Has Stratospheric HCl in the Northern Hemisphere Been Increasing Since 2005?

Stratospheric hydrogen chloride (HCl) is the main stratospheric reservoir of chlorine, deriving from the decomposition of chlorine-containing source gases. Its trend has been used as a metric of ozone depletion or recovery. Using the latest satellite observations, it is found that the significant increase of Northern Hemisphere stratospheric HCl during 2010–2011 can mislead the trend of HCl in recent decades. In agreement with previous studies, HCl increased from 2005 to 2011; however, when the large increase of stratospheric HCl during 2010–2011 is removed, the increasing linear trend from 2005 to 2011 becomes weak and insignificant. In addition, the linear trend of Northern Hemisphere stratospheric HCl from 2005 to 2016 is also weak and insignificant. The significant increase of HCl during 2010–2011 is attributed to a strong northern polar vortex and a weakened residual circulation, which slowed down the transport of HCl between the low-mid latitudes and the high latitudes, leading to an accumulation of HCl in the middle latitudes of the stratosphere. In addition, a weakened residual circulation leads to enhance conversion of chlorine-containing source gases of different lifetimes to HCl, thus increasing the levels of HCl. Simulations by both chemistry transport and chemistry-climate models support the result. It is further found that the joint effect of a La Niña event, the west phase of the quasi-biennial oscillation and positive anomalies of sea surface temperature in the North Pacific is responsible for the strong northern polar vortex and a weakened residual circulation.


INTRODUCTION
Hydrogen chloride (HCl) is important to the gas phase chemistry of ozone depletion because of its role as a reservoir species for chlorine in the stratosphere (Molina and Rowland, 1974;Jones et al., 2011;Kohlhepp et al., 2012;Mahieu et al., 2014;Rozanov, 2018;Zhang et al., 2018). Understanding the time variations of stratospheric HCl makes it possible to estimate the amount of active chlorinecontaining source gases, and further to monitor ozone recovery. Stratospheric HCl has increased from the 1950s due to the increasing trend of chlorofluorocarbon and chlorofluorocarbon photodissociation in the stratosphere as a result of human activities (Solomon, 1990). Owing to the effectiveness of the Montreal Protocol in 1987, and its amendments and adjustments, several studies have suggested a negative trend in the concentration of stratospheric HCl in the subsequent years (Brown et al., 2011;Jones et al., 2011). Using Halogen Occultation Experiment (HALOE) and Atmospheric Chemistry Experiment Fourier Transform Spectrometer data, Jones et al. (2011) found a negative trend of −5%/decade in HCl in the midlatitudes middle to upper stratosphere between 1997 and 2008. In addition, Brown et al. (2011) used Atmospheric Chemistry Experiment Fourier Transform Spectrometer data to diagnose the trend of HCl in the upper stratosphere and deduced a −7%/ decade declining trend from 2004 to 2010. Based on Microwave Limb Sounder (MLS) measurements, Froidevaux et al. (2006) derived a decrease in HCl of −0.78 ± 0.08%/year in the upper stratosphere for the years 2004-2006. In the 2014 ozone assessment report, Carpenter and Reimann (2014) summarized the results from these authors and concluded that the trend of HCl (50°N-50°S) in the middle and upper stratosphere is a mean decline of 0.6 ± 0.1%/year between 1997 and 2012. Besides the HCl trend in the middle and upper stratosphere, Kohlhepp et al. (2012) analyzed the total column HCl trend based on Fourier transform infrared (FTIR) data from 17 ground stations of the Network for the Detection of Atmospheric Composition Change (NDACC) and found a negative HCl trend ranging from −4 to −16%/decade from 2000 to 2009.
