Surface temperature trends according to the AMRC-AWS

Figure 3 differentiates average daily temperature taken from all of the stations in the WAP and EAP, respectively. The WAP’s mean temperatures vary from 0°C during DJF to −20°C during the JJA season, as shown in Fig. 3a. The mean maximum temperature is −5.6°C and the mean minimum temperature is −15°C. The annual mean temperature was calculated by combining all seven AMRC-AWS datasets. Data for the available AWS were selected based on them being > 75% complete for each and having more than 10 complete years.

Figure 3.

Daily maximum (red), mean (black) and minimum (blue) temperatures (°C) for the 1985–2020 period of the Antarctic Meteorological Research and Data Center (AMRC) and Reference Antarctic Data for Environmental Research (READER) automatic weather systems in a. West Antarctic Peninsula and b. the East Antarctic Peninsula.

Figure 3. Long description

A two-panel line graph titled Long Term 1985 to 2020 Daily A M R C A W S Observational Temperature Mean Variation over W A P and E A P.

Panel a, labeled W A P, shows temperature on the y-axis from negative 40 to 10 degrees Celsius and Day of the Year on the x-axis from 0 to 350. Three jagged lines follow a U-shaped seasonal trend. The top red line (maximum) starts near 5 degrees, dips to negative 10 around day 200, and returns to 5. The middle black line (mean) starts at 0, dips to negative 15, and returns to 0. The bottom blue line (minimum) starts at negative 5, dips to negative 20, and returns to negative 5.

Panel b, labeled E A P, uses the same axes but shows a deeper U-shaped curve indicating colder temperatures. The red line starts near 5, drops sharply to negative 20 around day 220, and rises back to 5. The black line starts at negative 2, drops to negative 30, and rises to negative 2. The blue line starts at negative 5, reaches a minimum near negative 35, and rises back to negative 5. The fluctuations in panel b are more pronounced than in panel a.

Based on the graphs in Fig. 3, the AP has shown two distinct features and characteristics regarding the EAP and WAP: one has a lower daily annual temperature, while the other has a higher daily annual temperature. This may be due to the topography and orography effects on the area, with a narrow mountain barrier extending northwards to Drake Passage from Ellsworth Land of 1300 km in length near 63°S (Carrasco, Reference Carrasco2013). This steep mountain range produces two distinct climate environments on each side of the AP. Therefore, the WAP has a warmer range for its mean annual temperature of between −13.74°C and 5.41°C compared to EAP, which has a lower mean annual temperature of −15.27°C. Climatological MSLP charts revealed cyclonic circulation on both sides of the AP, associated with distinct low-pressure centres: one located over the Amundsen Sea and the other farther south, near the Antarctic coast along the Atlantic Ocean. The former favours northerly to north-westerly flow, while the latter promotes southerly flow, including the occurrence of katabatic winds.

During DJF periods, the EAP’s AWSs recorded high daily maximum temperatures, such as 7.37°C at Larsen Ice Shelf and 7.28°C at Butler Island. These high temperatures are influenced by strong westerly winds and the high orography of the region, which affects temperature distribution more on the EAP than on the WAP (Turner et al. Reference Turner, Lu, Marshall, Phillips, Bannister and Colwell2021). The strengthening of the westerly winds is linked to the positive phase of the SAM, which has been increasingly observed in recent decades. This positive phase of the SAM results in a poleward shift of the westerly wind belt, enhancing the advection of warmer air masses towards the AP (Stenni et al. Reference Stenni, Curran, Abram, Orsi, Goursaud and Masson-Delmotte2017, Dickens et al. Reference Dickens, Kuhn, Leng, Graham, Dowdeswell and Meredith2019, Tarca et al. Reference Tarca, Guglielmin, Convey, Worland and Cannone2021, Zhang et al. Reference Zhang, Wang and Hou2022).

The AP’s climatic complexity is further highlighted by studies that demonstrate the influence of the SAM on temperature and precipitation patterns. For instance, a positive SAM phase is associated with increased precipitation on the WAP and decreased precipitation on the EAP, contributing to the distinct climatic conditions observed across the AP (Marshall Reference Marshall2003, Fogt & Marshall Reference Fogt and Marshall2020). These findings underscore the importance of considering global atmospheric patterns and teleconnections and regional orographic features for understanding the AP’s climate dynamics.

Figure 4 illustrates the long-term trend in the AP that fell within the El Niño years and during the positive and negative SAM to identify which of the years were likely to be influenced by the previously discussed climate drivers. Figure S1 shows the seasonal trend for the whole AP taken from the AMRC-AWS, which exhibits a warming trend, except in SON. There is also a noticeable condition: for the ‘very strong’ El Niño years (1997–1998), when overlapped with a positive SAM, there was an increase in temperature, while another episode (2015–2016) only shows an increase in temperature for JJA. Our study also found that, by using AMRC-AWS, the temperature recorded was increasing during the JJA seasons for most of the stations but not during DJF, as has been previously reported in other studies (Ding & Steig Reference Ding and Steig2013, Bozkurt et al. Reference Bozkurt, Carrasco, Cordero, Fernandoy, Gómez-Contreras, Carrillo and Guan2024). The same trend was also noticed by Thomas & Tetzner (Reference Thomas, Tetzner, Kanao, Toyokuni and Yamamoto2018) between 1950 and 2011, during which the most significant warming occurred during the JJA season, and rapid cooling happened during the DJF season.

Figure 4.

Antarctic Meteorological Research and Data Center automatic weather station (AMRC-AWS) mean temperatures (°C) across 1985–2020. Vertical shaded lines indicate the intensities of El Niño and La Niña years. The smooth magenta line represents the Southern Annular Mode (SAM) index.

Figure 4. Long description

The horizontal x-axis represents Year from 1985 to 2020. The primary vertical y-axis on the left represents Temperature in degrees Celsius from minus 20 to 0. The secondary vertical y-axis on the right represents S A M from minus 2 to 2.

Data Series:

* A solid dark blue line for A M R C dash A W S shows temperature fluctuating between minus 15 and minus 2 degrees Celsius. A dashed black linear trend line shows a gradual increase from approximately minus 12 in 1985 to minus 7 in 2020.

* A smooth magenta line represents the S A M index, oscillating frequently between minus 1.5 and 1.8.

Vertical Shaded Regions:

* Light orange bars indicate Strong El Niño years: 1987-88 and 1991-92.

* Red bars indicate Very Strong El Niño years: 1997-98, 1998-99, and 2015-16.

* Light blue bars indicate Strong La Niña years: 1999-2000, 2007-08, and 2010-11.

Notable peaks in temperature and S A M occur around the 1997-98 Very Strong El Niño event.

Recent studies have also further explored the impact of climate variability on the AP. For instance, Clem et al. (Reference Clem, Fogt, Turner, Lintner, Marshall, Miller and Renwick2020) examined the influence of the SAM and found that positive phases of the SAM contribute to warming on the AP, particularly during JJA. This adds another layer of understanding to the observed seasonal trends, and Fig. S1 highlights this within the context of the WAP and EAP more clearly.

