Polar stratospheric clouds (PSCs) form in the polar winter stratosphere at altitudes between 15 to 30 km. PSCs consist of liquid and solid particles and have been classified into three different types based on their composition and physical state: (1) supercooled ternary solutions (STS), (2) Nitric Acid Trihydrate (NAT) and (3) ice. The formation of PSCs is strongly temperature dependent. Liquid PSC cloud particles (STS) form by the condensation of water vapour (H2O) and nitric acid (HNO3) on the liquid stratospheric background sulfate aerosol particles at temperatures 2–3 K below the NAT existence temperature TNAT (∼195 K at 20 km) while for the formation of solid cloud particles (ice) much lower temperatures are required, usually 3–4 K below the ice frost point Tice (∼185 K at 20 km) e.g.. The formation of the liquid STS particles is quite well understood, however, the exact formation mechanism of NAT and ice PSC particles still leaves some unresolved questions and is still an active area of research.

Progress in understanding PSC particle formation processes has been made recently in the frame of the European project RECONCILE (Reconciliation of essential process parameters for an enhanced predictability of Arctic stratospheric ozone loss and its climate interactions) . For example, CALIPSO measurements for the Arctic winter 2009/10 presented by showed that widespread NAT formation occurred, albeit in low number densities, before ice clouds had been formed at temperatures well above Tice. Further, lidar measurements performed during recent years have also indicated that there must be a formation mechanism for NAT PSCs above the ice frost point Tice without ice particles necessarily serving as a nucleation kernel for NAT particles. Until the RECONCILE project, the only known pathway to form NAT was through heterogeneous nucleation on ice particles, forming NAT clouds downstream of mountain wave ice clouds .

Heterogeneous nucleation on particles such as meteoric smoke has been suggested to be a potential pathway for NAT formation . performed box model simulations along air parcel trajectories based on the observations by CALIPSO made during the Arctic winter 2009/10 applying a new parameterisation for heterogeneous NAT nucleation, assuming NAT formation on particles as e.g. meteoric smoke. The CALIPSO observations were well reproduced by the model simulations applying this new parameterisation thus indicating that NAT nucleation on other particles than ice is possible. Further, both the modelling study by and the one by using the Zurich Optical and Microphysical box Model (ZOMM) showed that small-scale temperature fluctuations usually not represented in meteorological data needed to be considered to reproduce the CALIPSO observations.

Denitrification, the permanent removal of HNO3 by sedimenting polar stratospheric cloud particles, limits the deactivation process of the ozone destroying substances in springtime and thus leads to a prolongation of the ozone destroying cycles. Stratospheric cooling caused by increasing greenhouse gas concentrations will have significant implications on denitrification and ozone loss. Model simulations predict that very large ozone losses will occur more frequently in the future in the Arctic and that the recovery of the ozone layer will be delayed by more than a decade due to increased greenhouse gas concentrations e.g..

Water vapour is one of the most important greenhouse gases and plays a key role in the chemistry and radiative balance of the upper troposphere and lower stratosphere (UT/LS). Several studies have been performed in the past investigating stratospheric water vapour trends using in situ and remote-sensing measurements e.g.. combined 10 data sets covering the time period 1954–2000 and found a 1 % yr-1 (0.054 ppmv yr-1) increase in lower stratospheric water vapour in the mid-latitudes. Long-term balloon-borne measurements at Boulder, Colorado (40∘ N/105∘ W) indicate an increase of lower stratospheric water vapour abundances, on average by 1 ppmv, during the last 30 years (1980–2010) . Recently, analysed a merged satellite time series spanning from the late 1980s to 2010, which did not confirm the findings from the Boulder data set, arguing the representativeness of these data on a larger spatial scale. In the lower stratosphere negative changes were dominating, while positive changes were found only in the upper part of the stratosphere. The decrease in the lower stratosphere was attributed to a strengthened lower stratospheric circulation. A decisive role here played a pronounced drop in water vapour in 2000 (also known as the millennium drop) , that first started to recover in 2004 to 2005. This drop was caused by a reduced transport of water vapour from the troposphere into the stratosphere in response to a colder tropical tropopause. The temperature decrease has been due to variations of the QBO (quasi-biennial oscillation), ENSO (El Niño Southern Oscillation) and the Brewer-Dobson circulation that collectively acted in the same direction lowering the tropopause temperatures. In 2011 such a drop happened again, however more short-lived .

