Reviewer #1 (Comments to Author (shown to authors):

In general it’s a thoughtful analysis of an important issue, the effect of El Niño on tidal oscillations in the mesosphere and lower thermosphere (MLT). It may be that this paper is more relevant to Annales Geophysique since, while it contains some detailed analysis of tropospheric effects, its ultimate intent is to understand the MLT. However, there is a larger issue- that of novelty. A recent paper by Vitharana et al., (JGR, 2021, doi:10.1029/2021JA029588) quite clearly states and demonstrates the anti-correlation between DW1 and El Niño. 

And they both attribute similar causes. Thus compare: Vitharana ”due to changes in tropospheric forcing” vs.  Cen “heating rates in the tropical troposphere”. And both analyze SABER data. 

Certainly there are areas where Cen’s analysis can be deeper than Vitharana so Vitharana should not be considered the last word. For example, while the negative correlation is now established, there are the relative roles of different components of the effect (heating, filtering by stratospheric winds, GW forcing) that the present work can contribute. Furthermore, Vitharana appears to misquote Pedatella and Liu, 2012 by saying that their results are consistent with that older reference. When in fact, I concur with the present authors in saying that Pedatella and Liu reached the opposite conclusion. But this present submission should be reworded and re-oriented to be following Vitharana’s analysis. This probably means more work on “fleshing out” the details of the causes, for example the GW effect (which seems pretty clear in Figure 6). Their conclusions presently seem more like a simple listing- but I think they could, and should, give more information on the relative importance- perhaps one cause is more important at one altitude for example? (relevant to 4th and 5th bullets below)

1.I do not see where Ramesh showed a positive correlation between MLT DW1 and El Niñoas stated on lines 94-95. Ramesh had lots of “predictors” and it wasn’t clear what was forcing what. Perhaps the authors could clarify if I’ve missed something.

Response: Thanks for your suggestion. As added in lines 101-105 in the revised manuscript, “As suggested by the WACCM version 6 simulations with self-generated QBO and ENSO, Ramesh et al. (2020) illustrates the linear response of latitude‐pressure variation of DW1-T to the seven predictors including ENSO in four seasons. They suggest that the response of DW1 to ENSO is significantly positive in the equatorial MLT region during the NH winter (Figure 5 in Ramesh et al. 2020).” 

Figure R1 (Figure 5 in Ramesh et al., 2020): The seasonal variation of latitude‐pressure distribution of ΔT24 responses to Niño3 averaged for three WACCM6 realizations. The responses in stippled regions are not significant at the 95% confidence level (p > 0.05). Contour intervals = 0.05 K/K, (The fourth row of figure 5 in Remash et al., 2020).

2.There is not a clear statement as to what SABER shows for the overall structure of the tide compared to WACCM. Do the authors agree with Vitharana? In which case, they can just state that, but also refer to the relevant figure in Vitharana. This is relevant to the 4th bullet below.

Response: To compare the distribution of the climatological mean DW1 tidal amplitude of SD-WACCM simulation with SABER observation, Figure R2 (Figure 1 in the revised manuscript) shows the average DW1 temperature amplitude in SABER observation and SD-WACCM during the winter from 2002 to 2013. The boreal winter (December-January-February, DJF) mean amplitude of DW1 temperature is the largest (~12 K) in the equatorial mesopause region extracted from TIMED/SABER observation. Although the mean amplitude in SD-WACCM is weaker than that in SABER, the distribution of the DW1 T amplitude in SD-WACCM simulation is quite similar to that derived from SABER observation, with the maximum at 90-100 km above the equator. There are some differences between SABER and SD-WACCM: SABER has a weaker peak above the equator at 70-80 km, but this peak cannot be seen in SD-WACCM. The interannual variation of the equatorial MLT DW1-T amplitudes weighted by phase difference in SABER and SD-WACCM also agree well during the northern winter of 2002-2014 (Figure 2a and 2b in the revised manuscript).

