Observation of large and all-season ozone losses over the tropics

This paper reveals a large and all-season ozone hole in the lower stratosphere over the tropics (30degN-30degS) since the 1980s, where an O3 hole is defined as an area of O3 loss larger than 25% compared with the undisturbed atmosphere. The depth of this tropical O3 hole is comparable to that of the well-known springtime Antarctic O3 hole, whereas its area is about seven times that of the latter. Similar to the Antarctic O3 hole, approximately 80% of the normal O3 value is depleted at the center of the tropical O3 hole. The results strongly indicate that both Antarctic and tropical O3 holes must arise from an identical physical mechanism, for which the cosmic-ray-driven electron reaction (CRE) model shows good agreements with observations. The whole-year large tropical O3 hole could cause a serious global concern as it can lead to increases in ground-level ultraviolet radiation and affect 50% of Earth's surface area, home to approximately 50% of the world's population. Moreover, the presence of the tropical and polar O3 holes is equivalent to the formation of three 'temperature holes' observed in the stratosphere. These findings will have significances in understanding planetary physics, ozone depletion, climate change, and human health.


INTRODUCTION
The successful execution of the Montreal Protocol and its revisions has led to the declining total level of tropospheric ozone-depleting substances (ODSs) (mainly chlorofluorocarbons--CFCs) since around 1994. Yet, the O3 hole over the Arctic in 2020 set a biggest record, 1 while the O3 holes over the Antarctic in 2020 and 2021 were among the largest, deepest and most persistent ones. As a matter of fact, these observations were not expected from photochemical models 2 but precisely predicted by the cosmic-ray-driven electron-induced-reaction (CRE) model. [3][4][5][6][7][8][9] There are also significant research interests in electron-induced reactions of halogen-containing molecules including CFCs in the gas phase and clusters and on ice surfaces [10][11][12][13][14][15] and in the effects of cosmic rays (CRs) on Earth's clouds, climate and space weather 16-20 . This study was motivated by the following observations. First, the author 9 has recently shown that the CRE-initiated O3-depleting reaction completely destroys or even overkills the O3 layer in the lower Antarctic stratospheric region at altitudes of 13.5-17.5 km corresponding to the CR ionization maximum. Second, CRs have a latitude effect due to the geomagnetic field, which leads to a maximum intensity in the polar stratosphere and a lower intensity by about 50% over the tropics, whereas CFCs have a spatial distribution anti-correlated with the intensity of CRs. [4][5][6][7][8] Third, critically there exist all-season low-temperature cyclones centered at the altitude of around 15 km over the tropics (30N-30S), in which the CR ionization intensity peaks and the temperatures are low at 190-200 K, as will be shown by observed data in this study. The latter is comparable to that in the polar stratospheric vortex over Antarctica in winter and cold enough for forming tropical stratospheric clouds (TSCs), similar to polar stratospheric clouds (PSCs) that are required for surface reactions leading to an O3 hole. 21 It is important to note that the composition and climate of the lower stratosphere over the tropics is very different from those over the polar regions. This should give rise to a unique active halogen evolution for the tropical lower stratosphere: with the constant presence of TSCs, there are no stable chlorine/bromine reservoirs (HCl and ClONO2) as they are rapidly destroyed on the surfaces of TSCs, while with the constant presence of intense sunlight, no ClO dimer (Cl2O2) can exist and O3-depleting reaction cycles are much more effective. Therefore, halogen-catalyzed reactions are much more efficient for tropical O3 destruction, and even a low level of active halogen can cause significant ozone depletion over the tropics in all seasons. The O3 holes over Antarctica 22 and the Arctic 23 have been well observed, whereas no O3 hole over the tropics has been reported. 2 Nevertheless, there have been reports of significant ozone losses over the tropics, implying the possibility of a tropical ozone hole. For example, Bian et al. 24 and Chen et al. 25 reported observations of an "ozone mini hole" over the Tibetan Plateau (29.6° N, 91.1° E) at times with colder winters in the 1990s-2000s, which is located at the edge of the tropics. Polvani et al. 26 observed significant O3 losses at the altitude of 18.5 km or 67/68 hPa over the tropics (30S-30N) from 1979 to 1997 from three independent datasets. Newton et al. 27 observed ozone-poor air (with very low O3 concentrations) in the tropical tropopause layer over the west Pacific by aircraft measurements from the three experiment campaigns based in Guam in January-March 2014. Basing on the above reasoning, the author hypothesized that there likely exists a large O3 hole over the tropics, which should be comparable to the well-known Antarctic O3 hole in depth and much larger than the latter in area. Therefore, this article is devoted to a search for the tropical O3 hole.
