| 29 Sep 2026
North Atlantic jet stream responses to AMOC tipping and global warming in regional climate simulations
Abstract. European weather variability is largely influenced by the position and intensity of the North Atlantic (NA) jet stream. Understanding how the NA jet may respond to global warming and an additional potential collapse of the Atlantic Meridional Overturning Circulation (AMOC) is therefore essential for assessing future European climate. Here, we investigate these changes using two fully coupled Earth System Models (MPI-ESM-HR and CESM1) together with high-resolution regional dynamical downscaling. Global warming is simulated by increasing atmospheric CO2 until a global mean warming of 2 K is reached, while AMOC collapse is induced through freshwater hosing from this 2 K state.
The regional simulations retain the main characteristics of the large-scale jet response simulated by the driving ESMs, while revealing regional modifications in both winter and summer that are more pronounced in summer. Under global warming, the response is highly model dependent, with an equatorward displacement of the winter jet in the MPI-driven simulations and a poleward displacement in the CESM-driven simulations. Following AMOC collapse, these contrasting responses persist, with the MPI-driven simulations retaining an equatorward-displaced jet and the CESM-driven simulations exhibiting a marked jet intensification.
These contrasting circulation changes arise from differences in the simulated AMOC response to global warming. MPI simulates substantially stronger AMOC weakening than CESM, leading to a more pronounced North Atlantic warming hole and stronger Arctic amplification. The resulting differences in sea surface temperature modify the meridional temperature gradients and baroclinicity, providing a physical explanation for the contrasting jet responses. Our results demonstrate that uncertainty in the simulated AMOC response propagates to regional atmospheric circulation and highlight the added value of regional dynamical downscaling for assessing the impacts of AMOC weakening and collapse on the North Atlantic jet stream.
This manuscript examines the response of the North Atlantic eddy-driven jet to CO2 increase, followed by AMOC collapse, using two Earth system models, dynamically downscaled to regional 15 km resolution. They find contrasting jet responses in the two models, which they attribute mainly to differences in AMOC changes and the associated changes in SST gradients and baroclinicity. The regional simulations largely reproduce the responses of the driving models, while revealing some regional modifications, particularly during summer. While the topic is important, I have several major concerns regarding (1) the scientific novelty and added value of regional downscaling, (2) the attribution of the contrasting jet responses to AMOC changes, (3) the equilibration of the simulated climate states and the robustness of the results given the use of a single realization and the substantial internal variability of the North Atlantic jet, and (4) the methodology used to quantify changes in jet position and intensity. Major comments
1. Scientific novelty and added value of regional downscaling
The impact of AMOC weakening on the North Atlantic jet stream has already been investigated in several previous studies (e.g., Jackson et al., 2015; Bellomo et al., 2021; Orbe et al., 2023; Meccia et al., 2024). In particular, Bellomo et al. (2021) examined the relationship between AMOC weakening and jet displacement across climate models, finding that stronger AMOC weakening was associated with a poleward displacement of the midlatitude jet (in contrast to the current findings). The main novelty of the present study therefore appears to be the application of high-resolution regional downscaling. However, the authors show that the main characteristics of the jet response are already present in the driving ESMs and are largely reproduced by WRF. Although some regional differences emerge, it is unclear whether these represent physically meaningful improvements or simply differences between the models. The authors should better demonstrate what new scientific understanding is gained from the regional simulations beyond providing higher spatial resolution, or change their conclusions throughout accordingly.
2. Attribution to AMOC changes
The authors attribute the contrasting jet responses in the two models primarily to their different AMOC responses. However, the models also differ in their atmospheric physics, background circulation, and climate feedbacks, which may independently affect the jet response. To quantify the specific contribution of AMOC weakening to the jet changes under global warming, either the AMOC weakening is suppressed while the model experiences the same greenhouse-gas forcing or the models are forced with the same AMOC weakening under the full 2k experiment. Such approaches have already been employed by Bellomo and Mehling (2024), who compared elevated-CO₂ simulations with and without AMOC weakening to isolate its contribution to the climate response. Comparing the jet response in such an experiment with that in the full 2 K warming experiment would help distinguish the contribution of AMOC weakening from other effects of global warming. Without such an experiment, the proposed attribution remains speculative.
3. Internal variability and robustness of the jet response
The analysis is based on a single realization of each experiment, with only 20 years used to characterize the jet response. The North Atlantic eddy-driven jet exhibits substantial internal variability on interannual and decadal timescales, and it is therefore unclear whether the diagnosed changes in jet position, intensity, and regime frequency represent robust forced responses or could arise largely from internal variability. This is particularly important because the contrasting responses of MPI-ESM-HR and CESM1 form the basis of the manuscript's main conclusions. The statistical significance tests presented in the manuscript do not adequately address this issue, as they do not establish the robustness of the forced response across independent realizations. At a minimum, the authors should use their long control simulations to estimate the distribution of jet changes between independent 20-year periods and assess whether the diagnosed responses exceed the expected range of internal variability. Ideally, multiple ensemble members would be needed to establish the robustness of the responses and the differences between the two models. This is also true for other variables in the system (e.g., AMOC).
4. Quantification of jet position and the use of clustering
I am not convinced that the clustering approach used in Sect. 2.4 is the most appropriate way to quantify changes in the North Atlantic jet. Since one of the main objectives of the study is to investigate changes in jet position and intensity, why not directly calculate the latitude and strength of the jet from the maximum of the zonally averaged low-level zonal wind at each time step, as was done in my previous studies? This would provide a continuous and more direct measure of jet displacement, allowing the authors to quantify changes in the mean position, variability, and distribution of jet latitude across the different experiments. In contrast, the clustering approach reduces the continuous variability of the jet to three discrete regimes and may obscure shifts occurring within individual regimes. Using canonical metrics would help compare the analysis to previous studies.