Figure R2 (Figure 1 in the revised manuscript). (a) The average DW1 temperature amplitude of SABER observation during 2002-2013 winter (DJF, Dec-Jan-Feb). (b) the same as (a), but for SD-WACCM.

     As mentioned above, Vitharana et al. (2021) have stated that the MLT DW1 negatively responds to ENSO by calculating the multi-linear regression for all the months from 2003-2016. However, it is noted that the responses of MLT DW1 tide vary quite a bit to ENSO among different seasons (e.g. Zhou et al., 2018; Kogure et al., 2021). Thus, calculating the regression by binning the data among different months together may underestimate the actual response of MLT DW1 tide during the particular season due to the masking between different responses with each other. Indeed, the negative responses of MLT DW1 to ENSO during the winter are approximately five times larger than those estimated from all the months' data (Vitharana et al., 2021), while the location of the most significant responses is also different. It should be pointed out that the MEI indices were "normalized" by setting the maximum value to 1 in Vitharana (2021), which is different from the method we used (set the standard deviation to 1). If we also adopt the same method as Vitharana et al. (2021) did, the regression coefficients of MLT DW1 amplitude response to ENSO in winter (Figure R3) is over -3 K per index at 90 and 100 km, while the maximum of regression coefficients in Vitharana (2021) is only -0.6 K per index at the same region. The comparison between our results with those of Vitharana et al. (2021) has been rewritten in lines 82-93 and 248-253 in the revised manuscript.

Figure R3. The linear regression coefficient of normalized Niño3.4 in SABER (a) and SD-WACCM (b) DW1-T. The contour interval is 0.6 K for SABER and 0.3 K (N3.4) for SD-WACCM. Red represents positive response and blue represents negative; the gray regions denote confidence levels below 95% for F-test.

3. I notice the authors use WACCM4, not WACCM6 which is the latest. While this is probably acceptable, they should at least note this and offer any comments on possible differences. For example, WACCM6 uses a self-consistent QBO (which might allow for better characterization of feedbacks?) and a different (better?) GW scheme as well as higher spatial resolution.

       a. I’m not sure I fully understand line 249, but it does seem to speak to the question of feedbacks between QBO and ENSO which, if so, is relevant to the question of the WACCM model version number. Can they clarify?

Response: Thanks for your suggestion. WACCM is a high-top model that can be used as the atmospheric component of the Community Earth System Model (CESM1) of the National Center for Atmospheric Research. WACCM4 is based on the Community Atmosphere Model, version 4 with the vertical model domain extended to ~145 km. 

The “Specified dynamics” version of WACCM4 is based on WACCM4 and nudged to meteorological fields from Modern-Era Retrospective Analysis for Research and Applications (MERRA) reanalysis data in the troposphere and stratosphere (from the surface up to 1 hPa) and then is freely run in the MLT (above 0.3 hPa) (Kunz et al., 2011). With the relaxation, the atmospheric variables such as QBO are consistent with the reanalysis in the troposphere and stratosphere. 

WACCM6 is the latest version of WACCM in CESM2, transitions to higher horizontal and vertical resolution, which can simulate finer structures in the atmosphere. WACCM4 includes a representation of the quasi-biennial oscillation (QBO), achieved by relaxing equatorial zonal winds between 86 and 4 hPa to observed interannual variability (Marsh et al. 2013). Compared with WACCM4, the QBO in WACCM6 is self-generated (Gettelman et al. 2019). WACCM6 is able to simulate, albeit imperfectly, the QBO, the equatorial zonal wind propagation, which in observations has an average period of ~28 months. Similar to WACCM5 (Mills et al., 2017), WACCM6 simulates a reasonable QBO with 70 levels in a free running configuration.

Both WACCM4 and WACCM6 can be run fully coupled to active ocean and sea ice model components or use specified SST (Marsh et al., 2013). Pedatella & Liu (2012 and 2013) utilize the WACCM4 with self-generated ENSO and no QBO signal, while Ramesh et al. (2020) use the WACCM6 simulations with self-generated ENSO and QBO. Different from these two simulations, the SST which follows the observation is prescribed in SD-WACCM. And the SD-WACCM atmospheric variables such as QBO are consistent with the reanalysis in the troposphere and stratosphere.

