Recent autumn sea ice loss in the eastern Arctic enhanced by summer Asian-Pacific Oscillation

The Arctic has warmed at least twice as fast as the global average since the mid-20th century, with an accelerating decline of sea ice, particularly in autumn. While anthropogenic forcing explains the long-term downward trend of Arctic sea ice, internal climate variability also makes substantial contributions, yet its role remains incompletely understood. Prior work has linked atmospheric circulation patterns such as the Arctic Dipole, Arctic/North Atlantic Oscillations, and the Pacific-North American pattern to Arctic sea ice variability. This study investigates whether the Asian-Pacific Oscillation (APO)—a summertime mode characterized by a zonal seesaw of upper-tropospheric temperatures between Asia and the North Pacific—acts as a driver of the following autumn sea ice variability in the eastern Arctic. The central hypothesis is that positive summer APO phases precondition autumn atmospheric thermodynamics in the eastern Arctic, leading to enhanced sea ice loss via North Atlantic sea surface temperature (SST) persistence and teleconnections.

The impacts of Arctic sea ice loss on regional and remote climate are widely documented, including links to Eurasian cold winters, storminess, East Asian monsoon variability, and precipitation changes, although some studies contest the strength and robustness of these teleconnections. Anthropogenic forcing is identified as the primary cause of the secular decline in Arctic sea ice, but internal variability can markedly modulate the rate and spatial pattern of loss, with summertime Arctic circulation anomalies reportedly explaining up to 60% of September sea ice decline in some analyses. Several internal modes (Arctic Dipole, AO/NAO, PNA) have been implicated in sea ice variability. The APO has known influences on large-scale circulations across Asia–North Pacific and beyond, with interannual to decadal variability, but its role in Arctic sea ice variability had not been clearly established. North Atlantic and North Pacific SST anomalies, including those tied to Atlantic multidecadal variability and PNA-like patterns, have been shown to modulate Arctic sea ice via atmospheric teleconnections. These strands of literature motivate testing a mechanistic APO–SST–Arctic sea ice linkage.

Data and period: 1950–2019 analysis using linearly detrended fields. Sea ice concentration (SIC): HadISST1 (1.0°×1.0°). Sea surface temperature (SST): NOAA ERSST v5 (2.0°×2.0°). Atmosphere: NCEP/NCAR reanalysis (2.5°×2.5°). Model: CESM Large Ensemble (CESM-LE), 40-member fully coupled runs at 0.9°×1.25°, historical forcing 1920–2005. Indices and domains: Summer (JJA) upper-tropospheric eddy temperature (UTT; 500–200 hPa departure from zonal mean) over Asian-Pacific sector (25°–65°N, 75°E–120°W). SIC domain: eastern Arctic (70°–83°N, 30°E–180°). APO index defined as the time series of the leading MCA mode of UTT (highly correlated with UTT EOF1 PC1; r=0.98) for observations; in simulations, APO computed as UTT difference between Asia (30°–55°N, 80°–135°E) and North Pacific (30°–55°N, 170°E–125°W). SIC index (SICI) taken as MCA temporal expansion for autumn SIC (highly correlated with area-weighted SIC over the eastern Arctic; r=−0.98). Mid-latitude SST indices: NA_sst (30°–50°N, 75°–25°W) and NP_sst (30°–50°N, 145°E–150°W). Analyses: Maximum Covariance Analysis (MCA) between summer UTT and autumn (SON) SIC to extract coupled modes; Empirical Orthogonal Function (EOF) of UTT to identify APO; regression and correlation analyses; lead–lag correlations between APO and SST indices; squared wavelet coherence for APO–SIC time scales; partial correlations to separate NA_sst vs NP_sst contributions; wave activity flux (Takaya–Nakamura formulation) and streamfunction diagnostics for stationary Rossby wave propagation; lower-tropospheric (e.g., 925 hPa) humidity, downwelling longwave radiation (DLR), and surface air temperature (SAT) regressions. CESM-LE ensembles: Members selected based on the sign of correlation between summer APO and autumn SIC (1950–2005). BMME (Best Member Multi-ensemble): 3 members with significant negative correlation (consistent with observations). WMME (Worst Member Multi-ensemble): 4 members with significant positive correlation (contrary to observations). Ensemble means computed across selected members. Additional ocean subsurface temperature analysis examined vertical extent (to ~150 m) and seasonal persistence of APO-related warming.

