Radioisotopes demonstrate changes in global atmospheric circulation possibly caused by global warming

In this paper, we present a new method to study global atmospheric processes and their changes during the last decade. A cosmogenic radionuclide measured at ground-level, beryllium-7, is utilized as a proxy to study atmospheric dynamics. Beryllium-7 has two advantages: First, this radionuclide, primarily created in the lower stratosphere, attaches to aerosols that are transported downwards to the troposphere and travel around the globe with the general atmospheric circulation. By monitoring these particles, we can provide a global, simple, and sustainable way to track processes such as multi-annual variation of the troposphere, tropopause heightening, position and speed of atmospheric interface zones, as well as the poleward movement and stalling patterns of jet streams. Second, beryllium-7 is a product of cosmic rays which are themselves directly linked to solar activity and the earth magnetic field. This study shows whether beryllium-7 observed concentration changes are correlated with such natural processes or independent of them. Our work confirms that major changes in the atmospheric circulation are currently ongoing, even though timeseries are too short to make climatological assessments. We provide solid evidence of significant and progressive changes of the global atmospheric circulation as well as modifications of tropopause heights over the past decade. As the last decade happened to be the warmest on record, this analysis also indicates that the observed changes are, at least to some extent, attributable to global warming.

The global circulation is weakening and this is linked to extra-tropic forcing 6 , meaning that warming is enhanced at the mid-latitudes (strong emissions of CO 2 ) and the polar region (melting of ice shield) compared with the tropics. Simulations show that injection of greenhouse gases has proven to be much more effective when delivered at the mid latitudes because they decrease the temperature differences between the tropical and extra-tropical regions 2,10,12 and therefore induce a slowdown of the general circulation. However, these changes in global circulation are rather complex and cannot be monitored as easily as changes in temperature, precipitation, and surface solar radiation 25 .
It is certainly common sense that beryllium-7 concentrations are not suitable to directly measure atmospheric circulation. They are a useful proxy to observe the annual and inter-annual dynamics of atmospheric circulation cells.
www.nature.com/scientificreports www.nature.com/scientificreports/  (3) weakening of the atmospheric circulation in the mid-latitudes due to the decreasing temperature difference between the equator and the mid-latitudes. Poleward extension and weakened circulation extend the area where cosmogenic radionuclides are detected and increase the period of radionuclide detection during summer months. In the mid-latitudes, weaker circulation and increased vertical temperature differences due to the rising of the tropopause have two effects: stationary weather system and increased convective energy, resulting in more frequent and stronger convective storms. (a,b) are simplified with respect to the STE (Stratosphere-Troposphere exchange), as we do not represent tropopause folding (or duplication) in spring (Northern Hemisphere) and its role in the STE 43 . See Figure S7 for flowchart representing step-by-step the process shown in (a,b). (c) The beryllium-7 concentration profile from Delaygue et al. 26 shows the relationship between tropopause height (blue line) and beryllium-7 concentrations, indicated by isolines 26 . www.nature.com/scientificreports www.nature.com/scientificreports/ production and transport of cosmogenic radioisotopes. Beryllium-7 with a half-live of 53.2 days is particularly suitable as an atmospheric tracer 43,44 . This isotope is produced by the spallation of oxygen and nitrogen due to cosmic radiation in the upper troposphere and lower stratosphere (UTLS) region. Once created, the radionuclides attach to aerosols and are subject to atmospheric circulation until they decay. Currently, atmospheric concentration data can be readily obtained from ground level measurements by utilising the monitoring results from an existing global network of radionuclide stations 45 (currently 71 in operations). These stations are part of the International Monitoring System (IMS) established and maintained by the Provisional Technical Secretariat of the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) 45 located in Vienna, Austria. Air samples are continuously collected for 24 hours and then analysed for environmental radioactivity.
Datasets currently available from 70 global IMS monitoring stations cover up to the last 20 years and can therefore be used to analyse atmospheric characteristics and their changes for this time period 30,31,36 .