As suggested by Figure 5 in Ramesh et al. (2021), the equatorial MLT DW1 positively responds to ENSO in the northern hemispheric winter, which is the opposite of that in the SABER observations and SD-WACCM simulations. The difference in generating ENSO and QBO among different versions of the model could play a role in the divergence in the ENSO-DW1 relationship.

Two main adjustable parameters in the frontal gravity wave source specification have been changed since WACCM4 due to the increased horizontal resolution in WACCM6: The frontogenesis threshold in WACCM6 is set to 0.108 K² (100 km)⁻² ·h⁻¹ and the source stress of frontally generated waves is set to τ_b=3×10⁻³ Pa (Gettelman et al. 2019). In WACCM4, the frontogenesis threshold is set to 0.045 K² (100 km)⁻² ·h⁻¹ and the source stress of frontally generated waves is set to τ_b=1.5×10⁻³ Pa. Different parameterization schemes and resolutions may have some effects on DW1-ENSO relationship. However, the completely opposite DW1-ENSO relationship in SD-WACCM and WACCM6 should not be simply attributed to the gravity waves. Indeed, DW1-ENSO relationship in WACCM4 are similar to those in WACCM6 (Pedatella & Liu, 2012 & Ramesh et al., 2021). Other factors which differ in different version of WACCM could affect the DW1-ENSO connection, more investigation and comparison between models are needed to determine the effect of the gravity wave parameterization is in the future work.

4. The effect of R on DW1 seems to maximize at latitudes below the peak of the DW1 (reference is to Figure 5 but this is where a statement or a figure as to the overall structure of DW1 would be helpful). As a result, I wonder whether it is really relevant. Or at least not at the peak- this is where going beyond a simple listing of causes could be useful.

Response: Thanks for the suggestion. The ratio of the absolute and planetary vorticity R is equivalent to changing the planet rotation rate. In classical theory, the vertically propagating DW1 is restricted near the equator due to the planet's rapid rotation. Therefore, a faster rotation rate (positive R anomalies) will suppress the latitudinal band (i.e., waveguide) where DW1 can propagate vertically. On the other hand, the slower rotation rate (negative R anomalies) favors the vertical propagation and is thus able to enhance the amplitude of DW1 at the low latitudes (Mclandress. 2002b). When the ratio of the absolute and planetary vorticity R-value at a certain height becomes larger, the upward propagation of tide is suppressed, which lead to weaker  tides above there. We modified Figure R4 (Figure 6 in the revised manuscript) to show significant areas of the multivariate linear regression (MLR) coefficient of R on Niño 3.4. The green thick solid line represents the sum of the equatorial ratio of the absolute and planetary vorticity R values (15-30°N and 15-30°S), and the thick lines indicate the area where the regressed coefficients are significant. The mean R value (15-30°N and 15-30°S) response to ENSO is significantly positive at 60-90 km, which would lead to the suppressed propagation of DW1 above these areas.

Figure R4 (Figure 6 in the revised manuscript). The anomaly of the ratio of the absolute and planetary vorticity, δR. The thin, dashed red, blue and green lines denote the averages of the Northern Hemisphere (from 15°N to 30°N), Southern Hemisphere (from 15°S to 30°S) and the whole (15-30°N and 15-30°S) , respectively. The thick, solid lines denote confidence levels below 95% for the F test.

5. In general, I think the GW analysis could use more detail. Overall, I think it’s believable, but I would like more information- specifically I think they should put more effort on teasing out the effects of source forcing and filtering that they allude to in lines 377 and 378. If the gravity waves in WACCM are linked to convection, then shouldn’t they be able to quantify the change in GW forcing more rigorously? Presumably there are certain phase speeds which are more or less relevant here?

