Fig. 3. (a) The zonally resolved correlation coefficients between tropical tropopause temperature anomalies, tropical SWV entry anomalies and global SWV anomalies. The orange area indicates the IPWP region. The bold line indicates that significance at the 90% confidence level. (b) SST trends in the tropical Pacific and Indian Ocean from 1984 to 2020, with the black box indicating the area of the IPWP (30°S to 30°N, 30°E to 180°). Dots indicate significant trends at the 90% confidence level. (c) The time series of the IPWP averaged SST anomalies from 1984 to 2020. This solid line indicates SST anomalies in the IPWP. The dashed line is the linear regression line. The trend and uncertainty are shown in the top right-hand corner. The uncertainties are expressed by 2σ errors.
Fig. 4 shows the distribution of zonal mean SWV decreasing trends and tropical tropopause cooling trends, with the left column showing the raw trends and the right showing the trends after linearly regressing out the warming IPWP signal. In SWOOSH, the significant dehydration trends in the SWV below 30 hPa induced by the cooling trend in the tropical tropopause are coherent with that shown before, and the only decreasing trend is located above 30 hPa at high latitudes (Fig. 4a). And this is consistent with Hegglin et al. (2014). In ERA5, the distribution of the trend is significantly different. The decreasing trend is mainly in the southern hemisphere, while the decreasing trend in the northern hemisphere is weak (Fig. 4b). After removing the IPWP warming signal, decreasing SWV trends become weaker in both data sets and even increasing trend dominates the northern hemisphere in ERA5 (Figs. 4b, 4d). This implies that the recent decadal SWV trends may be connected with IPWP warming. The IPWP warming signal is also shown in the cooling trends of tropical tropopause. After removing the IPWP warming, the significant cooling trends in tropical tropopause are relatively weak in both ERA5 and JRA55 (Figs. 4f, 4h).