The manuscript presents significant efforts to produce a high-resolution (30 m) dataset of key ecosystem services (ES) in China over 20-year period. This work addresses an important gap in fine-resolution ES data, which is critical for local-scale policy planning and evaluation—such as targeted restoration under China’s Ecological Redline Policy and the Grain for Green Program. The use of established process-based models (CASA, RUSLE, RWEQ, and InVEST) coupled with high-accuracy land cover data (GlobeLand30) is generally appropriate. Validation against existing datasets and limited in situ measurements enhance credibility. However, to strengthen the paper and improve the broader applicability of the dataset, several critical improvements are needed, particularly in methodological transparency, validation rigor, uncertainty quantification, and discussion of limitations.

Specific comments:

1. Please clarify the temporal coverage of the dataset. The title and abstract suggest a continuous 2000–2020 dataset, but results focus on 2000, 2010, and 2020 only. This is misleading if only three years are included. Specify in the title/abstract that data covers decadal intervals. If possible, discuss feasibility of annual data generation or provide suggestions or methods for users to further generate the annual dataset. I recommend making it clear about the spatial resolution of the ES dataset in the title.
2. Please enrich the detailed processing in terms of the methodologies and model parameters. While the models used (e.g., CASA, RUSLE, RWEQ, InVEST) are well-known, key assumptions, parameterization details, and calibration procedures are insufficiently described. For example, the RUSLE model—designed for plot to watershed scales—may overestimate erosion at 30 m due to neglected deposition processes and micro-topographic effects. It is essential to include a table summarizing input data sources, sample sizes, spatial coverage, and parameter values. Sensitivity or uncertainty analyses for critical parameters (e.g., ε_max in CASA) would greatly strengthen the methodology. Furthermore, please address the suitability of each model for high-resolution applications and provide relevant citations to support their use at 30 m.
3. Currently, only net primary productivity (NPP) is validated through cross-comparison with existing datasets. Other ES outputs lack validation, which limits confidence in their reliability. Where possible, incorporate in situ measurements or site-level observed data for validating additional ES variables (e.g., soil erosion, water yield). Multiple open-source NPP datasets are available and should be utilized for more robust validation. Please provide more comprehensive validation for the published dataset.
4. The spatial patterns of ES are well illustrated, but the drivers behind temporal changes (e.g., increased NPP due to afforestation, CO₂ fertilization, or climate influences) are not adequately discussed. Similarly, changes in water yield should be explicitly linked to potential drivers such as urbanization or climate variability. The figures currently rely on qualitative descriptions; adding quantitative summaries—such as provincial averages, standard deviations, and statistical significance tests for change maps—would greatly enhance interpretability.
5. The Discussion section should be expanded to address limitations more thoroughly, including model simplifications (e.g., InVEST’s omission of groundwater processes), data scarcity in remote regions, and remote sensing artifacts (e.g., NDVI distortions due to cloud cover). Additionally, suggestions for future improvements—such as annual dataset generation, inclusion of additional ES (e.g., biodiversity), or integration with socioeconomic data—would help outline pathways for further development and application.

In general, this work represents a substantial effort and provides a valuable baseline for the quantification of ecosystem services across the entire country. However, I recommend that the manuscript be revised before acceptance to improve two major components:

1. Justification of models and data sources: Different models and datasets are used for the production of different ecosystem service maps. A brief justification of how these input data sources and models were selected, and why they are appropriate, would help readers better understand their role in generating the maps.

2. Validation and discussion: Building on the first point, the results and discussion section would benefit from more detailed validation and interpretation of the different ecosystem service products. Given the comprehensiveness of the dataset—covering multiple services and a large geospatial region—such validation could be presented separately for each service, and, if data permit, also stratified by land cover type. This would improve readers’ confidence in the dataset and highlight its applicability.

Given the comprehensiveness of the dataset, it is precisely this breadth and richness of information that calls for transparent communication of its limitations, uncertainties, advantages, and disadvantages. Additional, more specific comments can be found in the attached PDF file.