.5×10 14 cm −2 till mid March. During that period in several occasions (10, 20 February, 16 March) air masses with slightly enhanced OClO SCDs appear to be mixed outside the polar vortex in air masses with low PV values (8, bottom panel). Also the opposite happens at 21-26 February 295 when enhanced OClO SCDs appear only at very high PV values. In the last two weeks of March the stratosphere starts to heat up, there is also no evidence of PSCs in the CALIOP data reported anymore and the OClO SCDs decrease reaching almost zero at the end of the month although there is still a small area with temperatures below T NAT at lower altitudes.  below T NAT . An indication for a local OClO activation would however be the PSC evidence values that were slightly above zero since the beginning of May (Fig. 12, middle panel). These values (at longitudes around 15°E -60°W) seem however not to have a clear relation with the collocated OClO SCDs (Fig. 12, top panel) which are larger at other longitudes (60°-120°E) than at the collocated longitudes. However, when also the local temperatures drop below T NAT (starting with 20 May), clearly 315 enhanced OClO SCDs appear, despite the local PSC evidence being above zero only once in these days at the end of May and at a single longitude ( 10°E) where at the same time the polar mean and maximum PSC evidence increases distinctively.

Area with T < T NAT
Here also the time series of OClO SCDs resolved with respect to temperature shows larger OClO SCDs at temperatures close to T NAT . Even 'trails' with increased OClO SCDs starting at locations with elevated surface heights (black contourlines in the longitudinally resolved time series of OClO SCDs plot in Fig. 11) and transported eastwards with time are observed indicating 320 chlorine activation induced by a possible PSC formation due to mountain wave activity. A more consistent PSC evidence in 20 https://doi.org/10.5194/acp-2021-600 Preprint.      (boundary) are observed. Finally, at the end of September to the beginning October a rather quick chlorine deactivation occurs despite the fact that the temperatures are still below T NAT and the polar vortex is stable. Besides a relation with the decrease in PSCs evidence as observed by CALIOP (or at least PSCs descending to lower altitudes not covered by the considered altitude range of >4 km above the tropopause) at the end of September, also the mechanism of chlorine deactivation as described by Grooß et al. (2011) can play a role: when an almost complete destruction of ozone occurs, almost all chlorine becomes bound 340 in HCl and cannot be reactivated.

Winter 2019
The winter 2019, however, was quite unique as a minor sudden stratospheric warming was observed, which was just a bit weaker than the major sudden stratospheric warming in 2002 Klekociuk et al., 2021). Also a very small ozone  The daily mean and maximum OClO SCDs (see Fig. 13) show a similar temporal development as in 2018 until 6 September.
Also clearly increased OClO SCDs at local temperatures below T NAT (middle May) and even more increased OClO SCDs at 350 local temperatures below T' NAT (from the beginning of June) are observed (Fig. 14). From beginning of June also evidence for PSCs at the locations with increased OClO SCDs are consistently observed (Fig. 15). After the stratospheric warming (6 September), the area with temperature below T NAT decreases rapidly and the hemispheric minimum temperature rises above T NAT (at PT 475 K) by the end of the third week of September. The decrease and the rise are accompanied by a strong decrease of the OClO SCDs with a rather constant rate till the end of September. After 6 September also the PSC evidence (both local, 355 as well as the polar mean and maximum) observed by CALIOP becomes almost zero. At the beginning of October the OClO SCDs decrease further at a lower rate. Interestingly, two distinct temperature drops at lower altitudes (at PT around 400 K) lead to two small short-term increases in the mean and maximum OClO SCDs. Looking on the parameter (longitude, temperature and PT) resolved time series (Fig. 14) one can notice that the high OClO SCDs appear at rather high local temperatures and low PV values already on 11 August and more clearly on several days 360 after 18 August. Also a mixing towards low PV values after 5 September can be seen being especially strong at the beginning of the second week of this month which coincides with the sudden warming episode. The small chlorine activation events at the beginning of October can be seen well distinguished in all parameter resolved time series of OClO SCDs occurring at the lowermost temperatures and the highest PV values. We can speculate that this potential for a further chlorine activation indicates that not all ozone in the polar vortex was destroyed by the initially activated chlorine. This indicates that chlorine 365 could in principle be reactivated again if the temperatures become low enough, as it is usually the case in the Arctic.