Despite the rationales given above, the search for the tropical O3 hole is challenging. This is due to some intrinsic challenges. First, unlike polar O3 holes that are seasonal and mainly appear in spring, a tropical O3 hole if exists is essentially unchanged across the seasons and is therefore invisible in original observed data. Second, unlike in polar regions, the tropical O3 layer mainly lies in the middle stratosphere (25 km) and only a relatively small percentage (25-30%) of total O3 is distributed in the lower tropical stratosphere (at altitudes of 10-25 km). Thus, the amplitude of O3 depletion in the lower tropical stratosphere is not explicitly reflected in measured total O3 data. An associated fact that led to no previous observations of any O3 hole over the tropics is the conventional definition of an O3 hole, which is defined as an area in which total O3 values drop below the historical threshold of 220 Dobson Units (DU) (the total O3 value measured by the NASA TOMS satellite in 1979). With this definition, no tropical O3 holes could be observed in satellite images of total O3 even if a large percentage up to ~65% of the O3 amount in the lower tropical stratosphere is depleted. Moreover, much attention was drawn to the studies of O3 depletion over the polar regions, as no O3 hole was expected to appear over the tropics from current photochemical models. Fortunately, ground-based measurements since the 1960s combined with satellite measurements since 1979 of O3, CFCs and temperature in the stratosphere still provide sufficient data to examine the above hypothesis. Here an O3 hole is defined as an area with O3 loss in percent larger than 25%, with respect to the undisturbed O3 value when there were no significant CFCs in the stratosphere (~ in the 1960s). Despite its difference from the conventional definition of an O3 hole, this new definition is supported by the observed O3 loss (~25% in spring) over the Arctic (shown later) where and when an O3 hole was reported. In other words, an area with O3 loss by 25% is approximately the threshold of observing an Arctic O3 hole.

DATA AND METHODS
A standard troposphere-stratosphere O3 climatology in altitude relative to the ground level in the Trajectory-mapped Ozonesonde dataset for the Stratosphere and Troposphere (TOST) obtained from the World Ozone and Ultraviolet Radiation Data Centre (WOUDC), latitude-altitude CFC distribution, time-series total ozone, Ozonesonde and Umkehr datasets, and lower stratospheric temperature climatology from multiple ground-based or satellite measurements are used for this study. TOST is a global 3-D (latitude, longitude, altitude) climatology of tropospheric and stratospheric O3, derived from the WOUDC global O3 sounding record by trajectory mapping and uses approximately 77,000 ozonesonde profiles from some 116 stations worldwide since 1965, with details given by Liu et al. 28 The long-term zonal mean latitude-altitude distribution of the O3 climatology measured over the past five decades (1960s-2010s) is shown in Fig. S1 in Supplementary Material (SM).
To reveal the tropical O3 hole, we show the differences of the O3 climatology in 1970s, 1980s, 1990s, 2000s and 2010s with respect to that in 1960s. It is critical to note that the distribution of O3 over the tropics is highly ununiform in altitude. Thus, the differences in zonal mean latitudealtitude distribution of the O3 climatology do not simply reflect the value of O3 depletion by a physical mechanism, and there are larger O3 losses in absolute values at higher altitudes. Thus, we plot the relative O3 changes in percent (%) in 1970s-2010s with respect to that in the 1960s to reveal the tropical ozone hole. Furthermore, to confirm the robustness of the observation, we also show the seasonal O3 changes in percent in 2000s with respect to those in the 1980s since the latter ozone data should be more reliably measured and more abundant than those in the 1960s. Moreover, the observed results of the present study are also compared with those of Polvani et al. 26 using TOST 28 and other datasets from combined ground-based and satellite measurements 29,30 .
