1. Introduction
The Antarctic ozone hole has formed each austral spring since the early 1980s where up to half of the total column ozone (TCO) is depleted (Solomon 1999; WMO 2014). Although it has little impact on global temperatures, this long-term ozone depletion has likely influenced the Southern Hemisphere (SH) atmospheric circulation and thus the surface climate. It cools the SH polar stratosphere and strengthens the polar vortex and is associated with a summertime poleward shift and strengthening of the midlatitude jet (Lee and Feldstein 2013; Seviour et al. 2017), which are strongly associated with the positive phase of the southern annular mode (SAM), the leading mode of climate variability in the SH extratropical circulation (Trenberth 1979; Rogers and van Loon 1982). While increasing greenhouse gases (GHGs) also force a positive summer SAM trend (e.g., Arblaster and Meehl 2006; McLandress et al. 2011; Grise and Polvani 2017), model experiments that have compared the influence of both factors individually have suggested that ozone depletion is likely the dominant factor (e.g., Arblaster and Meehl 2006; McLandress et al. 2011; Polvani et al. 2011; Stone et al. 2016).
In addition to the long-term trend, the size of the ozone hole varies substantially between years due to dynamical processes (Salby et al. 2011, 2012). Years with anomalously small ozone holes are usually associated with stronger winter planetary wave forcing that transports more ozone to the polar region and warms the Antarctic stratosphere, thus weakening the polar vortex. The warmer temperatures inhibit the formation of polar stratospheric clouds that deplete ozone via chemical reactions and hence reduce ozone loss (Salby et al. 2011, 2012). This year-to-year variability in the size of the ozone hole has been linked to variability in the SAM and surface temperatures in the SH. Son et al. (2013) reported a statistically significant negative correlation between September ozone concentration and the October SAM index. Bandoro et al. (2014) further reported a significant relationship between November TCO and seasonal mean summer surface temperatures in the SH midlatitudes, including Australia, with unusually hot summers associated with anomalously small ozone holes (higher TCO) in the previous spring. The connection between spring ozone and summer temperature over Australia is thought to arise due to the link between ozone and the SAM. A negative SAM in summer (associated with high spring ozone) causes anomalous westerly surface winds that lead to decreased precipitation and warmer surface temperatures over subtropical eastern Australia in summer (Hendon et al. 2007; Son et al. 2013; Bandoro et al. 2014).
Australian summer temperature extremes are influenced by large-scale modes of climate variability including El Niño–Southern Oscillation (ENSO), the Indian Ocean dipole, and SAM (Hendon et al. 2007; Risbey et al. 2009; Arblaster and Alexander 2012; Min et al. 2013). ENSO has some predictability on seasonal time scales and has, therefore, traditionally been the main component considered in operational seasonal forecasts (McBride and Nicholls 1983) before the implementation of a dynamical seasonal forecast system (Hudson et al. 2016). The observed connection between spring Antarctic ozone and Southern Hemisphere climate suggests that including real-time stratospheric ozone variability could potentially improve skill in seasonal outlook systems. This is particularly timely as extreme summers in Australia are likely to become more common under future emission scenarios (e.g., Perkins et al. 2015; Perkins-Kirkpatrick et al. 2016). Improved seasonal forecasting could, therefore, be an important adaptation tool for mitigating the impacts of extreme heat events.
However, climate models must be able to reliably simulate ozone behavior and stratospheric–tropospheric dynamics to produce accurate forecasts. Chemistry–climate models (CCMs) are perhaps the most useful model to address these interactions as chemistry is fully interactive and coupled to dynamics and radiation; CCMs, therefore, tend to simulate the impacts of ozone on the circulation and climate better than models with prescribed ozone (e.g., Son et al. 2008; Li et al. 2016). Since chemical reactions cause the ozone hole, it is critical that interactive chemistry is included in the model to capture and predict these interannual relationships.
Many climate model studies have examined the long-term impact of Antarctic ozone depletion on stratospheric and tropospheric circulation and climate (e.g., Gillett and Thompson 2003; McLandress et al. 2011; Polvani et al. 2011) but few have addressed the impact on interannual time scales. Fogt et al. (2009) and Li et al. (2010) reported that a CCM captures observed interannual ozone–SAM and SAM–Brewer-Dobson circulation relationships, respectively. However, a deficiency common to these models is related to a “cold pole” bias present in many CCMs (Eyring et al. 2006), which further delays the breakdown of the polar vortex and likely causes the model to overpredict the impacts of the ozone hole (Lin et al. 2017). Moreover, Seviour et al. (2014) reported that the October mean SAM could be forecast from midstratospheric anomalies at the beginning of August, and Dennison et al. (2015) showed that during the period of ozone depletion, the tropospheric circulation is influenced for up to two months following a stratospheric SAM extreme event. To date, there has yet to be a study that has examined whether climate models can simulate the interannual link between ozone and surface temperatures, leading to improved seasonal forecasts.
The purpose of the present study is to investigate the potential for predicting summer surface temperature extremes using ozone variability. This involves examining historical simulations from the Whole Atmosphere Community Climate Model (WACCM) and other CCMs to assess whether these models can capture the influence of the interannual variability in the Antarctic spring ozone hole on summer temperatures, with a focus over the Australian continent. This is a necessary first step in examining the potential for its inclusion in a seasonal prediction system.
2. Data and analysis method
a. Observational and reanalysis data
Multiple observational datasets for TCO and surface temperature are used in this study to examine sensitivity to observation and reanalysis uncertainty. Monthly mean TCO fields have been obtained from the NIWA-BS (National Institute of Water and Atmospheric Research–Bodeker Scientific) database (Bodeker et al. 2005; http://www.bodekerscientific.com/data/total-column-ozone). The NIWA-BS data averaged over the polar cap (63°–90°S), are mainly compared to TCO from the Halley Station (herein Halley), which measures ozone variability at a single grid point (75°S, 26°W; https://legacy.bas.ac.uk/met/jds/ozone/data/ZOZ5699.DAT). TCO from the South Pole (90°S, 25°W) and Syowa (69°S, 39°E) stations (http://www.woudc.org) are also examined. Monthly mean surface temperatures are obtained from the European Centre for Medium-Range Weather Forecasts interim reanalysis (ERA-Interim; Dee et al. 2011). ERA-Interim is compared to monthly surface temperature from the Australian Water Availability Project (AWAP; Jones et al. 2009), which is a gridded dataset based on station data. Monthly mean maximum and minimum temperatures for AWAP were averaged to produce monthly mean temperature. The Marshall (2003) SAM index (http://www.nerc-bas.ac.uk/public/icd/gjma/newsam.1957.2007.txt) and ENSO are used to examine links between ozone and modes of climate variability. Gridded observed monthly SSTs from the Hadley Centre Ice and Sea Surface Temperature dataset (Rayner et al. 2003) were used to calculate the Niño-3.4 index (described in section 2d).
b. Model output
This study uses the output from version 1 of WACCM, conducted as part of the Chemistry–Climate Model Initiative (CCMI; Eyring et al. 2013; Hegglin and Lamarque 2015; Morgenstern et al. 2017). WACCM is a fully interactive CCM where chemistry is coupled with dynamics and radiation, and this, therefore, permits chemistry–climate feedbacks. WACCM was chosen as the primary model analyzed as it has been shown to have excellent agreement with observations in the evolution of the Antarctic ozone hole (Marsh et al. 2013) and is one of a limited number of CCMs that is coupled to an ocean (Morgenstern et al. 2017), which is an important characteristic for seasonal prediction. The model domain extends from the surface to 140 km with 66 hybrid sigma-pressure levels, and horizontal resolution of 1.9° latitude by 2.5° longitude (Marsh et al. 2013).
Four WACCM experiments are analyzed in this study to examine the role of ocean coupling for stratospheric–tropospheric relationships and the influence of ozone-depleting substances (ODSs) and GHGs individually. Each experiment has an ensemble of three or five transient simulations that have slightly different initial conditions (Eyring et al. 2013; Morgenstern et al. 2017) and cover 1960–2005:
REF-C1 (or uncoupled): uses an atmosphere-only model configuration forced with observed SSTs and sea ice and historical radiative forcings (GHGs, ODSs, tropospheric ozone and aerosols, quasi-biennial oscillation, very short-lived species, volcanic aerosols, and solar variability).
REF-C2 (or coupled): uses the same atmospheric configuration and historical radiative forcings as REF-C1 but is fully coupled to an interactive ocean and sea ice component, and extends to 2100 following the A1 scenario for ODSs (WMO 2014) and representative concentration pathway (RCP) 6.0 scenario (Meinshausen et al. 2011).
SEN-C2-fODS1960 (herein ODS1960): the same as REF-C2 but with ODSs containing chlorine and bromine set at 1960 levels. Thus, interannual variations in the size of the Antarctic ozone hole will still occur due to dynamic variability, but no ozone depletion is simulated.
SEN-C2-fGHG (herein GHG1960): the same as REF-C2 but with anthropogenic GHGs fixed at 1960 levels.
The WACCM uncoupled experiment has five simulations available; unless stated otherwise, for consistency only the first three were used to avoid a larger sample size biasing the results compared to the coupled model ensemble. The other two members were tested and produced quantitatively similar results to the three members used in this study.