The WAP is also considered a highly climate-sensitive area, as no other part of the SH faces such a rapid warming trend, as shown in Fig. S2. The warming rate per year for the WAP is 0.19°C (P < 0.01), and the highest increase per year occurs during JJA, at 0.40°C. The results from the EAP show the opposite condition, with significant cooling (P < 0.01) for all of the seasons, with the most significant decrease shown during MAM (−0.51°C), followed by SON (−0.48°C) and JJA (−0.30°C). This cooling period coincides with the overall cooling trend observed over the AP from 1999 to 2000 to ca. 2014–2015, as well as with the global warming hiatus. Although a slightly different approach was taken by Carrasco et al. (Reference Carrasco, Bozkurt and Cordero2021) in their paper, the same cooling period was observed using only two AWS: Larsen Ice Shelf and Butler Island (P < 0.05). This is supported by the annual mean from the ERA5 and ERA-Interim analyses. These differences will be highlighted in more details later in this paper.

The seasonal trends for the WAP here have been warming since 1985. AMRC-AWS data indicated the influence of SAM, as seen in Fig. S2. During the ‘very strong’ El Niño years (1997–1998 and 2015–2016) and ‘strong’ El Niño years (1991–1992), combined with a positive SAM phase, there were a peak in the trends for both the WAP and EAP. Studies such as those by Stammerjohn et al. (Reference Stammerjohn, Martinson, Smith, Yuan and Rind2008) found that El Niño events are linked to decreased sea-ice extent and increased surface temperatures in the WAP.

The interaction between the ENSO and SAM can amplify the warming trends observed in the WAP. During periods when both the ENSO and SAM are positive, the combined effect can lead to more pronounced warming. Clem & Fogt (Reference Clem and Fogt2013) discussed how the concurrent positive phases of the ENSO and SAM lead to enhanced poleward heat transport and reduced sea ice, which intensify the warming in the WAP.

Further supporting these findings, Stammerjohn et al. (Reference Stammerjohn, Martinson, Smith, Yuan and Rind2008) examined the relationship between sea-ice retreat and ENSO/SAM phases, finding that the combined influence of these patterns leads to significant sea-ice decline and warming in the WAP. Many other earlier studies (Meredith & King Reference Meredith and King2005, Turner et al. Reference Turner, Colwell, Marshall, Lachlan-Cope, Carleton and Jones2005) have also shown that warming in the WAP is strongly linked to the frequency and intensity of El Niño events and related changes in atmospheric circulation patterns driven by ENSO and SAM dynamics.

The seasonal variability for the EAP depicted in Fig. S2 is consistent throughout all of the seasons, showing a decrease in temperature over the years recorded (P < 0.01). The temperature dip can be seen for all seasons during the negative phase in 1999–2000, while in 2010–2011 only the JJA and SON seasons show any cooling. Engel et al. (Reference Engel, Láska, Nývlt and Stachoň2018) also discussed these cooling conditions in the EAP, where these authors associated these conditions with the enhanced easterly flow across the Weddell Sea towards the AP, as well as there being possible contributions from cold southerly winds originating from the interior of Antarctica. In contrast to the conditions in the WAP, the temperature dip in the EAP can also be seen during DJF, MAM and SON for strong El Niño years.

Dickens et al. (Reference Dickens, Kuhn, Leng, Graham, Dowdeswell and Meredith2019) discussed in their paper that the cooler atmospheric temperatures that prevailed on the EAP developed during the late Holocene (~1500–500 years ago), which caused the contrasting temperature patterns, are associated with developing Antarctic dipole-like conditions. These conditions arise due to temperature and sea-ice cover anomalies in the Weddell Sea and ABS. During the late Holocene period, these dipole-like conditions were influenced by and can be explained by assessing the ENSO.

Figure 5 shows the decadal average for daily surface air temperatures. The latest decade shows an increase in the average annual cycle for air temperatures ranging between −1°C and −12°C, while the earliest record shows the minimum temperature for the AP was approximately −17°C.

Figure 5.

Decadal averages of the annual cycle of the near-surface temperature of all Antarctic Meteorological Research and Data Center automatic weather station (AMRC-AWS) for 1985–2020. Straight lines indicate complete decadal years and the dots represent incomplete years. AP = Antarctic Peninsula.

Figure 5. Long description

The graph plots Temperature (C) on the Y-axis from -24 to 4 against Day of the Year on the X-axis from 1 to 361.

* The data follows a U-shaped seasonal cycle with peaks at the beginning and end of the year and a trough around day 196.

* A pink solid line represents the 1991-2000 decade, showing a smooth curve peaking near 0 degrees Celsius and dipping to approximately -17 degrees Celsius.

* A green solid line represents the 2001-2010 decade, following a similar path but consistently lower than the pink line, dipping to approximately -19 degrees Celsius.

* Blue diamond dots represent the incomplete decade 1985-1990, showing scattered data points that generally align with the lower green curve.

* Red diamond dots represent the incomplete decade 2011-2018, showing data points that are generally higher than both solid lines, particularly during the winter trough where they remain above -14 degrees Celsius.

* The legend at the bottom identifies the four time periods: 1985-1990 (blue diamond), 1991-2000 (pink line), 2001-2010 (green line), and 2011-2018 (red diamond).

The decadal changes in the air temperature analysis show overall warming for the AP. However, the decadal average also shows that the AP encountered a significant cooling between 1985 and 2018, based on the fluctuations shown in Fig. 5. This may be due to the number of incomplete years during these periods, as 1985–1990 and 2011–2020 do not have sufficient data to represent an entire decade. However, the findings are supported by the second and third decadal changes, as these periods have complete data. This is another example of how an incomplete year dataset affects the analysis due to having insufficient years to compare decadal variability, but despite this we can nevertheless see a similar cooling trend between 1999 and 2014 from the stations’ data (Turner et al. Reference Turner, Lu, White, King, Phillips and Hosking2016, Carrasco et al. Reference Carrasco, Bozkurt and Cordero2021).

The decadal changes between the WAP and EAP are further analysed in Fig. 6. The WAP warmed between the two decades from 1985–1990 to 1991–2000 but did not demonstrate a significant increase in the following decade. There was a warming to 0–2°C within DJF (day 335–336 to 1–60), but then the temperature dropped to approximately −8°C for JJA (day 152–243). The years 2011–2018 only show small shifts from the data for 2001–2010.

Figure 6.

Decadal averages of the annual cycle of the near-surface temperature for the Antarctic Meteorological Research and Data Center automatic weather station (AMRC-AWS) in a. the West Antarctic Peninsula (WAP) and b. the East Antarctic Peninsula (EAP) for 1985–2020. Coloured lines indicate complete decadal years and the dots represent incomplete years.

Figure 6. Long description

Two side-by-side line graphs labeled a and b. Both share a horizontal x-axis representing Day of the Year from 1 to 361 and a vertical y-axis representing Temperature in degrees Celsius. A legend at the bottom identifies four periods: 1985 to 1990 (blue diamonds), 1991 to 2000 (magenta line), 2001 to 2010 (green line), and 2011 to 2018 (red diamonds).