analysed satellite data together with a trajectory model. They did not see any firm evidence of trends (neither positive nor negative) in the data since the mid 1980s. However, they cannot rule out that a trend exists that is just too small to be identified given the large inter-annual and inter-decadal variability. presented model simulations of 18 coupled Chemistry Climate Models (CCMs) in the tropical tropopause layer (TTL). The models simulate decreases in the tropopause pressure in the 21st century, along with ∼1 K increases per century in cold point temperature and 0.5–1 ppmv per century increases in water vapour above the tropical tropopause.

Any changes in atmospheric water vapour bring important implications for the global climate. Increases in stratospheric water vapour cool the stratosphere but warm the troposphere. Both the cooling of the stratosphere and the increase in water vapour enhance the potential for the formation of polar stratospheric clouds. More than a decade ago it was already suggested that a cooling of stratospheric temperatures by 1 K or an increase of 1 ppmv of stratospheric water vapour could promote denitrification . During the two Arctic winters 2009/10 and 2010/11, the strongest denitrification in the recent decade was observed . In the latter winter, denitrification led also to severe ozone depletion with a magnitude comparable to the Antarctic “ozone hole” .

In this study, the correlation between observed water vapour variability and the recent temperature evolution in the Arctic together with PSC observations are considered to investigate a possible connection between the increase in stratospheric water vapour and polar stratospheric cloud formation/denitrification. This study aims at (1) performing a sensitivity study on how an increase in water vapour and decrease in temperature will affect PSC formation and existence and (2) on assessing the H2O variability during the 15-year period 2000–2014.

A sensitivity study is performed to investigate what effect changes in water vapour and temperature (due to a trend or variability) would have on PSC formation and occurrence. Therefore, air parcel back trajectories are calculated according to PSC observations by CALIPSO during the Arctic winter 2010/11. On the basis of this trajectory ensemble the increase in time the air parcels would be exposed to temperatures below TNAT or Tice along the trajectories are considered for a water vapour increase of up to 1 ppmv and a temperature decrease of up to 1 K in the stratosphere.

To investigate a possible water vapour trend as well as water vapour variability in the polar lower stratosphere, satellite observations of water vapour from the Odin Sub-Millimetre Radiometer (Odin/SMR), the Aura Microwave Limb Sounder (Aura/MLS), the Envisat Michelson Interferometer for Passive Soundings (Envisat/MIPAS), the Scanning Imaging Absorption spectrometer for Atmospheric Chartography (Envisat/SCIAMACHY), the SCISAT Atmospheric Chemistry Experiment Fourier Transform Spectrometer (SCISAT/ACE-FTS) and the UARS Halogen Occultation Experiment (UARS/HALOE) are used. A short description of these satellite instruments will follow below. A detailed intercomparison of water vapour derived from these instruments can be found in . For performing case studies along air parcel trajectories that are based on PSC measurements we apply measurements from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) on board of CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations).

The Arctic winter 2010/11 was one of the coldest in the last 2 decades . The 2010/11 winter was characterised by an anomalously strong vortex with an atypically long cold period that was persistent from mid-December to mid-March . The polar vortex formed at the end of November 2010 and remained stable until the end of April. The long cold period, lasting over 4 months, was interrupted by short warmer periods in the beginning of January, February and March due to minor warmings. In February and March, temperatures were colder than in previous years of the last decade. The final warming during the 2010/11 Arctic winter occurred in mid-April, thus later than usual .

