Fig. 4: Summer-half westerly jet positions and multi-field MCA analysis.

a Mean jet latitude composited for El Niño (red) and La Niña (blue) years. The dotted contour (blue) shows the 2500 m elevation, approximating the boundary of the Tibetan Plateau. The average latitudinal position of the jet, the mean difference (Δ) between the two composites, and the associated 95% confidence intervals and p-values are reported on the figure. Significance was assessed using both a two-tailed Student’s t-test and bootstrap resampling (see “Methods”). b Same as (a), but for high (red) and low (blue) CEC δ¹⁸O_p index years after regressing out the Niño 3.4 influence (see “Methods”). c, d Multi-field maximum covariance analysis (MCA) between tropical Pacific SST (shaded, left panel, c) and East Asian δ¹⁸O_p (shaded) and 200 hPa zonal winds (contours) (right panel, d) for September–October. The SST data is from ref. ⁶⁸. Shown are homogeneous regression maps for SST (units: °C per normalized SST time series) and heterogeneous regression maps for δ¹⁸O_p (‰ per normalized SST time series) and zonal winds (m s⁻¹ per normalized SST time series). Contours are plotted at 0.5 intervals, with negative values dashed, positive values in solid blue, and the zero contour in thick black. The Tibetan Plateau (≥2500 m) is outlined with a blue dotted line. The leading MCA mode explains 83% of the squared covariance fraction.