General comments:

The manuscript by Hao et al. (egusphere-2022-1097) evaluated the performance of E3SM land model in simulating a number of snow parameters, e.g., snow cover fraction, surface albedo, snow water equivalent and snow depth, and snow phenology using in situ, remote sensing, and reanalysis data. The paper presented a comprehensive model evaluation and thoroughly discussed the sources of the model uncertainties and biases. I think the results of the paper are useful for the related researchers to understand the capacity and drawbacks of E3SM land model in snow simulations, and to get insights of how to further improve the model.

Specific comments:

1. I would suggest revising the title of the paper. As this paper mainly evaluated the performance of the E3SM land model in simulating snow processes against a number of observation datasets. Basically, this is not an evaluation of snow processes.
2. L208-209. “The snow accumulation and snowmelt seasons are defined as the periods from September to January and from February to August, respectively.” However, as I know, many regions in the Northern Hemisphere have the peak SWE in February. Please discuss more about the rationality of this definition in snow season.
3. Fig. 3. I notice that the temporal correlations between the simulated and observed snow fractions in winter are obviously lower than those in spring. Please explain the reason. Are they caused by different parameterizations of the model for the two seasons? Normally, it is more challenging for the snow models to simulate the complex snow processes during the melt season. Thus, the results in Fig. 3 are confusing. Pleas add more discussions.
4. Table 2. Some simulated snow variables showed small correlations with some observations (R=-0.2~0.2) but obviously higher correlations with other observations (R>0.5). Please explain the reason.

Technical corrections:

1. L264-265. “The regional average f_sno is 0.41 and 0.15, respectively for spring and winter.” However, Fig. 3 shows winter has higher f_sno than spring. Please recheck whether it is a typo.
2. Fig. 3. I would prefer using blue for areas having more snow and using red for snow-rare areas in figures.
3. Figure captions. I would suggest giving the full names of the variables in the captions, instead of abbreviations.
4. Table 2. It is likely the typesetting of Table 2 is problematic. It is not easy to match the products with the corresponding error metrics. Please improve.

This manuscript evaluated the performance of ELM in simulating snow-related properties across the Western United States. The authors conducted a 50-year offline regional land simulation and evaluated the modeled snow properties against various observational and reanalysis datasets. The experiment is well-designed, and the discussions are well-presented. Such comprehensive model evaluations are valuable for further studies on improving climate simulations, especially for studies based on E3SM.

General comments:

1. The authors should consider revising the title of this manuscript. Snow processes on land imply snow metamorphism, i.e., how snow changes over time. This manuscript focuses more on the accuracy of ELM-simulated snow properties rather than evaluating the ELM snow metamorphism schemes.
2. The authors gathered a lot of observational data to evaluate ELM model simulations. They presented many well-organized figures, including model-minus-observation and their temporal correlations for each snow property. Yet, they need to add more discussions on how these properties interact. For example, is the bias in snow grain size contributing to snow albedo and further influencing SWE and snow-covered fraction? Such discussions are crucial and will help the users to understand the snow simulations in ELM.
3.  Lastly, the authors conducted an offline ELM experiment, presumably for computational efficiency. Would the results differ with coupled simulations considering various snow-related feedbacks?