The above studies indicate that stratospheric HCl has decreased since 1997, which is consistent with the observed decline in chlorine source gases at the surface and model calculations due to the effectiveness of the Montreal Protocol (Kohlhepp et al., 2012). However, some recent studies have reported a significant increase of HCl in the lower stratosphere of the Northern Hemisphere (Mahieu et al., 2014;Stolarski et al., 2018;Han et al., 2019), which is in contrast to the ongoing monotonic decrease of near-surface source gases and has attracted a great deal of attention. Based on the measurements at eight NDACC ground stations and modeling results, Mahieu et al. (2014) found that stratospheric HCl exhibited an increasing trend in the Northern Hemisphere from 2005 to 2011. The authors attributed this trend to a slowdown of the Northern Hemisphere atmospheric circulation (i.e., the residual circulation) occurring over several consecutive years, transporting older air into the lower stratosphere, and characterized by a larger relative conversion of source gases to HCl. Subsequently, Stolarski et al. (2018) and Han et al. (2019) confirmed the increase of Northern Hemisphere stratospheric HCl from 2005 to 2011 using MLS measurements. They further pointed out the contribution of the dynamical variability of the atmosphere to the increase in HCl.
Here, we revisit the Northern Hemisphere stratospheric HCl variations from 2005 to 2012 using MLS measurements and Global Ozone Chemistry and Related trace gas Data Records (GOZCARDS) for the Stratosphere HCl observational data ( Figure 1). Consistent with previous results, we find that the Northern Hemisphere stratospheric HCl exhibited an increasing trend from 2005 to 2011 ( Figure 1A, black and red lines). However, when the variations in stratospheric HCl from 2005 to 2016 are considered, we found that HCl exhibits a weak decreasing trend ( Figure 1B, black lines). Another outstanding feature is a significant increase of Northern Hemisphere stratospheric HCl during (from December 2010to March 2011. When this significant increase of stratospheric HCl is removed, the increasing linear trend from 2005 to 2011 becomes weak and insignificant ( Figure 1A, blue lines). This illustrates that the significant increase in 2010-2011 plays a crucial role in the trend of Northern Hemisphere stratospheric HCl and can mislead the trend of HCl in recent decades. The remainder of this paper explores the relevant mechanisms for the significant increase of Northern Hemisphere stratospheric HCl during 2010-2011.

DATA, METHODS AND MODEL
The primary trace gas dataset used in this study is the version 4.2x MLS Level 2 data (Livesey et al., 2015), extending from 2005 to 2016. The MLS measures daily atmospheric chemical species which have a global coverage from 82°N to 82°S and a vertical resolution of ∼3 km (Livesey et al., 2015). We use the gridded MLS data at a 4°latitude × 5°longitude resolution and the quality screening rules for this dataset can be found in Livesey et al. (2015). The zonal-mean HCl from satellite-based Global Ozone Chemistry and Related trace gas Data Records for the Stratosphere (1991Stratosphere ( -2012 is produced from high quality data from past missions (e.g., HALOE data) as well as ongoing missions (ACE-FTS and Aura MLS) (Froidevaux et al., 2013). Its meridional resolution is 10°with 25 pressure levels from 147 up to 0.5 hPa.
In addition, the long-term simulations  from TOMCAT/SLIMCAT three-dimensional offline chemical transport model (Chipperfield, 2006) are used to diagnose the mechanisms that resulted in the significant increase of Northern Hemisphere stratospheric HCl during 2010-2011. The model has identical stratospheric chemistry and aerosol loading, solar flux input and surface mixing ratios of long-lived source gases as Chipperfield et al. (2018). The model uses horizontal winds and temperature from the reanalysis data of the European Centre for Medium-Range Weather Forecasts (ERA-Interim, Dee et al., 2011). Previous studies have found that the wind and temperature fields from the ERA-Interim reanalysis agree well with those from Modern-Era Retrospective analysis for Research and Applications-Version 2 (MERRA2), especially in middle and high latitudes (Rienecker et al., 2011;Lindsay et al., 2014). The long-term simulation  was performed with a coarse horizontal resolution of 5.625°latitude × 5.625°longitude and 32 levels from the surface to 60 km. The model uses a hybrid σ-p vertical coordinate (Chipperfield, 2006) with detailed tropospheric and stratospheric chemistry. Vertical advection is calculated from the divergence of the horizontal mass flux (Chipperfield, 2006), and chemical tracers are advected, conserving second-order moments (Prather, 1986). The TOMCAT/SLIMCAT model has been extensively evaluated against various HCl satellite and sounding datasets and provides a good representation of stratospheric chemistry (e.g., Chipperfield, 2006;Feng et al., 2007Feng et al., , 2011. The meterological fields used in this study are from ERA-Interim reanalysis data set with a horizontal resolution of 1°l atitude × 1°longitude and 37 vertical pressure levels. More details about ERA-Interim reanalysis data are described by Dee et al. (2011). In addition, the variations in North Pacific sea surface temperature (SST) from the Hadley Center HadISST dataset (Rayner et al., 2006) are analyzed, which has a 1°by 1°h orizontal resolution. In this study, the North Pacific (40-50°N, 160-200°E) SST anomalies are defined as the December, January, and February (DJF) mean SST removing the 1979-2016 climatology [please refer to Nakamura et al. (1997) and Hurwitz et al. (2011)].