In addition, the relative O3 change in percent in an O3 hole is given by the equation derived from the CRE mechanism: 6-8 where [O3]0 is the normal O3 amount when there were no significant stratospheric effects of CFCs (i.e., [O3]0[O3] in the 1960s), [C] is the equivalent effective chlorine (EECl) in the stratosphere, I is the CR intensity, and k is a fitting constant determined by the best fit to past measured O3 data. Thus, we also use Eq. (1) to analyze the observed ozone and temperature data in the stratosphere.

Ozone climatology
The large Antarctic O3 hole is well known to appear in the spring, while the hypothesized tropical O3 hole is expected to appear in all four seasons. Thus, we first show the observed results for the Antarctic spring season (September, October and November-SON) in order to compare the tropic O3 hole if exists with the well-known Antarctic O3 hole. As shown in Fig. S2, the thus plotted result shows that from the 1960s to the 1970s, there were almost no ozone holes but small ozone decreases at 10-20 km over the Antarctica and at 16-25 km over the tropics and there was a tropospheric-stratospheric region of significant ozone increases at altitudes below 16 km and latitudes of 10N-40S, which was likely caused by volcanic effects. These small O3 decreases in the Antarctic and tropical lower stratospheres are expected, as it is known that the reaction of stratospheric CFCs remained insignificant up to the 1970s. Despite that there were arguably more sparse measured data available for constructing the TOST data for the 1960s, the result in Fig. S2 shows the validity of the 1960s' TOST data.
As plotted in Fig. 1, in contrast to Fig. S2, the observed results for the 1980s-2010s with respect to that of the 1960s show clearly a deep and large tropical O3 hole with significant O3 loss since the 1980s, along with the well-known Antarctic O3 hole. The Antarctic O3 hole reached the largest and deepest in the 1990s, while the tropic hole continued to grow and deepen in the 1990s and reached its maximum in the 2000s. For both largest and deepest Antarctic (1990s) and tropical (2000s) O3 holes, the same maximum (decadal averaged) O3 loss of ~80% at the hole centers was observed. However, it is worthwhile noting that the geometric area of the tropical O3 hole at latitudes of 30S-30N is approximately 7.5 times that of the Antarctic hole at 60-90S. Both holes in the 2010s were slightly smaller than those in the 2000s. whereas the tropical O3 hole shows a weak sensitivity to seasons, existing in all seasons. The Arctic O3 hole in MAM with a maximum O3 loss of ~25% is far less and smaller in both depth and area than the Antarctic hole in SON or the tropical hole in any season. This is not unexpected, as the winter stratosphere of the Arctic is well known to be much warmer than that of the Antarctic or tropics and the Arctic vortex is not as well formed and as persistent as in the Antarctic or tropics.
Moreover, to show the robustness of the observations in Figs. 1-2, the relative seasonal O3 changes in percent in 2000s are plotted with respect to those in the 1980s in Fig. 3. The latter solidly confirms the observation of the large and all-season ozone hole over the tropics, which was even slightly deeper than the springtime Antarctic ozone hole in SON (62% versus 49%). In either the tropical or Antarctic ozone hole, ozone losses in Fig. 3 are smaller than those in Fig. 2. This is reasonable and expected, as ozone depletion was well known to become significant since 1980 (the beginning of the 1980s) and therefore there was a significant loss of ozone in average from the 1960s to the 1980s. The plotted results in Fig. 3 clearly confirm the validity of the data presented in Figs. 1-2 using the 1960s' data as the undisturbed ozone climatology.  To show time-series depths of the decadal mean Antarctic, Arctic and tropical O3 holes in SON, MAM and annual respectively, averaged O3 changes in % at the lower stratospheric altitudes of 14-21 km are shown in Figs. 4a-c, which also include the data for the sum of concentrations of main ODSs (CFCs and CCl4) measured in the troposphere (no transport lag times considered). It is interesting to show quantitatively that the Antarctic O3 hole reached its maximum O3 loss by about 60% in the 1990s, whereas the tropical hole reached its maximum loss by about 56% in the 2000s. By contrast, the Arctic O3 hole reached its maximum O3 loss by 20-25% in the 1990s, less than the half of the maximum O3 loss in the Antarctic or tropical hole. Both tropical and polar ozone hole trends closely follow that of the total level of ODSs, having significant O3 decreases in the 1980s and 1990s only with no significant trends over the past ~25 years.
