Fig. 4: Correlation of temperatures and circulation with stratospheric water vapor across climate models.

Shown are the inter-model correlations for CCMI–2022 and CMIP6 models, between the mean water vapor mixing ratio in the lowermost stratosphere (LMS, Northern hemisphere black, Southern hemisphere red) and a LMS temperature, b subtropical meridional temperature gradient, c subtropical jet intensity (zonal wind at upper jet flank), and d tropical upwelling (only available for a few models). For details on selected regions and available models see Methods. Data shown are from CCMI–2022 models (circles) and CMIP6 (crosses), with EMAC (in CCMI–2022) highlighted with stars. Pearson correlation coefficient values are given in each plot for Northern/Southern hemisphere data, together with the associated p-values estimating statistical significance (in brackets, rounded to 2 decimals). e Pearson correlation coefficients between mean LMS water vapor mixing ratios and local zonal wind across CCMI–2022 and CMIP6 models at each pressure–latitude grid point for December–February, and f for June–August. Climatological zonal wind for the multi-model mean is shown as black contours, the WMO lapse rate tropopause (from ERA5 reanalysis) as a thick gray line, and the regions for calculating mean LMS water vapor are highlighted as gray dashed boxes. Correlation coefficients in the Northern hemisphere are calculated with Northern hemisphere LMS water vapor, and respectively for the Southern hemisphere. Dots indicate where correlations are not significant at 95% confidence level.