This is an interesting work finding an interpretation and a remedy to one of the serious known limitations of climate models. The manuscript is well written and I have only a limited number of minor comments and questions to be considered by the authors.

Solomon et al. (2010) wrote that variations of lower stratospheric water vapour may account for up to 30 % of the greenhouse gases modification of the radiative budget on a decadal scale. They do not write that they account for 30 % of the total variation since 1850 as suggested by the sentence on lines 13-14 in the manuscript. Please correct to avoid such confusion.

It is quite clear that the excessive numerical diffusivity of the CMIP6 models should be related to their spatial resolution and their transport scheme, which display large differences among the ensemble. It is quite frustrating that they are here all put in the same bag without any attempt to draw a distinction. For instance, it would be very interesting to know whether it is the horizontal rather than the vertical resolution that matters as very different choices have been made among the ensemble. In terms of horizontal resolution, the T42 resolution f the EMAC used here put is at the lower end of the CMIP6 ensemble but its vertical resolution puts it at the upper end. This is perhaps an answer to the above question as fig.1 shows it does much worse than the CMIP6 mean although what probably matters is not the total number of levels but the number of those which span the UTLS, a parameter which is badly documented.

My main concern is about the figures and their readability. Black contours and labels on dark blue and red are hardly visible and readable. This is not good on the screen and it is terrible on a printed version. This needs to be improved. There is no reason to use a divergent color map to show water vapour in fig2(a-c). The two first rows of fig.2 show redondantly PV and U on panels which are overcrowded. Please reorganise these two rows to show only 3 variables in each panel. In figs 1 and 2, some quantities (wind, temperature) would be better displayed as differences between EMAC-ClaMS and EMAC. Adding a grid would overload the figures but ticks can be put on the upper and right sides of the figures to improve readability. In figure 3, I do not think that the temperature gradient contours are very useful, and they have no labels and there is no indication of contour intervals in the caption. I would prefer to have some contours for the quantities displayed in color as it is almost impossible to read the values from the color map (or choose a better indexed color map).

It is very hard to appreciate from the first two rows of fig. 2 that the isentropic PV gradient is strengthened around the tropopopause and whether vapour contours are following or not the PV structure, although it is quite clear from the third row.

Figure 3 shows that the zonal wind incease it shifted by 30°E with respect to the water vapour anomaly. So the temperature drop should also be shifted which means that the response to the water vapour is not the simple local process advocated in the manuscript but involves also transport and delay.

It should be noted that the differences between EMAC and EMAC-CLaMs are much smaller than those between ERA5 and EMAC except perhaps for the PV in the lower stratosphere. In this respect it would be useful to see the curves for EMAC and EMAC-ClaMS in fig.4 (c-d) to appreciate the improvement in regard to the current dispersion and bias of the models.

The manuscript is not totally clear about the effect on the monsoon circulation. It is indicated that the equatorward branch is broadened and strengthened on the eastern side but this is not a mechanism which by itself is able to modify the closed monsoon circulation as dicussed in section 3 since this modification is correlated to its internal PV budget.

I am unsure the proper way to refer to ERA5 data is a link to Lawrence Livermore National Laboratory.

Ploeger et al. explore the connection between lower stratospheric water vapour anomalies and dynamical biases previously identified by Charlesworth et al. in CMIP6 models using a combination of CMIP6 model output, observations, and specifically designed simulations using the EMAC model run with different transport schemes. Their finding that Pacific UTLS water vapour anomalies impact regional circulation builds on the findings of Charlesworth et al. by exploring inn more detail local water vapour anomaly-circulation bias connections, particularly with relation to the Asian summer monsoon and transport from Asia into the Pacific. I found the manuscript to be well written and the analysis clear. I feel the paper fits the scope of ACP and explores and important topic. However, I would recommend the authors address the comments below before publication.