To examine the strength of the findings in WACCM and evaluate the impact of model biases, nine other CCMI models are included [ACCESS-CCM, CESM1 CAM4-Chem, CMAM, EMAC-L47MA, EMAC-L90MA, GEOSCCM, MRI-ESM, NIWA-UKCA, and SOCOL; model specifics are available in Morgenstern et al. (2017)]. In total, this facilitates analysis of five models with a coupled ocean for REF-C2 and five with an uncoupled ocean that prescribe SSTs and sea ice concentrations using simulations from another climate model. For the CCMs without an ocean, different SSTs and sea ice concentrations were used for REF-C1 and REF-C2. All ensemble members available on the British Atmospheric Data Center were included. For the additional models, the lowest model level was used for the temperature at the surface. We compared the difference between the surface temperature field and lowest model level in the correlation analysis for WACCM, and the difference was negligible.
c. Analysis period
This study examines the period 1979–2005, which represents the overlap period for the satellite data and the model historical period. These years are selected as studies have reported that the relationship between interannual variations in Antarctic ozone and SH surface climate is strengthened during the period of ozone depletion (Fogt et al. 2009; Bandoro et al. 2014) as the ozone hole delays the polar vortex breakdown and leads to increased coupling between the stratosphere and troposphere (Shaw et al. 2011). The Antarctic ozone layer has also shown signs of recovery since 2000 (e.g., Solomon et al. 2016; Chipperfield et al. 2017), and the years before this, therefore, were when stratospheric ozone depletion was largest overall. November ozone and summer (December–February) surface temperatures are the focus of this study for comparison with Bandoro et al. (2014) and when temperature extremes arguably cause more impact. Ozone variability throughout the year in WACCM is in good agreement with observations and tends to peak in October–November, similar to observations (see Table A1; Roff et al. 2011; Son et al. 2013; Bandoro et al. 2014). Apart from section 3e, all model analysis is conducted for WACCM.
d. Indices
The ozone hole is defined as the weighted area average TCO over the polar cap (63°–90°S), after similar studies (e.g., Son et al. 2013). The ozone index is calculated for September to April only, as observations are unavailable in other months due to polar night. Figures 1a and 1b show the time series of the ozone index in November for the first member of the WACCM uncoupled and coupled experiments, respectively. The interannual variabilities of the ozone hole and all indices are obtained by first removing the long-term linear trend (Figs. 1c,d), following Bandoro et al. (2014). Detrending the data also removes the linear influence of GHG increases.
Time series (1979–2004) of the November ozone index for the first ensemble member of WACCM (a) uncoupled (REF-C1) and (b) coupled (REF-C2) experiments, and the detrended November ozone index for the (c) uncoupled and (d) coupled experiments. Years with high/low polar cap (63°–90°S) averaged TCO are identified as those that exceed ±one standard deviation (red/blue horizontal lines). Note that ±one standard deviation is calculated across the three members.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Time series (1979–2004) of the November ozone index for the first ensemble member of WACCM (a) uncoupled (REF-C1) and (b) coupled (REF-C2) experiments, and the detrended November ozone index for the (c) uncoupled and (d) coupled experiments. Years with high/low polar cap (63°–90°S) averaged TCO are identified as those that exceed ±one standard deviation (red/blue horizontal lines). Note that ±one standard deviation is calculated across the three members.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Time series (1979–2004) of the November ozone index for the first ensemble member of WACCM (a) uncoupled (REF-C1) and (b) coupled (REF-C2) experiments, and the detrended November ozone index for the (c) uncoupled and (d) coupled experiments. Years with high/low polar cap (63°–90°S) averaged TCO are identified as those that exceed ±one standard deviation (red/blue horizontal lines). Note that ±one standard deviation is calculated across the three members.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
For WACCM, the SAM index is defined as the difference in standardized zonal mean sea level pressure (SLP) between 40° and 65°S, following Gong and Wang (1999). Strong SAM events are identified when the value is greater or less than one standard deviation (after detrending). The Niño-3.4 index (5°S–5°N, 170°–120°W; Trenberth 1997) is used to analyze the ENSO influence on ozone and SAM.
e. Correlations and composites
Pearson correlation coefficients were calculated between ozone and other variables, after first removing the annual cycle and detrending. These links are examined throughout the year using lag correlations, with the ozone index correlated to each 3-month overlapping surface temperature or SAM period for up to 6 months. To explore the relationships further over Australia, we focus on area-averaged surface temperature in eastern Australia (10°–44°S, 141°–156°E; where at least 50% of each grid box had to comprise of land surface to be included in the calculation) as Bandoro et al. (2014) found that the relationship between November ozone and summer surface temperatures was largest in this region. For analysis with WACCM, the three 26-yr time series from each experiment were concatenated (unless stated otherwise) to provide a larger sample than is possible with the observations and improve the signal-to-noise ratio.
To investigate the influence of large ozone (SAM) anomalies on stratospheric and tropospheric climate, years with high and low November ozone (summer SAM) were identified as years that exceed one standard deviation (after first removing the annual cycle, detrending, and concatenating the three members) (Figs. 1c,d). Composites were then created for the difference between years with high and low November ozone (summer SAM). Statistical significance of correlations and composites were assessed using a two-sided Student’s t test with the degrees of freedom reduced based on the lag-1 autocorrelation, following Bretherton et al. (1999) and Santer et al. (2000).
3. Results
a. Ozone–SAM relationship
The main interest of this paper is the interannual impact of the Antarctic ozone hole on surface temperatures. As the impacts of ozone depletion on surface climate resemble the SAM (see, e.g., Thompson et al. 2011), the interannual link between ozone and SAM is first explored. Figure 2 shows lag correlations between ozone and SAM (the time reference is based on the ozone index) for Halley ozone (75°S, 25°W; Fig. 2a), NIWA-BS ozone (averaged 63°–90°S; Fig. 2b), and the four WACCM experiments (averaged 63°–90°S; Figs. 2c–f) where the three ensemble members for each experiment were first concatenated.
Lag correlation between the detrended ozone index and detrended SAM index for each 3-month overlapping period, for 1979–2004 (1980–2005 for the ozone index in January–April). (a) Halley ozone (75°S, 25°W) and (b) NIWA-BS ozone (63°–90°S) with the Marshall (2003) SAM index, and (c)–(f) TCO (63°–90°S) and SAM from WACCM uncoupled, coupled, GHG1960, and ODS1960 experiments, respectively. The horizontal axis indicates the ozone index month. The vertical axis shows the 3-month overlapping average SAM [e.g., September ozone correlated with SAM in +0 SON, +1 OND, +2 NDJ, +3 DJF, and +4 JFM (SON is September–November, OND is October–December, and so on)]. The correlation coefficients that are statistically significant at the 90%, 95%, and 99% confidence levels are bound by yellow, green, and white contour lines, respectively. A two-tailed t test is used to test significance, with the degrees of freedom reduced based on the lag-1 autocorrelation coefficient.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Lag correlation between the detrended ozone index and detrended SAM index for each 3-month overlapping period, for 1979–2004 (1980–2005 for the ozone index in January–April). (a) Halley ozone (75°S, 25°W) and (b) NIWA-BS ozone (63°–90°S) with the Marshall (2003) SAM index, and (c)–(f) TCO (63°–90°S) and SAM from WACCM uncoupled, coupled, GHG1960, and ODS1960 experiments, respectively. The horizontal axis indicates the ozone index month. The vertical axis shows the 3-month overlapping average SAM [e.g., September ozone correlated with SAM in +0 SON, +1 OND, +2 NDJ, +3 DJF, and +4 JFM (SON is September–November, OND is October–December, and so on)]. The correlation coefficients that are statistically significant at the 90%, 95%, and 99% confidence levels are bound by yellow, green, and white contour lines, respectively. A two-tailed t test is used to test significance, with the degrees of freedom reduced based on the lag-1 autocorrelation coefficient.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Lag correlation between the detrended ozone index and detrended SAM index for each 3-month overlapping period, for 1979–2004 (1980–2005 for the ozone index in January–April). (a) Halley ozone (75°S, 25°W) and (b) NIWA-BS ozone (63°–90°S) with the Marshall (2003) SAM index, and (c)–(f) TCO (63°–90°S) and SAM from WACCM uncoupled, coupled, GHG1960, and ODS1960 experiments, respectively. The horizontal axis indicates the ozone index month. The vertical axis shows the 3-month overlapping average SAM [e.g., September ozone correlated with SAM in +0 SON, +1 OND, +2 NDJ, +3 DJF, and +4 JFM (SON is September–November, OND is October–December, and so on)]. The correlation coefficients that are statistically significant at the 90%, 95%, and 99% confidence levels are bound by yellow, green, and white contour lines, respectively. A two-tailed t test is used to test significance, with the degrees of freedom reduced based on the lag-1 autocorrelation coefficient.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
The observations and WACCM experiments all capture strong negative correlations between spring ozone and SAM in the following months, implying that smaller (larger) spring ozone holes are associated with decreases (increases) in the SAM. This association is consistent with long-term ozone depletion leading to a more positive SAM in summer (Thompson et al. 2011). However, it is unclear if interannual ozone variations drive variations in the SAM through the same mechanism by which stratospheric ozone depletion influences the SAM, as other factors, such as winter–spring wave driving, also influence SAM and ozone variations (Thompson et al. 2005; Son et al. 2013; Seviour et al. 2014) and it is difficult to separate cause and effect. Note that while we are focused on the use of ozone for prediction, Fogt et al. (2009) previously found significant negative correlations between observed ozone and SAM also at negative lags, indicating that when the SAM is weak, more ozone is transported to the polar vortex.
There are substantial differences between observational datasets and between model experiments. In the observations, correlations are largest for September–October ozone, whereas the WACCM experiments peak one month later in November–December. A possible cause for the delayed onset in the model experiments could be due to the cold pole bias. For example, Sheshadri and Plumb (2016) found in an idealized atmosphere model that the surface response to polar stratospheric cooling (indicative of ozone depletion) is sensitive to the timing of the cooling. The ozone–SAM link is weaker and less persistent for Halley (Fig. 2a) and could be a consequence of this station being located at the edge of the polar vortex in some parts of the year. Both observational datasets and the WACCM coupled and uncoupled experiments capture a band of positive correlations in March–April, which have been linked to natural variability in the polar vortex (Fogt et al. 2009; Smith and Polvani 2017). Furthermore, the SAM response to ozone concentrations seems to be too persistent in the WACCM coupled experiment (Fig. 2d) compared to the atmosphere-only configuration (Fig. 2c), especially in summer.
