| 21 Aug 2025
Improving Snow Simulations through Improved Representations of Vegetation Conditions: Insights from High Resolution Simulations over California
Abstract. Most hydrological models underestimate snow water equivalent (SWE) in the mountainous western US. Key limitations may be due to coarse resolution precipitation input and inadequate representations of snow-vegetation interactions. Vegetation affects snow dynamics through snow interception, throughfall/unloading, and energy transfer through the canopy, yet basin-scale studies on the vegetation effects are limited. To address this issue, we applied the Noah-MP version 5.0 with a dynamic vegetation module to the Sacramento and San Joaquin River Basins in California at 1 km resolution driven by forcing from Analysis of Record for Calibration (AORC) and NLDAS-2, downscaled to 1 km using the WRF-Hydro Meteorological Forcing Engine. We carried out eight model experiments driven by the two forcing datasets and two vegetation schemes over the two Basins. Compared to 4 km PRISM and 1 km AORC, the downscaled 1 km NLDAS-2 forcing data shows higher precipitation and lower temperatures over the mountains, resulting in more snowfall and SWE, which is more consistent with in-situ SWE observations. Using the 1-km NLDAS-2 forcing, the default vegetation scheme with prescribed leaf area index (LAI) and vegetation cover fraction produces too much SWE on the ground due mainly to the strong canopy shading effect despite more snow intercepted by the canopy. The dynamic vegetation module produces smaller LAI than the prescribed (more consistent with MODIS data) and thus leads to more shortwave radiation reaching the snow surface, thereby enhancing episodic melting during the accumulation season. Validation against in-situ snow data suggests that the use of dynamic vegetation model and the downscaled NLDAS-2 data performs the best. These findings highlight the importance of vegetation effects and downscaling of atmospheric forcing to increase the accuracy of snow modeling over the Southwest US.
The paper by A. Gupta et al. addresses the representation of vegetation conditions and their effects on snow simulations based on high-resolution modeling over California. Several model experiments were conducted using different forcing datasets and vegetation schemes. While the paper contains several elements of interest to The Cryosphere readership, it currently does not sufficiently highlight its main contributions, and substantial revisions are required. Below, I outline my main concerns, followed by line-by-line comments.
Main comments
My main concerns are related to the evaluation data. The authors rely heavily on gridded products for model evaluation, which they treat as ground truth. For instance, a 4 km daily gridded climate product is insufficient to evaluate the downscaled NLDAS-2 data, because it cannot resolve topographic gradients and local effects and introduces its own biases. Station observations (e.g., GHCN-d) are available and should be used to validate the downscaling and provide many evaluation points within the study region.
A similar issue applies to the UA-SWE product: please include an error analysis against actual SNOTEL sites within the study region, rather than validating against a gridded or assimilated product. Section 3.2 should be revised for clarity-it is currently not obvious which datasets are used for which evaluation. I recommend adding a summary table listing, for each variable, the datasets used for validation.
Regarding the evaluation of SWE at the selected CDWR station in the main manuscript: why was this specific station chosen? While I acknowledge that results from other stations appear in the supplementary material, it would be valuable to present more than a single station in the main text.
For the first part of the results (meteorological downscaling): since the downscaling was not performed or developed by the authors, it should not be presented as a primary result. While such an evaluation is important context, it would be more appropriate in an appendix or supplementary section, serving as supporting material for the discussion. Moreover, the assessment itself is rather limited (mostly qualitative or daily means); please expand it to include more quantitative analysis.
The discussion is extremely brief and does not adequately engage with the caveats and implications of the study. It needs to be substantially expanded and rewritten to interpret results critically in light of known uncertainties.
The abstract should also be updated to better reflect the findings actually supported by the results. For example, the statement: ‘Using the 1-km NLDAS-2 forcing, the default vegetation scheme with prescribed leaf area index (LAI) and vegetation cover fraction produces too much SWE on the ground due mainly to the strong canopy shading effect despite more snow intercepted by the canopy.’ is not properly supported by any quantitative analysis in the manuscript. Please ensure that claims in the abstract are clearly supported by the results.