The PSC season during the 2010/11 winter can be divided into four PSC phases according to the four cold phases that occurred over the 4-month period from December 2010 to March 2011. The time periods of these four phases and the PSC types that occurred during each phase were derived from CALIPSO observations and are as follows : (1) 23 December 2010 to 8 January 2011: STS, Mix 1/2 and ice clouds. (2) 20–28 January 2011: mainly Mix-1 and Mix-2 with some STS and ice. (3) 5–27 February: STS, Mix-1 and Mix-2 as well as ice clouds. (4) 5–19 March: STS clouds (Note: no CALIPSO data are available from 8 to 13 March).

A graphic presentation of the temporal evolution of time, VPSC, T-TNAT and several trace gases during this winter can be found in and . performed a similar analysis on PSC occurrence as we did, but used MIPAS observations for PSC detection. They also derive four PSC phases from MIPAS, however with somewhat different time periods for each phase as the ones we derive from CALIPSO. Differences in PSC detection between both instruments are caused by the different measurement principle (active lidar in the visible vs. passive spectroscopy in the infrared spectral region). CALIPSO generally detects PSCs in a greater fraction than MIPAS. This can be explained by the patchier nature of PSCs in the Arctic and the different spatial resolutions of the two instruments, which makes the clouds more likely to be detected by the CALIPSO lidar . The Arctic winter 2010/11 has been well analysed, especially with respect to ozone loss while the dynamical perspective, thus the exceptional dynamical conditions of this winter so far were only discussed in detail by .

The sensitivity study is performed based on measurements of PSCs by CALIPSO during the Arctic winter 2010/11. The basic approach is demonstrated on single trajectories, but the final results rely on a statistical assessment of a trajectory ensemble. Based on the PSCs observed by CALIPSO during the Arctic winter 2010/11 air parcel trajectories were calculated 6 days backwards with the NOAA HYSPLIT (Hybrid Single Particle Lagrangian Integrated Trajectory Model) model based on GDAS (Global Data Assimilation System) analyses

http://ready.arl.noaa.gov/HYSPLIT.php

The Arctic winter 2010/11 has been chosen for the sensitivity study because it was one of the coldest Arctic winters leading to a high number of PSC occurrences. This makes the statistics more reliable than if we would have chosen a warmer winter with less PSC occurrences. During the Arctic winter 2010/11 PSCs were detected by CALIPSO on 47 days on 259 orbit tracks. In total, 738 trajectories were calculated based on the CALIPSO observations and considered for the sensitivity study on the trajectory ensemble. For the sensitivity study on single trajectories we selected two trajectories, one trajectory where temperatures below TNAT were reached, but not below Tice and one where temperatures below both, TNAT and Tice, were reached along the trajectory. Figure  shows a map with locations where the trajectories were started according to the PSCs detected by CALIPSO, colour coded by the four cold phases during the Arctic winter 2010/11. PSCs were observed around Greenland during Phase 1, during Phase 2 PSCs were observed over Russia and during Phase 3 over the entire Arctic. During Phase 4 only a few PSCs were detected which were located around Greenland

Note: no CALIPSO observations are available from 8 to 13 March.

In the previous section we demonstrated that a water vapour increase and temperature decrease would increase the potential for PSC formation. More than a decade ago it was already suggested that a cooling of stratospheric temperatures by 1 K or an increase of 1 ppmv of stratospheric water vapour could promote denitrification . During the two Arctic winters 2009/10 and 2010/11, the strongest denitrification in the recent decade was observed .

Here, we investigate the variability of Arctic water vapour and temperature since the new millennium to see if there is a connection to the severe denitrification observed in the past years. For that we used observations from UARS/HALOE, Odin/SMR, Envisat/MIPAS, SCISAT/ACE-FTS, Envisat/SCIAMACHY and Aura/MLS as well as ERA interim reanalysis data. In a first step these data sets were interpolated onto a regular potential temperature grid with a resolution of 25 K. Temperature and pressure data needed for the conversion from the native vertical coordinate to potential temperature were taken from the individual data sets themselves. Then for every profile the equivalent latitude as function of altitude was derived from ERA interim potential vorticity data. Data within high equivalent latitudes, i.e. between 70 to 90∘ N, were subsequently binned into monthly and zonally averaged time series. Finally the resulting time series were de-seasonalised to make variability on inter-annual to decadal scales more visible. For a given month a multi-year average was calculated from the individual monthly averages which was then subtracted from the latter. The multi-year averages were based on the entire time period covered by the individual data sets.