The differences between the sensitivity and all forcing experiments provide some indications of the forcings driving observed ozone–SAM links. The GHG1960 experiment (Fig. 2e) looks like the all-forcing (coupled) experiment, indicating that ozone depletion is the main driver of this interannual relationship. There are still significant correlations when ODSs are fixed at 1960 levels (Fig. 2f), although the correlations are less persistent, suggesting that long-term ozone depletion has increased the strength of the ozone–SAM relationship, as also found by Fogt et al. (2009).
b. Ozone–temperature relationship
Figure 3 is similar to Fig. 2; however, it shows lag correlations between ozone and eastern Australia surface temperature. Eastern Australia was chosen for the reference region as the observational study by Bandoro et al. (2014) showed that the correlation between November ozone and summer surface temperatures was largest in this region. The observations capture significant positive correlations between spring ozone months and seasonal eastern Australia surface temperature, where years with smaller (larger) ozone holes are typically associated with warmer (cooler) temperatures in spring and summer (Figs. 3a–d). However, as mentioned earlier, this result does not demonstrate causality. It is difficult to separate the roles of the polar vortex, wave driving, and ozone concentrations as they are closely related; however, substituting 10-hPa geopotential heights averaged over the polar cap for ozone leads to weaker correlations with Australian temperature in the model. In complementary results to ours, a recent study by Lim et al. (2018) showed that an index of the SH polar vortex is correlated with October and November ozone and with Australian October–January surface temperature.
As in Fig. 2, but for the lagged correlation between the ozone index and eastern Australia surface temperature. Two surface temperature datasets are used for the observations: ERA-Interim (a) with Halley ozone and (c) NIWA-BS ozone, and AWAP with (b) Halley and (d) NIWA-BS.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
As in Fig. 2, but for the lagged correlation between the ozone index and eastern Australia surface temperature. Two surface temperature datasets are used for the observations: ERA-Interim (a) with Halley ozone and (c) NIWA-BS ozone, and AWAP with (b) Halley and (d) NIWA-BS.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
As in Fig. 2, but for the lagged correlation between the ozone index and eastern Australia surface temperature. Two surface temperature datasets are used for the observations: ERA-Interim (a) with Halley ozone and (c) NIWA-BS ozone, and AWAP with (b) Halley and (d) NIWA-BS.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
There are differences between observational datasets, especially when using Halley ozone (Figs. 3a,b), and correlations are less significant overall for AWAP and Halley, compared to ERA-Interim and Halley. A distinct separate band of significant positive correlations is seen for the Halley and ERA-Interim correlations for February ozone (Fig. 3a), which is absent in the other observational correlations. These positive correlations are largest in autumn (p < 0.01) and could be related to a trend toward the positive SAM in April–May (Thompson and Solomon 2002; Ivy et al. 2017). As this second band of significant positive correlations is missing for the other datasets, this suggests that there is some uncertainty in the ozone–Australian surface temperature relationship.
It is in the link between ozone and surface temperature that noticeable differences between the WACCM experiments begin to appear. The uncoupled experiment captures significant correlations between spring ozone and eastern Australian surface temperature in the subsequent seasons (Fig. 3e), broadly like the observations, although it peaks slightly later and has significant positive correlations during more months of the year and for longer lags than observed. Studies have found that CCMs tend to overpredict interannual stratosphere–troposphere relationships due to the polar vortex breaking down later than observed (Fogt et al. 2009; Li et al. 2010). This bias may contribute to the overestimated response in the uncoupled experiment (see also section 3a) and likely has implications for improved seasonal forecasting using ozone. Unlike the uncoupled experiment and observations, the coupled experiment does not capture a significant relationship between spring ozone and spring–summer temperatures (Fig. 3f).
The WACCM sensitivity experiments again provide some insight as to the forcings contributing most to the ozone and Australian temperature teleconnection. In the GHG1960 experiment (Fig. 3g), there are strong positive correlations between the year-to-year size of the ozone hole and eastern Australia surface temperature. These correlations are largest in November but occur for more months in the year than observed, like the uncoupled experiment (Fig. 3e). In the ODS1960 experiment (Fig. 3h), there are weak and insignificant correlations between ozone and surface temperature for most months in the year, similar to the coupled experiment. Thus, the impact of GHGs alone results in a weaker response, consistent with previous results (e.g., Fogt et al. 2009; Bandoro et al. 2014) that suggest that long-term ozone depletion has led to an increase in interannual ozone variability (Table A1) and is therefore more able to produce a signal that can influence the surface. The GHG1960 experiment uses the identical configuration to the coupled experiment but with GHGs fixed at 1960 levels. This experiment captures significant correlations between ozone and eastern Australia surface temperature (Fig. 3g), unlike the all-forcing experiment (Fig. 3f), indicating that the WACCM coupled model can simulate this observed connection, but time-evolving GHGs appear to weaken the relationship. We speculate that this could be related to the interactive impact of increasing GHGs on sea ice and SSTs in this model. For example, similarly to most coupled climate models, Antarctic sea ice extent undergoes large declines over the historical period in the coupled experiment, in contrast to the observed increase over the satellite era (Marsh et al. 2013). This decline would likely impact interannual variability in the SAM (e.g., Kidston et al. 2011; Raphael et al. 2011) and hence Australian surface temperatures, although this hypothesis requires further investigation with additional models.
To provide a global view, surface temperature from ERA-Interim is used for all observational analysis conducted herein. Figure 4 shows the spatial pattern of correlation coefficients of November ozone and summer surface temperatures. These months are examined in detail to evaluate whether WACCM can capture the observed link between November ozone and summer temperatures over Australia found by Bandoro et al. (2014). The two observational ozone datasets (Halley and NIWA-BS) have very similar regional structures (Figs. 4a,b), despite differences in the ozone hole definition and data collection method, indicating that the large-scale patterns are mostly unaffected by these factors. In the Australian region, correlations are largest over southern and eastern Australia. The observed correlations for the period 1979–2004 are not statistically significant over Australia like in the 1979–2012 period (not shown) used by Bandoro et al. (2014), but the pattern is similar.
Correlation coefficients between detrended November ozone and detrended summer surface temperatures (1979–2004) for (a) Halley ozone and (b) NIWA-BS ozone with ERA-Interim surface temperatures, and (c),(d) ozone and surface temperatures from WACCM uncoupled and coupled experiments, respectively. Hatching indicates correlations that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Correlation coefficients between detrended November ozone and detrended summer surface temperatures (1979–2004) for (a) Halley ozone and (b) NIWA-BS ozone with ERA-Interim surface temperatures, and (c),(d) ozone and surface temperatures from WACCM uncoupled and coupled experiments, respectively. Hatching indicates correlations that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Correlation coefficients between detrended November ozone and detrended summer surface temperatures (1979–2004) for (a) Halley ozone and (b) NIWA-BS ozone with ERA-Interim surface temperatures, and (c),(d) ozone and surface temperatures from WACCM uncoupled and coupled experiments, respectively. Hatching indicates correlations that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
The WACCM experiments (Figs. 4c,d) capture similar relationships to the observations over the Antarctica polar cap regarding sign and magnitude; however, away from this region, there are noticeable differences in the spatial pattern. The uncoupled experiment has more significant correlations in the tropical Pacific and Indian Oceans and tends to simulate stronger correlations overall than observed (Fig. 4c). Notably, both model experiments incorrectly simulate the sign of the correlations over the Indian Ocean and have weaker magnitude over the Southern Ocean. The uncoupled experiment captures strong positive and significant correlations over Australia, with the largest correlations in the south and southeast, consistent with the observations, whereas the coupled experiment only has low correlations over Australia, as expected from Fig. 3f.
Figure 5 displays the correlation coefficients between November ozone and summer surface temperature in eastern Australia. For the observations, we compared correlations calculated using Halley and NIWA-BS ozone with TCO from the Syowa and South Pole stations (Fig. 5). The relationship is weaker using South Pole ozone and strongest using Syowa ozone instead of Halley ozone, consistent with van Ommen and Morgan (2010), who found a significant relationship between Antarctic snowfall in the Indian Ocean sector and southwest Australian rainfall. The correlations between November ozone and eastern Australia summer surface temperature in the WACCM coupled experiment are weaker overall than the uncoupled experiment and observations (Fig. 5). Using the correlation coefficients for each ensemble member of the WACCM coupled and uncoupled experiments, we conducted an unpaired two-sample t test to assess the significance of the difference of the means. This analysis shows that the difference in the correlation coefficients between the WACCM uncoupled and coupled experiments is statistically significant at the 5% level.
Correlation coefficients for detrended November ozone and detrended eastern Australia summer surface temperature (1979–2004). Column 1 shows the observations: ERA-Interim surface temperature and Halley (red cross; 75°S, 26°W), Syowa (yellow; 69°S, 39°E), South Pole (blue; 90°S, 25°W), and NIWA-BS (green; 63°–90°S) ozone. Columns 2 and 3 show the WACCM uncoupled and coupled experiments, respectively. Individual ensemble members are shown with a cross and the ensemble mean with a circle. Columns 4, 5, and 6 show all available members for the CCMI models in three groups: CCMI-REF-C1, CCMI-REF-C2-uncoupled where SSTs and sea ice are prescribed from another climate model, and CCMI-REF-C2-coupled (ACCESS-CCM = red cross, CESM1 CAM4-Chem = blue, CMAM = magenta, EMAC-L47MA = gray, EMAC-L90MA = dark green, GEOSCCM = purple, MRI-ESM = pale green, NIWA-UKCA = yellow, SOCOL = orange).