Similar to our results shown in Fig. 4b, Polvani et al. 26 made an interesting observation. They observed large tropical ozone loss by about 300 ppbv from 1979 to 1997, though they only showed ozone anomalies (not O3 loss in percentage) relative to the post-1998 mean at the single altitude of 18.5 km or at the pressure of 67 or 68 hPa over the tropics (30S-30N) from 1979 to 2013. They showed such a large tropical O3 loss, consistently from three gridded datasets: the TOST for the period 1979-2013 (the same dataset as used for the present study), 28 the binary database of ozone profiles (BDBP) based on a compilation of observations obtained from several sources of high vertical resolution ozone profile measurements for the period 1979-2006, 29 and the NASA's global ozone chemistry and related trace gas data records for the stratosphere (GOZCARDS) merged satellite dataset based on a compilation of data from a number of satellite measurements for the period 1979-2012. 30 For a comparison, the results of the present study and Polvani et al. 26 are co-plotted in Figs. 5a-b. It is evidently shown that these observed data from multiple datasets of ground-based and combined ground-based and satellite measurements exhibit excellent agreements.
It is also worthwhile noting that the results in Fig. 4b indicate that no ozone hole over the tropics would be observed by the conventional definition of an ozone hole, as the total ozone over the tropics were above 220 DU even with a loss by 56% of the column ozone at altitudes of 14-21 km, given that the undisturbed total ozone was about 270 DU over the tropics, of which only approximately 25% is distributed at 14-21 km. In contrast, with the new definition of an ozone hole in the present study, a pronounced tropical O3 hole is observed, as clearly seen in Figs. 1-4.  29 and the NASA's global ozone chemistry and related trace gas data records for the stratosphere (GOZCARDS) for the period 1979-2012. 30 3.2 CFC-12 distribution and tropical O3 hole The author 6-8 has previously shown that significant decompositions of CFCs and N2O but not CH4 occur in the lower Antarctic stratosphere during winter. The data provided solid evidence of the CRE reactions of CFCs and N2O, but not CH4, in the winter polar stratosphere, though CH4 is not even a secure 'inert trace gas' because it can also react with radicals generated from the reactions of CFCs or N2O. Here zonal mean latitude-altitude distribution of CF2Cl2 (CFC-12) in 1992 from the dataset of the NASA UARS (CLEAS) and the relative difference averaged over the seasons of the O3 climatology in the 1990s with respect to the 1960s are respectively plotted in Figs. 4d-e. The CFC-12 images over the four seasons are shown in Fig. S3. The data clearly show the depletion of CFC-12 in the winter and early spring lower polar stratosphere, as observed previously [6][7][8] . Most importantly, the data in Figs. 4d and S3 show that in all seasons, the CFC-12 concentration was depleted in the lower stratosphere below 25 km over the tropics (at latitudes of 30S-30N), most significant in the zone at 16-20 km and at 20S-20N, in which correspondingly the circularly symmetric annual mean tropical O3 hole is centrally located (Fig. 4e).