General comments:

Care should be taken throughout the paper to make a better distinction between water vapour anomalies and biases. To me, the anomaly is the relative abundance of local water vapour with respect to the zonal mean, whereas a bias is specifically a comment about the local abundance of water vapour in the model vs observations. These quantities may be related to an extent, but it is possible to imagine a very dry model that has a large water vapour anomaly over the Pacific having a small bias with respect to observations. I feel throughout the paper these two terms are used interchangeably, and this should be addressed.

In their analysis the authors use different lengths of time, over different years, for the different datasets used in this study. The EMAC model data used for analysis covers 2000-2009, the CMIP6 model data covers 2000-2014, and the ERA5 data covers 2005-2015. Given that the authors speak about the variability of water vapour on decadal timescales in the opening sentences of the introduction, I wonder to what extent the features identified in the paper are dependent on the choice of these relatively short time periods. For the CMIP6 data it is conceivable that averaged over multiple ensemble members 15 years represents something of a robust climatology, but can the same be said of 10 years of ERA5 or EMAC data? If the authors had used a different time period (e.g., 2000-2009), or a longer time period (2000-2020), would the ERA5 data show the same results? Can the authors say anything about this that strengthens the arguments made in the paper?

Many of the figures are difficult to read and crowded with information. Figures 1 and 2 have a lot going on, and it is difficult to see the black contours and arrows on the very dark red and blue shading. In figure 3 it is very difficult to tell the values form the blue and red shading. I would recommend replotting many of the figures to improve readability, and where possible separating the figures into more panels.

It is clear from the CMIP6 multi-model mean and figures S3/S4 that the Pacific water vapour anomalies are common across CMIP6 models. However, not all models show large positive anomalies with respect to the zonal mean above the region of the Asian summer monsoon (most notably the CNRM model, some configurations of the CESM model, and the SAM0-UNICON model). Can anything be inferred about links between Pacific water vapour anomalies and regional circulation biases in these models that look quite different to the others? Going further, can anything be said about model structural differences (e.g., resolution, transport scheme) in driving the biases explored in this study, or is it the case that whatever the model resolution, global models are too susceptible to processes like numerical diffusion? I think I’d like to see some discussion exploring model differences in the manuscript.

Specific comments:

L13-14: “Variations in stratospheric water vapour have been shown to modify past global warming by up to 30% (Solomon et al., 2010)” This sentence is misleading – the Solomon et al. paper says that variations in lower stratospheric water vapour on a decadal timescale can have a significant radiative impact, but this sentence could be read as 30% of past climate change can be attributed to water vapour changes, which is not the case.

L19: My understanding is that the 0.1-0.26 W m-2 K-1 range from Banerjee et al. is calculated using CMIP5 models. Is there an estimate from CMIP6 models? I know that a recent paper by Nowack et al. (2023) has tried to constrain this estimate and comes up with a slightly reduced range of 0.086-0.201 W m−2 K−1, which may be worth noting here.

L36-38: “Analysis of the distribution of water vapour in the lower stratosphere simulated by a climate model has shown a substantial model bias in the Asian monsoon region” I was not sure here if the authors are talking about analysis of a single climate model, or making a statement that is true for all models.

L52: Make it clear the anomalies are with respect to the zonal mean distribution of water vapour

L63-64: Here, when saying the EMAC anomaly is more pronounced than the CMIP6 multi-model mean, I feel it is worth explicitly stating whether the EMAC model is an extreme case, or within the CMIP6 model spread, citing figure S3/S4 as evidence.

L71-72: When speaking about temperatures here, it would be great if the authors could show the temperature difference between EMAC and EMAC-CLAMS in a separate figure/panel (especially given general comment above about figure readability).

L114-116: “The CMIP6 inter-model correlation between local meridional v–wind velocity in the monsoon anticyclone and a water vapour index measuring the strength of the Pacific moisture anomaly (Methods) shows a significant anticorrelation eastward of 100◦E (Fig. 3d).” I am unsure here what feature I should be looking at. Firstly, I think this should be figure 3c, not 3d. If so, is it the statistically significant blue shading between 100-50 hPa at around 150 degrees? Please provide more specific description in the text. Additionally, it is very hard to read from the figure the correlation in this region. Perhaps it could be given in the text?