Figure  shows the de-seasonalised time series for temperature (upper panel) and water vapour (lower panel) covering the time period 2000–2014 averaged over the potential temperature range 475–525 K (∼ 18–22 km). Temperature information is provided by ERA-Interim, Aura/MLS, Envisat/MIPAS and SCISAT/ACE-FTS; for water vapour there is in addition also data from UARS/HALOE and Envisat/SCIAMACHY. In terms of temperature there is a very good agreement among the different data sets. The overall variability is dominated by the winter season. Strong inter-annual variability can be found for this season with pronounced cold and warm events that show absolute anomalies of more than 10 K. For water vapour there is more scatter and less consistency between the individual data sets. Deviations exceed occasionally 0.5 ppmv, which is a typical level of uncertainty of these observations . Unlike temperature the water vapour variability is not as apparently dominated by the winter season. The pronounced events seen in temperature can be still observed in water vapour in an anti-correlated sense, but not as obvious. There are indications of a substantial decrease throughout 2003, however not in all data sets. A similar feature can be observed in 2011. Likewise there are indications of an increase in 2006 and 2007. Overall this demonstrates significant changes over short time periods.

Figure  is of the same kind as Fig.  however it considers the potential temperature range between 525 and 825 K (∼ 22–28 km). Please note that there are no water vapour data from Envisat/SCIAMACHY available here. For temperature very similar characteristics can be observed as in Fig. . This is also true for water vapour. The scatter is somewhat reduced providing a clearer picture on inter-annual to decadal variability. In particular the variations in the aftermath of sudden stratospheric warmings in early 2009 and 2013 are very pronounced here .

Visually, both Figs.  and do not indicate any obvious linear changes in temperature or water vapour overall. To investigate this aspect more rigorously we performed separately a regression analysis of the Envisat/MIPAS and the Aura/MLS time series. The selection of these two data sets was based on their extensive and regular observational coverage. The regression model considered an offset, a linear term, periodic variations of 3, 4, 6 and 12 months as well as the QBO in the form of Singapore winds at 50 and 30 hPa

http://www.geo.fu-berlin.de/met/ag/strat/produkte/qbo/index.html

In the present work we were focusing on the polar regions to understand how water vapour and temperature changes in the lower polar stratosphere affect PSC formation which eventually can also have an impact on denitrification. The influence of water vapour and temperature on PSCs is two-fold. There is a background component and on top there is inter-annual variability. The long-term changes of low-latitude water vapour described in the introduction eventually also influence the high latitudes. On the other hand vortex dynamics play an essential role for PSC variations from year to year. Over the time period from 2002 to 2012 Envisat/MIPAS observations indicate a significant change in water vapour only between 375 and 450 K in form of an increase. Aura/MLS observations considering the time period from 2004 to 2014 show positive changes over a wider altitude range, many of them even significant. Figure  indicates that the later start and end of the Aura/MLS measurements relative to the MIPAS observations could play a decisive role for this difference. In principle an increase is expected. After the long-standing millennium drop a new increase has been observed in the tropics, starting around 2004 to 2005 . This increase was only interrupted by the 2011 drop that was however much more short-lived than the millennium drop. By early 2014 the volume mixing ratios had more or less recovered . For the time period since 2002 we cannot find any significant changes in temperature in the altitude range between 350 K and 1000 K potential temperature.