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Correlation coefficients for detrended November ozone and detrended eastern Australia summer surface temperature (1979–2004). Column 1 shows the observations: ERA-Interim surface temperature and Halley (red cross; 75°S, 26°W), Syowa (yellow; 69°S, 39°E), South Pole (blue; 90°S, 25°W), and NIWA-BS (green; 63°–90°S) ozone. Columns 2 and 3 show the WACCM uncoupled and coupled experiments, respectively. Individual ensemble members are shown with a cross and the ensemble mean with a circle. Columns 4, 5, and 6 show all available members for the CCMI models in three groups: CCMI-REF-C1, CCMI-REF-C2-uncoupled where SSTs and sea ice are prescribed from another climate model, and CCMI-REF-C2-coupled (ACCESS-CCM = red cross, CESM1 CAM4-Chem = blue, CMAM = magenta, EMAC-L47MA = gray, EMAC-L90MA = dark green, GEOSCCM = purple, MRI-ESM = pale green, NIWA-UKCA = yellow, SOCOL = orange).
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Correlation coefficients for detrended November ozone and detrended eastern Australia summer surface temperature (1979–2004). Column 1 shows the observations: ERA-Interim surface temperature and Halley (red cross; 75°S, 26°W), Syowa (yellow; 69°S, 39°E), South Pole (blue; 90°S, 25°W), and NIWA-BS (green; 63°–90°S) ozone. Columns 2 and 3 show the WACCM uncoupled and coupled experiments, respectively. Individual ensemble members are shown with a cross and the ensemble mean with a circle. Columns 4, 5, and 6 show all available members for the CCMI models in three groups: CCMI-REF-C1, CCMI-REF-C2-uncoupled where SSTs and sea ice are prescribed from another climate model, and CCMI-REF-C2-coupled (ACCESS-CCM = red cross, CESM1 CAM4-Chem = blue, CMAM = magenta, EMAC-L47MA = gray, EMAC-L90MA = dark green, GEOSCCM = purple, MRI-ESM = pale green, NIWA-UKCA = yellow, SOCOL = orange).
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Figure 5 also shows correlations for the ensemble mean of the WACCM uncoupled and coupled experiments. The ensemble mean reduces the natural variability through averaging and thus helps to isolate the forced response. The ensemble mean correlation is larger than the original correlation in both WACCM experiments, indicating that the forcings (i.e., the historical forcings as well as the SSTs and sea ice in the uncoupled experiment) are enhancing the interannual signal. Given the time series are detrended, it indicates that either some portion of the GHG or ODS forced changes has not been removed through linear regression or that additional forcings are contributing to the interannual relationships found.
Given that the ensemble mean of the uncoupled experiment has larger correlations than the individual ensemble members (Fig. 5), this suggests that part of the ozone–temperature relationship is due to the boundary conditions (SSTs and sea ice) driving both interannual variations in ozone and variations in Australian temperature. To test this hypothesis, we subtracted the ensemble mean ozone and Australian temperature from each ensemble member and repeated the calculation with the concatenated ensemble members. This removes the response to the historical forcings and driving SSTs, and the resulting anomalies represent the response to internally generated ozone variations. The correlations were reduced to a similar magnitude as the coupled experiment (Table 1), therefore confirming our hypothesis.
It is somewhat surprising that the ensemble mean correlation is also enhanced in the coupled model experiment (Fig. 5) since the influence of SSTs will be removed through averaging. Given the time series are detrended, this points to the role of a nonlinear external forcing. Large volcanic eruptions have been shown to impact global mean temperature and significantly deplete stratospheric ozone over Antarctica (e.g., McCormick et al. 1995; Solomon et al. 2016; Stone et al. 2017). When the years corresponding to the El Chichón (1982) and Mount Pinatubo (1991) eruptions were removed from the temperature and ozone time series and the ensemble mean was recalculated, in the WACCM coupled experiment, the ensemble mean correlations are substantially reduced (Table 1). This suggests that the impact of these eruptions on the ozone hole and Australian temperatures is reinforced in the ensemble mean and the ozone hole and Australian temperatures are responding to the volcanic forcings. Most of the signal in the uncoupled experiment appears to be coming from the SSTs as the ensemble mean correlations only show minor decreases when the major volcanic eruptions are removed (Table 1).
Correlation coefficients for detrended Nov ozone and detrended summer surface temperature over eastern Australia from 1979–2004 in the WACCM experiments. The ensemble mean and ENSO were removed from the 5 (3) concatenated ensemble members from the uncoupled (coupled) experiment, and the two major volcanic eruptions [El Chichón (1982) and Mount Pinatubo (1991)] were removed from the model ensemble mean. Correlations for the concatenated members are also shown; refer to Fig. 5 for the ensemble mean values. An asterisk indicates correlations statistically significant at the 90% confidence level, with italics for the 95% level and bold for the 99% level. A two-tailed t test is used to test significance with the degrees of freedom reduced based on the lag-1 autocorrelation.
The WACCM uncoupled experiment appears to have a very strong ENSO response (Fig. 4c), and the summer Niño-3.4 index is significantly correlated with November ozone unlike in the observations or coupled experiment (Table 2). However, strong correlations are still obtained after the ENSO signal is removed from surface temperatures (Table 1), via linear regression against the summer Niño-3.4 index, consistent with the observational study of Bandoro et al. (2014). In the coupled experiment, the relationship between ozone and temperature is strengthened after ENSO is removed from surface temperatures (Table 1).
Correlation coefficients for detrended Nov ozone with the detrended summer Niño-3.4 index, for the period 1979–2004. An asterisk indicates correlations statistically significant at the 90% level, italics for the 95% level, and bold for the 99% level.
c. Analyzing differences between model experiments
Section 3b demonstrated that a CCM (WACCM) could capture the ozone–temperature teleconnection over Australia, including the observed link between November ozone and summer surface temperature. However, this is not the case for the WACCM coupled experiment, as it only captures weak correlations that are not significant (Figs. 4d and 5). The analysis conducted in sections 3c and 3d, therefore, focuses on understanding why the WACCM coupled experiment cannot capture the observed relationship.
To assess the differences between the WACCM uncoupled and coupled experiments, composites are now used; taking the differences between years with high and low November ozone, defined as years greater than one standard deviation. Figure 6 shows the vertical profile of polar cap geopotential height as a function of month. Higher geopotential heights are observed over Antarctica in years with high ozone (Fig. 6), consistent with the negative phase of the SAM. The difference between the two experiments is largest in the troposphere rather than the stratosphere (tropopause located at approximately 200 hPa over Antarctica). In the uncoupled experiment, stratospheric anomalies appear to be followed by similar-signed anomalies in the troposphere, shown by the significant differences between high and low years, and these anomalies reach the surface in late spring to early summer (Fig. 6a). These tropospheric composite differences in December–January are consistent with observations (Thompson and Solomon 2002) where surface anomalies lag stratospheric anomalies by one season, but appear to reach the surface too early in late spring. In the WACCM coupled experiment, there is also downward migration in summer (Fig. 6b); however, it is weaker and not significant. Less downward influence in the coupled experiment could also be related to the somewhat weaker interannual variability in the SAM (see Table A2).
Time–height evolution of the composite differences (high − low ozone) between the years with the highest and lowest (magnitude exceeds one standard deviation; number of years indicated at the top left of each column) polar cap (63°–90°S) averaged November ozone values (1979–2004) for vertically resolved polar cap average geopotential height (m). Composite differences for the WACCM (a) uncoupled and (b) coupled experiment. Hatching indicates differences that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Time–height evolution of the composite differences (high − low ozone) between the years with the highest and lowest (magnitude exceeds one standard deviation; number of years indicated at the top left of each column) polar cap (63°–90°S) averaged November ozone values (1979–2004) for vertically resolved polar cap average geopotential height (m). Composite differences for the WACCM (a) uncoupled and (b) coupled experiment. Hatching indicates differences that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Time–height evolution of the composite differences (high − low ozone) between the years with the highest and lowest (magnitude exceeds one standard deviation; number of years indicated at the top left of each column) polar cap (63°–90°S) averaged November ozone values (1979–2004) for vertically resolved polar cap average geopotential height (m). Composite differences for the WACCM (a) uncoupled and (b) coupled experiment. Hatching indicates differences that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Despite the differences shown in Fig. 6, the coupled experiment can simulate the link between ozone and the SAM (Fig. 2d and Table 3). Thus, the relationship between ozone and surface temperatures appears to break down in the link between the circulation and temperatures, rather than in the link between ozone and the circulation.
Figure 7 shows the difference in the tropospheric and surface response in summer between years with high and low November ozone in the observations. The responses are similar between the Halley and NIWA-BS ozone datasets and ozone hole indices but are weaker overall for NIWA-BS. Years with high ozone are associated with easterly wind anomalies over Australian latitudes (Figs. 7c,d) and warmer temperatures across southern and southeastern Australia (Figs. 7a,b), and the SLP field resembles the negative phase of the SAM (Figs. 7e,f).
Composite differences (high − low ozone) in summer between the years with the highest and lowest (magnitude exceeds one standard deviation; number of years indicated at the top left of each column) November ozone values (1979–2004): (a),(b) surface temperature (K), (c),(d) zonal wind at 500 hPa (m s−1), and (e),(f) sea level pressure (SLP; hPa). Composite differences are shown for (left) Halley ozone and (right) NIWA-BS ozone. Hatching indicates differences that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Composite differences (high − low ozone) in summer between the years with the highest and lowest (magnitude exceeds one standard deviation; number of years indicated at the top left of each column) November ozone values (1979–2004): (a),(b) surface temperature (K), (c),(d) zonal wind at 500 hPa (m s−1), and (e),(f) sea level pressure (SLP; hPa). Composite differences are shown for (left) Halley ozone and (right) NIWA-BS ozone. Hatching indicates differences that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Composite differences (high − low ozone) in summer between the years with the highest and lowest (magnitude exceeds one standard deviation; number of years indicated at the top left of each column) November ozone values (1979–2004): (a),(b) surface temperature (K), (c),(d) zonal wind at 500 hPa (m s−1), and (e),(f) sea level pressure (SLP; hPa). Composite differences are shown for (left) Halley ozone and (right) NIWA-BS ozone. Hatching indicates differences that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Figure 8 is like Fig. 7 but for the WACCM uncoupled and coupled experiments. In the uncoupled experiment, years with high November ozone are associated with significant warm anomalies of up to 2 K over Australia and a warming of the equatorial Pacific Ocean in summer (Fig. 8a). A strong signal can be seen over the Southern Ocean in the 500-hPa zonal wind in the uncoupled experiment (Fig. 8c), corresponding to an equatorward shift and change in the strength of the 500-hPa midlatitude jet during high ozone years or the negative SAM. In comparison, the coupled experiment does not produce a clear surface temperature difference between high and low ozone years and does not exhibit an ENSO signature (Fig. 8b).