The continuous monitoring of cosmogenic beryllium-7 at these stations shows in the long-term a regular annual cycle, but with significant daily variations that amount to a factor of two or more. There are multiple causes for the variability of beryllium-7 activity concentrations in ground-level air [46][47][48] . Two main processes determine the abundance of these isotopes at lower altitudes: (1) the production rate in the UTLS region, and (2) the subsequent transport to the surface. At lower altitudes during horizontal transport, the beryllium-7 concentration may be diminished by dilution and rainout. There is a clear dependency between the transport processes investigated in this study and some modified atmospheric circulation patterns during the last decade 34 . production through cosmic Rays. The production of cosmogenic radioisotopes in the UTLS region is governed by incoming cosmic rays 35,46,[48][49][50][51] . The incoming flux of cosmic rays is not constant but modulated by the approximately 11-year cycle of solar activity 52 . An increased solar activity as indicated by a higher number of sunspots is accompanied by strong magnetic fields that retain charged particles. Thus, the flux of cosmic rays reaching the Earth, is anti-correlated with the number of sunspots 52 .
Since an increased influx of cosmic rays leads to a proportionally increased spallation activity, an analysis of the radionuclide concentrations at ground level with focus on atmospheric circulation must include a correction against these variations in the production. In a comparison of the annual averages of 7 Be concentrations with the annual average of cosmic rays, only 18 out of the 62 stations studied in this paper show a significant positive correlation (α = 0.05) for the years 2004 to 2018. The other stations show weak positive (37 stations) or even negative (7 stations) correlations not significant at the α = 0.05 level. At these stations the correlation is overprinted by atmospheric processes and due to the mixing of old and fresh air masses during transport 34,35 .
After linear correction for the cosmic ray variations, the remaining variability reflects the influence of atmospheric processes and long-term trends thereof, see Fig. 2.
Looking at the corrected time series in Fig. 2, two effects are highlighted: the troposphere multi annual variation 17 (up and down movement of grey dotted line) and the overall trend towards a heightening or lowering of the tropopause (continuous black line). www.nature.com/scientificreports www.nature.com/scientificreports/ Tropopause instability show different frequencies and patterns depending on the station location. As mentioned by Hakuba et al. 17 the troposphere multi annual variation is influenced by seasonal, inter-annual, multi-annual variations associated with Brewer Dobson Circulation, Quasi Biannual Oscillation (QBO), El Nino Southern Oscillation (ENSO), etc.
Besides the 11-year solar activity cycle, a systematic decrease of the Earth's magnetic field strength and a corresponding increase in cosmogenic radiation is being observed over the last 150 years 53 . This effect accounts for a systematic increase of only +0.4% in the beryllium-7 production over a decade 54,55 which is taken into account by the above described correction for the measured cosmic ray flux. transport through atmospheric processes. The second process, which is the transport of beryllium-7 from the UTLS region to the surface, is mainly governed by the systematic vertical downward movement taking place within the Hadley-Ferrel Interface Zone (HFIZ) 30,36,43 . As shown in Fig. 1(a,b), the HFIZ features the downward movement of air masses, while the tropical and polar interface zones feature convergence and upward movements. The HFIZ follows an annual cyclic latitudinal movement due to the poleward expansion of the Hadley cell during summer seasons in the respective Hemispheres 56 .
Since the isotopes are attaching to particles, the effective radionuclide downward transport is also dependent on the stratospheric aerosol loading 26 . The aerosol load is variable depending on external factors; e.g., it can increase by more than two orders of magnitude following large volcanic eruptions. The day-to-day variations were smoothened out by the postprocessing of the data, as described in the next section. Longer time scales, as in the focus of this work, are more robust against such fluctuations.
Taking all these dependencies and variations into account, the observed trends in the cosmogenic radionuclide concentrations measured at ground level can be used as proxies to investigate changes in the global tropospheric circulation possibly triggered by climate change effects 25 . concentrations of cosmogenic isotopes at ground level. The measurements of beryllium-7 activity concentrations exhibit significant daily and weekly fluctuations 44,47,57,58 . Time series were post-processed with a moving average of 120 days to smooth out these fluctuations and normalized to make data more comparable (supplementary file Dataset1).
Concentration data from 70 radionuclide monitoring stations were analysed. Some time series go back as far as 2000; other stations were later established, but only stations (62 stations) with at least seven years of data starting from 2003 were considered for this study (Table S1). In total, over 270,000 radionuclide samples were selected. Most of the samples analysed contained between 1000 and 9500 µBq/m³ of beryllium-7. Figure 3 shows time series of the averaged and normalised concentration data for the selected stations, with the orange and darker bandwidths indicating above-average beryllium-7 concentrations. Normalisation was obtained by dividing absolute values with the overall timeseries average. Stations exhibit phases of above-average concentrations followed by phases of below-average concentrations, resulting in a sinusoidal time series with a period of one year. The time series from stations in opposing hemispheres are phase-shifted by six months due to the simultaneous extension and contraction of the Northern and Southern Hemisphere Hadley cells 31,36 . This annual variation pattern is particularly visible in mid-latitudes where the stations are strongly affected by the HFIZ movement ( Figure S2). For a given monitoring location in the mid-latitudes, the concentration rises when the HFIZ approaches from the equator and falls when the HFIZ retreats towards the equator. This effect is even visible at monitoring locations not directly below the HFIZ pathway due to the horizontal transport of air masses at lower altitudes (supplementary file surface charts).
