This article has made valuable contributions by investigating the strong coupled data assimilation (SCDA) of land and atmospheric models. The experiments conducted in this study examined the effectiveness of SCDA in improving the simulation of soil moisture in land models using multiple observational data. The paper has yielded useful findings, e.g., analyzing atmospheric variables by assimilating SM data improves SM analysis and forecasts and mitigates a warm temperature bias in the lower troposphere where a dry SM bias exists; whereas analyzing SM by assimilating atmospheric observations has detrimental impacts on SM analysis and forecasts. These results hold significant implications for future improvements in numerical modeling and forecasting.

The couled models and data assimilation system employed in this paper build upon previous work by the research team, providing a solid foundation. The experimental results demonstrate significant improvements in the representation of soil moisture variables through strong coupling assimilation of soil moisture data. The paper effectively elucidates the reasons behind these improvements through comparative analysis. Additionally, it highlights the adverse effects and their underlying causes when assimilating atmospheric data for SCDA of soil moisture variables. The organization of the paper is reasonable, and the presentation is clear. The appropriate use of figures further enhances the interpretation of the results.

The following are some comments based on my personal observations.

1. The title of the paper, "Comparative Study of Strongly and Weakly Coupled Soil Moisture Data Assimilation with a Global Coupled Land-Atmosphere Model," does not accurately reflect the comprehensive discussion presented in the paper. While the paper does investigate both strongly and weakly coupled data assimilation within the Coupled Land-Atmosphere Model, assimilating both atmospheric and land observations. Furthermore, the paper conducts experiments specifically examining the SCDA/WCDA on SM variable with atmospheric observations and discusses the results. Therefore, the title appears to be somewhat inadequate in capturing the full extent of the paper's content.

2. The paper introduces multiple scenarios to compare the strong/weak coupling assimilation of different model variables with different observational data. Although the authors employ their defined IDs (such as A^x_xL^x_-) to represent these scenarios, as depicted in Figure 2, the representation of these IDs is not particularly user-friendly. The correspondence between the "yes" and "no" labels in Figure 2 and the "x" and "-" symbols in the IDs is unclear. Additionally, the order of A and L in Equation (8) and the naming of cases (c - f) as "one-way SCDA" scenarios may lead to confusion. I suggest using simpler expressions. For reference, (e) could be referred to as "AtoL SCDA", and (f) as "LtoA SCDA". Referring to Penny and Hamill (2017), case (c) wich strongly assimilated only atmospheric observations could be considered quasi-SCDA, while the CTRL scenario would be quasi-WCDA. The (d) scenario is quite distinctive, representing the assimilation of atmospheric variables using both atmospheric and land observations. Therefore, describing it as one-way SCDA would be inaccurate, and a more appropriate abbreviation is required. By adopting more suitable and concise terminology, the reliance on cumbersome IDs can be minimized, leading to enhanced readability.

1. The paper assimilates reanalysis data from GLDAS, and the validation data used for comparison also comes from GLDAS. However, the authors' explanation of GLDAS is not sufficiently detailed. It is unclear whether GLDAS incorporates results from multiple models, and if so, whether the authors only assimilate a subset of those models while using the rest for validation. The mention of "Noah model-based SM data" on line 224 suggests that it could be one of the models within GLDAS. However, it remains uncertain whether GLDAS includes results from other models alongside Noah.

1. In Section 2.2, the authors briefly describe how coupled data assimilation is implemented using the LETKF method. However, this description does not apply to all seven scenarios. In Section 2.2, Equation (8) is used to represent the covariance of the entire system, suggesting that x includes variables from both components. However, in scenarios (a) and (d), x should not include the land component, as it would result in a singular matrix for the covariance P. This should be clarified somewhere in the paper.

2. I am not entirely clear about the purpose of Figure 3. Does it explain the rationale behind assimilating reanalysis data?

1. The reference cited in line 336, Kang et al. 2011, is not listed in the final bibliography.

2. The statement made in line 429 suggests that the advantages of SCDA increase as the peak of precipitation moves northward. However, the results may not demonstrate the extent of this effect as explicitly as indicated by the authors. Could this be quantified using relevant numerical measures, such as the difference in regional average RMSE?

3. Figure 12 shows that the difference between strong and weak coupling is not significant when compared to ERA5, as long as both observations are assimilated. Does this indicate that the earlier results heavily rely on using the same data for assimilation and validation?

The above are just some of my questions and suggestions regarding the paper. Overall, the work conducted in the paper is meticulous and comprehensive. The authors have also provided detailed discussions and future prospects. I support the publication of this paper and hope that the authors will consider my suggestions for minor revisions.

In this study, the authors investigated the effect s of weakly and strongly-coupled DA experiments by assimilating atmospheric observations and SM data simultaneously using a coupled land-atmosphere model. The topic of this paper is interesting, which is expected to provide more evidence for improving NWP by the land-atmosphere DA. However, some detailed descriptions about the DA experiments are missing. Therefore, I would recommend this work for publication after minor revisions. Here are my detailed comments.

1. My main criticism is how to calculate the cross-component error covariance between atmosphere and land variables in the strongly coupled DA (Eq. 8). How to examine the impact of the generation of ensemble samples on the cross-component error covariance or assimilated results?
2. Abstract: More detailed conclusions are missing in the abstract.
3. L280-285: Details about the soil moisture state variables are needed. Since the top two soil layers of MATSIRO are 0-0.05 and 0.05-0.25 meters while 0-10cm for GLDAS, how to set the soil moisture state variables?
4. L256: What is RTPS?
5. Updating SM by assimilating atmospheric observations had detrimental impacts on SM due to spurious error correlations between atmospheric observations and land model variables. However, the error correlation between SM observations and atmospheric model variables is more reliable when updating atmospheric model variables by assimilating SM observations. It is hard to understand this conclusion. More detailed interpretations are needed.