As in Fig. 7, but for WACCM, showing composite differences for the (left) uncoupled and (right) coupled experiment.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
As in Fig. 7, but for WACCM, showing composite differences for the (left) uncoupled and (right) coupled experiment.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
As in Fig. 7, but for WACCM, showing composite differences for the (left) uncoupled and (right) coupled experiment.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
The contrast between the observations and WACCM and the difference between the WACCM uncoupled and coupled experiments is largest in the SLP field (Figs. 7e,f and 8e,f). In the uncoupled experiment, there is a Pacific–South American (PSA) wave train (Karoly 1989). The SAM has been shown to strongly resemble the PSA pattern in the Pacific (Ding et al. 2012) and the PSA is related to ENSO on interannual time scales (Mo 2000), indicating that variability in SSTs in the equatorial central-eastern Pacific is linked to the SAM (i.e., ozone) and may strengthen the link between ozone and Australian temperatures. The uncoupled experiment still produces strong correlations between ozone and Australian temperatures when the ENSO signal is removed (Table 1), despite looking like a typical El Niño response (Figs. 8a,e; Zubiaurre and Calvo 2012), indicating that ozone variability can sufficiently induce changes in the SAM and impact surface temperatures. In contrast, the coupled experiment looks more like zonal wavenumber 3 (Fig. 8f; Raphael 2004), which alters the wind patterns and temperature response over Australia. Unlike the observations (Figs. 7e,f), the SAM signal is less distinct in WACCM (Figs. 8e,f). However, the PSA/ENSO and zonal wavenumber 3 patterns are linked to the SAM: ENSO and SAM are strongly correlated in the late spring and early summer (e.g., L’Heureux and Thompson 2006; Lim et al. 2013) and the amplitude of wavenumber 3 is related to the phase of SAM (Turner et al. 2017). The different responses between the uncoupled and coupled experiments could also be related to the ENSO response to ozone being too strong in the uncoupled case (Table 2).
d. Model biases
A coupled ocean in WACCM appears to change the atmospheric and surface response to interannual ozone variability. The coupled experiment may not capture the observed ozone–Australian temperatures teleconnection because the evolution of observed SSTs may be crucial to the relationship. Given that the ensemble mean of the uncoupled experiment shows a higher correlation coefficient, we suggest that the SSTs could be driving both interannual variations in ozone and Australian temperatures, although it is likely that the overall ability of a model to reproduce this relationship is also influenced by model biases.
Figure 9 shows the observed and simulated correlations for eastern Australia summer surface temperatures with SSTs and SLP, respectively. Australian summer surface temperatures are influenced by ENSO, SAM, and the Indian Ocean (Figs. 9a,b). The WACCM uncoupled experiment broadly captures these correlations (Figs. 9c,d) in all basins. In the coupled experiment, however, eastern Australia summer temperatures are dominated by strong anomalies in the tropical Pacific and Indian Oceans (Figs. 9e,f). These tropical model biases may be overwhelming the SAM response and inhibiting the interannual link between ozone and eastern Australian surface temperatures via the SAM. This is supported by the fact that removing the ENSO signal from surface temperatures in the WACCM coupled experiment (Table 1) slightly increases the strength of the relationship between November ozone and summer surface temperature in eastern Australia. Model biases in the Indian Ocean (e.g., Lim and Hendon 2015) may also result in interference in the surface response in the coupled case.
Correlation coefficients between detrended eastern Australia summer surface temperatures and detrended summer SSTs and SLP (1979–2004) for (a),(b) observations, and WACCM (c),(d) uncoupled and (e),(f) coupled experiments. Hatching indicates correlations that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Correlation coefficients between detrended eastern Australia summer surface temperatures and detrended summer SSTs and SLP (1979–2004) for (a),(b) observations, and WACCM (c),(d) uncoupled and (e),(f) coupled experiments. Hatching indicates correlations that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Correlation coefficients between detrended eastern Australia summer surface temperatures and detrended summer SSTs and SLP (1979–2004) for (a),(b) observations, and WACCM (c),(d) uncoupled and (e),(f) coupled experiments. Hatching indicates correlations that are statistically significant at the 95% level.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
Table 4 shows the correlation coefficients between summer ENSO and SAM with eastern Australia summer surface temperature. In the WACCM uncoupled experiment, ENSO explains up to 22% (p < 0.01) of the interannual variability in Australian temperatures, which is close to the observations where ENSO explains 20% (p < 0.1). However, the influence of ENSO in the coupled experiment is too strong and explains more than 50% (p < 0.01; Table 4). The difference between the uncoupled and coupled experiments is highlighted particularly in the impact of SAM on Australian temperatures. While both experiments capture a strong and significant relationship between ozone and SAM (Figs. 2c,d and Table 3), the coupled model poorly simulates the connections between ozone and SAM with Australian temperatures (Figs. 3f,4d, and 5 and Table 4). Similar results for the low-top version of the WACCM model (CCSM4; Table 4) indicate that this bias is not related to the inclusion of interactive chemistry or a more resolved stratosphere in WACCM but rather to a likely breakdown in tropical–extratropical interactions in this version of the atmosphere, potentially related to the overestimated magnitude of ENSO (Deser et al. 2012; Marsh et al. 2013). The most recent version of the low-top model (CESM1-CAM5) has a much improved relationship between SAM and Australian temperatures (Table 4).
Correlation coefficients for detrended summer ENSO and summer SAM indices with detrended eastern Australian summer surface temperature, for the period 1979–2004. An asterisk indicates correlations that are statistically significant at the 90% level, with italics for the 95% level and bold for the 99% level.
The different SAM responses between the WACCM uncoupled and coupled experiments are further highlighted in Fig. 10, the composite differences in summer between years in the high and low phases of the summer SAM. Over Australia, the positive phase of SAM is associated with cooler temperatures (Fig. 10a) related to the poleward shift of the midlatitude jet (Fig. 10d). The WACCM uncoupled experiment broadly resembles the observations, although with stronger temperature differences over Australia (Fig. 10b). In the coupled experiment, SAM does not appear to make a strong contribution to Australian temperatures, shown by the weak temperature differences between high and low SAM (Fig. 10c). The composite temperature differences for the SAM for the coupled experiment also appear to have an ENSO signature (Fig. 10c). Although this warming in the equatorial Pacific Ocean is not significant, it is not seen in the observations or WACCM uncoupled experiment and indicates that ENSO has a strong influence on the SAM during summer in this model experiment. The coupled model SLP composites (Fig. 10i) also indicate positive anomalies in the tropical Indian Ocean and western Pacific Ocean that are not observed and likely interfere with the response to SAM over Australia.
As in Figs. 7 and 8, but for the composite differences (high − low SAM) in summer between the positive and negative phases of the summer SAM (when the magnitude exceeds one standard deviation; number of years is indicated at the top left of each column). Results are shown for (a),(d),(g) observations and for WACCM (b),(e),(h) uncoupled and (c),(f),(i) coupled experiments.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
As in Figs. 7 and 8, but for the composite differences (high − low SAM) in summer between the positive and negative phases of the summer SAM (when the magnitude exceeds one standard deviation; number of years is indicated at the top left of each column). Results are shown for (a),(d),(g) observations and for WACCM (b),(e),(h) uncoupled and (c),(f),(i) coupled experiments.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
As in Figs. 7 and 8, but for the composite differences (high − low SAM) in summer between the positive and negative phases of the summer SAM (when the magnitude exceeds one standard deviation; number of years is indicated at the top left of each column). Results are shown for (a),(d),(g) observations and for WACCM (b),(e),(h) uncoupled and (c),(f),(i) coupled experiments.
Citation: Journal of Climate 32, 11; 10.1175/JCLI-D-18-0273.1
e. Additional models
Figure 5 also shows the correlation coefficients between November ozone and summer surface temperature in eastern Australia for the nine additional CCMI models. These models are organized in three groups: CCMI REF-C1, CCMI REF-C2-uncoupled (SSTs and sea ice prescribed from another climate model), and CCMI REF-C2-coupled. Overall, most models capture the correct sign for the correlation between November ozone and summer surface temperatures in eastern Australia, but there is large intermodel variability in the strength of the correlation coefficients. Unlike in WACCM, there does not appear to be a systematic difference between REF-C1 (uncoupled) and REF-C2 (coupled) for the other models. This suggests that the SSTs might not be primarily driving the response, although they may contribute in part; rather, model biases are likely impacting most models’ ability to reproduce the observed interannual relationship between ozone and Australian summer surface temperature.
4. Discussion
This paper is the first to investigate the possibility of predicting seasonal temperatures in Australia with ozone using a climate model. We have demonstrated that a climate model with interactive chemistry can capture observed connections between interannual variability in Antarctic TCO and Australian temperatures. Although CCMs are computationally expensive, traditional models that prescribe an ozone climatology can severely underestimate the effects of the ozone hole on climate (Li et al. 2016) and will not be able to capture this interannual relationship.
a. Interpretations of discrepancies between observations and models
Section 3 demonstrated that the WACCM REF-C1 (observed SSTs and sea ice) and REF-C2 (coupled ocean) experiments are both able to capture the interannual relationship between ozone and SAM. However, the coupled experiment cannot simulate the interannual relationships between ozone and Australian temperatures and SAM and Australian temperatures, indicating that the relationship breaks down at the surface. Based on the analysis conducted as part of this study, there are currently three plausible interpretations:
A strong relationship is only seen when the model is forced with observed SSTs, suggesting that much of the observed signal could be due to the SSTs rather than the Antarctic ozone hole, and the Australian temperatures and ozone hole are simultaneously responding to the SSTs.