Stations located in mid-latitudes, like. Canada and Japan, showed a clear pattern, namely a gradual extension of the high beryllium-7 season (Fig. 3). The annual period with above-average concentrations starts earlier and ends later in the season. Similar patterns were found especially along the sub-tropical zones in both hemispheres.
The station in Resolute, Canada, showed lower levels of beryllium-7 and the transit time to reach the yearly maxima increased over the years, which indicates that the atmospheric circulation and trade winds are weakening. Another example, the station in Stockholm, Sweden, showed a 40% net increase over the past three years as well as 54 additional days with concentrations above average. In general, most stations showed a similar increase in the beryllium-7 concentrations. changes in the tropopause height. The tropopause is the thermally stable boundary layer between the troposphere and the stratosphere. The tropopause height decreases from about 17 km in the equator region to about 7 km in the polar region. Observations have demonstrated that the tropopause height has been changing over the recent years, due to climate change processes 5,22,59 . Between 2001 and 2007, a height increase of 20-50 meters per year (Fig. 4) has been measured at around 30-40 degrees latitude and the poles, while a decline of 10-30 meters per year has been measured around the equator and 60 degrees latitude 20,59-61 . However, these observations, obtained using reanalysis data, were not evenly distributed globally, were conducted for a limited time and are therefore subject to significant uncertainty.
The beryllium-7 concentrations rise with altitude in the troposphere and then increase sharply above the tropopause 26,32 . Therefore, an increase in the tropopause height integrates areas of enhanced beryllium-7 concentrations into the troposphere. As vertical mixing within the troposphere is much stronger compared with troposphere-stratosphere exchange, this causes an increase of the amount of beryllium-7 being transported to the surface, see Fig. 1c. www.nature.com/scientificreports www.nature.com/scientificreports/ Tropopause height, is defined by a thermodynamic gradient. The tropopause height increases if the critical temperature lapse rate moves to higher altitudes. Though this is accompanied by some air mass movement, the production of beryllium-7 is fairly robust against these changes and mainly depends on the height from ground level.
Tropopause trends presented here are in agreement with the work of Santer et al. 59 . A global interpolation of beryllium-7 concentration from 2003 to 2019 as troposphere multi annual variation or tropopause instability is available as supplementary file.
The datasets from re-analysis data and beryllium-7 are from different periods as data from 2001 to 2007 are not available for all the stations needed to build the latitudinal profile shown in Fig. 4a. Figure 4b shows the beryllium-7 growth rate of all stations with a significant (α = 0.05) linear trend over the past 7-14 years compared to re-analysis data 20 . The highest increases are found at mid-latitudes corresponding with the locations of tropopause heightening. Thirteen out of 62 stations show a significant (α = 0.05) linear correlation between beryllium-7 and calendar year. Eleven of these stations show a positive correlation with the steepest increases of 7Be being observed at Okinawa, Azores, and Midway Islands. Two stations (Cocos Island and Zalesovo) show a significant negative correlation. Beryllium-7 trends from all stations, all available years is shown in Fig. 4c.
Global warming is also associated with decreasing lower-tropospheric temperature differences between the equator and the polar region (enhanced polar warming), which may cause a weakening of the geostrophic (westerly) circulation. Furthermore, it affects the poleward movement of the HFIZ. While the former may lead to more stable weather patterns, such as the blocking situations discussed below, the latter extends the time period as well as geographical area where high beryllium-7 concentrations are observed. The growth is greater in the Southern Hemisphere and lesser in the Northern Hemisphere, as indicated by the data. In general, the global beryllium-7 data correlated with re-analysis results.