Response: Thanks for your suggestion. In this study, the GW analysis with respect to the response of diurnal tide to ENSO is based on the parameterized GWs drag from the “specified dynamics” WACCM. In the standard setup of SD-WACCM4, the gravity wave source spectrum includes wave components with phase velocities in the range from -80 to +80 m s^-1, at intervals of 2.5 m s^-1 (Beres et al. 2005). In SD-WACCM, the GW drags are separated with respect to different excitation sources, while the detailed information such as the phase speed are not available in the model output.

The discussion between GW drag generated by frontal systems and convection has been added in lines 387-388 in the revised manuscript as “The GW in the tropics is primarily induced by the convection, while the GW in the middle to high latitudes is mainly generated by the frontal systems (Figure S5, S6).”.

Response to the comments are also presented in the  pdf file as supplementary.

Reference

Beres, J., Garcia, R. R., Boville, B. A., & Sassi, F. (2005). Implementation of a gravity wave source spectrum parameterization dependent on the properties of convection in the Whole Atmosphere Community Climate Model (WACCM). Journal Of Geophysical Research-Atmospheres, 110, D10108. http://doi.org/10.1029/2004JD005504

Gettelman, A., Mills, M. J., Kinnison, D. E., Garcia, R. R., Smith, A. K., Marsh, D., … Randel, W. (2019). The Whole Atmosphere Community Climate Model Version 6 (WACCM6). Journal Of Geophysical Research: Atmospheres, 124, 12380-12403.

https://doi.org/10.1029/2019JD030943

Kogure, M., & Liu, H. (2021). DW1 tidal enhancements in the equatorial MLT during 2015 El Niño: The relative role of tidal heating and propagation. Journal of Geophysical Research: Space Physics, 126, e2021JA029342. https://doi.org/10.1029/2021JA029342

Kunz, A., Pan, L. L., Konopka, P., Kinnison, D., & Tilmes, S. (2011). Chemical and dynamical discontinuity at the extratropical tropopause based on START08 and WACCM analyses. Journal Of Geophysical Research, 116, D24302. https://doi.org/10.1029/2011JD016686

Marsh, D., Mills, M., Kinnison, D. E., & Lamarque, J. -F. (2013). Climate change from 1850 to 2005 simulated in CESM1(WACCM). Journal Of Climate, 26, 7372-7391.

https://doi.org/10.1175/JCLI-D-12-00558.1

Mills, M. J., Richter, J. H., Tilmes, S., Kravitz, B., MacMartin, D. G., Glanville, A. S., … Kinnison, D. E. (2017). Radiative and chemical response to interactive stratospheric sulfate aerosols in fully coupled CESM1(WACCM). Journal Of Geophysical Research: Atmospheres, 122, 13061-13078. https://doi.org/10.1002/2017JD027006

Ramesh, K., Smith, A. K., Garcia, R. R., Marsh, D. R., Sridharan, S., & Kishore Kumar, K. (2020). Long‐term variability and tendencies in migrating diurnal tide from WACCM6 simulations during 1850–2014. Journal of Geophysical Research: Atmospheres, 125, e2020JD033644. https://doi.org/10.1029/2020JD033644

Vitharana, A., Du, J., Zhu, X., Oberheide, J., & Ward, W. E. (2021). Numerical prediction of the migrating diurnal tide total variability in the mesosphere and lower thermosphere. Journal of Geophysical Research: Space Physics, 126, e2021JA029588. https://doi.org/10.1029/2021JA029588

Zhou, X., Wan, W., Yu, Y., Ning, B., Hu, L., and Yue, X. (2018). New approach to estimate tidal climatology from ground‐and space‐based observations. Journal of Geophysical Research: Space Physics, 123, 5087– 5101. http://doi.org/10.1029/2017JA024967

Reviewer #3 (Comments to Author (shown to authors):

1. I agree with the first reviewer that this paper should compare with Vitharana et al. (2021), where the negative correlation between DW1 and ENSO is clearly established, Vitharana also used SABER data.