Model biases might hinder the ability of some climate models to simulate this interannual relationship reliably.
Uncertainty in the observations could indicate that the connection between ozone and Australian surface temperatures is not robust and could also be influenced by natural decadal variability.
The first possibility is based on the results from WACCM where the observed SSTs and sea ice appear to have an important role in the ozone and Australian temperatures relationship. The results from the WACCM uncoupled experiment are consistent across all ensemble members and increase for the ensemble mean, suggesting that SSTs could be driving variations in both Australian temperatures and the ozone hole. This hypothesis is supported by the results from REF-C2, where the coupled ocean and freely evolving SSTs and sea ice result in no significant relationship between ozone and Australian temperatures. However, there does not seem to be as clear a difference between REF-C1 and REF-C2 pairs in the four other CCMI models that have an interactive ocean for REF-C2 (Fig. 5); this hypothesis, therefore, requires further investigation.
The second possibility is that the models are unable to represent the key processes necessary to simulate the ozone–temperature relationship correctly. In the WACCM coupled experiment, for example, significant correlations (p < 0.1) for November ozone and summer surface temperatures in south and southeast Australia are only obtained after linearly removing the ENSO signal (not shown) and are still much weaker than observed (Table 1). As noted in section 3d, the large ENSO amplitude (Deser et al. 2012; Marsh et al. 2013) in the WACCM coupled model may be interfering with the SAM response and impacting the relationship between ozone and surface climate. Furthermore, in the coupled experiment, sea ice is interactive (compared to the uncoupled experiment where it is prescribed from observations); therefore, it is also possible that the coupled model could be influenced by a sea ice feedback (Magnusdottir et al. 2004) that may interfere with the SAM. The preliminary analysis of the fixed GHG experiments (section 3b) highlights that the coupled model shows an improved simulation of the ozone–surface temperature relationship when the long-term warming associated with increased GHGs is omitted. This suggests that the warming acts to interfere with the interannual variability in ozone and surface climate relationship in the coupled experiment. One hypothesis is that the unrealistic Antarctic sea ice declines and different SST patterns could push the climate system into a different state to that observed and will be the subject of future work.
The third possibility is that the observed connection between ozone and Australian temperatures is not particularly robust. While Bandoro et al. (2014) reported a statistically significant relationship between November Halley ozone and ERA-Interim summer surface temperatures in eastern Australia, correlations are largest when these two datasets are used and are weaker or more variable for other combinations. The correlations in this study for the period 1979–2004 are also weaker overall than for 1979–2012 (not shown). After 2000 the TCO trend is less negative and even starting to become positive due to initial signs of ozone recovery (WMO 2014; Solomon et al. 2016), although detection of recovery is hindered by limited data records and large atmospheric variability (Chipperfield et al. 2017). Studies have linked the summer positive SAM trend since 2000 to recent changes in SSTs and decadal variability (Pacific decadal oscillation; e.g., Schneider et al. 2015) in addition to ozone depletion and these changes could also be influencing the observed relationship. The possible time-varying nature in the strength of the ozone–temperature connection makes it difficult to compare the observations to model output directly. Further analysis is required to examine the strength and linearity of this relationship and its applicability to additional datasets, to assess whether model results fall within error estimates.
b. Predicting Australian summer temperatures with ozone
This study aimed to examine whether interannual Antarctic spring ozone variability could be used as an indicator of Australian summer surface temperature variability in climate models. To this end, this work has shown that some CCMs can capture the observed relationship between ozone and surface temperatures, and has, therefore, indicated a potential benefit of incorporating ozone variability in seasonal forecasting systems. Operational seasonal forecasting systems such as the Australian Bureau of Meteorology’s seasonal climate forecast system, the Predictive Ocean and Atmosphere Model for Australia (POAMA), are coupled but initialized with observed atmosphere and ocean conditions (Lim et al. 2016). Therefore, the ability of some uncoupled models in this study, including WACCM (Fig. 5), to capture a strong relationship between ozone and Australian summer surface temperatures is encouraging to eventually use real-time ozone variability to improve skill in season outlook systems. However, the reliability and accuracy of modeling this relationship is hindered by model biases. Further research to assess model biases will assist in understanding why some models cannot correctly simulate this observed connection with a view to eliminating model biases and eventually improving seasonal prediction.
Current operational seasonal forecasting models typically have a poorly resolved stratosphere (Maycock et al. 2011). For example, POAMA only has five levels above 200 hPa, and the ozone concentration is set to climatological values (Lim et al. 2016). It may be unable to capture links between the stratosphere and troposphere, and thus there is a large scope for improving prediction of tropospheric interannual variability. Roff et al. (2011) found that improvements in the stratosphere in a forecasting model, such as a higher stratospheric resolution and better representation of stratospheric dynamics and thermodynamics, led to significant improvements in tropospheric forecast skill. Hence, even if spring Antarctic ozone levels do not prove to be a reliable predictor of SH summer temperature extremes, there is still potential benefit in including time-varying ozone and improving stratospheric representation in operational forecasting systems.
5. Conclusions
This study examined the ability for WACCM as well as other CCMs to simulate observed links between the spring Antarctic ozone hole and summer surface temperatures over Australia. A systematic difference is found between the uncoupled and coupled experiments in WACCM and three possible interpretations are provided to explain the discrepancy in simulating the ozone–temperature teleconnection: 1) SSTs play a dominant role and drive interannual variations in both the ozone hole and Australian temperatures, 2) the CCMI models are unable to represent key processes, and/or 3) the observed relationship has some uncertainty and is time-varying. While the Australian temperatures and ozone hole may be responding primarily to the SSTs in WACCM, there is some indication that CCMI uncoupled experiments also capture the relationship with more fidelity than the coupled experiments. However, there is not as clear a difference between other coupled and uncoupled experiment pairs among the CCMI models, and this hypothesis requires further investigation. It is also possible that the models are unable to capture the observed relationship due to biases, such as in the ENSO amplitude. Furthermore, long-term GHG-induced warming also seems to interfere with the response in the WACCM coupled model. This paper has also highlighted that there is some observational uncertainty regarding the strength of the ozone–temperature teleconnection.
The results of this study are encouraging for incorporating ozone variability to improve seasonal predictions, although more work is needed to identify causality in the link between spring ozone and SH surface climate. An experiment, that compares the predictive skill in a seasonal forecasting model that is initialized with and without observed ozone, would be the next step to demonstrating useful seasonal skill from Antarctic ozone. In addition, targeted modeling experiments which separate the role of SSTs and interannual ozone variations would help to elucidate the mechanism by which ozone impacts the surface climate.
Acknowledgments
Zoe Gillett was funded by the Grains Research and Development Corporation (UHS11005). Portions of this study were supported by the Regional and Global Climate Modeling Program (RGCM) of the U.S. Department of Energy’s Office of Biological and Environmental Research (BER) Cooperative Agreement DE-FC02-97ER62402 and the National Science Foundation (NSF) as well as the Australian Research Council (ARC) Centre of Excellence for Climate Extremes (CE170100023). Andrea Dittus acknowledges support from the ARC Centre of Excellence for Climate System Science (CE110001028) and the U.K. Natural Environment Research Council (NERC) Project SMURPHS (NE/N006054/1). This research was undertaken with the assistance of resources and services from the National Computational Infrastructure, which is supported by the Australian Government. The National Center for Atmospheric Research (NCAR) Command Language (NCL) was used for data analysis and visualization. We acknowledge the modelling groups for making their simulations available for this analysis, the joint WCRP SPARC/IGAC Chemistry–Climate Model Initiative (CCMI) for organizing and coordinating the model data analysis activity, and the British Atmospheric Data Centre (BADC) for collecting and archiving the CCMI model output. We acknowledge high-performance computational support for the WACCM simulations from Yellowstone (ark:/85065/d7wd3xhc) provided by the Climate Simulation Laboratory at NCAR’s Computational and Information Systems Laboratory, sponsored by NSF and other agencies. NCAR is funded by NSF. The EMAC simulations have been performed at the German Climate Computing Centre (DKRZ) through support from the Bundesministerium für Bildung und Forschung (BMBF). DKRZ and its scientific steering committee are gratefully acknowledged for providing the HPC and data archiving resources for the consortial project ESCiMo (Earth System Chemistry integrated Modelling). Laura Revell acknowledges partial support from the Deep South National Science Challenge (Contract C01X1412) and China Southern. Eugene Rozanov acknowledges partial support from the Swiss National Science Foundation under Grants 200021 169241 (VEC) and 200020 182239 (POLE) and the gained information will be used to improve the CCM SOCOL. Robyn Schofield and Kane Stone acknowledge support from the ARC Centre of Excellence for Climate System Science (CE110001028), the Australian Government’s National Computational Merit Allocation Scheme (q90), and Australian Antarctic science grant program (FoRCES 4012). We also acknowledge Bodeker Scientific, supported through the Deep South National Science Challenge, for providing the combined total column ozone database. We thank Dan Marsh for useful discussions during the course of this study, and three anonymous reviewers whose comments helped to significantly improve the manuscript.
APPENDIX
Standard Deviation of the Ozone and SAM Indices
Table A1 shows the standard deviation of the monthly ozone index for the observations and WACCM experiments. Table A2 shows the standard deviation of the seasonal SAM index for the observations and WACCM experiments.
Standard deviation of the ozone index over 26 years of data, for the observed Halley and NIWA-BS datasets and the three uncoupled, coupled, GHG1960, and ODS1960 WACCM ensemble members. Years are 1979–2004 for Sep–Dec and 1980–2005 for Jan–Apr.