Time-series total ozone and Umkehr datasets
Time-series total O3 data at the springtime Antarctic (60-90S) and the annual tropics (30S-30N) measured by satellites since the 1979 are shown in Figs. 6a-b. It is seen that the decreases in absolute total O3 values over the tropics are smaller than those over the Antarctic. This is expected due to the difference between the undisturbed O3 distributions over the Antarctic and the tropics (as mentioned above). The 11-year cyclic variation of total O3 over Antarctica has been well reproduced by the CRE equation previously. [6][7][8] Here, we use the CRE Eq. (1) to give a quantitative analysis of observed total O3 data. Given the observed distributions of the CR ionization rate and CFCs versus latitude and the use of the CR intensity measured at the Antarctic and the concentrations measured at the lower troposphere of CFCs which are well mixed globally as inputs to calculate O3 loss over both the Antarctic and tropics, the reaction constant k in Eq. (1) for the Antarctic is approximately double that for the tropics, i.e., kTro½kAnt. Another factor must also be considered: the transport lag times of CFCs from the troposphere to the lower stratosphere over the Antarctic and the tropics, which are about 1 year and 10 years respectively determined from analyses of observed data in terms of the CRE mechanism. 6-9 Moreover, we make a rough assumption that only the O3 layer in the lower tropical stratosphere below 25 km (i.e, ~25% of the total O3) is affected by the CRE mechanism. Taking all these factors into account, the CREmodeled results are shown in Figs. 6a-b. It is seen that given the fluctuation level of observed data, the results fitted by Eq. (1) for the Antarctic O3 hole are excellent. This gives rise to the best fitting constant kAnt, half value of which (kTro½kAnt) is used to calculate the O3 loss over the tropics with no adjustments. We notice that the modeled results for tropical O3 loss is not as good as for the Antarctic. Given several rough approximations made in this modeling, however, the modeled results reproduce the observed O3 trends surprizingly well and are able to capture the main characteristic of 11-year cyclic variations. Moreover, the Umkehr dataset for the Mauna Loa (19.5°N, 155.6°W, HI, USA) is shown in Figs. 6c-d. For the Antarctic O3 hole, the 11-year cyclic oscillation of total O3 was mainly due to the CRE-caused O3 loss in the stratospheric layers at which the CR ionization intensity was strong, with the most pronounced effect at the Umkehr layer of 63-32 hPa (21-25 km) slightly above the CR-peak layer of 126-63 hPa (15-21 km) in which the CRE mechanism overkilled the O3 layer. 9 In contrast, no 11-year cyclic variations were observed for O3 in the upper stratospheric layer where the photochemical mechanism dominates. 9, 31 A similar behavior is now observed for O3 loss over the tropics, as shown in Figs. 6c-d. The results in Figs. 4a-b and 6a-d indicate that O3 depletion over the Antarctica and the tropics must arise from an identical physical mechanism.

Lower stratospheric temperature (LST)
It is well known that O3 depletion causes stratospheric cooling; temperature drop in the lower stratosphere is a direct measure of O3 loss. The seasonal zonal mean latitude-altitude distribution of the temperature at altitudes of 8-40 km for 2002-2010, which is made from the zonally averaged monthly means on a latitude-altitude grid of radio occultation (RO) satellite datasets since 2002, obtained from the high-quality ROM SAF's Climate Data Record (CDR) datasets, 32 is shown in Fig. S4. Remarkably, the data show that the temperature in the low-temperature cyclones over the tropics over the seasons is 190-200 K, being only 5-8 K above that in the Antarctic polar vortex in winter (JJA) but 13-16 K below that in the Arctic polar vortex in winter (DJF). This can well explain why the O3 hole over the Antarctic or tropics is much larger and deeper than the Arctic O3 hole and why the tropical O3 hole is constantly formed over the seasons. The annul mean zonal mean latitude-altitude distribution of the temperature averaged over the four seasons is shown in Fig. 4f. The latter shows a remarkable phenomenon that the annul mean LST over the tropics is the lowest at 193-198 K, compared with 209-217 K over the Antarctic and 217-220 K over the Arctic. As shown in Fig. 4e, correspondingly the annul mean O3 depletion is the largest in the tropical O3 hole, which is 77% versus 47% for the Antarctic and versus ~10% for the Arctic in the 1990s when the polar holes peaked. Consistent with the observation from ozone datasets shown in Figs. 1-5, the stratospheric temperature data in Figs. 4f and S4 exhibit a pronounced temperature 'hole' over the tropics, providing strong evidence of a deep and large tropical O3 hole, along with the well-known Antarctic and Arctic O3 holes.
