GoKawiilData
Ice mass variability was quantified using GRACE and GRACE-FO gravimetry data50 spanning 2003–2024. SMB (precipitation minus meltwater runoff, sublimation, evaporation and wind erosion) outputs from RACMO2.4p1 (ref. 51) and MARv3.14 (refs. 52,53) were used to link precipitation variability to observed ice mass changes. Precipitation variability was assessed using several independent datasets, including ERA5 (ref. 54) and MERRA-2 (ref. 55) reanalyses and RACMO2.4p1 regional climate-model precipitation, to ensure consistency across data sources. The Law Dome ice-core snow-accumulation record49 was used as extra observational input. Large-scale circulation and moisture transport were analysed primarily using ERA5 fields, including SST, Z300, evaporation and integrated water vapour transport (IVT). Velocity potential was obtained from NCEP-NCAR Reanalysis 1. The Niño 3.4, IOD, IPO, zonal wave-3 and Southern Annular Mode56 indices were used to examine their relationships with TWP variability. To assess the frequency of TWP warming events, we analysed both ERA5 (1950–2025) and the CESM1 Large Ensemble57, including the 1,800-year fully coupled pre-industrial control simulation and the 40-member historical and RCP8.5 simulations. To provide an independent long record check of the TWP–East Antarctic dipole–QW precipitation relationship, we also used two paleo-reanalysis products. EKF400v2 (ref. 58) provides monthly fields of Z500, 2-m air temperature and precipitation from 1603 to 2003. LMR v.2.1 (ref. 59) provides last-millennium ensemble reanalysis fields of Z500, SST and precipitation on an annual basis. Because EKF400v2 does not provide SST, its 2-m air temperature field was used as an indicator of SST variability. Monthly anomalies were generally calculated by removing the climatological seasonal cycle.
Guided by the spatial pattern of mass and cumulative precipitation anomalies, QW was defined as the region spanning 70–145° E and 65–75° S. The TWP was defined as a 15° clockwise-rotated ellipse centred at 135° E, 5° S, with semi-major and semi-minor axes of 35° and 15°, respectively, on the basis of the observed relationships between EA precipitation and tropical SSTs. We also defined an EA dipole index as the area-mean Z300 anomaly over the EA high-pressure centre (55°–70° S, 120°–160° E) minus that over the low-pressure centre south of Australia (30°–50° S, 80°–150° E).
Significance testing
Statistical significance of correlations, regressions and trends was assessed using two-tailed Student’s t-test at the 95% confidence level, with the effective sample size corrected for autocorrelation.
$${N}^{* }=N\frac{1-{r}_{1}{r}_{2}}{1+{r}_{1}{r}_{2}}$$ (1)
Here N is the total number of samples, and r 1 and r 2 are the lag-1 autocorrelation coefficients of the two correlated time series. For spatial fields, statistical significance was further controlled using false discovery rate correction60.
MCA decomposition
MCA61 is a multivariate statistical technique that uses singular value decomposition on the cross-covariance matrix of two fields to extract paired spatial patterns and time series that maximize their covariance. Each MCA mode represents a linear combination of the original variables that captures a specific covarying structure shared by the two fields. The importance of each mode is measured by the squared covariance fraction, which indicates the proportion of total squared covariance explained by that mode. Higher squared covariance fraction values denote leading modes of coupled variability. In this study, MCA was applied to deseasonalized and detrended monthly anomalies of ERA5 Z300 south of 23° N and Antarctic precipitation for 1979–2024.
WAF diagnosis
... continue reading