Response: The comparison between our results with those of Vitharana et al. (2021) has been rewritten in lines 82-93 and 248-253 in the revised manuscript.

2. The significance of this paper lies in the physical mechanisms. I am surprised to see all three sources (tropospheric heating, wind filtering and gravity waves) are pulling in the same direction, making DW1 amplitude smaller in the El NiñoFor tropospheric heating, the Hough mode analysis is well done. The tropospheric heating however presents positive and negative correlations with DW1 amplitude at different heights and latitudes. The authors averaged the heating between 0-16 km and 35 N and 35 S and found there is an overall decreasing heating rate during El Niño. How do you justify the choice of the altitude and latitude range? Apparently, if you calculate the correlation with a different range, you can get a totally different conclusion.

Response: Thanks for your comment. According to the tidal theory, Not only HR near the equator that affects DW1 hough (1, 1) but the heating in global troposphere. For example, the major heating for H₂O in equinox is associated with the symmetric (1, -2) mode, and the (1, 1) and (1, -4) symmetric modes are excited with about equal strength with amplitudes ~20-25% of the (1, -2) heating rates (Volland and Hans, 1988; Forbes 1982). Therefore, it is more reasonable to calculate the mass weighted heating rate covering all the area of tropical troposphere (35°S-35°N, 0-16 km) to investigate the effect of tropospheric heating on tides. The correlation coefficient between 15 km DW1-T and the mass weighted HR of the whole tropical troposphere (35°S-35°N, 0-16 km) is 0.45 (significance at the 95% level according to the Student’s T test). The correlation between DW1-T and the HR over the whole tropical troposphere is higher than those between DW1-T and the HR over the regions suggested by previous studies (Table 2). Although the correlations with MLT DW1 tide are weaker or even insignificant, the HR averaged over different region as selected by previous studies also suggests negative response during the El Niño winters.

The comparison between HR calculated in different areas has been discussed in lines 328-343 in the revised manuscript.

Table 2 in the revised manuscript: The correlation coefficient between the DW1 T amplitude at 15 km and the mass-weighted HR in different areas during the winters of 1979-2014. The bold numbers indicate that the correlation coefficients are significant at the 95% level. The MLR coefficient on the normalized Niño3.4 index (10^-3 mw m^-3 index^-1) is also exhibited.

3. Similar scenario happened to R, the range is chosen between 15 and 35 degrees in each hemisphere, how is this range chosen? does the conclusion change if a different range is chosen? R is positive and negative several times below the MLT, how does that affect DW1 propogation? 

Response: Thanks for your suggestion. To investigate the effect of the wave guide on the upward propagation of tide near the equator, the R near the tropics (15-30°) are considered. However, Since f tends to 0 near the equator, the  tends to infinity. The R over the similar range are also adopted to McLandress et al. (2002b) and Wu et al. (2017).

We modified Figure 6 to show the significance of the regressed R on Niño index. The green thick solid line represents the mean value of the ratio of the absolute and planetary vorticity (R) at the subtropics (15-30°N and 15-30°S), and the thick lines indicate the area where the regressed coefficients are significant at 95% level. The R response to ENSO is positive at 60-100 km in the northern subtropics and 65-100 km in the southern subtropics, which would suppress the upward propagation of the DW1 tide in the mesosphere and contribute to the negative response of the DW1 tidal wind.

Liu (2015) chose 10-40°N (S) to calculate the ratio of the absolute and planetary vorticity R. Figure R2 shows the R value between 10-40°N (S). It can be seen that only the mean R values at 60-80km are continuously significant, which is consistent with figure 6 (15-30°).

Figure R1 (Figure 6 in the revised manuscript). The anomaly of the ratio of the absolute and planetary vorticity, δR. The thin, dashed red, blue and green lines denote the averages of the Northern Hemisphere (from 15°N to 30°N), Southern Hemisphere (from 15°S to 30°S) and the whole (15-30°N and 15-30°S) , respectively. The thick, solid lines denote confidence levels below 95% for the F test.