Panel a, titled W A P: The y-axis ranges from minus 20 to 4. All data follows a U-shaped seasonal curve with peaks at the start and end of the year and a trough around day 211. The 1991 to 2000 and 2001 to 2010 continuous lines are closely aligned, peaking near 0 degrees Celsius and dipping to minus 9. The 1985 to 1990 blue dots show significantly lower temperatures, dipping to minus 18. The 2011 to 2018 red dots are generally warmer than the 1980s data but cooler than the continuous decadal lines.

Panel b, titled E A P: The y-axis ranges from minus 34 to 6. The seasonal cycle is more pronounced than in W A P. The 1991 to 2000 magenta line is the warmest, peaking near 0 and dipping to minus 22. The 2001 to 2010 green line follows a similar path but is consistently 4 to 5 degrees cooler. The 1985 to 1990 blue dots and 2011 to 2018 red dots are scattered around the green line, with the red dots reaching the lowest recorded temperatures near minus 30 around day 196.

The EAP, on the other hand, shows a consistent cooling pattern from 1985–2000 to 2010–2018. From Fig. 6b, we can see that the EAP experienced a decrease in temperature per decade of between 2°C and 4°C.

Climatology of the AP based on AMPS

The simulations from AMPS were used to examine how orography influences the climate of the AP. AMPS Domain 6 data initialized at 00h00 UTC were used in this study, and 3 h forecast outputs were analysed. The ERA5 data are compared with the simulation results to determine their differences in representing the meteorological parameters affected by the Antarctandes. The AMPS output is used due to its higher resolution than the ERA5 data and its ability to capture the orographic influence on climatology. It effectively demonstrates the climatic differences between the WAP and EAP. Bromwich et al. (Reference Bromwich, Nicolas, Monaghan, Lazzara, Keller, Weidner and Wilson2012) also highlighted that AMPS serves as a benchmark for assessing the reliability of polar weather research and forecasting in Antarctic studies.

Uniform spatial resolutions of Domain 6 were produced by interpolating all of the data sources used, as they had different horizontal resolutions throughout the operational years of 2006/2008 (6.6 km), 2008–2011 (5 km) and 2013 onwards (3 km).

A distinct temperature gradient is identifiable within a few hundred kilometres of Palmer Land, and it extends almost towards Graham Land (68°S). The disparity is most noticeable during MAM, reflecting warmer temperatures on ABS and colder temperatures from the Weddell Sea moving towards the Larsen and Ronne ice shelves. The higher temperature on the WAP and slight temperature increase over the EAP were also noticeable in the models studied by Bozkurt et al. (Reference Bozkurt, Bromwich, Carrasco and Rondanelli2021), which can be explained by the low adiabatic conditions warming the leeward side due to the weak condition of the south-eastward advection of the warm mid-latitudes over the orography. This produces dipole-like conditions in the AP, which we can see from the trends as discussed previously (Fig. S2) and from the spatial distribution shown in Fig. 7.

Figure 7.

Annual means of temperature (°C) from a. Antarctic Mesoscale Prediction System (AMPS) Domain 6 (3 km resolution; March 2006–December 2016) initialized at 00h00 UTC every 3 h and b. European Centre for Medium-Range Weather Forecasts (ERA5; 31 km resolution; 1979–2019).

Figure 7. Long description

A multi-panel figure showing two geographical maps of the Antarctic Peninsula with a shared color scale. Both maps use a latitude grid from 60 degrees South to 76 degrees South and a longitude grid from 80 degrees West to 60 degrees West.

* Panel a, titled A M P S, shows the annual mean temperature from the Antarctic Mesoscale Prediction System. The northernmost regions and coastal edges are colored in red and orange, indicating temperatures between 0 and minus 5 degrees Celsius. Moving South along the peninsula, colors transition through yellow and light blue. The interior and southern regions show deep blue, indicating temperatures dropping to minus 24 degrees Celsius.

* Panel b, titled E R A 5, shows the annual mean temperature from the European Centre for Medium-Range Weather Forecasts. The spatial distribution is similar to panel a but appears smoother due to the 31 kilometer resolution. The temperature gradient follows the same North-to-South cooling trend, with the warmest temperatures at the northern tip and the coldest temperatures in the South and high-elevation interior.

* A vertical color scale on the far right is labeled Temperature in degrees Celsius. It ranges from 4 degrees Celsius at the top (dark red) to minus 24 degrees Celsius at the bottom (dark blue), with 1 degree Celsius increments.

The average spatial output in Fig. 7 shows differences in temperature experienced by the WAP and EAP, with the WAP’s temperature ranging between −9°C and −1°C, while the EAP’s temperature ranges from −18°C to −5°C as we move gradually to the north. Other than the blocking effect of the orography, these differences could be explained by the variation in the SAM (Fig. S2) and ENSO that affects temperature distributions (Fig. 7) and wind patterns across the AP (Ding & Steig Reference Ding and Steig2013, Sato et al. Reference Sato, Inoue, Simmonds and Rudeva2021, Zhang & Duan Reference Zhang and Duan2023).

For instance, while experiencing the orographic effect and under the influence of the climate drivers, warm air can be advected towards the WAP from the mid-latitudes, while cold air is funnelled into the EAP. This creates a seesaw effect in temperatures, whereby the WAP experiences warming trends while the EAP remains colder. Studies by Vaughan et al. (Reference Vaughan, Marshall, Connolley, Parkinson, Mulvaney and Hodgson2003) have documented a pronounced warming trend in the WAP, particularly since the 1970s, and this continues to increase (Mar Flexas et al. Reference Mar Flexas, Thompson, Schodlok, Zhang and Speer2022), attributed to changes in atmospheric circulation and the influence of warmer ocean currents. In contrast, the EAP has shown a cooling trend, further accentuating the dipole effect.

The results also corroborate other studies, including that by Turner et al. (Reference Turner, Marshall, Clem, Colwell, Phillips and Lu2019), showing that the AP experienced a warm DJF season averaging approximately −5°C and a cold JJA season with a mean of −20°C. However, a further distinction can be seen in Fig. 7, which shows how orography plays a role in determining the weather conditions of the maritime environment of the west coast compared to the colder east coast continental area.

This is prominent during the JJA period, as shown in Fig. S3e,f. Leeson et al. (2017) also identified that the EAP was affected by atmospheric circulation, especially in the Weddell Sea, resulting in a warming trend before 2014, and AMPS can detect the point of change in 2010 with a significance threshold of ± 1.96 (P = 0.05), which then later cooled down.

ERA5 is able to represent the orographic effects on the AP, revealing the contrasting climatic conditions experienced by the WAP and EAP. However, due to its coarser horizontal resolution, ERA5 does not fully capture orographic influences when compared with AMPS. In contrast, AMPS successfully identifies the colder region over Palmer Land, which is not resolved in ERA5. Nonetheless, both AMPS and ERA5 show that the NAP experiences warmer temperatures than the south and overall warming in the WAP and in the NEAP tip, which is similar to findings from several previously conducted studies (Leeson et al. 2017, Carrasco et al. Reference Carrasco, Bozkurt and Cordero2021).