Therefore, global beryllium-7 monitoring provides an opportunity to support the tropopause height monitoring efforts. More information on the role of intensified STE associated with this process can be expected from sodium-22/beryllium-7 ratios. The 19-year time series from Switzerland shows an overall positive trend (i.e. increasing contribution of stratospheric air), however, with relatively high year to year variability ( Figure S2).  Table S1). The black arrows indicate the trends; they are only for visual guidance. White areas indicate periods of n/a data.
www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Figure 5 shows tropopause height calculated from gridded monthly global positioning system radio occultation (GPS-RO) data 62,63 versus the beryllium-7 concentrations at the same location (equal coordinates). The post processing applied to create the chart is mean annual records. In both cases GPS-RO and beryllium-7 data are not smoothed additionally. Each location shows a high degree of correlation between the 2 parameters confirming connections between tropopause and beryllium-7.
Standard deviation is low at the mid latitudes and increases at the equator and at the very high latitudes, those areas where beryllium-7 is most affected by wash out. Years with lower correlation values need to be further investigated as possibly associated to additional drivers such as advection or stratospheric tropospheric exchange. the slowdown and poleward movement of the ferrel cell. Changes of atmospheric cell progression are also visible through isotopic ratio changes over time. As seasons have characteristic isotopic ratios due to the associated position of the circulation zones, isotopic ratios of sodium-22 and beryllium-7 can also indicate the circulation trends 64 . Sodium-22 has similar properties as beryllium-7 except that is formed by spallation with Argon but has a significantly longer half-life (2.6 years). Figure S3 displays how much the radioisotope ratio seasonal gradient ( 22 Na versus 7 Be) varies over time, meaning that a seasonal peak is expression of fast transition of the atmospheric cell convergence zone over the point of measurement (IMS Station). A flattening of the ratio indicates a slowdown of the transition.
The resulting graphs ( Figure S3) with the squared ratio of sodium-22 and beryllium-7 depicted a slow velocity of the isotopic ratio gradient over time, suggesting a weakening of the circulation.
The time series of the beryllium-7 concentrations (and sodium-22 for figure S3) were clearly affected by the changing seasonal movement of the HFIZ. Along with the increasing height of the tropopause, the Hadley cell expands not only vertically but also in the latitudinal direction, accompanied by a possible slowing down of the Ferrel circulation. Zhang and Song provide evidence that reduction of surface pressure gradient support predictions of weakening of vertical circulation in a warmer climate, including the weakening of trade winds 65 . Over the years, this effect was reflected in the time series by a prolongation of the periods with high isotope concentrations (supplementary file Table S3). Most of the locations followed this pattern, as indicated by the black dotted lines in Fig. 3.
The ongoing extension of the high beryllium-7 season is, therefore, highly likely directly associated with longer and warmer summers. The Hadley cell is susceptible to temperature effects as it is driven, like the polar cells, by the thermal circulation of warm and cold air and the creation of high-and low-pressure areas. The poleward extension of the HFIZ movement, (together with a widening of the tropical circulation, poleward shift of the tropospheric jet streams) can be linked to global temperature increase issues 4,7,8,22,46,65 . Static weather situations. Comparing the annual curves of beryllium-7 at a given station indicated that the observed shape of the curves has been significantly changing over the years. While in the early years of the available data sets-i.e., the 2000s-the curves showed rather smooth and steady increases and decreases, closely resembling an idealized sine curve, the more recent data showed plateaus where the beryllium-7 concentration stayed roughly constant for a few days or weeks per year (see Fig. 6). www.nature.com/scientificreports www.nature.com/scientificreports/ The beryllium timeseries in both Sweden and Germany show the transition from sinusoidal (dotted line) to plateau (continuous red -purple line). In terms of HFIZ movement the sinusoidal curve indicates a smooth progression northward in summer and southward in winter. Once the timeseries flattens out into a plateau pattern we can translate it as stalling of the HFIZ.  Fig. 6). With decreasing temperature differences between the equator and mid-latitudes, the circulation slows down, and we see higher frequencies of stalling situations, plateaus, or step features where the interface zone seems to pause its latitudinal movement 6,10,65,66 . These stalling conditions facilitate static weather patterns which correlate with the general trade wind deceleration 10,12,14 and the increase of convective storms at mid-latitudes over the past few years 1,2,9,10,13 .
A warmer and higher tropopause along with a weaker westerly circulation would cause more tropical weather during summertime as well as more steady weather situations. Longer episodes of large-scale atmospheric blocking and high-pressure areas with low pressure gradients have been frequently observed globally over the last few years.