In this study, the transformed Eulerian mean residual circulation (v p , w p ) is used to diagnose the residual circulation (Andrews et al., 1987). In the pressure coordinates, they are defined by Here, overbars donate zonal means and primes are deviations from the zonal mean of a given variable. z is logpressure height and φ is the latitude. v and w are meridional and vertical velocity, respectively. ρ 0 is density which is defined as ρ 0 ρ s exp(−z/H) with H 7.0 km θ is potential temperature and a is Earth's radius.
To identify planetary wave fluxes in the atmosphere, Eliassen-Palm fluxes (EP fluxes) are employed and are calculated using the formula given by Andrews et al. (1987).
We also use version 4 of the Whole Atmosphere Community Climate Model (WACCM4) in this study since WACCM has been shown to have a good performance in simulating the stratospheric circulation, temperature and chemical composition variations (Garcia et al., 2007). WACCM4 is part of the Community Earth System Model framework developed by the National Center for Atmospheric Research (NCAR). WACCM4 uses a finite-volume dynamical core, with 66 vertical levels extending from the ground to 4.5 × 10 -6 hPa (145 km geometric altitude), and a vertical resolution of 1.1-1.4 km in the tropical tropopause layer and the lower stratosphere (below a height of 30 km). The simulations presented in this paper are performed at a horizontal resolution of 1.9°× 2.5°and with interactive chemistry (Garcia et al., 2007). More details regarding WACCM4 are provided in Marsh et al. (2013). The time-slice simulations presented in this paper (Table 1) are performed at a resolution of 1.9°× 2.5°, with interactive chemistry.

ATTRIBUTION OF THE SIGNIFICANT INCREASE OF HCL DURING 2010-2011 Observational Results
After the Montreal Protocol in 1987, the near-surface total chlorine concentration showed a maximum in 1993, followed by a decrease of half a percent to one percent per year (WMO, 2010), in line with expectations. In addition, Mahieu et al. (2014) found that there is no evidence that unidentified chlorine-  (Rayner et al., 2006), are averaged over the period 1979-2016. QBO phase signals for 28 months (fixed cycle) were included in WACCM4 as an external forcing of zonal wind. Monthly mean climatologies of surface emissions used in the model were obtained from the A1B emissions scenario developed by the Intergovernmental Panel on Climate Change, averaged over the period 1979-2016 R2 As for R1, but with La Niña SST anomalies added to the SST forcing in all 12 months of the year R3 As for R2, but with QBO phase signals removed R4 As for R1, but with warmer SST anomalies over the NP (40-50°N, 160-200°E) added to the SST forcing in all 12 months of the year a Experiments were performed for a period of 43 years, with the first 3 years excluded for model spin-up. Only the remaining 40 years were used for the analysis.