That research also highlighted that the READER research stations in the NAP tip recorded a significant warming trend ranging between +0.26°C to +0.51°C (10 yr)⁻¹. Based on these findings, along with the results from AMRC-AWS, we suggest defining the EAP by exclusively including the area from Larsen Ice Shelf and below ~66–78°S, as the three AWSs evidently show a decrease in temperature. This would explain the three distinguishable climates of the AP.

Figure 8 displays the trends obtained using the AMPS data, ranging only from 2006 to 2016. Figure 8 shows that the AP experienced a cooling temperature for this period, coinciding with the well-documented overall cooling trend starting from the middle of 1998 to early 1999 as the turning point from the warming period until ca. 2014. Comparing AMRC-AWS and the selected READER stations database, the decrease in temperature is more aligned with the overall AP temperature trend from the READER database, and the SAM indices are also aligned with these trends, as shown in Figs 8 & S4.

Figure 8.

Antarctic Mesoscale Prediction System (AMPS) interannual temperature variability (°C) across 2006–2016. Vertical shaded lines indicate the varying intensities of the El Niño and La Niña years. The smooth magenta line represents the Southern Annular Mode (SAM) index. AP = Antarctic Peninsula.

Figure 8. Long description

The x-axis represents Year from 2006 to 2016. The primary y-axis on the left shows Temperature in degrees Celsius ranging from minus 17 to minus 5. The secondary y-axis on the right shows S A M index values from minus 4 to 4.

* A solid blue line representing A M P S temperature starts at minus 8.5 in 2006 and shows a fluctuating downward trend, reaching its lowest point near minus 15 in 2015 before rising slightly in 2016.

* A dashed black linear trend line overlays the blue line, showing a steady decrease from minus 10 to minus 15 over the decade.

* A smooth magenta line representing the S A M index oscillates across the center of the graph, generally staying between minus 1 and 2.

* Two vertical light blue shaded regions indicate Strong La Niña years, occurring roughly between 2007 to 2008 and 2010 to 2011.

* One vertical light red shaded region on the far right indicates a Very Strong El Niño year during 2015 to 2016.

The SAM’s positive phase is associated with stronger south-westerly winds that commonly shifted poleward and intensify over the Bellingshausen Sea, and this caused a uniform effect across the AP, leading to cooling over the AP, specifically on its eastern side. With low-pressure systems moving south-eastward across the AP, the westerly winds prevail and predominantly become stronger near the northern tip (Oliva et al. Reference Oliva, Navarro, Hrbáček, Hernández, Nývlt and Pereira2017). This resulted in local warming due to adiabatic heating as air descended, which caused the EAP to become drier and warmer. However, this would be counterbalanced by the stronger influence of cold air masses flowing northward from the Weddell Sea. These cold, southerly winds bring cold and dry conditions along the eastern coast of the AP, causing a thermal asymmetry between the two sides, whereby the WAP, influenced by moist westerly winds, remains warmer and wetter, while the EAP, exposed to the cold southerly winds, remains colder and drier due to the intrusion of cold southerly air masses.

Figure S4 breaks down the overall trends into seasonal groupings and shows that the SAM indices more visibly impact AMPS overall interannual variability during JJA and SON. Marshall et al. (Reference Marshall, Thompson and van den Broeke2017) explored the relationship between the SAM and temperature variability, finding that the positive phase of the SAM during the early twenty-first century contributed to the observed cooling on the AP. Additionally, the ozone hole recovery has influenced atmospheric circulation patterns, impacting temperature trends (Solomon et al. Reference Solomon, Ivy, Kinnison, Mills, Neely and Schmidt2016, Turner et al. Reference Turner, Lu, White, King, Phillips and Hosking2016). Solomon et al. (Reference Solomon, Ivy, Kinnison, Mills, Neely and Schmidt2016) found that the healing ozone hole affected the jet stream, contributing to the cooling trend observed during the early 2000s. Turner et al. (Reference Turner, Lu, White, King, Phillips and Hosking2016) confirmed this cooling trend, attributing it to natural variability rather than the reversal of the long-term warming trend.

Other studies have corroborated these findings. Oliva et al. (Reference Oliva, Navarro, Hrbáček, Hernández, Nývlt and Pereira2017) noted a cooling trend on the AP between 1999 and 2014, particularly in the NAP. They attributed this to changes in atmospheric circulation and sea- ice extent. Carrasco (Reference Carrasco2013) also reported a cooling trend from 1999 to 2010, with significant variability in different parts of the AP, while Clem et al. (Reference Clem, Renwick, McGregor and Fogt2016) analysed the impacts of the ENSO on Antarctic climate, showing how La Niña conditions during the early twenty-first century could have contributed to the cooling trend on the AP. However, as previously noted, the conditions stated here could have influenced the EAP as a location, as can be seen in Figs 7 & S3.

Figure 9.

Spatial correlation between the European Centre for Medium-Range Weather Forecasts (ERA5) global sea-surface temperature data and observational data for the West Antarctic Peninsula (WAP) and East Antarctic Peninsula (EAP) temperature (°C) across 1980–2020.

Figure 9. Long description

A two-panel global map display with a shared vertical color bar on the right.

Top Panel (W A P):

This map shows the correlation for the West Antarctic Peninsula. Positive correlations (red-orange) are dominant across the Northern Hemisphere, particularly in the North Atlantic and North Pacific. Stronger positive correlations are also visible in the tropical Indian Ocean and parts of the Southern Ocean near Australia. Negative correlations (blue) are concentrated in the equatorial Pacific and the Southern Ocean south of 50 degrees South.

Bottom Panel (E A P):

This map shows the correlation for the East Antarctic Peninsula. It displays a starkly different pattern with widespread negative correlations (blue) across most of the Northern Hemisphere and the tropical Pacific. A prominent band of strong positive correlation (dark red) is located in the Southern Ocean between 50 degrees South and 70 degrees South, encircling Antarctica. This band is marked with black stippling and contour lines labeled 0.50.

Axes and Scale:

Both maps use a latitude scale from 70 degrees North to 70 degrees South and a longitude scale from 180 degrees West to 180 degrees East. The color bar on the right indicates the Correlation Coefficient, ranging from negative 0.9 (dark blue) through 0.0 (white) to positive 0.9 (dark red).

The ENSO and SAM’s roles in the AP

The correlations between the ENSO and SAM indices and the annual, regional and seasonal AMRC-AWS surface temperature data are listed in Tables S6 & S7. The maximum temperatures recorded for Uranus Glacier correspond with the very strong El Niño events for 1997–1998 (Table S3) and show a weak but positive correlation. Minimum temperatures recorded in 1989 and 2006 at Butler Island, Racer Rock and Limbert AWSs occurred during the strong and weak events of La Niña. The peaks and troughs for some of the temperature variability exhibited a similar pattern for the AP from 1999 to 2002, which then continued from 2006 to 2011 despite showing only weak correlations with one another. The AP’s DJF seasonal variability also shows the same pattern of having a weak correlation, and 1997–1998’s very strong ENSO event overlapped with the SAM positive shift. This may have had an influence on the WAP, showing a positive correlation of r = 0.520. The greatest dip, however, was seen between 2010 and 2011, corresponding with the strong episode of La Niña for DJF, MAM and JJA. This supports the 2010s finding that the cold temperature had countered the melting of the ice shelves near the Amundsen Sea.

