Baroclinic annular mode (BAM) represents a leading mode of variability in extratropical eddy activity, which owes its existence to the two-way feedback between baroclinicity and the eddy fluxes of heat. BAM is linked to the position and strength of midlatitude storm track activity at subseasonal timescales, enabling it to influence midlatitude precipitation patterns and the hydrological cycle. Despite its significance, the physical processes affecting its dynamics and variability are not fully understood.

Here, we propose that cloud radiative effects (CREs) and their driving dynamics can influence the spatiotemporal characteristics of BAM. A suite of experiments using the DOE’s E3SM atmospheric model is performed to test this hypothesis by either applying a cloud-locking technique to disable interactive CREs without altering the mean states, or by making clouds transparent (radiatively inactive) to longwave LW radiation. Our results indicate that interactive CREs have a statistically significant damping effect on the intensity of BAM and its contribution to the total variability. Specifically, interactive CREs reduce BAM intensity by up to ~15% and decrease its explained variance by up to ~6%. These effects primarily arise from the radiative modification of clouds through LW and condensational heating, which strengthen the low-level static stability and weaken meridional temperature gradients. The reduced BAM variability is also consistent with CRE-reduced two-way feedback between baroclinicity and eddy heat fluxes, following the charge-discharge mechanism (baroclinic oscillator). Interestingly, the opposite results are found when the clouds are made transparent to LW CRE. The clouds instead increase BAM intensity by up to ~25%. This effect results from the mean LW CREs, which decrease the background static stability and increase the temperature gradients. This result is supported by multi-model experiments within the Cloud Feedback Model Intercomparison Project (CFMIP), which demonstrate a notable ~10%-22% increase in BAM intensity using a similar experimental protocol.

Overall, the results suggest that both interactive CREs and changes in mean CREs can significantly influence the dynamics and variability of the BAM. The opposite results obtained from the two experimental setups do not suggest that one method is more accurate than the other. Rather, they underscore the necessity of selecting the appropriate methodology based on the specific research question being addressed. Regardless of the methodology used, the results emphasize that a proper representation of CREs is essential to accurately capture the leading mode of variability in extratropical storm track and to improve its future projections.