Furthermore, the author has well demonstrated that over the Antarctic, the LST regulated by stratospheric cooling due to O3 loss has pronounced 11-year cyclic variations and can be well reproduced by the CRE equation. [6][7][8][9] Here, Figs. 6e-f show time-series temperature anomaly datasets at the lower stratospheres (100-30 hPa) over the Antarctic and the tropics from multiple satellite and ground-based measurements, including NOAA's MSU UAH 33 and RSS 34 satellite datasets, ROM SAF's RO satellite datasets 32 , and NOAA's radiosonde-based Ratpac-B time-series dataset 35 . Note that the LST anomalies in original datasets were normalized to various reference temperatures and had therefore to be offset to compare with each other, but this offsetting has no effects on their long-term trends. These datasets for the Antarctic solidly confirm 11-year cyclic LST variations reported previously, [6][7][8][9] while the LST over the tropics shows similar cyclic variations. Moreover, all the datasets show that significant O3 or LST reductions only occurred in the 1980s and 1990s with no significant trends over the past ~25 years. This is similar to the observation by Polvani et al 26 and those reported in the newest IPCC AR6 36 . The latter states: "most datasets show that lower stratospheric temperatures have stabilized since the mid-1990s with no significant change over the last 20 years" (see page 2-48, chap. 2 of the Report 36 ). Using the same relationship of kTro½kAnt as that used for modeling the O3 data in Figs. 6a-b, we now show that the CRE equation fits to the observed LST data in Figs. 6e-f reasonably well. Again, these results clearly confirm that the Antarctic and tropical O3 holes must arise from an identical physical mechanism, for which the CRE mechanism has given the best agreements with observations. So far, the CRE mechanism is the only physical mechanism that has well reproduced the observed 11-year cyclic stratospheric O3 and LST variations. [4][5][6][7][8][9]

DISCUSSION
The present finding of a tropical ozone hole is closet to the observation by Polvani et al. 26 of large O3 losses at the altitude of 18.5 km or 67/68 hPa over the tropics (30S-30N) from 1979 to 1997 with data from three independent datasets (TOST, BDBP and GOZCARDS) 28-30 , though they did not reveal the tropical O3 hole. Polvani et al. 26 also showed that global and tropical LST cooling had disappeared since 1997, while the tropical O3 concentration reached the minimum around 2005. In explaining their observed results, however, Polvani et al. argued for the low abundance of active chlorine and hence no local chemical O3 destruction in the tropical lower stratosphere. Instead, they argued that the observed tropical O3 and cooling trends were primarily driven by tropical upwelling caused by ODSs rather than greenhouse gases (GHGs) (mainly nonhalogenated GHGs, as widely believed). As we have noted in Introduction and will further discuss later, however, even a low level of active halogen can cause significant ozone depletion in the tropical lower stratosphere.
Interestingly, Polvani et al. 26 also performed simulations from a chemistry-climate model (CCM) with incrementally added single forcings (sea surface temperatures-SSTs, GHGs, ODSs, volcanic eruptions and solar variations) to detail the contribution of each forcing to tropical ozone and LSTs. Although their simulated results showed that ODSs were the dominant forcing of tropical ozone loss over GHGs, it must be pointed out that their simulated values of sum ozone loss (28±13 ppbv per decade, see their Table 1) were about five times smaller than their observed results (~150 ppbv per decade for the 1980 and 1990s), even ignoring that not all individual ensemble members showed statistically significant trends (see their Table 2). Moreover, in contrast to their claim that tropical lower stratospheric ozone would be closely tied to tropical upwelling w*, their simulated value of the w* increase by ODSs is not dominant but very close to that by GHGs, each force contributing to an increment by ~0.1 km yr 1 decade 1 at 85 hPa, namely 0.04±0.09 for SSTs, 0.10±0.11 for SSTs+GHGs and 0.21±0.11 for SSTs+GHGs+ODSs (see their Table 1). The latter results were also consistent with their simulated results of tropical LST trends (see their Fig. 3).