Frontiers in Earth Science | www.frontiersin.org December 2020 | Volume 8 | Article 609411 containing source gases are responsible for the increase of HCl from 2005 to 2011. Thus, the significant increase in Northern Hemisphere stratospheric HCl during 2010-2011 is unlikely to have been caused by chemical decomposition due to increased emission sources. Thus, dynamical processes are mainly considered in this study. Figure 2 shows the month-to-month evolution of Northern Hemisphere stratospheric HCl anomalies and zonal-mean wind ( (Figures 2A-D). In April 2011, the negative U in the high latitudes implies the collapse of the polar vortex ( Figures 2E,F). According to the evolution of stratospheric HCl anomalies and U in Figure 2, a possible mechanism for the increase of stratospheric HCl during 2010-2011 is that the strong polar vortex during 2010-2011 slows down the exchange of air between low-mid latitudes and high latitudes. Transport of the HCl-rich air in the low-mid latitudes into the high latitudes is reduced, and transport of the HCl-poor air in the high latitudes into the low-mid latitudes is also reduced. This explains why we find an increase in HCl in the middle latitudes of the middle stratosphere but a decrease in the high latitudes of the lower stratosphere from December 2010 to March 2011 (Figures 2A-D). By April 2011, the polar vortex had collapsed, speeding up the exchange of HCl rich air between low-mid latitudes and high latitudes. This may explain why the center of positive middle stratospheric HCl anomalies moves to the high latitudes of the lower stratosphere ( Figures 2E,F).
Previous studies have documented that the residual circulation transports chemically and radiatively important trace gases from the tropics to the polar regions, and thus has a significant impact on the global-scale distribution of stratospheric trace gases (Butchart and Scaife, 2001;Austin and Wilson, 2006;Randel et al., 2006;Youn et al., 2006;Xie et al., 2008;Hegglin and Shepherd, 2009;Austin et al., 2010;Eyring et al., 2010;Shu et al., 2011;Seviour et al., 2012;Wang and Waugh, 2012;Abalos et al., 2013;Lin and Fu, 2013;Butchart, 2014;Remsberg, 2015;Han et al., 2018). Therefore, the other possible relevant process is the air mass transport by the residual circulation. The month-tomonth evolution of stratospheric residual circulation anomalies from December 2010 to May 2011 is shown in Figure 3. Starting in December 2010, there were upwelling anomalies in the high  Figure 5A shows the time series of the strength of the northern polar vortex index, which is defined as PV averaged over Siberia 60°-75°N, 60°-90°E between the isentropic layer 430-600 K, referring to Zhang et al. (2016). Deep residual circulation is driven by the breaking of planetary waves (EP flux convergence) in the extratropical stratosphere in winter via the "downwardcontrol principle" (Haynes et al., 1991;Holton et al., 1995). As the vertical EP flux reflects the upward propagation of planetary waves, the vertical component of EP flux (Fz) in the lower stratosphere region 45-75°N may serve as a proxy for the residual circulation (Newman et al., 2001;Li and Thompson, 2013). Thus, we use Fz at 50 hPa in 45-75°N to represent the intensity of the residual circulation and the wave activity ( Figure 5B). As expected, in 2010-2011 the northern polar vortex index is strongest and the derived residual circulation Fz is weakest within the period from 2005 to 2016. This indicates a strong northern polar vortex and weakened residual circulation in 2010-2011, corresponding to the large increase of HCl as analyzed above.
The above analysis suggests that the strengthened northern polar vortex and weakened residual circulation together contribute to the large increase of HCl in 2010-2011. Next, we will try to diagnose the main processes responsible for the strong polar vortex and weakened residual circulation during 2010-2011. Figure 1 suggests that the time series of HCl anomalies is mainly characterized by interannual variability, which may be Frontiers in Earth Science | www.frontiersin.org December 2020 | Volume 8 | Article 609411 5 caused by factors with interannual variations; e.g., the El Niño-Southern Oscillation (ENSO), the Quasi-Biennial Oscillation (QBO), and Stratospheric Sudden Warming (SSW). SSW, El Niño, and the east phase of the QBO (EQBO) generally correspond to strong tropospheric wave activity and a weaker polar vortex, while La Niña and the west phase of the QBO (WQBO) correspond to weaker tropospheric wave activity and a stronger polar vortex (Taguchi and Hartmann, 2006;Wei et al., 2007;Garfinkel and Hartmann, 2008;Calvo and Garcia, 2009;Chen and Wei, 2009;Xie et al., 2012Xie et al., , 2020Richter et al., 2015;Zhang et al., 2015). In addition, warming of the North Pacific SST (SST NP ) is also associated with weak tropospheric wave activity and a stronger polar vortex (Hurwitz et al. (2011)). Therefore, La Niña events, WQBO and positive SST NP are considered in this study and their occurrences are shown in Table 2. It is found that only during 2010-2011 did strong La Niña, the WQBO and positive SST NP events occur simultaneously. This suggests that a joint effect of La Niña, WQBO and positive SST NP events may have caused the weakened wave activity (Figure 5