Standard deviation of the SAM index over 26 years of data for each 3-month overlapping period, for the Marshall (2003) SAM index and the three uncoupled and coupled WACCM ensemble members; 1979–2004 for periods beginning in Sep–Dec and 1980–2005 for Jan–Apr.
REFERENCES
Arblaster, J. M., and G. A. Meehl, 2006: Contributions of external forcings to southern annular mode trends. J. Climate, 19, 2896–2905, https://doi.org/10.1175/JCLI3774.1.
Arblaster, J. M., and L. V. Alexander, 2012: The impact of the El Niño–Southern Oscillation on maximum temperature extremes. Geophys. Res. Lett., 39, L20702, https://doi.org/10.1029/2012GL053409.
Bandoro, J., S. Solomon, A. Donohoe, D. W. J. Thompson, and B. D. Santer, 2014: Influences of the Antarctic ozone hole on Southern Hemispheric summer climate change. J. Climate, 27, 6245–6264, https://doi.org/10.1175/JCLI-D-13-00698.1.
Bodeker, G. E., H. Shiona, and H. Eskes, 2005: Indicators of Antarctic ozone depletion. Atmos. Chem. Phys., 5, 2603–2615, https://doi.org/10.5194/acp-5-2603-2005.
Bretherton, C. S., M. Widmann, V. P. Dymnikov, J. M. Wallace, and I. Bladé, 1999: The effective number of spatial degrees of freedom of a time-varying field. J. Climate, 12, 1990–2009, https://doi.org/10.1175/1520-0442(1999)012<1990:TENOSD>2.0.CO;2.
Chipperfield, M. P., and Coauthors, 2017: Detecting recovery of the stratospheric ozone layer. Nature, 549, 211–218, https://doi.org/10.1038/nature23681.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Dennison, F. W., A. J. McDonald, and O. Morgenstern, 2015: The effect of ozone depletion on the southern annular mode and stratosphere–troposphere coupling. J. Geophys. Res. Atmos., 120, 6305–6312, https://doi.org/10.1002/2014JD023009.
Deser, C., and Coauthors, 2012: ENSO and Pacific decadal variability in the Community Climate System Model version 4. J. Climate, 25, 2622–2651, https://doi.org/10.1175/JCLI-D-11-00301.1.
Ding, Q. H., E. J. Steig, D. S. Battisti, and J. M. Wallace, 2012: Influence of the tropics on the southern annular mode. J. Climate, 25, 6330–6348, https://doi.org/10.1175/JCLI-D-11-00523.1.
Eyring, V., and Coauthors, 2006: Assessment of temperature, trace species, and ozone in chemistry–climate model simulations of the recent past. J. Geophys. Res., 111, D22308, https://doi.org/10.1029/2006JD007327.
Eyring, V., and Coauthors, 2013: Overview of IGAC/SPARC Chemistry-Climate Model Initiative (CCMI) community simulations in support of upcoming ozone and climate assessments. SPARC Newsletter, No. 40, WMO-WRCP, Geneva, Switzerland, 48–66.
Fogt, R. L., J. Perlwitz, S. Pawson, and M. A. Olsen, 2009: Intra-annual relationships between polar ozone and the SAM. Geophys. Res. Lett., 36, L04707, https://doi.org/10.1029/2008GL036627.
Gillett, N. P., and D. W. J. Thompson, 2003: Simulation of recent Southern Hemisphere climate change. Science, 302, 273–275, https://doi.org/10.1126/science.1087440.
Gong, D., and S. Wang, 1999: Definition of Antarctic Oscillation index. Geophys. Res. Lett., 26, 459–462, https://doi.org/10.1029/1999GL900003.
Grise, K. M., and L. M. Polvani, 2017: Understanding the time scales of the tropospheric circulation response to abrupt CO2 forcing in the Southern Hemisphere: Seasonality and the role of the stratosphere. J. Climate, 30, 8497–8515, https://doi.org/10.1175/JCLI-D-16-0849.1.
Hegglin, M. I., and J.-F. Lamarque, 2015: The IGAC/SPARC Chemistry-Climate Model Initiative Phase-1 (CCMI-1) model data output. NCAS British Atmospheric Data Centre, accessed 5 March 2018, http://catalogue.ceda.ac.uk/uuid/9cc6b94df0f4469d8066d69b5df879d5.
Hendon, H. H., D. W. J. Thompson, and M. C. Wheeler, 2007: Australian rainfall and surface temperature variations associated with the Southern Hemisphere annular mode. J. Climate, 20, 2452–2467, https://doi.org/10.1175/JCLI4134.1.
Hudson, D., A. G. Marshall, O. Alves, G. Young, D. Jones, and A. Watkins, 2016: Forewarned is forearmed: Extended-range forecast guidance of recent extreme heat events in Australia. Wea. Forecasting, 31, 697–711, https://doi.org/10.1175/WAF-D-15-0079.1.
Ivy, D. J., C. Hilgenbrink, D. Kinnison, R. A. Plumb, A. Sheshadri, S. Solomon, and D. W. J. Thompson, 2017: Observed changes in the Southern Hemispheric circulation in May. J. Climate, 30, 527–536, https://doi.org/10.1175/JCLI-D-16-0394.1.
Jones, D. A., W. Wang, and R. Fawcett, 2009: High-quality spatial climate data-sets for Australia. Aust. Meteor. Oceanogr. J., 58, 233–248, https://doi.org/10.22499/2.5804.003.
Karoly, D. J., 1989: Southern Hemisphere circulation features associated with El Niño–Southern Oscillation events. J. Climate, 2, 1239–1252, https://doi.org/10.1175/1520-0442(1989)002<1239:SHCFAW>2.0.CO;2.
Kidston, J., A. S. Taschetto, D. W. J. Thompson, and M. H. England, 2011: The influence of Southern Hemisphere sea-ice extent on the latitude of the mid-latitude jet stream. Geophys. Res. Lett., 38, L15804, https://doi.org/10.1029/2011GL048056.
Lee, S., and S. B. Feldstein, 2013: Detecting ozone- and greenhouse gas–driven wind trends with observational data. Science, 339, 563–567, https://doi.org/10.1126/science.1225154.
L’Heureux, M. L., and D. W. J. Thompson, 2006: Observed relationships between the El Niño–Southern Oscillation and the extratropical zonal-mean circulation. J. Climate, 19, 276–287, https://doi.org/10.1175/JCLI3617.1.
Li, F., P. A. Newman, and R. S. Stolarski, 2010: Relationships between the Brewer-Dobson circulation and the southern annular mode during austral summer in coupled chemistry–climate model simulations. J. Geophys. Res., 115, D15106, https://doi.org/10.1029/2009JD012876.
Li, F., Y. V. Vikhliaev, P. A. Newman, S. Pawson, J. Perlwitz, D. W. Waugh, and A. R. Douglass, 2016: Impacts of interactive stratospheric chemistry on Antarctic and Southern Ocean climate change in the Goddard Earth Observing System, version 5 (GEOS-5). J. Climate, 29, 3199–3218, https://doi.org/10.1175/JCLI-D-15-0572.1.
Lim, E.-P., and H. H. Hendon, 2015: Understanding the contrast of Australian springtime rainfall of 1997 and 2002 in the frame of two flavors of El Niño. J. Climate, 28, 2804–2822, https://doi.org/10.1175/JCLI-D-14-00582.1.
Lim, E.-P., H. H. Hendon, and H. Rashid, 2013: Seasonal predictability of the southern annular mode due to its association with ENSO. J. Climate, 26, 8037–8054, https://doi.org/10.1175/JCLI-D-13-00006.1.
Lim, E.-P., H. H. Hendon, D. Hudson, M. Zhao, L. Shi, O. Alves, and G. Young, 2016: Evaluation of the ACCESS-S1 hindcasts for prediction of Victorian seasonal rainfall. Bureau Res. Rep. 019, 43 pp., http://www.bom.gov.au/research/publications/research-reports.shtml.
Lim, E.-P., H. H. Hendon, and D. W. J. Thompson, 2018: Seasonal evolution of stratosphere-troposphere coupling in the Southern Hemisphere and implications for the predictability of surface climate. J. Geophys. Res. Atmos., 123, 12 002–12 016, https://doi.org/10.1029/2018JD029321.
Lin, P., D. Paynter, L. Polvani, G. J. P. Correa, Y. Ming, and V. Ramaswamy, 2017: Dependence of model-simulated response to ozone depletion on stratospheric polar vortex climatology. Geophys. Res. Lett., 44, 6391–6398, https://doi.org/10.1002/2017GL073862.
Magnusdottir, G., C. Deser, and R. Saravanan, 2004: The effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part I: Main features and storm track characteristics of the response. J. Climate, 17, 857–876, https://doi.org/10.1175/1520-0442(2004)017<0857:TEONAS>2.0.CO;2.
Marsh, D. R., M. J. Mills, D. E. Kinnison, J.-F. Lamarque, N. Calvo, and L. M. Polvani, 2013: Climate change from 1850 to 2005 simulated in CESM1(WACCM). J. Climate, 26, 7372–7391, https://doi.org/10.1175/JCLI-D-12-00558.1.
Marshall, G. J., 2003: Trends in the southern annular mode from observations and reanalyses. J. Climate, 16, 4134–4143, https://doi.org/10.1175/1520-0442(2003)016<4134:TITSAM>2.0.CO;2.
Maycock, A. C., S. P. E. Keeley, A. J. Charlton-Perez, and F. J. Doblas-Reyes, 2011: Stratospheric circulation in seasonal forecasting models: Implications for seasonal prediction. Climate Dyn., 36, 309–321, https://doi.org/10.1007/s00382-009-0665-x.
McBride, J. L., and N. Nicholls, 1983: Seasonal relationships between Australian rainfall and the Southern Oscillation. Mon. Wea. Rev., 111, 1998–2004, https://doi.org/10.1175/1520-0493(1983)111<1998:SRBARA>2.0.CO;2.