The simulated results of CCMs by Randel and Thompson 37 and others 38 had some differences from but were overall similar to the above-mentioned CCM results by Polvani et al. 26 In most simulated results of CCMs, the strength of tropical upwelling was projected to increase from 1960 to 2100 by ~2% per decade with the largest trends occurring in JJA, corresponding to tropical O3 reductions at 50 hPa of 0.15-0.35 ppmv (11-25 ppbv per decade) 38 . This ozone loss trend resulting from CCM simulations is about 10 times less than the observations by Polvani et al. 26 and the present observations shown in Figs. 1-5. More crucially, the observed data have robustly shown that significant tropical ozone loss and LST cooling occurred in the 1980s and 1990s only, which is in drastic discrepancy from the simulated results of CCMs. Polvani et al. 26 was then led to the open question: How could ODSs affect the stratospheric circulation? They conceded that the underlying mechanism for ODSs being a key forcing for tropical lower stratospheric O3 and temperature trends remained largely unexplored. Knowing the wide belief in CCMs that the key drivers of tropical upwelling and thus tropical O3 or LST trends since 1960 are non-halogenated GHGs (mainly CO2), Polvani et al. 26 were forced to suggest that polar O3 depletion caused by ODSs would cause tropical upwelling and hence large tropical O3 losses. This explanation cannot be correct either, as the polar O3 hole is seasonal and appears only in the springtime, whereas the tropical O3 hole is all-season and has no changes in its central location over the seasons and over the decades since its appearance in the 1980s (Figs. 1-3 and 4e).
The present observed results in Figs. 1-6 and S1-S4 strongly indicate that, like the Antarctic O3 hole that was once incorrectly explained by the misconceived air transport mechanism ('dynamical theory'), the tropical O3 hole must not result from changes in normal atmospheric circulation patterns over the tropics since the 1960s or 1970s but result from an identical physical/chemical mechanism to that for the polar O3 hole. Obviously the tropical O3 hole varies closely with the atmospheric level of CFCs (as seen in Fig. 4b), so it must originate from a CFC-related mechanism. The postulated stratospheric cooling and tropical upwelling effects of increasing non-halogenated GHGs have disappeared in observed O3 and temperature data for the Antarctic lower stratosphere [6][7][8] and for the tropical lower stratosphere 26 . It is obvious that the simulated results from CCMs 26, 37, 38 do not agree with the observed results shown in Figs. 4b, 5a-b and 6f, which show that the negative O3 trends were about 10 times larger (−25 to −30% per decade) in the 1980s and 1990s and there have been no significant O3 or LST trends in the tropical/Antarctic since the mid-1990s. The latter is actually consistent with the observations summarized in the newest IPCC Report 36 . Moreover, the proposed enhanced tropical upwelling directly contradicts with the observed CFC depletion in the lower tropical stratosphere (Fig. 4d), as increased upward motion would transport CFC-rich air from the troposphere. Indeed, the observed data robustly show no shifts in the positions of both Antarctic and tropical O3 holes that have constantly been centered at the altitude region corresponding to the CR ionization peak since the 1960s/1980s and circularly symmetric O3 depletion cyclones are formed with the largest depletion at the centers ( Fig. 1-3 and 4e). These major features cannot be explained by tropical upwelling due to non-halogenated GHGs (mainly CO2) that have kept rising since the 1960s. All the observed data strongly indicate that tropical upwelling cannot be the major mechanism for the observed large, deep and all-season tropical O3 hole. The simultaneous depletions of both CFCs and O3 in the lower tropical stratosphere are most likely due to a physical reaction mechanism that occurs locally. For the latter, the CRE mechanism, supported by the observed data in Figs. 1-6 as well as the substantial datasets obtained from both laboratory and atmospheric measurements 3-9, 39-41 , has provided the best and predictive model.
It is well known that the presence of PSCs is crucial for formation of the Antarctic O3 hole. [42][43][44][45] It was proposed that on the surfaces of PSCs, chlorine reservoir molecules (HCl and ClONO2) are converted into photoactive forms (Cl2) that can then undergo photolysis to destroy O3. There are two types of PSCs, namely Type I and Type II PSC. The composition of Type II PSC is water ice, while Type I PSC is composed of mixtures of nitric acid (HNO3), water vapor (H2O), and sulfuric acid (H2SO4). The temperatures required for formation of Type I and II PSCs are 195K and 188K, respectively. Thus, it is very likely that TSCs, at least Type I PSC-like TSCs, can also form in the tropical lower stratosphere over the seasons due to the observed low temperatures of 190-200 K (Figs. S4 and 4f). CRs may also play a certain role in forming PSCs and PSC-like TSCs. 16,19 Note that the tropical lower stratosphere is very different from the polar lower stratosphere in both composition and climate. The former is rich of CFCs and other halogen-containing gases, whereas the latter is composed of inorganic chlorine species and lower-level CFCs. However, the CRE mechanism put forward two decades ago has proposed that O3-depleting reactions of both CFCs and inorganic halogen species can effectively occur on the surfaces of PSCs. [3][4][5][6][7][8][9][39][40][41] Therefore, there are required and sufficient conditions for O3-depleting reactions to occur on the surfaces of proposed TSCs in the tropical lower stratosphere. As noted in Introduction, the constant copresence of low-temperature TSCs and intense sunlight should lead to a unique active halogen evolution in the tropical lower stratosphere, in which halogen-catalyzed reactions are much more efficient for O3 destruction than that in the polar lower stratosphere.