The seasonal temperature variability for the WAP and EAP is shown in Fig. S2, with them both showing different trends from one another. We can see how both locations experience different types of temperature variability, as explained previously, which was probably due to the orographic barrier of the AP (a mountain chain between 1500 and 2000 m high, 250 km wide and 1250 km long; Campo et al. Reference Campo, Wellner, Domack, Lavoie and Yoo2017) having a blocking effect (Marshall et al. Reference Marshall, Orr, van Lipzig and King2005). Figure S2 highlights the atmospheric circulation impact on the WAP and EAP’s seasonality. Figure S3 shows that in 1997–1998 the region experienced among the strongest episodes of the ENSO and SAM in terms of WAP temperature variability, but this is noticeably absent during JJA. The EAP, by contrast, exhibited patterns that more closely follow La Niña’s weak to strong events than El Niño events, but it still showed high temperatures in 1991–1992 (strong episode) and 1997–1998.

While the analysis in Table S6 shows a lack of significance level for the correlation between the SAM and the temperature in the WAP, Table S7 demonstrates that there are moderate correlations between the SAM and the temperature for the WAP (P < 0.05) and the AP (P < 0.10) in MAM.

To understand further the influence of the ENSO on the AP’s climate, we explore the spatial correlation between the ERA5 sea-surface temperature (SST) data and both AMRC-AWS and READER surface air temperature data from 1980 to 2020, as depicted in Figs 9 & 10. The two sides of the AP can be seen to respond differently to the tropical SST during different seasons (Fig. S5a,b). For the WAP, positive air temperatures are associated with La Niña events in SON, whereas no such relationship exists for the EAP. La Niña events in DJF corresponded better with the EAP’s low temperatures.

Figure 10.

Spatial correlation of the European Centre for Medium-Range Weather Forecasts (ERA5) global sea surface temperature anomaly and seasonal observational data for the West Antarctic Peninsula (WAP) and East Antarctic Peninsula (EAP) temperature (°C) across 1980–2020.

Figure 10. Long description

A two-panel geographical map displays correlation coefficients. A vertical color bar on the right scales from negative 0.9 in dark blue to positive 0.9 in dark red, with white representing zero.

Both panels show the Antarctic Peninsula coastline with a latitude grid from 62 degrees South to 76 degrees South and a longitude grid from 80 degrees West to 50 degrees West.

Left Panel (W A P): The map shows predominantly light orange and red hues across the Southern Ocean surrounding the peninsula, indicating a weak to moderate positive correlation. The highest positive correlations are concentrated in the Northeast and Northwest corners of the mapped area.

Right Panel (E A P): This map shows a more polarized distribution. A strong positive correlation (dark red) is visible in the Northwest corner, marked with a dashed contour line labeled 0.50. Conversely, a significant area of negative correlation (light to medium blue) is concentrated along the Western coast of the peninsula and extending Southward toward 72 degrees South. The Eastern side of the peninsula shows scattered light orange patches of weak positive correlation.

Figure S5 shows a positive correlation between WAP temperature and SST during MAM in the western Pacific. This condition persisted in JJA until we the La Niña pattern becomes visible in SON, with the moderate negative correlation concentrated in the eastern Pacific. The opposite can be seen for the EAP during SON, where little correlation is evident; however, the La Niña pattern is clearer in DJF. A negative correlation between the ERA5 SST data and the observational data for EAP can clearly be seen in the western Pacific, and a weak positive correlation can be seen in the eastern Pacific.

La Niña events typically lead to cooler-than-average SSTs in the central and eastern tropical Pacific Ocean (Sassi et al. Reference Sassi, Kinnison, Boville, Garcia and Roble2004). These cooler conditions can influence large-scale atmospheric circulation patterns, resulting in global anomalous weather conditions across various regions. In the case of the WAP, the influence of La Niña manifests in the form of positive air temperature anomalies during SON. This could be due to altered atmospheric circulation patterns associated with La Niña, such as changes in wind patterns and atmospheric pressure systems, which indirectly affect temperature distributions in the WAP region. Studies have shown that La Niña can alter the position and strength of the polar jet stream, leading to changes in atmospheric circulation that affect temperature distributions in high-latitude regions such as the AP (Stammerjohn et al. Reference Stammerjohn, Martinson, Smith, Yuan and Rind2008).

During La Niña events, the strengthened polar jet stream can shift atmospheric pressure systems and wind patterns. These shifts often result in the advection of warmer air masses from lower latitudes into the WAP region. The altered circulation patterns are characterized by increased northerly winds, which bring relatively warmer air from the South Atlantic towards the AP (Fogt & Bromwich Reference Fogt and Bromwich2007). This can lead to the observed positive temperature anomalies during SON.

By contrast, our analysis did not reveal a similar relationship between ENSO events and air temperature anomalies in the EAP region during SON. This suggests that the influence of the ENSO on temperature dynamics in the EAP during this season is neither as pronounced nor as consistent as are observed in the WAP. The lack of a significant relationship between La Niña events and air temperature anomalies in the EAP during SON may be attributed to various factors, including the complex orography and oceanographic features of the region, which can modulate the impact of large-scale climate phenomena such as the ENSO.

Furthermore, our analysis highlighted a notably stronger influence of ENSO events on air temperatures in the WAP during JJA, as evidenced by the higher spatial correlation coefficient of 0.7 depicted in Fig. S6e. This heightened influence during the JJA season could be attributed to the amplification of atmospheric and oceanic teleconnections associated with ENSO, leading to more pronounced temperature anomalies in the WAP region.

In contrast, while the EAP did not exhibit a direct relationship between La Niña events and air temperature anomalies during SON, our analysis revealed interactions in surrounding areas, particularly around the Drake Passage and the Weddell Sea, as illustrated in Fig. S6b. These interactions suggest that the influence of the ENSO on temperature variability in the EAP region during SON may be mediated by local or regional factors, such as ocean currents and sea-ice extent, which can modulate the propagation of ENSO-related signals and their impact on air temperatures.

Surface pressure, MSLP and resultant wind influencing the WAP and EAP

The surface pressure varies from 960.56 to 994.28 hPa at the AP over the period 1985–2020, and the mean values for the maximum and minimum pressure recorded at the AP are 976.08 and 961.74 hPa, respectively. The regional long-term daily surface pressure values for all seven AMRC-AWSs are segregated into the WAP and EAP in Fig. 11a,b.

Figure 11.

Daily maximum (red), mean (black) and minimum (blue) surface pressure (hPa) for 1985–2020 at the Antarctic Meteorological Research and Data Center automatic weather stations (AMRC-AWS) in a. the West Antarctic Peninsula (WAP) and b. the East Antarctic Peninsula (EAP).

Figure 11. Long description

The figure consists of two side-by-side line graphs. Both share a horizontal x-axis labeled Day of the Year ranging from 0 to 350 and a vertical y-axis labeled Pressure h P a ranging from 930 to 1000.

Panel a, titled W A P on the left, shows three jagged lines with high daily variability. The top red line representing maximum pressure fluctuates between 960 and 975 h P a. The middle black line representing mean pressure fluctuates between 955 and 970 h P a. The bottom blue line representing minimum pressure fluctuates between 935 and 965 h P a. The overall trend is relatively flat across the year.