Frontiers in Earth Science | www.frontiersin.org December 2020 | Volume 8 | Article 609411 7 vertical lines, respectively. For comparison, the variations of HCl from 2005 to 2016 from MLS satellite measurements are also presented here (yellow line). It is evident that the changes of HCl derived from the SLIMCAT simulations are in overall agreement with those derived from MLS satellite measurements and the correlation coefficient between them is 0.89 in the common period. As expected, the HCl anomaly is largest in 2010-2011 during the period from 1979 to 2016, which may be caused by a joint effect of La Niña, WQBO and positive SST NP events (blue + red + green vertical lines). In addition, HCl shows an increase in both La Niña years (blue vertical line), WQBO years (red vertical line) and positive SST NP years (green vertical line), implying that the independent effect of La Niña, WQBO or positive SST NP events can increase Northern Hemisphere mid-latitude stratospheric HCl. This feature is more evident in the latitude-height cross sections of composite stratospheric HCl for the independent influences of La Niña, WQBO and positive SST NP events (Figure 7). The patterns of the composite HCl anomalies associated with each of WQBO, La Niña and positive SST NP events (Figure 7) are in overall agreement with those in Figure  We have investigated the independent influences of La Niña, WQBO and positive SST NP events on stratospheric HCl in the NH based on composite results from the SLIMCAT model simulation. Here, we further verify the results using time-slice simulations performed with WACCM4. For the model configurations, please refer to Table 1. Figure 8 shows the simulated stratospheric HCl anomalies and zonal-mean zonal wind for independent forcing by WQBO, La Niña and positive SST NP events. Note that R1 is the control run. R2 is as for R1, but with La Niña SST anomalies added to the SST forcing in all 12 months of the year. R3 is as for R2, but with QBO phase signals removed. R4 is as for R1, but with warm SST anomalies over the NP (40-50°N, 160-200°E) added to the SST forcing in all FIGURE 8 | Same as Figure 7, but from the WACCM4 experiments. (A-F) Differences in HCl (color shading) and U (contours) between R2 and R3, between R2 and R1 and between R4 and R1, respectively. The WQBO is identified when the zonal-mean zonal wind in R1 at 30 hPa in 10°S-10°N is larger than a positive standard deviation.
Frontiers in Earth Science | www.frontiersin.org December 2020 | Volume 8 | Article 609411 8 12 months of the year. Thus, the differences between experiments R2 and R3 represent the independent influence of the QBO on stratospheric HCl, the differences between experiments R2 and R1 represent the independent influence of the La Niña events, and the differences between experiments R4 and R1 represent the independent influence of the positive SST NP events. Consistent with Figure 7, the positive zonal-mean zonal wind at high latitudes caused by La Niña, WQBO and positive SST NP events implies a strong polar vortex in February and March ( Figures  8A,C,E), while negative zonal-mean zonal wind in April and May implies the breakup of polar vortex ( Figures 8B,D,F). In the simulations of stratospheric HCl, La Niña activity, WQBO phase and positive SST NP events can cause an increase of HCl in the middle latitudes and a decrease of HCl in the high latitudes in  Table 2. February and March ( Figures 8A,C,E), and the increased midlatitude HCl moves to the high latitudes of the lower stratosphere until April and May ( Figures 8B,D,F). Note though that the increase of HCl in the middle latitudes induced by WQBO in the sensitivity simulations ( Figures 8A,B) is greater than that in the chemistry transport model simulation ( Figures 7A,B), and the increase of HCl in the middle latitudes induced by La Niña and positive SST NP events in the sensitivity simulations ( Figures 8C-F) is smaller than that in the chemistry transport model simulation ( Figures 7C-F). These differences between SLIMCAT and WACCM4 results are likely due to the different types of model used.
