Fig. 2. EEMD for the tropical (30°S-30°N) averaged SWV entry from 1984 to 2020 (units: ppmv; at 82 hPa in SWOOSH, at 70 hPa in ERA5). (a) and (b) are the original time series. (c), (d), (e), (f), (g) and (h) are the first three IMFs. (i) and (j) are the final residuals after seven IMFs extracted.

In sum, a robust drying trend in the lower tropical stratosphere is confirmed by both the linear and nonlinear long-term trend analysis. The freezing-dry model is generally applicable in the decrease rate, indicating the dominating role of the tropical pathway in the process.

As a key factor modulating the transport through the tropical pathway, the SST long-term variations in tropical oceans, especially the tropical Pacific and tropical Indian Ocean, are further investigated on its linkage to the drying trend in the lower stratosphere. Fig. 3a shows the zonally resolved correlation coefficients between tropical SST anomalies and the tropical tropopause temperature and SWV. To remove the strong interannual variations, we apply a 5-year running mean on tropical tropopause temperature and SWV time series before the linear relationship analysis. A strong negative linear relationship is shown between the long-term change of SST of the IPWP and the tropical tropopause temperature, with a weak positive relationship over the Niño3 region. Because tropical tropopause temperature is a great indicator of SWV, the SST in IPWP also has a greatly negative correlation with SWV, for both the tropical and global averages from both data sources. During 1984-2020, the IPWP has been substantially warming, at a higher rate (0.12 ^oC per decade) than the eastern Pacific (Figs. 3b, 3c). The strong correlation between the SST in IPWP and SWV indicates that there may be some connection between the warming IPWP and the drying stratosphere.