Panel b, titled E A P on the right, shows the same three metrics but at higher pressure values. The red line fluctuates between 980 and 995 h P a. The black line fluctuates between 975 and 990 h P a. The blue line fluctuates between 965 and 985 h P a. Unlike the left panel, these lines show a slight seasonal arc, peaking around day 150 to 200 before dipping toward the end of the year.

The two regions of the AP have two different pressure features. The EAP’s AMRC-AWSs recorded a higher daily surface pressure than those of the WAP. In his study, Harangozo (Reference Harangozo2000) identified and explained this as a dipole structure for sea-level pressure. This is due to the orographic feature differences on either side of the AP, similarly to the temperature distribution (Figs 7 & S3). The annual surface pressure for the WAP is 962.15 hPa, and the EAP’s annual MSLP is 982.98 hPa. As the WAP is closer to the Bellingshausen Sea and the Amundsen Sea, the Amundsen Sea Low (ASL) - a persistent low-pressure system - influences the climate of the AP, especially in the WAP, by lowering the MLSP (Turton 2017). The daily pressures shown in Fig. 11a,b show that the different climates between the WAP and EAP are the result of the orographic features of the AP. The strengthened northerly flow also causes pressure variability difference due to the increasing South Atlantic pressure and the strong circumpolar westerlies caused by the positive SAM to the east (Xu et al. Reference Xu, Yu, Liang, Vhima, Bozkurt, Hu and Yang2021).

The SAM has a stronger influence over the EAP than the WAP because of lee-side compressional warming during periods dominated by strong westerly (upwind) flow (Russell & McGregor Reference Russell and McGregor2009).

Uranus Glacier AWS, located in the WAP, experienced the lowest sea-level pressure ever recorded for the AP (853.61 hPa) in May 1989, followed by the second lowest ever recorded at Bonaparte Point AWS (885.25 hPa) in May 2012, also in the WAP. The highest pressure ever recorded was at Larsen Ice Shelf AWS in the EAP (1034.80 hPa) in September 2000.

Similarly to Costanza et al. (2016), this study also found that the AP is characterized by higher surface pressure between March and June (days 92–182) and lower surface pressure from October until mid-December (days 274–345). The deepening of the circumpolar trough and the ASL can explain the shifts observed during SON and DJF.

The MSLP data obtained from the AMPS and ERA5 are shown in Fig. 12, highlighting a significant difference in pressure (hPa) across the AP. The area with predominantly low pressure experienced by the AP in the AMPS dataset is clearer than in the ERA5 dataset, with values < 988 hPa (Fig. 12a). The low pressures are concentrated in the EAP and near the main part of the mountainous region. This is mainly influenced by the orography effect.

Figure 12.

Annual means of mean sea-level pressure (MSLP; hPa), wind speed (m s⁻¹) and direction from a. Antarctic Mesoscale Prediction System (AMPS) Domain 6 (3 km resolution; March 2006–December 2016) and b. European Centre for Medium-Range Weather Forecasts (ERA5; 31 km resolution; 1979–2019).

Figure 12. Long description

Two panels, labeled a and b, show M S L P in h P a and wind vectors. Both maps use a color scale from dark blue (982 h P a) to dark red (995 h P a) and cover a grid from 60 degrees South to 76 degrees South and 80 degrees West to 60 degrees West.

* Panel a, A M P S: Shows a deep low-pressure system (dark blue, 982 to 984 h P a) concentrated in the West and Southwest. Wind vectors show a strong cyclonic circulation around this low. Higher pressure (orange, 991 to 992 h P a) is visible in the Northeast corner. The central Antarctic Peninsula acts as a sharp boundary between the low pressure to the West and moderate pressure to the East.

* Panel b, E R A 5: Displays a more zonal pressure gradient. High pressure (red, 994 to 995 h P a) is concentrated in the North and South, while a band of lower pressure (light blue, 986 to 988 h P a) stretches across the center-West. Wind vectors are predominantly West-to-East across the top and bottom of the frame, with more complex, localized flow around the southern tip of the peninsula.

* Legend: A vertical color bar on the right indicates M S L P values in increments of 1 h P a, ranging from 982 (dark blue) at the bottom to 995 (dark red) at the top.

The spatial distributions for the AMPS and ERA5 data highlight the seasonal variability of the pressure patterns. The MSLP shown in Fig. S7 demonstrates the conditions brought over by the surrounding low-pressure system, especially from the ASL. Antarctica is surrounded by a region of low pressure called the ‘circumpolar trough’. However, the main continent itself is dominated by high pressure, which enables the formation of katabatic winds on top of the orography effect. The AMPS data performed better at demonstrating that the EAP experienced lower pressure than the WAP. The warming of the surface temperature, as in Figs 7 & S3, further impacted the MLSP in the region.

During the DJF and JJA seasons, the low pressure persisted in the WAP, especially within the Bellingshausen Sea, as seen in Fig. S7a(i),e(ii), where the recorded MSLP values were well below 985 hPa, similarly to the conditions under which low pressure is commonly found in the WAP and shifting towards the Ross Sea (Raphael et al. Reference Raphael, Marshall, Turner, Fogt, Schneider and Dixon2016). The recurring pressure system, along with the strong surface wind, impacted and drove the sustained west-north-westerly airflow across the EAP (Fig. 12), delaying the autumnal ice-edge advance and hastening the SON ice retreat (Fogt et al. Reference Fogt, Wovrosh, Langen and Simmonds2012).

Fogt et al. (2012) also reported that the local climatological pressure minimum migrates east during JJA as its amplitude is reduced, as can be seen in Fig. S7e,f. The interannual variability in the pressure field, which occurs within the proximity of the ABS Low (ABSL), has been coined the ‘pole of variability’, which may also contribute to the high and low pressure values. This situation may also be due to the Antarctic orography, which can generate divergence flow around the AP, and this area is also highly influenced by the SAM and ENSO. These large-scale atmospheric climate phenomena are well documented as being able to increase the interannual pressure variability in the region.

The minimum pressure is usually present at the tip of the NAP and EAP during the DJF season, while it shifts to the south and west during the JJA season. These conditions are evident in the AMPS data (Fig. S7a(ii),e(ii)) but are less prominent in the ERA5 data. This MLSP seasonal cycle is well attested in the Antarctic circumpolar trough and is associated with the Semi-Annual Oscillation (SAO). Bozkurt et al. (Reference Bozkurt, Bromwich, Carrasco and Rondanelli2021) found this cycle to follow a dipole-like pattern characterized by a decrease in MSLP and a weakening of the ASL, which further weakened the westerlies, especially in the JJA season (Fig. S7e(i),f).

During both the DJF and JJA seasons, a transition between warm and relatively warm cool air masses can be seen in the EAP and NAP but not in the WAP, with its average temperature being below −3°C (Figs 7 & S3), as the DJF and JJA seasons are influenced by the barrier winds arriving from the south. The southerly was shown to move away more strongly in the AMPS data, showing almost cyclonic activity near the Weddell Sea (Fig. S7e(i),g). The westerly is also represented better in the AMPS data: it can be seen to have travelled over to the EAP because the wind is shown more clearly in the data. The influence of orography was not apparent in the ERA5 data. The effects of orography on the winds and their passage across the Antarctandes could not be observed in the ERA5 output.