The inter-annual variability component is driven by the vortex-related dynamics and other short-term variations, like QBO and drops in water vapour. During wintertime we found in the satellite data a significant correlation between cold/warm winters and enhanced/reduced water vapour mixing ratios. This correlation indicates a connection between dynamical processes that influence the polar winter dynamics. On the one hand there is the subsidence within the polar vortex and its variations. On the other hand there are sudden stratospheric warmings that break up the polar vortex and completely revert the dynamical conditions. During polar winter vigorous descent occurs within the polar vortex, transporting air masses from the upper stratosphere and mesosphere down to the lower stratosphere . As water vapour typically exhibits a maximum around the stratopause this descent also transports moister air towards the lower stratosphere. analysed SCIAMACHY data from 2002–2009 and found that the QBO west phase is associated with larger PSC occurrences and stronger chemical ozone destruction than the QBO east phase. Their findings are in agreement with the Holton-Tan mechanism which relates the QBO west phase to a colder and more stable vortex. During cold Arctic winters, as 2010/2011, the subsidence within the polar vortex is strongly enhanced as shown e.g. by , causing positive water vapour anomalies. This explains already qualitatively the correlation we observed. As an amplifying effect act sudden stratospheric warmings, like in early 2009 or 2013, that led to high positive temperature anomalies and low water vapour. The latter are mainly explained by the poleward transport of dry air once the horizontal mixing barrier in the form of the polar vortex edge is removed.

The signatures of the water vapour drops in 2000 and 2011 are not easily distinguishable in the Arctic. In the altitude range between 475 to 525 K the decrease throughout 2003 may be attributed to the millennium drop. Arctic observations of POAM III indicated the drop already in early 2001 . This seems to be consistent with studies by that showed a delay of up to 12 months between the drop occurrence in the tropics and at 50 ∘ latitude at these low altitudes. The UARS/HALOE observations employed here do not show a clear sign of a decrease in 2001, however admittedly the measurement coverage of this instrument has not been optimal for these high latitudes. The decrease in the Arctic in 2011 may correspond to the drop observed in the tropics. Yet, the length of the decrease is shorter than observed at the low latitudes. Higher up, between 525 K and 825 K potential temperature, a longer delay to the drop occurrence in the tropics can be expected . Thus, the decrease observed here in 2002 and 2003 is more likely attributed to the millennium drop. The decrease in 2011 on the other hand is unlikely to be connected to the tropical event.

Overall, based on the observations since the new millennium we can conclude that such strong denitrification events as in 2010/11 are for the time being driven by inter-annual variability than any changes in water vapour and/or temperature since the millennium.

By performing sensitivity studies we, on one hand, investigate what implications water vapour and temperature changes and/or variability would have on PSC formation and existence. On the other hand this sensitivity study also shows what implications uncertainties in water vapour from measurements and temperature measurements and/or reanalyses would have on Arctic studies. Gravity waves can affect PSC occurrence and composition in the Arctic and Antarctic . Temperature perturbations that are caused by gravity waves are usually not represented in meteorological analyses. Further, meteorological analyses tend to have cold or warm biases as was shown by e.g. . PSC formation and existence is quite sensitive to temperature and water vapour changes as well as uncertainties of these. An increase of e.g. 1 ppmv in water vapour and a decrease of 1 K in temperature would significantly alter the estimates of PSC volume and area (VPSC and APSC, respectively) and thus would affect the estimates on e.g. ozone loss and chlorine activation . However, irrespective of if there is a cold or warm bias in the trajectory temperature or in the water vapour mixing ratio in the stratosphere, an increase in water vapour mixing ratios or a cooling of temperature will definitely result in a prolongation of the potential for PSC existence as shown in our sensitivity study.

Our sensitivity study was performed on the basis of the 2010/11 winter, which was the coldest Arctic winter in the recent decade. The anomalously strong polar vortex and the atypically long cold period that persisted from mid-December to mid-March led to extensive PSC formation. PSCs were observed by CALIPSO over 47 days during the Arctic winter 2010/11. During the Arctic winter 2009/10, a cold period extended over 4 weeks, from mid-December to mid-January, and PSCs were only observed on 26 days. Thus, if this study would have been performed on the basis of another (warmer) winter less PSC observations would have been served as a basis for the trajectory calculations and thus as a basis for the statistic. As a consequence the total times where the temperature was below TNAT or Tice, respectively, would have been shorter as for the Arctic winter 2010/11. However, the resulting increase in time due to a decrease in temperature and an increase in water vapour can be expected to be similar, thus as dramatic as for the 2010/11 winter.

