General comments

This manuscript extends the widely-used benchmarking tool ILAMB to include soil moisture and uses it to perform a comprehensive evaluation of soil moisture Earth System Models from CMIP6, both near the surface and to a depth of 1m. It represents a substantial contribution to the field. The evaluation is extensive and reproducible; of particular note is that a variety of different metrics are used, including the relationship between soil moisture and other variables. The presentation quality is high throughout.

Specific comments

Add some sentences mentioning that soil moisture below 1m can be important too (particularly in areas with deep roots), even though evaluating it is beyond the scope of this analysis.

Line 160: this is an important point you make here – refer back to it in section 3.2

Line 212-3 This statement needs citations. Even better would be to add a brief justification of this statement and mention the limitations of these datasets, so that the reader can bear these in mind when interpreting your results.

Line 249 Add a description of how Overall Score is calculated from Bias Score, RMSE Score, Seasonal Cycle Score, and Spatial Distribution Score. Add few words to clarify the “Seasonal Cycle Score” and “Spatial Distribution Score”.

Line 272: I couldn’t see anywhere what time resolution was used to calculate the Taylor plots (e.g. annual, monthly or daily?). Same for fig. 6 and 7.

Line 299: what do you mean by “parameters that mask deficiencies in SM representation”? Maybe add a clarifying phrase, or an example.

Line 300: I disagree with this statement because I would characterize transpiration as a vegetation-related process. The representation of water flux through the canopy and carbon flux through the canopy typically both rely structurally on very similar parts of the code.

Line 319: “However, it is important to note that the ILAMB spatial climatology used in Figures 6 and 7 may be affected by ESA-CCI’s inconsistent spatiotemporal coverage” I don’t understand this sentence – what does the phrase “ILAMB spatial climatology” mean here?

Technical corrections

Line 43: consider replacing “, most commonly relying” with “. The majority rely” (the existing sentence has an ambiguity about whether “most” refers to “ESMs” or “commonly” which hampers the sentence flow)

Line 45: consider replacing “by incorporating” with “using” or “incorporating it into”

Line 121: the minus sign in both units needs to be in the superscript

Consider putting mrsol and mrsos in monospace font, given that it is the name of a variable

Equation 1: It is more standard to not use italics if the variable contains more than one letter i.e. consider making all letters non-italic apart from n, l, z, w, ρ. Consider renaming mrsol in equation 1 with something briefer e.g. m_SM, θ_m

Line 137: put the numbers in the volumetric SM units into superscript

The paper extends ILAMB to benchmark soil moisture (SM) in CMIP6 Earth System Models, evaluating both surface and root-zone depths and relating SM skill to ecohydrologic variables (ET, GPP, LAI), including climate-region analysis. It is a great effort to conduct this type of benchmarking, but I hope to see a more thorough description and interpretations of the differences in performance among these ESM models to help guide model improvement.

Then, the motivation to relate soil moisture (SM) skill to ecosystem functioning is well taken, but the choice of GPP and LAI as primary ecohydrological variables raises concerns regarding interpretability. GPP and LAI are highly integrated, bulk ecosystem variables that reflect multiple interacting processes beyond soil moisture. As a result, discrepancies in SM–GPP or SM–LAI relationships cannot be straightforwardly attributed to deficiencies in soil hydrology, right? In contrast, hydrological variables more directly linked to soil moisture dynamics—such as runoff, drainage, evaporation and transpiration partitioning, and canopy interception—would provide a clearer and more mechanistic pathway for diagnosing soil-moisture–related model behaviour. Including such variables would strengthen the causal interpretability of the proposed benchmarking framework and its relevance for guiding land-model development.

A further limitation of the current benchmarking framework is the lack of systematic documentation of key land-surface input differences among the evaluated models. In particular, it is unclear whether—and how—soil texture, soil hydraulic parameters, rooting depth distributions, or lower boundary conditions differ across models. Because soil moisture dynamics are highly sensitive to these prescribed or semi-prescribed inputs, benchmarking model outputs without explicitly accounting for such differences makes it difficult to attribute performance differences to model structure rather than to input choices. Providing a concise overview of relevant soil and hydrological inputs for each model, or at least discussing their expected variability and implications, would improve the interpretability of the results and strengthen the paper’s ability to inform model development.

Some Detailed comments:

P2 L45-47: It needs more details about the ILAMB framework and why it suits evaluating ESMs.

P3, l92-94: The authors first augue the drawback for the Qiao et al., 2022 data is that they used reanalysis data as reference to evaluating the SM, but instead here, they said they used the Wang et al., 2021 data which is averaged across both observational and also model simulation outputs, why the Wang et al., 2021 data is better that the reanalysis data?

PA6, section 2.3, please describe why you chose these different GPP, LAI and ET products

Move Figure 3 to the Results section.

L250-254, you don’t need to describe what each figure is about, but focus on what you found from different figures.

Section 3.1 did not provide a clear summative description of the overall benchmark score. The description of what each figure is about is not what you have have in the Results sections

L299-301: unclear to me why it is important to benchmark GPP and LAI in this study. Yes, they are linked to the SM, but their variations are also affected by many other factors.