Antarctic sea-ice expansion between 2000 and 2014 driven by tropical Pacific decadal climate variability Gerald A. Meehl 1, Julie M. Arblaster 1,2, Cecilia M. Bitz 3, Christine T.Y. Chung 4, and Haiyan Teng 1 1. National Center for Atmospheric Research, Boulder, CO, USA 2. Monash University, Melbourne, Australia 3. University of Washington, Seattle, WA, USA 4. Bureau of Meteorology, Melbourne, Australia NATURE GEOSCIENCE www.nature.com/naturegeoscience 1
Supplementary Fig. 1: a) Observed annual mean Antarctic sea ice extent (106 km2, solid line, NASA data31 ) and annual mean multi-model mean (from CMIP5 models) Antarctic sea ice extent (106 km2, dashed line), shading indicates the 95% confidence interval, vertical solid line marks transition of the IPO from positive to negative, solid lines are linear trends fit to observations from 1979-1999, and 2000-2014; b) geographical designations for regions near Antarctica: BS = Bellingshausen Sea, AS = Amundsen Sea, RS = Ross Sea, WPO = Western Pacific Ocean, IO = Indian Ocean, WS = Weddell Sea; c) observed SST trend ( C decade-1; NOAA-OI) during the recent negative phase of the IPO (2000-2013), stippling indicates significance at the 10% significance level from a two sided t test; d) June-July-August (JJA) precipitation anomalies for the 2000-2013 average, (mm day-1) from ERA-Interim, stippling indicates significance at the 10% significance level; e) same as (d) except for GPCP precipitation; f) IPO pattern defined as second EOF of low pass filtered observed SSTs (first EOF is trend), HadISST data from 1870-2014, sign convention shown here for positive IPO phase; g) IPO index (also shown as inset in Fig. 1 of main text) defined as the PC time series of the second EOF in (e); h) regression of IPO index in (g) on to observed GPCP precipitation data (compare to similar calculation done with ERA-Interim data in main text Fig. 1a). 2 NATURE GEOSCIENCE www.nature.com/naturegeoscience
Supplementary Fig. 2: Seasonal mean Antarctic sea ice extent trends (10 6 km 2 ), for other CMIP5 ensemble members with positive Antarctic sea ice trends other than those with observed globally averaged surface air temperature trends as observed, error bars are 95% confidence interval. NATURE GEOSCIENCE www.nature.com/naturegeoscience 3
Supplementary Fig. 3: Observed rainfall and sea level pressure (SLP) for 1979-2014 regressed on to area-average SLP in the ASL region (denoted by box, 55 S-65 S, 70 W-130 W). The sign of the regressions is for negative values of ASL SLP. Contours are SLP regressions (negative are blue dashed, positive are red solid), and colored shading is for precipitation. Statistical significance at the 95% level for precipitation is indicated by stippling, and for SLP by thickened contour line segments. Largest precipitation values associated with a deepened ASL are in the equatorial Pacific east of about 160 E; e) regression of annual mean observed SLP onto the observed IPO index shown in Supplementary Fig. 1g, 1979-2014; sign convention of regression values is for negative IPO, with positive SLP values in the eastern tropical Pacific, negative in the west producing an enhanced east-west SLP gradient that drives stronger trade winds, cooler eastern equatorial SSTs and thus negative IPO, negative precipitation and convective heating anomalies in the eastern equatorial Pacific, and negative SLP values in the ASL region. 4 NATURE GEOSCIENCE www.nature.com/naturegeoscience
Supplementary Fig. 4: a) Observed Antarctic sea ice extent trends for MAM season during era of positive IPO, 1979-99 (trends greater than 2.5%/decade are plotted as color shading, values significant at the 5% level are stippled; MAM is the only season with significant increases of sea ice extent in Fig. 1 in main text); b) SLP anomalies (hpa) from a specified convective heating experiment with a positive heat source in the equatorial Atlantic region (30 W, equator), negative SLP anomalies shaded blue, positive orange, stippling indicates significance at the 5% significance level, scaling vector for surface winds (m sec -1 ) at lower right (same as Supplementary Fig. 5b but included here to compare directly to observed sea ice trends to show connection between northward anomaly winds in model simulation in Ross Sea region and increases in sea ice in that region). NATURE GEOSCIENCE www.nature.com/naturegeoscience 5
Supplementary Fig. 5: SLP anomalies (hpa) from a specified convective heating experiment with a positive heat source in the equatorial Atlantic region (30 W, equator), negative SLP anomalies shaded blue, positive orange, stippling indicates significance at the 5% significance level, scaling vector for surface winds (m sec -1 ) at lower right of each panel. Negative values indicating a deepening of the ASL are denoted by L. 6 NATURE GEOSCIENCE www.nature.com/naturegeoscience
Supplementary Fig. 6: SLP anomalies (hpa) from a specified convective heating experiment with a positive heat source in the SPCZ region (170 E, 20 S), negative SLP anomalies shaded blue, positive orange, stippling indicates significance at the 5% significance level, scaling vector for surface winds (m sec -1 ) at lower right of each panel. Negative values indicating a deepening of the ASL are denoted by L. NATURE GEOSCIENCE www.nature.com/naturegeoscience 7
Supplementary Fig. 7: Same as Fig. 2 in main text but shading indicates sea level pressure trends with white stippling denoting 5% significance. 8 NATURE GEOSCIENCE www.nature.com/naturegeoscience