The tropics (30°N-30°S) constitutes 50% of Earth's surface area, which is home to about 50% of the world's population. O3 depletion in the tropics could cause a large global concern. In areas where O3 depletion is observed to be smaller in absolute O3 value, UV-B increases are more difficult to detect as the detection can be complicated by changes in cloudiness, local pollution, and other difficulties. However, it is generally agreed that the depletion of the O3 layer leads to an increase in ground-level UV radiation because ozone is an effective absorber of solar UV radiation. Exposure to enhanced UV-B levels could increase the incidence of skin cancer and cataracts in humans, weaken human immune systems, decrease agricultural productivity, and negatively affect sensitive aquatic organisms and ecosystems 46 . Indeed, there was a report called HIPERION published by the Ecuadorian Space Agency in 2008 47 . The study using ground measurements in Ecuador and satellite data for several countries over 28 years found that the UV radiation reaching equatorial latitudes was far greater than expected, with the UV Index as high as 24 in Quito. This Ecuadorian report concluded that O3 depletion levels over equatorial regions are already endangering large populations in the regions. Further delicate studies of O3 depletion, UV radiation change, increased cancer risks and other negative effects on health and ecosystems in the tropical regions will be of great interest and significance.
Another important result is that the global lower stratospheric temperature is essentially governed by the O3 layer, which is expected as ozone is the main and dominant molecule that absorbs solar radiation in the stratosphere. As a result, the presence of the tropical and polar O3 holes will play a major role in stratospheric cooling and regulating the global lower stratospheric temperature, as seen previously [6][7][8][9]26 and in the results shown in Figs. S4, 4f, and 6e-f. As seen in Figs. S4 and 4f, this is equivalent to the formation of three temperature 'holes' in the stratosphere, corresponding to the Antarctic, tropical and Arctic O3 holes respectively. This result will be further explored in a subsequent paper.

CONCLUSIONS
In summary, a critical review of the CRE mechanism led to a hypothesis of an ozone hole in the lower stratosphere over the tropics. Substantial datasets of troposphere-stratosphere ozone climatology since the 1960s, latitude-altitude CFC distribution, time-series total ozone, Ozonesonde and Umkehr datasets, and lower stratospheric temperature climatology from multiple ground-based or satellite measurements have provided strong evidence of the existing whole-year ozone hole over the tropics. At the hole center, up to approximately 80% of the ozone concentrations relative to the normal value in the 1960s is depleted. The averaged depth of the annual tropic ozone hole is comparable to that of the springtime Antarctic ozone hole, while the size of the tropic hole is about seven times that of the Antarctic hole. The annul mean ozone depletion in the lower stratosphere over the tropics is strikingly large, which is 77% versus 47% over the Antarctic and versus ~10% over the Arctic in the 1990s when the polar holes were in their peaks. The observed results strongly indicate that both the Antarctic ozone hole and the tropical ozone hole must arise from an identical physical mechanism and the CRE mechanism has provided the best predictive model which exhibits good agreements with observations. The tropical lower stratosphere is rich of CFCs and other halogenated ODSs and CR ionization events, and critically its temperature for all seasons is sufficiently cold for formation of PSC-like TSCs. Therefore, there is every reason to conclude that the CRE mechanism of ozone-depleting reactions have most likely taken place on the surfaces of TSCs present in the lower stratosphere over the tropics, leading to the formation of the tropical ozone hole in the whole year.