As shown in this study, increases in stratospheric water vapour as well as decreases in stratospheric temperature can prolong PSC formation and existence. An increase of water vapour of 1 ppmv and a decrease of temperature of 1 K increased the times where the temperature is below TNAT by 38 %. A much stronger increase in time was found for ice. The increase in time where temperatures were below Tice were enhanced by a factor of 12 for an increase of H2O by 1 ppmv and a temperature decrease of 1 K. Generally, temperatures sufficiently low for ice formation are rarely reached in the Arctic. Therefore, this strong increase in time where temperatures would be below Tice would mainly be of importance for very cold, extreme Arctic winters as e.g. the Arctic winter 2010/11. However, if cold Arctic winters will become colder in the future as suggested by , ice formation will also become more common in the Arctic. Thus, changes in stratospheric H2O mixing ratio and temperature can significantly alter PSC formation and existence and thus the chemistry of the polar stratosphere, as e.g. increasing denitrification and thus ozone loss.

The Arctic winter 2010/11 was one of the coldest in the last 2 decades. The 2010/11 winter was characterised by an anomalously strong vortex with an atypically long cold period that was persistent from mid-December to mid-March. During the Arctic winter 2010/11 the strongest denitrification during the recent decade was observed. More than a decade ago it was already suggested that a cooling of stratospheric temperatures by 1 K or an increase of 1 ppmv of stratospheric water vapour could promote denitrification, the removal of HNO3 by sedimenting PSC particles.

Based on the 2010/11 winter a sensitivity study was performed to investigate how a change of up to 1 ppmv in water vapour and temperature decrease of up to 1 K (due to a trend or variability) would affect PSC formation and occurrence. Air parcel trajectories were calculated 6 days backward according to PSC observations by CALIPSO. In total, 738 trajectories were calculated. On the basis of this trajectory ensemble the increase in time the air parcels would be exposed to temperatures below TNAT or Tice along the trajectories was calculated. Measurements from several different satellites derived for the 15-year period 2000–2014 were used together with temperatures from ECMWF to investigate water vapour trends and variability in the polar stratosphere. So far trend studies on stratospheric water vapour have focused on the tropics and mid-latitudes. Here, for the first time such an analysis has been performed for the polar stratosphere.

Our sensitivity studies based on air parcel trajectories confirm that PSC formation is quite sensitive to water vapour and temperature changes. Increased H2O (and further cooling of the stratosphere) would increase the potential for PSC formation and prolong PSC existence and thus the chemistry of the polar stratosphere, as e.g. increasing denitrification and thus ozone loss. On the other hand an increase in temperature and a decrease in water vapour will reduce PSC formation and existence. From the Envisat/MIPAS (2002–2012) and Aura/MLS (2004–2014) observations we derive predominantly positive changes in the altitude range between 350 to 1000 K potential temperature. Those from Envisat/MIPAS observations are largely insignificant, while those from Aura/MLS are mostly significant. For the temperature neither of the two instruments indicate any significant changes. Given the strong inter-annual variation we observed in water vapour and particular temperature the severe denitrification observed in 2010/11 cannot be directly related to any changes in water vapour and/or temperature since the millenium. However, we found from the satellite observations that cold winters coincide with high water vapour mixing ratios, as e.g. for the winters 2006/07, 2007/08 and 2010/11. This correlation is quite significant in the lower stratosphere at 475–525 K and indicates a connection between dynamical and radiative processes that govern water vapour and temperature in the Arctic lower stratosphere.

This work is dedicated to our highly valued colleague Jo Urban who passed away much too early. Without his devoted work on the Odin/SMR data processing over many years this work would not have been possible. In particular the retrieval of water vapour from the SMR observations and the combination of these data with other data sets to understand the long-term development of this trace constituent comprised a large part his life's work. With his death, we lost not only a treasured colleague and friend, but also a leading expert in the microwave and sub-millimetre observation community.