In general, the language of this manuscript requires improvement. There are numerous incomplete or unclear sentences. Acronyms are inconsistently introduced or redefined (e.g., CONUS is never introduced, LSM several times, ..). Some hyperlinks are broken, and several cited papers are missing from the bibliography. The figures also need attention: improve overall quality, adjust color maps for clarity and accessibility, and ensure captions are complete and precise.
Line by line comments:
L50: SNOTEL needs to be introduced and defined.
L50: NCAR CLM v4.0 is an outdated version; please contextualize results or comparisons using more recent model developments.
L53–58: Add appropriate citations.
L66–68: Add citations.
L88: The acronym LSM was already defined in line 45.
L92: Clarify: “best by what metric?” Specify performance criteria.
L93: Cho et al., 2022 is missing from the references.
L96: Clarify the aim of the study. Define what “high resolution” means in this context and specify the dynamic vegetation model used.
L103: Revise - “significant snow cover” is vague; use a more quantitative or physically meaningful term.
L114: Since one forcing is considered superior, briefly justify why.
L116: CONUS acronym not introduced.
L126: The section structure is difficult to follow. Consider reorganizing by variable evaluated (e.g., temperature, precipitation, SWE) rather than by dataset name/acronym.
L127: Why are gridded products used for validation? Comparison outside of measurement sites lacks meaning; please justify or reconsider.
L135: “Averaged to 4 km resolution” — elaborate on the method used for upscaling.
L138: Why are actual SWE measurements not used here?
L144 / L152: Additional gridded products are introduced—justify their relevance or quality.
L223 ff: Add more detail on the radiation transfer model, including its assumptions and parameterizations.
L234 ff: Was there a model spin-up? If so, describe procedure and duration.
L247: Add a reference and brief explanation for the mountain mapper algorithm.
L248: Elaborate on how the correction factor is applied and justified.
Tables & Figures
Table 1: Do not reuse LAI both as an acronym for Leaf Area Index and as a simulation label - rename simulation identifiers.
Tables 2 & 3: Consider moving to the appendix or expand explanations in-text.
Figure 2: Improve colormap - differences in row 2 are hard to see. Since comparisons depend on elevation, include elevation contours.
Figure 3: Update colormap for clarity.
Figure 4: Clarify: what does “independent SWE estimates, SNODAS” mean? What are the spatial units?
Figure 5: The location of the label “GOL 2057 m” is confusing. Improve colormap and rewrite caption to make the number of stations and variables shown clear.
Figure 6: Revise colormap for readability.
Figure 7: Clarify whether values are averaged across the basin.
L337: Why was this single station chosen? Provide justification or include multiple stations for robustness.
L343: “AORC simulations are underestimated by ~500 mm” — be more specific (relative to what? annual mean? which period?).
L345: Explain why the GRM station performs poorly.
L388–389: The described “substantial modifications” are not visually evident in Figure 8. Consider including quantitative comparisons to support this claim.
Discussion
L410: Sentence incomplete (“more than” missing?).
L416: “This is favourable when compared to…” — rephrase in a more objective scientific tone.
L421: Revise for clarity.
L424: Omit “apparently”; use formal language.
Summary / Conclusions
L437: Specify the exact model version (e.g., CLM vX.Y) used.
L437: Toure et al., 2016 missing from references.
L442 ff: “The Noah-MP predicted SWE is most sensitive to the high-resolution…” — support this with a robust statistical test.
L453: MODIS LAI may not be an ideal reference; justify this choice.
L459–460: Revise sentence for clarity and conciseness.
L470: The link currently directs to the generic ArcGIS StoryMap site - please provide the actual map or dataset link. If possible, also provide SWE data for reproducibility.