McCormick, M. P., L. W. Thomason, and C. R. Trepte, 1995: Atmospheric effects of the Mt. Pinatubo eruption. Nature, 373, 399–404, https://doi.org/10.1038/373399a0.
McLandress, C., T. G. Shepherd, J. F. Scinocca, D. A. Plummer, M. Sigmond, A. I. Jonsson, and M. C. Reader, 2011: Separating the dynamical effects of climate change and ozone depletion. Part II: Southern Hemisphere troposphere. J. Climate, 24, 1850–1868, https://doi.org/10.1175/2010JCLI3958.1.
Meinshausen, M., and Coauthors, 2011: The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change, 109, 213, https://doi.org/10.1007/s10584-011-0156-z.
Min, S.-K., W. Cai, and P. Whetton, 2013: Influence of climate variability on seasonal extremes over Australia. J. Geophys. Res. Atmos., 118, 643–654, https://doi.org/10.1002/jgrd.50164.
Mo, K. C., 2000: Relationships between low-frequency variability in the Southern Hemisphere and sea surface temperature anomalies. J. Climate, 13, 3599–3610, https://doi.org/10.1175/1520-0442(2000)013<3599:RBLFVI>2.0.CO;2.
Morgenstern, O., and Coauthors, 2017: Review of the global models used within phase 1 of the Chemistry–Climate Model Initiative (CCMI). Geosci. Model Dev., 10, 639–671, https://doi.org/10.5194/gmd-10-639-2017.
Perkins, S. E., D. Argüeso, and C. J. White, 2015: Relationships between climate variability, soil moisture, and Australian heatwaves. J. Geophys. Res. Atmos., 120, 8144–8164, https://doi.org/10.1002/2015JD023592.
Perkins-Kirkpatrick, S. E., and Coauthors, 2016: Natural hazards in Australia: Heatwaves. Climatic Change, 139, 101–114, https://doi.org/10.1007/s10584-016-1650-0.
Polvani, L. M., D. W. Waugh, G. J. P. Correa, and S.-W. Son, 2011: Stratospheric ozone depletion: The main driver of twentieth-century atmospheric circulation changes in the Southern Hemisphere. J. Climate, 24, 795–812, https://doi.org/10.1175/2010JCLI3772.1.
Raphael, M. N., 2004: A zonal wave 3 index for the Southern Hemisphere. Geophys. Res. Lett., 31, L23212, https://doi.org/10.1029/2004GL020365.
Raphael, M. N., W. Hobbs, and I. Wainer, 2011: The effect of Antarctic sea ice on the Southern Hemisphere atmosphere during the southern summer. Climate Dyn., 36, 1403–1417, https://doi.org/10.1007/s00382-010-0892-1.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, https://doi.org/10.1029/2002JD002670.
Risbey, J. S., M. J. Pook, P. C. McIntosh, M. C. Wheeler, and H. H. Hendon, 2009: On the remote drivers of rainfall variability in Australia. Mon. Wea. Rev., 137, 3233–3253, https://doi.org/10.1175/2009MWR2861.1.
Roff, G., D. W. J. Thompson, and H. Hendon, 2011: Does increasing model stratospheric resolution improve extended-range forecast skill? Geophys. Res. Lett., 38, L05809, https://doi.org/10.1029/2010GL046515.
Rogers, J. C., and H. van Loon, 1982: Spatial variability of sea level pressure and 500 mb height anomalies over the Southern Hemisphere. Mon. Wea. Rev., 110, 1375–1392, https://doi.org/10.1175/1520-0493(1982)110<1375:SVOSLP>2.0.CO;2.
Salby, M. L., E. A. Titova, and L. Deschamps, 2011: Rebound of Antarctic ozone. Geophys. Res. Lett., 38, L09702, https://doi.org/10.1029/2011GL047266.
Salby, M. L., E. A. Titova, and L. Deschamps, 2012: Changes of the Antarctic ozone hole: Controlling mechanisms, seasonal predictability, and evolution. J. Geophys. Res., 117, D10111, https://doi.org/10.1029/2011JD016285.
Santer, B. D., T. M. L. Wigley, J. S. Boyle, D. J. Gaffen, J. J. Hnilo, D. Nychka, D. E. Parker, and K. E. Taylor, 2000: Statistical significance of trends and trend differences in layer-average atmospheric temperature time series. J. Geophys. Res., 105, 7337–7356, https://doi.org/10.1029/1999JD901105.
Schneider, D. P., C. Deser, and T. T. Fan, 2015: Comparing the impacts of tropical SST variability and polar stratospheric ozone loss on the Southern Ocean westerly winds. J. Climate, 28, 9350–9372, https://doi.org/10.1175/JCLI-D-15-0090.1.
Seviour, W. J. M., S. C. Hardiman, L. J. Gray, N. Butchart, C. MacLachlan, and A. A. Scaife, 2014: Skillful seasonal prediction of the southern annular mode and Antarctic ozone. J. Climate, 27, 7462–7474, https://doi.org/10.1175/JCLI-D-14-00264.1.
Seviour, W. J. M., D. W. Waugh, L. M. Polvani, G. J. P. Correa, and C. I. Garfinkel, 2017: Robustness of the simulated tropospheric response to ozone depletion. J. Climate, 30, 2577–2585, https://doi.org/10.1175/JCLI-D-16-0817.1.
Shaw, T. A., J. Perlwitz, N. Harnik, P. A. Newman, and S. Pawson, 2011: The impact of stratospheric ozone changes on downward wave coupling in the Southern Hemisphere. J. Climate, 24, 4210–4229, https://doi.org/10.1175/2011JCLI4170.1.
Sheshadri, A., and R. A. Plumb, 2016: Sensitivity of the surface responses of an idealized AGCM to the timing of imposed ozone depletion-like polar stratospheric cooling. Geophys. Res. Lett., 43, 2330–2336, https://doi.org/10.1002/2016GL067964.
Smith, K. L., and L. M. Polvani, 2017: Spatial patterns of recent Antarctic surface temperature trends and the importance of natural variability: Lessons from multiple reconstructions and the CMIP5 models. Climate Dyn., 48, 2653–2670, https://doi.org/10.1007/s00382-016-3230-4.
Solomon, S., 1999: Stratospheric ozone depletion: A review of concepts and history. Rev. Geophys., 37, 275–316, https://doi.org/10.1029/1999RG900008.
Solomon, S., D. J. Ivy, D. Kinnison, M. J. Mills, R. R. Neely III, and A. Schmidt, 2016: Emergence of healing in the Antarctic ozone layer. Science, 353, 269–274, https://doi.org/10.1126/science.aae0061.
Son, S.-W., and Coauthors, 2008: The impact of stratospheric ozone recovery on the Southern Hemisphere westerly jet. Science, 320, 1486–1489, https://doi.org/10.1126/science.1155939.
Son, S.-W., A. Purich, H. H. Hendon, B.-M. Kim, and L. M. Polvani, 2013: Improved seasonal forecast using ozone hole variability? Geophys. Res. Lett., 40, 6231–6235, https://doi.org/10.1002/2013GL057731.
Stone, K. A., O. Morgenstern, D. J. Karoly, A. R. Klekociuk, W. J. French, N. L. Abraham, and R. Schofield, 2016: Evaluation of the ACCESS chemistry–climate model for the Southern Hemisphere. Atmos. Chem. Phys., 16, 2401–2415, https://doi.org/10.5194/acp-16-2401-2016.
Stone, K. A., and Coauthors, 2017: Observing the impact of Calbuco volcanic aerosols on South Polar ozone depletion in 2015. J. Geophys. Res. Atmos., 122, 11 862–11 879, https://doi.org/10.1002/2017JD026987.
Thompson, D. W. J., and S. Solomon, 2002: Interpretation of recent Southern Hemisphere climate change. Science, 296, 895–899, https://doi.org/10.1126/science.1069270.
Thompson, D. W. J., M. P. Baldwin, and S. Solomon, 2005: Stratosphere–troposphere coupling in the Southern Hemisphere. J. Atmos. Sci., 62, 708–715, https://doi.org/10.1175/JAS-3321.1.
Thompson, D. W. J., S. Solomon, P. J. Kushner, M. H. England, K. M. Grise, and D. J. Karoly, 2011: Signatures of the Antarctic ozone hole in Southern Hemisphere surface climate change. Nat. Geosci., 4, 741–749, https://doi.org/10.1038/ngeo1296.
Trenberth, K. E., 1979: Interannual variability of the 500 mb zonal mean flow in the Southern Hemisphere. Mon. Wea. Rev., 107, 1515–1524, https://doi.org/10.1175/1520-0493(1979)107<1515:IVOTMZ>2.0.CO;2.
Trenberth, K. E., 1997: The definition of El Niño. Bull. Amer. Meteor. Soc., 78, 2771–2777, https://doi.org/10.1175/1520-0477(1997)078<2771:TDOENO>2.0.CO;2.
Turner, J., J. S. Hosking, T. J. Bracegirdle, T. Phillips, and G. J. Marshall, 2017: Variability and trends in the Southern Hemisphere high latitude, quasi-stationary planetary waves. Int. J. Climatol., 37, 2325–2336, https://doi.org/10.1002/joc.4848.
van Ommen, T. D., and V. Morgan, 2010: Snowfall increase in coastal East Antarctica linked with southwest Western Australian drought. Nat. Geosci., 3, 267–272, https://doi.org/10.1038/ngeo761.
WMO, 2014: Scientific Assessment of Ozone Depletion: 2014. Global Ozone Research and Monitoring Project, World Meteorological Organization, 416 pp.
Zubiaurre, I., and N. Calvo, 2012: The El Niño–Southern Oscillation (ENSO) Modoki signal in the stratosphere. J. Geophys. Res., 117, D04104, https://doi.org/10.1029/2011JD016690.