Figure R2. The linear regression coefficient of normalized Niño3.4 in δR (the anomaly of the ratio of the absolute and planetary vorticity). The thin, dashed red, blue and green lines denote the averages of the Northern Hemisphere (from 10°N to 40°N), Southern Hemisphere (from 10°S to 40°S) and the whole (10-40°N and 10-40°S), respectively. The thick, solid lines denote confidence levels below 95% for the F test.

4. I also agree with reviewer 1 that the third mechanism about gravity wave drag needs further investigation. I don't understand why the correlation between gravity wave drag and DW1 is negligible or even negative while the correlation between gravity wave forcing and DW1 is positive in the MLT region at all latitudes.

Response: Thanks for your suggestion. When gravity wave drag variability and tide tendency are close to orthogonal, the correlation between the two can be ignored, but after calculating the phase weight, the orthogonal changes of the two have been excluded, and only the gravity wave drag variability and the tide tendency in the same phase are calculated. So gravity wave forcing and tide may be significantly correlated, albeit at a time when gravity wave drag is less correlated with tide.

Similarly, if the GW drag in the same direction as the tidal tendency increases, which lead the tidal amplitude increases, then gravity wave forcing is positively correlated with the tide. But if the GW drag that is orthogonal to the tidal tendency decreases by a large margin in the same time, it is found that GW drag is negatively correlated with tide, although GW drag which is orthogonal to the tidal tendency does not actually change tide strength directly.

We have included more detailed analysis of GW drag in the main text, please see lines 379-407 in the revised manuscript.

Reference

Forbes, JM (1982). Atmospheric tides 1. Model Description and Results for the Solar Diurnal Component. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 87, NO. A7, PAGES 5222-5240,

https://doi.org/10.1029/JA087iA07p05222

Kogure, M., & Liu, H. (2021). DW1 tidal enhancements in the equatorial MLT during 2015 El Niño: The relative role of tidal heating and propagation. Journal of Geophysical Research: Space Physics, 126, e2021JA029342. https://doi.org/10.1029/2021JA029342

Lieberman, R. S., Ortland, D. A., & Yarosh, E. S. (2003). Climatology and interannual variability of diurnal water vapor heating. Journal of Geophysical Research: Atmospheres 108(D3): https://doi.org/10.1029/2002jd002308

McLandress, C. (2002b), The seasonal variation of the propagating diurnal tide in the mesosphere and lower thermosphere. Part II: The role of tidal heating and zonal mean winds, J. Atmos. Sci., 59(5), 907–922, https://doi.org/10.1175/1520-0469(2002)059<0907:Tsvotp>2.0.Co;2.

Pedatella, N. M., & Liu, H. L. (2013). Influence of the El Niño Southern Oscillation on the middle and upper atmosphere. Journal of Geophysical Research: Atmospheres 118(5):2744–2755, https://doi.org/10.1002/Jgra.50286

Vitharana, A., Du, J., Zhu, X., Oberheide, J., & Ward, W. E. (2021). Numerical prediction of the migrating diurnal tide total variability in the mesosphere and lower thermosphere. Journal of Geophysical Research: Space Physics, 126, e2021JA029588. https://doi.org/10.1029/2021JA029588

Wu, Z., T. Li, and X. Dou (2017), What causes seasonal variation of migrating diurnal tide observed by the Mars Climate Sounder?, J. Geophys. Res. Planets, 122, http://doi.org/10.1002/2017JE005277.

Zhang, X., Forbes, J. M., & Hagan, M. E. (2010). Longitudinal variation of tides in the MLT region: 1. Tides driven by tropospheric net radiative heating. Journal of Geophysical Research: Space Physics , 115, A06316, https://doi.org/10.1029/2009JA014897.