The warm conditions that persisted during the DJF season could be caused by föhn winds, which are driven by the positive trend of westerly winds (Turton et al. Reference Turton, Kirchgaessner, Ross and King2017).

The downslope wind that prevailed on the lee side of the AP, which is perpendicular to the mountains, could cause an increase in temperature, as seen in both the AMPS and ERA5 maps (Fig. 12). This happens because the narrow mountain range acts as a barrier to the SH westerlies, creating large differences in temperature and precipitation across the area (van Wessem et al. Reference van Wessem, Reijmer, van de Berg, van den Broeke, Cook, van Ulft and van Miejgaard2015). The comparison made for both forecasts from the AMPS data with reanalyses from the ERA5 data supported the later, simplified representation of the wind field, even though progress had been made on improving the modelling of this for the Antarctic (Nygård et al. Reference Nygard, Vihma, Birnbaum, Hartmann, King and Lachlan-Cope2016). The coarse resolution from the reanalyses did not fully represent the complex orography of the AP, as shown in Fig. 12.

The resultant winds blowing northward over the EAP (Andreas et al. Reference Andreas, Claffey and Makshtas2000) were also more prominent in the AMPS data. The strong south-easterly wind along the east coast of the AP picked up speed, as more wind caused a flow over on the other side of the Antarctandes that then diverted away (moving down the slope and to different direction (western region ridge) (Kwon et al. Reference Kwon, Kim, Kim and Kim2021), particularly at Loubet Coast and Fallières Coast (~67.5°S, ~67.0°W) and English Coast (~73.5°S, 72.0°W) (Fig. S7c,e(i),g).

For the NAP, however, the effect of föhn winds over the EAP prevails when the westerly consists of dry, warm air as it descends upon the leeward side (Fig. S7a(i),c,e(i),g), passing over the orographic barrier (Wiesenekker et al. Reference Wiesenekker, Munneke, van den Broeke and Paul Smeets2018). The AMPS simulations demonstrate how the westerlies over the AP and the orographic effect impact the wind direction.

The highest wind speed recorded for each station (Table S3) was during the JJA season, except for Larsen Ice Shelf and Bonaparte AWSs in January 1992 and January 2006, respectively. The highest recorded wind speeds were recorded in the EAP, except for the high wind speeds recorded at Racer Rock AWS, which is closer to the NAP. This is due to this AWS’s position on the lee side of the Antarctandes, where the wind could pick up speed during its descent.

In terms of seasonal change, the trend for the lowest wind speeds in Antarctica is less than 5 km s⁻¹ in the DJF season, a finding that is similar to that from previous research by Costanza et al. (2016) for the Ross Ice Shelf area.

The long-term daily wind speeds of all seven AWSs and the mean wind speed for the AP range from 3 to 8 m s⁻¹. The red lines in Fig. 13 indicate the maximum wind speed throughout the year. The mean value for the maximum wind speed is 12.65 m s⁻¹, while the mean minimum wind speed is 1.66 m s⁻¹. Racer Rock AWS recorded the highest wind speed in the AP at 55.80 m s⁻¹ (Table S2). The high frequency of wind recordings and wind speed record during the JJA months can be associated with strong low-level jet occurrences nearby (Heinemann & Zentek Reference Heinemann and Zentek2021). In their study, Ambrožová et al. (Reference Ambrožova, Laska and Kavan2019) also linked a west-north-westerly wind with the highest mean air temperature, whereas the coldest temperatures were associated with a south-south-westerly wind.

Figure 13.

Daily maximum (red), mean (black) and minimum (blue) wind speeds (m s⁻¹) for 1985–2020 in the Antarctic Meteorological Research and Data Center automatic weather station (AMRC-AWS) data for a. the West Antarctic Peninsula (WAP) and b. the East Antarctic Peninsula (EAP).

Figure 13. Long description

A two-panel line graph titled Long Term 1985-2020 Daily A M R C dash A W S Observational Wind Speed over W A P and E A P.

Panel a, labeled W A P for West Antarctic Peninsula, has an x-axis for Day of the Year from 0 to 350 and a y-axis for Wind Speed in meters per second m s super minus 1 from 0 to 24. Three jagged lines represent the data. The top red line shows maximum wind speeds fluctuating between 8 and 15 m s super minus 1. The middle black line shows mean wind speeds between 4 and 7 m s super minus 1. The bottom blue line shows minimum wind speeds between 1 and 4 m s super minus 1. The trends are relatively flat across the year.

Panel b, labeled E A P for East Antarctic Peninsula, uses the same axes. The data shows higher variability and a seasonal peak. The top red line for maximum wind speed starts around 12, peaks near 22 around day 240, and then declines. The middle black line for mean wind speed follows a similar curve, peaking near 8 m s super minus 1. The bottom blue line for minimum wind speed remains low, mostly between 1 and 5 m s super minus 1, with occasional spikes.

Figure 14.

Interdecadal Pacific Oscillation (IPO) index from 1940 to 2016 with Antarctic Meteorological Research and Data Center automatic weather station (AMRC-AWS) temperature anomalies and selected Reference Antarctic Data for Environmental Research (READER) station temperatures. Background colour indicates the IPO index: positive (orange) and negative (blue). AP = Antarctic Peninsula.

Figure 14. Long description

The horizontal x-axis represents Year from 1940 to 2020. The left y-axis shows the I P O Tripole Index from minus 0.6 to 0.6. The right y-axis shows Temperature Anomaly in degrees Celsius from minus 0.9 to 0.9.

Background shading indicates I P O phases.

- 1940 to 1945: Positive phase (light orange) labeled I P O plus.

- 1945 to 1977: Negative phase (light blue) labeled I P O minus.

- 1977 to 1999: Positive phase (light orange) labeled I P O plus.

- 1999 to 2014: Negative phase (light blue) labeled I P O minus.

- 2014 to 2016: Positive phase (light orange) labeled I P O plus.

Three data lines are plotted.

1. I P O Tripole Index (blue line): Shows a long-period oscillation. It reaches a minimum near minus 0.5 around 1950 and 1973, and a maximum near 0.4 around 1993.

2. Selected R E A D E R Stations Temperature (black line): Shows high-frequency annual variability. It generally fluctuates between minus 0.3 and 0.2 on the right axis, with a slight upward trend over the 80-year period.

3. A M R C A W S Temperature (orange line): Begins around 1985. It shows extreme variability, with sharp peaks reaching 0.8 around 2014 and deep troughs falling to minus 0.5 around 1997 and 2018.

When analysed further, the long-term daily wind speeds for the WAP and EAP (Fig. 13) do not differ much or deviate much from the daily mean wind speed. However, the WAP’s trend (varying between 3.26 and 8.97 m s⁻¹) does show a slightly higher daily wind speed value when compared to the EAP (ranging between 3.25 and 7.52 m s⁻¹). Regarding extreme wind speeds, AWSs in the EAP experienced higher values than those from the WAP, at up to 21.23 m s⁻¹ compared to 15.16 m s⁻¹, respectively.