L119-120: “Above the tropopause, this correlation pattern closely resembles the pattern of differences in meridional flow (v–wind) 120 between the control and modified–Lagrangian simulations in the EMAC model experiment (Fig. 3a)” Does it? There is some overlap in the blue shaded region in figure 3a and 3c, but the features themselves have very different shapes, horizontal extents, and extensions into the troposphere.

Technical corrections:

L62: remove ‘also’

L166 – oxydation should be oxidation

Figure caption for S4 – I believe it should say (Fig. S3 continued)

Ploeger et al. report on the impact of moisture biases in the Pacific UTLS on regional circulation patterns at the tropopause level, such as the anticyclone Monsoon circulation and its zonal extent over the Pacific. They show that a modified Lagrangian scheme improves a ubiquitous deficiency of the EMAC model, in terms of the UTLS climatology of water vapor (too "zonal" and moist bias over W-Pacific) and this improvement also has effects on the circulation, such as on the Monsoon anticyclone structure (reducing the zonally too broad and strong Monsoon anticyclone, and too strong westerlies over the tropical Pacific). I find this study really interesting, and the paper is well structured. I only have some minor suggestions. I recommend prompt publication upon addressing them.

General comments:

1) I think that for CLAMS, the role of stratospheric water vapor (SWV) can be indeed nicely isolated with the suite of experiments presented in the paper. However, I am less convinced about the "generalization" for all CMIP6 models: I am not entirely sure that the "cross correlation" of U-wind vs SWV (Fig. 4) can be really taken as "proof" that moist biases in other models have the same effects on the circulation as demonstrated for EMAC. This is essentially shown in Fig. 4c and 4d. I see quite a difference, for example, between the effects of CLAMS in EMAC (panel b - this should be the "impact of SWV improvements") and the relationship between inter-model spread in SWV and U in CMIP6 (panel d) - the location of these correlations is quite different. Can the authors maybe test the correlation between SWV and U (on e.g. inter-annual time-scales instead of "inter model") within CLAMS and EMAC directly, to support their inferences about dynamical impacts of SWV biases in other climate models?

2) While the role of SWV can be indeed nicely explained, the role of other radiatively active species in the stratosphere is a lot less clear. Among them, ozone is another major heating source in the tropical stratosphere, but it's not discussed at all. I would expect the implementation of CLAMS to also affect the ozone in e.g. indirect ways. Have the authors looked into changes in ozone between the regular EMAC and EMAC-Clams? Would these be big enough to also play a role in the differences seen in terms of the large-scale circulation?

3) While the role of the westerly wind duct is clear (at 100 hPa) in linking SWV biases and the anticyclone circulation, the role of other (prominent) dynamical features of the lower stratospheric circulation are less clear... such as, for example, the QBO jets, the cold point tropopause, the tape recorder, etc. I would recommend the authors to give a "broader" view of the effects of the implementation of CLAMS, aside from the localized effects at 100 hPa.

Specific comments

L4 I'd recommend adding a specific altitude range when talking about "regional circulation systems" (this also applies to L8).

L132 I'm not entirely convinced about the causality... as many things change across different CMIP6 models. What about, for example, the role of vertical resolution across them?

L143 What about the effects of ENSO on the water vapor? WOuld that relationship also change in the CLAMS version of EMAC?

Figure 4 Would it be possible to also see the lines for EMAC and EMAC-CLAMS in this figure?

Why is 250 hPa the lower boundary chosen for CLAMS? Are results sensitive to this choice?

General recommendation for 5 Appendix -> I think this info should be moved into the main text, as lots of essential information is "packaged" into the Appendix, Since there are no length limitations for this journal that I'm aware of, I'd strong recommend restructuring and move all this info into the main paper.

General comment: while the impact of the diffusive transport scheme is clear and nicely demonstrated, it would be nice if the authors could comment on the role of other features on SWV, such as convective overshooting. Are there any changes in the Monsoon Anticyclone that are also driven "from the troposphere" or do all the differences only originate in the stratosphere?