This manuscript presents a highly relevant and novel approach to mapping soil thickness on the Qinghai-Tibet Plateau (QTP). By departing from the purely empirical and machine-learning paradigms that currently dominate national-scale soil mapping, the authors implement a physically consistent mass balance model. This work addresses a critical bottleneck in Earth system modeling: the structural bias of observational data in remote, high-altitude regions. Overall, this is a worthy effort that offers a robust, physics-driven alternative to pervasive data-driven methods, offering a promising path forward for data-scarce environments. I believe this paper makes a strong contribution to the field. However, several methodological assumptions require deeper discussion before publication.

1) The core of the mathematical solution fundamentally relies on the assumption of a geomorphic steady state. The QTP, however, is a highly transient landscape. While the authors acknowledge this limitation in the discussion, relying on adjustable weighting coefficients to mathematically force a balance may inadvertently mask true mass imbalances across the region. I hope that the authors expand the discussion to thoroughly explore how this mathematical workaround affects the final outputs and the interpretation of the underlying geomorphic mechanisms.

2) The model uses the Effective Energy and Mass Transfer framework to estimate potential weathering rates, which relies strictly on Mean Annual Temperature (MAT) and Mean Annual Precipitation (MAP). Consequently, the model largely overlooks the specific mechanics of freeze-thaw cycles and active layer dynamics. While I recognize the immense difficulty of implementing these complex cryogenic components into the current model structure, the authors must provide a more thorough and critical discussion of this limitation.

3) The model is driven by several external inputs, including a global machine-learning Depth-to-Bedrock product and high-resolution outputs from the Revised Universal Soil Loss Equation and the Revised Wind Erosion Equation. Because these are fundamentally empirical products, they often also fail to capture the complex sediment transport dynamics specific to alpine meadows, freeze-thaw eroded slopes, and discontinuous permafrost zones. The uncertainties inherent in these driving datasets will inevitably propagate into the final solum product. The authors should discuss whether there are methods to quantify, or ideally further reduce, the uncertainties introduced by these foundational inputs.

4) In Section 3.3, the authors perform a Pearson correlation analysis and report MAT and elevation as the primary drivers of the regional thickness gradient. However, because the model's core soil production function is mathematically driven by MAT and MAP in the first place, the final results are practically guaranteed to correlate strongly with MAT and elevation. I recommend reframing this section. Rather than presenting these correlations as independent scientific discoveries about the region, they should simply be framed as an internal validation confirming that the model behaves as parameterized.

Specific comments

1. Please maintain consistency in terminology. The manuscript uses “solum thickness,” “soil thickness,” and “solum depth” interchangeably.
2. Lines 163–165: The text initially states that freeze-thaw erosion is not considered as a separate category, but immediately follows up by describing two main freeze-thaw erosion processes. This creates a contradiction.
3. Lines 160–170: The text repeatedly states that human activities are not considered, which is excessively lengthy.
4. Figure 1: Increase the font size to ensure clarity, and review the formatting of all remaining figures and tables to ensure consistency with the journal’s requirements.
5. Table 2: What is the basis for the range of each parameter in Table 2? Please add supporting references.
6. Lines 272–274: The description of the distribution of cluster 6 is inaccurate.
7. Figure 7: Insert the corresponding legend for soil thickness.
8. Figure 5-8: Please add legend to Fig. 5 and carefully review the captions for Figs. 6–8.

This manuscript presents a new 1 km resolution solum thickness dataset for the Qinghai-Tibet Plateau (QTP) developed using a revised geomorphic mass balance model. The topic is highly relevant and timely. The methodological innovations represent a thoughtful and non-routine adaptation of classical mass balance models to the QTP’s unique conditions. This work has strong potential value for the soil science, cryospheric, and Earth system modeling communities.

Overall, the scientific core has clear merit and novelty. However, several non-trivial weaknesses in validation strategy and uncertainty quantification prevent acceptance in the current form. With careful attention to the major and minor points below, a revised version has a good chance of acceptance in ESSD.

Major Comments

1 The performance assessment relies on 4-fold cross-validation using the same 552 soil profiles that overlap significantly with the data sources used to construct the benchmark maps (Shangguan and Liu). Therefore, the claims of outperforming existing products by 10–17% in RMSE and MRE are weakened by the lack of truly independent test data. Please provide a clearer acknowledgment of this shared-data limitation. Additionally, spatial pattern analysis (e.g., correlation of modeled thickness with topographic curvature, topographic position index, or slope) might be helpful in enhancing assessment.

2 The steady-state assumption is also a critical simplification on a tectonically active and permafrost-degrading plateau. While limitations are discussed, there is no quantitative propagation of uncertainty into the final dataset. I suggest the authors can add some sensitivity analysis. For example, vary key parameters ±20–50% or using Monte Carlo sampling for weighting coefficients and present spatial uncertainty estimates, then discuss how uncertainty varies across clusters or geomorphic zones.

Minor Comments

Abstract

• Lines 24–25: Specify the metric(s) in the performance claim (e.g., “by approximately 10–17% in RMSE and MRE”).

Study Area and Data

• Line 105: Briefly mention known limitations of the Shangguan et al. (2017) DTB dataset in the QTP interior when it is first introduced.

Methods

• Lines 165–170: Justification for excluding human activities is too brief. Note that effects are assumed negligible at 1 km resolution.
• Table 1: Move “Parent Material of Soil Formation” to a new “Geological” category.

Results

• Table 3: Add “(–)” for the unitless NDVI column.
• Table 4: Replace “-” for Cluster 8 RMSE with “N/A (n=4 insufficient for cross-validation)”. Add a footnote explaining n.
• Fig. 3: Consider adding a histogram inset or basin-specific summary statistics.
• Lines 330–340: explain why curvature shows near-zero correlation at plateau scale yet drives local differentiation in the model.

Discussion and Comparison

• Table 5: Explicitly note the shared validation data limitation. explain the higher MAE.
• Lines 470–485: Strengthen discussion of potential RUSLE/RWEQ biases in permafrost areas and suggest future use of cryogenic-specific erosion models.
• Lines 500–510: Rephrase the sentence about weighting coefficients mitigating steady-state bias to “partially mitigated”.