Please find my comment attached. 

Comments on “Case-to-Case Variability in the Tropospheric Response to Sudden

Stratospheric Warmings Revealed by Ensemble Re-Forecasts” by Loeffel et al.

Summary

This study uses a model to study the downward impact of SSW on the troposphere by performing free-run simulations and the reinitializations of the selected SSWs. The authors called those simulations reforecasts, which are actually new restarts. Therefore, I did not agree that they can use those simulations as forecast study. Those SSWs from models are generated in the model, rather than realistic SSWs. Using the modelled SSWs as a study, I am afraid that much uncertainty still exists. The factors that controlled the downward propagation of the SSWs have been discussed in previous studies. The novel finding in this study should be emphasized. I suggest a major revision at the present time.

Major Comments

Comment 1:

All the forecasts are not initialized from the real observation state, and the forecasts are ideal based on the model outputs. Due to this flaw, I can not agree that SSWs are forecasts. But they are simulated in model that does not depend on the boundary conditions and initial state. The S2S forecasts depend on both boundary conditions and initial states. Therefore, I did not see any advantage of initializing more times for the modelled SSWs.

Comment 2:

This study emphasizes that the stratospheric state at 100 hPa is more determined for the SSW signal downward propagation. This finding is not surprising in my understanding. The lower level you use, the larger the correlation will be. Since 100 hPa is lower and closer to the troposphere, the relationship is naturally better than for 10 hPa. The authors should be award that the stratospheric signals show a lagged downward propagation from the stratosphere to the troposphere, with the signals gradually decreasing. Although the correlation between 100 hPa and troposphere is larger, the time lag between maximum center is different. The maximum 10 hPa signal leads longer than the 100 hPa signal, although the correlation from 100 hPa to the troposphere is larger. Overall, the 100 hPa state is also directly affected by 10 hPa SSW.

Comment 3:

Reforecasts has an advantage that they provide larger ensemble to study the SSWs. However, the SSWs from the reforecasts are still very limited. Compared with existing studies using S2S forecasts, this study even uses less SSW samples. I mean that the selected SSWs from the free run are very limited for reforecasts. It is still not easy to find the novelty of this study as compared with existing literature.

Comment 4:

The relationship between SSW and blocking is also widely reported in literature. This study seems to be a model assessment. What is the new finding in this study? I also suggest to show more composite results for near surface temperature and precipitation, which have been shown for S2S models.

Minor Comments:

1. L27: The case-by-case difference of SSW predictability and surface impact have been reported in Rao et al. 2019 (doi: 1029/2019JD030900) and Karpechko et al. 2017.

1. L31, 45: The forecast of typical downward propagating SSW and non-downward propagating SSW has been reported in Rao et al. 2020 (doi: 1029/2019JD031919) and Butler et al. 2021QJ.

1. L70: Is this method first used in literature? If not, please add relevant references. I remember that several studies based on the NCAR model also use similar methods.

1. L75: Is the free run performed for other project or exclusively for this study?

1. L123: This AO index is weird. I expect to see the AO index that can be either positive or negative. Positive phase corresponds to strengthened vortex and negative AO phase corresponds to weakened vortex.

1. L167: This difference might be bias rather than the internal variability. How did you prove that this difference is not bias.

1. L174: We described AO positive or negative, rather than decreased or increased.

1. Figure 3: Both plots show similar pattern and significant downward impact following SSW onset. The difference is the composite intensity. Rao et al. 2020 have reported that the SSW intensity has a more deterministic contribution to the downward impact.

1. L204, 229: Lower stratosphere is also stratosphere, which is closer to the troposphere.

1. L227: Blockings also usually precede the SSW onset. Do you check if the blocking is also a driver for SSWs?

1. L250, 260: Of course it should be so. See major comments.

1. L269: I suggest you should clarify those studies are mainly based on S2S model reforecast. Also include Rao et al. 2021 (doi: 1029/2021JD035057).

1. L273: I am not sure if it is an advantage, which can be possible using other models.

1. Figure 8 still shows that stronger SSWs have downward propagating impact. SSW 2 is stronger than SSW 3.

1. Figure 12: How about the composite in week 1-3 before SSWs? See the free run. Forecasts have no backward time (negative time).

1. L317-355: The discussion is too long and should be shortened.

1. L359, L435: Did your show the reflective layer for each SSW?