Previous studies have shown that the strength of planetary wave activity in the stratosphere associated with the EP flux is strongly correlated with the strength of the residual circulation (e.g., Dhomse et al., 2008). Figure 9 shows EP flux anomalies and EP flux divergence anomalies during WQBO phases, La Niña events and positive SST NP (PNP) events. In all three cases, there are downward EP flux anomalies and positive EP flux divergence anomalies, implying weakened planetary wave activity in the Northern Hemisphere. This weakening of planetary wave activity suggests the weakening of the residual circulation.
The above results support the role of a strong polar vortex and weakened residual circulation, which result from the joint effect of a La Niña event, the west phase of the QBO and positive SST NP anomalies, in causing a significant increase of Northern Hemisphere stratospheric HCl during 2010-2011.

CONCLUSIONS AND DISCUSSION
Using the latest satellite observations, this study reveals that a significant increase of HCl during 2010-2011 has a large impact on the trend of HCl in the Northern Hemisphere, and this HCl increase can mislead trends of HCl in recent decade. In line with previous studies, the Northern Hemisphere stratospheric HCl exhibited an increasing trend from 2005 to 2011, but when the significant increase of stratospheric HCl during 2010-2011 were removed, the increasing linear trend from 2005 to 2011 becomes weak and insignificant. In addition, the HCl still exhibited a weak and decreasing trend from 2005 to 2016. Further analysis of chemistry transport model and chemistry-climate model simulations suggests that during the period from 1979 to 2016, a strong La Niña event, the WQBO and positive SST NP events only occurred at the same time in 2010-2011. The joint effect of the La Niña event, the WQBO and positive SST NP results in a strong northern polar vortex and a weakened residual circulation, and hence, a slowdown of the transport of HCl from the low-mid latitudes to the high latitudes, leading to a large increase of HCl in the middle latitudes of the stratosphere.
Note that the trend of HCl from 2005 to 2016 (Figure 1b) is insignificant, implying that stratospheric ozone has not recovered in recent years as expected. This is consistent with recent studies (e.g., Kyrölä et al., 2013;Gebhardt et al., 2014;Sioris et al., 2014;Nair et al., 2015;Vigouroux et al., 2015;Ball et al., 2018;Zhang et al., 2018). One possible reason for the insignificant HCl trend is that the decline in concentrations of ozone-depleting substances over the past two decades is also not significant. It may imply a need for further control of ozonedepleting substance emissions.
In addition, from Figure 5a, we note that several weak polar vortex events also occurred between 2005 and 2016, e.g., 2008-2009 and 2012-2013. However, there is no evident decrease of stratospheric HCl during 2008-2009 and 2012-2013 in Figure 1. Why then does a strong polar vortex increase stratospheric HCl while the weak polar vortex cannot significantly reduce stratospheric HCl? Figure 10 shows the stratospheric HCl anomalies, the residual circulation anomalies and U in February 2009 and February 2013. Although the polar vortex is weak, the magnitude of downward circulation anomalies in the high latitudes of the lower stratosphere ( Figures 10B,D) is smaller than that of upward circulation anomalies ( Figure 3C). The weakening of the residual circulation can increase HCl, partly offsetting the effect of the weakening of the polar vortex decreasing HCl.
Note that this paper only analyzes qualitatively, not quantitatively, the different factors that contribute to the significant increase of HCl during 2010-2011, since the event that occurred in 2011 was just a single coincidence of factors in the past 20 years. Future work will focus on how to predict the recurrence of events like the 2011 event based on the occurrence of these three factors (La Niña, WQBO and SST NP ) in the context of global warming.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://zenodo.org/.

AUTHOR CONTRIBUTIONS
YH conceptualization, investigation, and writing. FX supervision, investigation, review, and editing. JZ formal analysis and methodology.

ACKNOWLEDGMENTS
We thank Wuhu Feng for designing and performing the SLIMCAT runs and Hongying Tian for providing the data. We also acknowledge Meteorological fields from ERA-Interim, SST from the UK Met Office Hadley Centre for Climate Prediction and Research, HCl data from the MLS and GOZCARDS, WACCM4 from NCAR and SLIMCAT from UK National Centre for Atmospheric Science (NCAS).