The impacts of the Interdecadal Pacific Oscillation on the WAP and EAP

AWS temperatures were also correlated against the Interdecadal Pacific Oscillation (IPO), as shown in Fig. 14. The results show a statistically significant moderate negative correlation between temperature and IPO (r = −0.547, P < 0.05). These findings suggest a potential influence of the IPO on regional temperature variability. The negative correlation implies that temperatures in the selected stations tended to increase as the IPO index shifted into its negative phase. Such a relationship suggests that variations in the IPO may strongly influence regional temperature patterns, probably through its impact on atmospheric circulation patterns and oceanic heat distribution.

The IPO index indicated a stronger relationship with the AMRC-AWS temperature anomaly data from 1985 until the last available index in 2015, demonstrating a significant and strong positive correlation (r = 0.618, P < 0.05).

Figure 15 shows the results from a composite analysis carried out to investigate further the influence of the IPO (Henley et al. Reference Henley, Gergis, Karoly, Power, Kennedy and Folland2015). Here, the composites were compiled according to the positive and negative years in reference to the method by Welhouse et al. (Reference Welhouse, Lazzara, Keller, Tripoli and Hitchman2016), based on ERA5 data ranging from 1977 to 2020. The MSLP anomaly data utilized here were linked to the circulation patterns, supporting the trends shown in Fig. 14. These data were also used to explain the atmospheric circulation patterns separately from the temperature anomaly data (Yang et al. Reference Yang, Ma, Wu, Klingaman and Zhang2019, Yin & Zhou Reference Yin and Zhou2020, Zhu et al. Reference Zhu, Liu, Song, Wang and Hu2021).

Figure 15.

Mean sea-level pressure anomaly composite analysis of Antarctica and the Antarctic Peninsula for a. & c. negative Interdecadal Pacific Oscillation (1999–2013) and b. & d. positive Interdecadal Pacific Oscillation (1977–1998 and 2014–2020) using the European Centre for Medium-Range Weather Forecasts (ERA5) dataset.

Figure 15. Long description

The figure consists of four panels labeled a through d and a shared color scale.

* Panel a (Top-Left): A circular polar map of Antarctica during the negative I P O phase. The continent is dominated by deep red shading, indicating positive pressure anomalies between 0.3 and 0.5. A small area of blue, indicating negative anomalies down to -0.4, is visible in the South Pacific sector around 120 degrees West.

* Panel b (Top-Right): A circular polar map during the positive I P O phase. The pattern is nearly the inverse of panel a, with deep blue negative anomalies covering most of the continent, particularly in East Antarctica, and light red positive anomalies in the South Pacific sector.

* Panel c (Bottom-Left): A regional map of the Antarctic Peninsula during the negative I P O phase. The Northwest region shows blue negative anomalies, while the Southeast and southern tip show red positive anomalies.

* Panel d (Bottom-Right): A regional map of the Antarctic Peninsula during the positive I P O phase. The area is characterized by light blue negative anomalies across most of the peninsula, with very faint red anomalies in the Northwest corner.

* Color Scale: A vertical bar on the right labeled Anomaly. It ranges from 0.5 at the top (dark red) to -0.5 at the bottom (dark blue), with 0 represented by white in the center. Increments are marked every 0.1 units.

The widespread high anomalies during the negative phase of the IPO (1999–2013; Fig. 15a,c) indicates a cooling trend after the year 1999, while the positive phase (1977–1998 and 2014–2020; Fig. 15b,d) explains the warming activity indicated by the negative pressure anomaly data. However, more details analysis indicates that, during the negative phase, the Ross Ice Shelf (163–210°E, 78–85°S) experienced a positive pressure anomaly, and Byrd Station recorded a strong warming trend before 1999 that later decreases by ± 1.5°C per decade, as discussed previously by Clem et al. (Reference Clem, Lintner, Broccoli and Miller2019). The negative anomaly during the positive phase, on the other hand, can be associated more with the SAM’s negative phase, where westerly winds shift northward away from Antarctica, and this could also be seen in an observed period from 1989 to 2018 of the warming pattern extending towards the Weddell Sea (Clem et al. Reference Clem, Fogt, Turner, Lintner, Marshall, Miller and Renwick2020). This was also demonstrated by Garreaud & Battisti (Reference Garreaud and Battisti1999), who showed that decadal ENSO-like variability in Pacific SSTs produce large-scale circulation anomalies across the SH, providing a mechanism for the pressure and temperature shifts observed in Antarctica.

AMRC-AWS and the selected READER stations’ observations show some correlation despite a few having a weak association with El Niño and La Niña events and SAM indices for their temperature values. Figure 7 further highlights the differences between the WAP and EAP in this matter. Here, the ERA5 data also show that there is an opposite pattern between the two regions. During the negative phase of the IPO, a low MSLP anomaly extended from the Pacific Ocean towards the WAP. However, the presence of the Transantarctic Mountains contributed to an increased anomaly over East Antarctica, the Antarctic interior, the Weddell Sea and the Atlantic Ocean. The opposite occurred during the positive phase, but to a lesser extent, whereby the AP was covered with low MSLP anomalies and the high MSLP anomalies further away from the South Pole.

Further distinctions were made by extracting different seasons (Fig. S8) to differentiate the conditions for these two phases. Low MSLP is consistent between these phases in the Weddell Sea and the Ross Sea, which extended towards the AP, especially during the MAM and SON seasons. Different pressure systems can be observed between the positive and negative phases during the MAM, JJA and SON seasons (Fig. S8c–h), which can influence the warming during the IPO phase. The SON warming may be influenced by seasonal changes in sea-ice cover and solar radiation (Stammerjohn et al. Reference Stammerjohn, Martinson, Smith, Yuan and Rind2008) for the AP in general. The negative phase of the IPO is most prominent during the JJA season (Fig. S8f), with a high-pressure system being concentrated in the interior part of the Antarctic compared to during the positive phase, and this coincided strongly with the extreme cold temperatures experienced in the particular region.

These events, which highlight different phases of the IPO, coincided with the SAM events recorded in the region that caused the warming and cooling trends at the South Pole, and this distribution also caused the pressure to decrease over the Bellingshausen Sea and the western Weddell Sea, as well as causing pressure to increase over the Antarctic interior (Clem et al. Reference Clem, Fogt, Turner, Lintner, Marshall, Miller and Renwick2020). The AP, which was the most rapidly warming region on the planet since the 1950s, started to cool down after 1999, which increased the levels of sea ice; however, Ross Ice Shelf experienced a reversal of this trend (from cooling to warming). This was associated with the dipole in temperature anomalies between the regions based on the ASL; however, this does not fully explain the equatorial Pacific teleconnections. A study by Zhu et al. (Reference Zhu, Liu, Song, Wang and Hu2021) identified that during the negative phase of the IPO with positive Atlantic multi-decadal variability between 1979 and 2019, there was broad and significant warming over the entirity of East Antarctica. This was caused by the strong anomalies in Antarctic sea ice and the IPO that induced different responses across Antarctica, including in surrounding areas such as northern and eastern Australia and the southern part of South America.