Atmospheric blocking is a phenomenon that leads to a static weather pattern lasting from one to several weeks, often associated with extreme weather events -heat waves in summer and cold waves in winter. Frequently, the onset of a block starts with the rapid poleward transport of subtropical air. Weather and climate models have a tendency to underpredict blocks and to underestimate their duration 3 . It is so far not well established whether blocks are becoming more or less frequent under global warming conditions, although regional increases in blocking conditions have been identified 25 .
The fraction of economic losses caused by severe convective storms in Europe with respect to all-natural hazards combined went from 26% between 1989 and 2017 to almost 40% between 2003 and 2017 11 . The trend towards persistent static weather patterns is in agreement with the observations (plateau features) at the German and Swedish stations (Fig. 6).
A majority of the stations showed an extended high-beryllium-7 season, as shown by examples in Fig. 6. By calculating how much time it takes to reach its yearly maximum at each station (which is the peak of beryllium-7 concentration per year when HFIZ is closest), it is possible to check whether the progression speed of the convergence zone has changed. Between 2003 and 2019, the progression clearly slowed down. The average of the beryllium-7 growth and extension period of growth for the past three years in all the stations indicated a prolongation of the summer season by around 1 month and a generalised deceleration (slower transit time) of circulation, producing a delay of 3 weeks to reach the beryllium-7 maximum worldwide ( Figure S3). Figure S4 shows the increase in the number of days required per year to reach the maximum threshold from a baseline of the yearly mean. The trend is positive which indicates a weakening of cell progression and therefore suggest a slowing down of the general circulation.
As discussed above, the trends of increasing beryllium-7 concentrations (Fig. 7), extended periods of high concentrations and growing transit times can be best explained by changes in large-scale atmospheric circulation patterns, possibly associated with the continued increase in greenhouse gas concentrations 27,28 .
All years are available in the supplementary file but we chose to select 2 figures with 10 years difference to highlight how much the concentration increased over time. Moreover, both plots correspond to solar maxima where we should have the minimum levels of beryllium-7 concentration. The maps do not have any meteorological information rather they show only interpolation of beryllium-7 values.
Initial analysis on GPS RO scale allowed to build a preliminary comparative scale between the temperature and increase of beryllium-7 as proxy. The comparative scale built on the ratio of absolute monthly tropopause height versus absolute monthly beryllium-7 values. The average ratio is between 3 and 6: lower (~3) at the mid latitudes and highest (~6) at the equator where the height of tropopause is the highest and the wash out effect is stronger (Fig. 5).
In fact, if beryllium-7 observation data would be assimilated, weather models with integrated simulation of beryllium-7 production could assess the precision of this monitoring method. Furthermore, longer time series www.nature.com/scientificreports www.nature.com/scientificreports/ are needed to better distinguish between decadal-scale and longer-time cyclic circulation variations and the systematic effect of global warming.

Summary and conclusions.
Radioisotopes like beryllium-7 and sodium-22 are produced by the incoming cosmic radiation in the UTLS region. Their concentrations at ground level depend on the tropopause height and prevailing atmospheric circulation and are therefore a function of the measurement location, mostly the station latitude, and the time of the year. The activity concentrations are significantly influenced by systematic changes in the atmospheric circulation. Such ground-level concentration data can thus yield invaluable insights into key atmospheric processes, which are assumed/predicted to be affected by global warming.
Beryllium-7 data measured over the last 15-20 years confirm that major changes in the atmospheric circulation are currently ongoing. Even though timeseries are too short to make climatological assessments, these changes may well be triggered by progressive global warming. Natural causes to explain the observed beryllium-7 variations such as solar flux or the strength of the Earth's magnetic field are relevant but act on either a different order of magnitude or at different time scales and thus cannot explain the trend we observe.
This new monitoring method using beryllium-7 as a proxy for global atmospheric circulation supports ongoing research related to changes in tropopause heights as well as global circulation patterns, which includes some topics marked as "known with lower confidence" in the latest report of the Intergovernmental Panel on Climate Change (IPCC) 25 . Although data sets cover less than two decades, our method provides some indication on how strongly global warming may influence the atmospheric circulation worldwide already today. This may also indicate that the warming trend is more severe and rapid than many climate projections show, and that we cannot exclude that weather extremes predicted to occur between 2050 and 2100 already materialize during the next few decades.

Data availability
The data that were used for the current study and that support its findings are not publicly available, but access through the virtual Data Exploration Centre (vDEC) can be granted by the CTBTO via a cost-free confidentiality agreement. The application for vDEC access to data can be submitted for approval through a simple web form at https://www.ctbto.org/specials/vdec/. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.