- Summer APO and autumn eastern Arctic SIC co-variability: The leading MCA mode explains 62.9% of squared covariance, with summer positive APO (warm UTT over Asia, cool over North Pacific) linked to reduced autumn SIC over the eastern Arctic. The associated temporal series correlate at r=0.62 (p<0.01); the area-mean autumn SIC over the eastern Arctic correlates with summer APO at r=−0.59 (p<0.01). - APO index validation: The MCA-based APO index correlates with UTT EOF1 PC1 at r=0.98 (p<0.01); UTT EOF1 explains 31% of variance. - Time-scale coherence: APO–SIC show predominant in-phase coherence in 5–8, 8–12, and 16–24-year bands. - Autumn circulation and thermodynamics linked to positive summer APO: Barotropic anticyclonic anomalies over Barents–Kara–Laptev and lower-tropospheric cyclonic anomalies over the East Siberian Sea enhance warm/moist advection into the eastern Arctic; increases in 925 hPa humidity, DLR, and SAT accompany sea ice reduction. • Correlations over the eastern Arctic: low-level humidity vs DLR r=0.95; humidity vs SAT r=0.91 (both p<0.01). SICI vs humidity/DLR/SAT: r=0.66/0.70/0.74 (p<0.01). Summer APO vs autumn humidity/DLR/SAT: r=0.52/0.58/0.56 (p<0.01). - APO-induced SST anomalies and persistence: Positive summer APO intensifies subtropical highs, inducing mid-latitude SST warming in North Pacific and North Atlantic in summer that persists into autumn. • Lead–lag: Correlations indicate APO leads NA_sst and NP_sst more strongly than it lags them, supporting atmosphere-to-ocean forcing. - North Atlantic pathway dominance: Autumn NA_sst excites a wave train into the Barents–Kara–Laptev sector, yielding barotropic anticyclonic anomalies and weaker cyclonic anomalies over East Siberian Sea, favoring moisture intrusion and thermodynamic warming. • Simultaneous correlations with eastern Arctic means (SON): NA_sst vs humidity/DLR/SAT/SICI: r=0.49/0.50/0.45/0.47 (p<0.01). NP_sst vs humidity/DLR/SAT: r=0.39/0.39/0.33. • Partial correlations: Controlling for NA_sst, NP_sst links to humidity/DLR/SAT become insignificant (r=0.11/0.09/0.05). Controlling for NP_sst, NA_sst links remain significant (r=0.33/0.35/0.33), indicating a primary NA pathway. - CESM-LE support: In BMME, the leading MCA explains 50.1% covariance with APO–SIC series correlated at r=0.58 (p<0.01). APO-related summer mid-latitude SST warming persists into autumn; NA warming drives a wave train and anticyclonic anomalies over Barents–Kara–Laptev; increases in low-level moisture (r=0.34, p<0.05) and SAT (r=0.27, p<0.05) accompany reduced sea ice. In WMME, autumn NA SST persistence is weak and not significant, and APO–SIC linkage is of opposite sign. - Subsurface ocean memory: In BMME, APO-related NA warming extends to ~150 m in summer and persists into autumn, supporting SST persistence; in WMME, warming is shallow (~50 m) and does not persist. - Implications: The recent decadal shift toward positive summer APO since mid-2000s likely accelerated autumn sea ice loss in the eastern Arctic, superimposed on anthropogenic warming.

The study demonstrates that the Asian-Pacific Oscillation, a mode of internal atmospheric variability, exerts a significant lagged influence on eastern Arctic autumn sea ice through an ocean–atmosphere bridge. Positive summer APO phases precondition mid-latitude North Atlantic SST warming that persists into autumn, which then drives a teleconnection wave train toward the Barents–Kara–Laptev Seas. The resulting barotropic anticyclonic anomalies enhance moisture transport, increase lower-tropospheric humidity, strengthen downwelling longwave radiation, and warm surface air temperatures, collectively reducing sea ice. This process explains a substantial fraction of interannual-to-decadal sea ice variability and clarifies a mechanistic pathway of internal variability impacts. Model ensemble analyses (CESM-LE) corroborate the observed mechanism when the North Atlantic SST persistence is simulated, while failures to simulate this persistence lead to incorrect APO–SIC relationships. The findings highlight that internal variability can modulate the rate of Arctic sea ice loss on decadal scales, implying that future projections must account for APO variability. CMIP5/6 projections suggest a weakening summer APO by century’s end, which could partially counteract anthropogenic autumn sea ice loss in the eastern Arctic, introducing uncertainty in sea ice projections and underscoring the need for improved APO representation in models.

This work identifies and mechanistically substantiates the summer Asian-Pacific Oscillation as a driver of subsequent autumn sea ice variability and loss in the eastern Arctic. The APO influences mid-latitude North Atlantic SSTs whose seasonal persistence enables a teleconnection that alters eastern Arctic thermodynamics—enhancing humidity, downwelling longwave radiation, and surface warming—to melt sea ice. Observational diagnostics and targeted CESM-LE ensemble analyses jointly support the proposed APO–NA SST–Arctic pathway. Given a recent tendency toward positive APO phases, internal variability likely accelerated autumn sea ice declines in recent decades. Future research should: (1) quantify the relative roles of upstream vs downstream wave propagation in the North Atlantic response; (2) investigate processes governing seasonal persistence of APO-forced SST anomalies, including ocean mixed-layer depth evolution; (3) assess model fidelity in simulating APO variability and its coupling to North Atlantic SSTs; and (4) evaluate how projected APO changes modulate future Arctic sea ice under varying greenhouse gas scenarios.

Key limitations include: (1) reliance on reanalysis and observed datasets with potential uncertainties, especially pre-satellite SIC reconstructions; (2) detrended analyses isolate variability but may not fully account for nonlinear trend–variability interactions; (3) ambiguity in the relative importance of upstream vs downstream wave propagation into the North Atlantic remains unresolved; (4) CESM-LE member selection reveals model diversity—some members fail to capture the APO–NA SST persistence and the correct APO–SIC sign, indicating sensitivity to oceanic mixed-layer processes and coupling; (5) while North Atlantic SSTs emerge as the primary bridge, partial correlations suggest other processes may contribute to the APO–Arctic linkage; and (6) projections of future APO weakening and its compensating effect on sea ice involve uncertainties across models and scenarios.