[Synopsis]

This manuscript evaluates a Sentinel-1 repeat-pass InSAR approach for retrieving changes in snow water equivalent (ΔSWE) and then integrating those changes to estimate "total SWE" over several western U.S. SnowEx sites. The authors compare Sentinel-1 retrievals to two independent data sources: airborne LiDAR snow depth maps (SnowEx QSI and ASO) and in situ SNOTEL observations (SWE, snow depth, and near-surface air temperature). They report that 6-day Sentinel-1 acquisitions show moderate agreement with LiDAR snow depth patterns (site-level correlations ~0.42–0.66), while 12-day acquisitions perform poorly due to temporal decorrelation and phase ambiguity. Comparisons of Sentinel-1 ΔSWE with SNOTEL ΔSWE over selected frames yield correlations up to ~0.81 with sub-cm RMSE, and the paper further analyzes how coherence, temperature, snow conditions, and terrain/vegetation characteristics influence retrieval performance.

[Major comments]

Major concern: comparability and representativeness of the validation comparisons.

The paper’s validation strategy would benefit from a more thorough treatment of what each measurement/product represents and how differences in sensing geometry and scale affect "apples-to-apples" comparison. The study combines measurements from different sensors/platforms (Sentinel-1 InSAR, airborne LiDAR, and point-scale SNOTEL), but it does not sufficiently describe their measurement characteristics (support/footprint, geometry, and physical quantity observed) nor how these differences are accounted for in the evaluation.

1. Geometry and slope effects (nadir vs slant path):

Over flat terrain, geometric discrepancies are limited, but in mountainous terrain the differences can be systematic. Airborne LiDAR (nadir-looking) measures snow depth approximately along the z-axis (distance between a snow-free reference surface and snow surface). In contrast, the InSAR retrieval used here is based on interferometric phase delay and is interpreted as ΔSWE under dry-snow assumptions (Eq. 1). Because radar propagation is along the slant path (satellite line of sight) and interacts with terrain slopes and viewing geometry, it is uncertain whether LiDAR depth patterns and InSAR-derived ΔSWE/SWE patterns should be expected to match without explicitly treating slope/local incidence effects and the definition of "effective path length" through snow. As slope increases, the observed quantities may diverge even if the underlying snowpack is the same.

2. SWE vs snow depth (density variability):

A central issue is that the manuscript compares retrieved SWE (or integrated SWE proxy) to LiDAR snow depth and reports correlations (Section 4.1). Depth and SWE are not directly comparable without accounting for snow density variability. This needs explicit discussion and, ideally, an uncertainty treatment or conversion approach (e.g., using density estimates, or restricting validation to periods/areas with representative density constraints). Since SNOTEL provides both SWE and snow depth, the authors could leverage co-located time series to at least characterize plausible density ranges/variability and how that affects depth–SWE comparisons.

3. Spatial representativeness and scaling:

The paper compares (i) pixel-scale satellite retrievals, (ii) very high-resolution LiDAR maps, and (iii) point-scale SNOTEL measurements. The manuscript should state clearly how LiDAR is resampled/aggregated to Sentinel-1 resolution and how point measurements are mapped to radar pixels (or whether a neighborhood averaging strategy is used). Without this, the reported correlations may partly reflect representativeness mismatch rather than retrieval skill.

Given these comparability issues (geometry/slope, quantity mismatch, density variability, and representativeness), I recommend major revision: clearly define what each dataset measures, how it is made comparable (geometry and scale), and how these factors influence the interpretation of agreement/disagreement.

[Specific comments]

1. Equation (1): definition of incidence angle θ in mountainous terrain.

How is θ defined for each pixel? Is it the sensor incidence angle relative to the ellipsoid, or a local incidence angle accounting for terrain slope and aspect? If slope is not included, then the effective angle between the local surface normal (or relevant propagation direction through the snowpack) and the satellite viewing vector changes with terrain, and this could affect both the phase-to-ΔSWE scaling and differences between ascending/descending geometries. Please clarify and, if needed, quantify sensitivity.

2. L156 (6 a.m. vs 6 p.m. performance)

The manuscript states that retrieval works better for 6 a.m. acquisitions than 6 p.m. and therefore focuses on descending data. Please provide a physical explanation (e.g., observation geometry, diurnal melt/freeze state, wet snow occurrence, near-surface temperature timing, changes in coherence) and show supporting evidence (e.g., coherence/temperature statistics split by acquisition time). 

3. Figure 1 color choice (accessibility).

Using green and red together is not ideal for color-vision deficiency. Please consider an alternative color palette or add different line styles/patterns for frames and LiDAR extents.

4. Figure 6: clarify red lines in panels d1 and d2.

In Figure 6, what do the red line segments represent in (d1) and (d2)? If they mark a subset of stations, special periods, thresholds (e.g., >0°C), or data gaps, please add to caption/legend. As-is it is ambiguous.

5. L293: define variables used in parameter analysis.

Please define "vegetation height," "ground topography," "slope," and "aspect" explicitly: data source, how computed (e.g., DEM), spatial resolution, and whether they are averaged/binned at Sentinel-1 pixel scale. Without definitions, the discussion is difficult to follow.

6. L311: definition of temporal coherence.

Temporal coherence is introduced earlier, but please provide a concise definition where it becomes central to the analysis (e.g., complex correlation magnitude between acquisitions; whether it’s VV/VH; whether it’s averaged over a time window; and any masking). This will help readers interpret Figures 8-13.

7. References formatting.

The reference list is comprehensive and relevant, but there are numerous formatting errors (incorrect journal titles, missing article numbers, duplicate entries, corrupted author names). Please revise the bibliography carefully and standardize citation format.