Zhou, X., Wan, W., Yu, Y., Ning, B., Hu, L., and Yue, X. (2018). New approach to estimate tidal climatology from ground‐and space‐based observations. Journal of Geophysical Research: Space Physics, 123, 5087– 5101. http://doi.org/10.1029/2017JA024967

Reviewer #2 (Comments to Author (shown to authors):

This manuscript investigated the migrating diurnal tidal variability in the mesosphere and lower thermosphere due to the El Niño-Southern Oscillation, and the driving mechanism of this variability. This is one of the important issues related to the interannual variability in the MLT region. The authors showed the significant negative correlation between the residual of diurnal tidal amplitude in the MLT and the Niño3.4 index, and attributed this diurnal tidal variability to its tropospheric source forcing change, background wind effect, and the modulation of the gravity wave drag. Although this paper included some interesting results, overall I think that the paper only has decent scientific progress since it is already well established of the negative correlation between the SOI/Niño3.4 index and the DW1 amplitude in the MLT region. The analysis is a good start point but I think the results presented herein are incomprehensive. Additional analysis with deeper informative results is needed to justify publication in ACP. I will indicate a major revision for this manuscript and think this manuscript can make an excellent contribution after major revision.

Data and method

1. The archived model data from latest WACCM 6 and SD-WACCM-X version 2.1 runs are both publicly available on CESM website, with significant change from previous version. The authors should provide reasons why they chose an older version of model output.

Response: Thanks for your suggestion. The latest version of WACCM (WACCM6) has been adopted to investigate the DW1 tide with different predictors, including ENSO by Ramesh et al. (2020). The MLT DW1 tidal T is suggested to be a significantly negative response to Niño 3.4, which is, however, opposite to the negative DW1-ENSO relationship suggested by SABER observations. The different DW1-ENSO relationship between different versions of WACCM simulation may be attributed to the changed scheme utilized in generating ENSO and QBO (ENSO and QBO are self-generated in WACCM6 simulated by Ramesh et al. (2020), while are nudged to MERRA2 below 50 km in the SD-WACCM4) and the associated atmospheric variation. As a result, the variation associated with tidal excitation or propagation may not follow reality.

As WACCM-X is built upon the chemistry, dynamics, and physics of CAM4 and WACCM4, the tidal forcing and the middle atmospheric variability in SD-WACCM-X follow that in SD-WACCM4 below the thermosphere. Thus, a similar response of MLT tide to ENSO should be expected. However, on the CESM website, there are neither parameterized tidal variables nor the averaged variables with a time resolution of less than one day in the datasets of SD-WACCM-X version 2.1. Both CAM and WACCM have seen their own significant recent developments, including increased horizontal resolution. While CAM6 and WACCM6 have been released as part of CESM 2, WACCM-X will incorporate the recent improvements in the lower and middle atmosphere components of CESM in the future versions. (Liu et al., 2018). Given the agreement with SABER observations and the availability of data, the simulation from SD-WACCM4 is adopted in this study to investigate the mechanism how ENSO could modulate the MLT DW1.

Tidal forcing

2. The author stated that the amplitude and phase of DW1 in the MLT could be potentially modulated by the ENSO and used a DW1 vector amplitude to combine their anomaly related to the Niño3.4 index. I think it will be better to assess the ENSO impact on the DW1 amplitude and phase separately.

Response: Thanks for your suggestion. Figure R1 shows the average DW1 temperature amplitude of SABER observation during 2002-2020 winter (Figure R1a) and the climatology average DJF DW1-T phase (Figure R1b). Figure R2 shows the SABER DW1-T amplitude and phase anomaly during El Niño winter. The amplitude of DW1 in the equatorial region is significantly reduced, while the phase anomaly is not obvious (less than 1 hour in most areas) during El Niño winter. 

The discussion assesses the ENSO impact on the DW1 amplitude and phase separately has been discussed in lines 245-247 as “The amplitude of DW1 in the equatorial region is significantly reduced, however the phase anomaly is not drifted much (less than 1 hour) during El Niño winter. (figure S1, S2)” in the revised manuscript.

Figure R1 (Figure S1 in the revised supplement). (a) The average DW1 temperature amplitude of SABER observation during 2002-2020 winter (DJF, Dec-Jan-Feb). (b) the same as (a), but for phase.