Inter-relationship between subtropical Pacific sea surface temperature, Arctic sea ice concentration, and North Atlantic Oscillation in recent summers

Abstract

The inter-relationship between subtropical western-central Pacific sea surface temperatures (STWCPSST), sea ice concentrations in the Beaufort Sea (SICBS), and the North Atlantic Oscillation (NAO) in summer are investigated over the period 1980-2016. It is shown that the Arctic response to the remote impact of the Pacific SST is more dominant in recent summers, leading to a frequent occurrence of the negative phase of the NAO following the STWCPSST increase. Lag-correlations of STWCPSST positive (negative) anomalies in spring with the negative (positive) NAO and SICBS loss (recovery) in summer have increased over the last two decades, reaching r = 0.4-0.5 with significance at the 5 percent level. Both observations and the atmospheric general circulation model experiments suggest that the positive STWCPSST anomaly and subsequent planetary-scale wave propagation act to increase the Arctic upper-level geopotential heights and temperatures in the following season. This response extends to Greenland, providing favorable conditions for developing the negative phase of the NAO. Connected with this atmospheric response, SIC and surface albedo decrease with an increase in the surface net shortwave flux over the Beaufort Sea. Examination of the surface energy balance (radiative and turbulent fluxes) reveals that surplus energy that can heat the surface increases over the Arctic, enhancing the SIC reduction.

Figure 1

(a) Time series of the SST anomaly over the subtropical western-central Pacific (130°E–180°E, 0°–35°N) (black), the Arctic sea ice concentration (SIC) anomaly over the Beaufort Sea (180°–120°W, 75°–90°N) (red), and the phase of the NAO (blue) over the period 1980–2016. The NAO and SIC anomalies are averages for June, July, and August (JJA), while the SST anomaly is an average for April, May, and June (AMJ), which is two months before JJA to present time-lagged impact of the SST that leads the variation of SIC and NAO phase. Two spatial patterns represent (b) simultaneous correlations between the NAO indices and SIC for JJA, and (c) two month lag correlation between the northern hemispheric SST in AMJ and the NAO in JJA. Green dots are plotted, where correlations are statistically significant at 5 percent.

Figure 2

Interannual variation of lag–correlations (with 20 year running window) between the NAO, sea ice concentration over Beaufort Sea (SICBS), and the subtropical western–central Pacific SST (STWCPSST) over the period 1980–2016. Years on the x–axis denote the centers of the individual 20–year windows. Each panel shows time variation of the lag–correlations for (a) STWCPSST × (−1) in AMJ, MJJ, and JJA vs. JJA NAO, (b) STWCPSST × (−1) in AMJ, MJJ, and JJA vs. JJA SICBS, and (c) SICBS in MJJ, JJA, and JAS vs. JJA NAO. Trends are removed from the variables before calculating correlations. Short and long dashed lines in each panel represent a statistical significance limit at 5 percent (short-dash) and 10 percent (long-dash), respectively.

Figure 3

Upper panel: Differences in the recent boreal summer (the 21st century) upper–tropospheric geopotential height [m] (250hPa, left) and mid–tropospheric temperature [0.1 K] (500hPa, right) between the subtropical western–central Pacific SST (STWCPSST) positive and negative spring anomalies. Composite of geopotential height and temperature in summer (JJA) preceded by cooler than average STWCPSST (detrended) in spring is subtracted from the composite of the geopotential height and temperature in summer preceded by warmer than average STWCPSST in spring. Green dots are plotted, where the difference values are significant at 10 percent. Lower panel: Same as the upper panel but for wave activity flux vector distribution in summer at 250 hPa. Shaded is the difference in surface temperature (AMJ average) between the STWCP warming and cooling in spring. Box with dashed line denotes the STWCP region.

Figure 4

Same as Fig. 3 but for differences in (a) the Arctic sea ice concentration (shaded) and surface albedo (contoured) in percentage, (b) surface net shortwave downward flux [W m⁻²], (c) surface net longwave flux [W m⁻²] (shaded) and total cloud area fraction (contoured), (d) surface net radiation [W m⁻²], (e) turbulent fluxes (latent + sensible) [W m⁻²], and (f) surface net radiation minus turbulent fluxes ((d) minus (e)) to assess the energy balance [W m⁻²].

Figure 5

Responses of geopotential height and temperature to the subtropical western–central Pacific SST (STWCPSST) increase simulated by the atmospheric general circulation model. Upper panel: Difference in 20 member averaged Jun–August geopotential height and temperature between Exp. SPW, where the observed warming (SST average over the years when the SST is warmer than climatological average) is imposed over the STWCP in spring, and Exp. CTL forced by climatological SST everywhere. Lower panel is the same as upper panel but for the response in winter to the STWCPSST increase in fall. Green dots are plotted, where the difference values are significant at 10 percent.