the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 May 2026
Technical note: Evaluation of snow water equivalent from large-scale land-surface products over Italy
Abstract. Snow water equivalent (SWE) is a critical hydrological variable for water resource management in mountainous regions, where seasonal snowpacks function as natural reservoirs regulating streamflow and water supply. While high-resolution, observation-constrained regional or national snow products provide reliable daily SWE estimates at fine spatial resolution (less than 1 km), their limited temporal coverage often restricts their use for long-term hydro-climatological studies. Large-scale land-surface products, in which SWE is derived from land-surface model simulations driven by atmospheric reanalyses or regional dynamical downscaling systems, provide multi-decadal coverage, but their reliability may be affected by biases in meteorological forcing, limited topographic representation, and simplified snow process parameterisation, requiring rigorous regional evaluation. This study evaluates the SWE estimates of three large-scale land-surface products, i.e., the global ERA5-Land, the European CERRA-Land, and the Italian VHR-REA_IT, against the national reference dataset IT-SNOW across Italy. The analysis combines grid-scale bias assessment of mean annual SWE and snow cover duration with temporal correlation analysis of daily SWE series, and is complemented by an evaluation of precipitation and temperature biases in each product, providing insight into how each product represents the atmospheric conditions governing snow accumulation and ablation and thereby supporting the interpretation of the identified SWE discrepancies. The results show clear regional differences in product performance, with no single product performing best across all metrics. ERA5-Land shows the strongest temporal correlation with IT-SNOW, but tends to overestimate mean annual SWE and snow cover duration in the Alps. CERRA-Land shows more moderate biases than ERA5-Land in the Italian Alps, but generally underestimates mean annual SWE across most subregions, where autumn and winter precipitation deficits in the forcing limit snow accumulation. Across the Apennines, both ERA5-Land and CERRA-Land tend to underestimate SWE and snow cover duration, particularly in the northern sectors. VHR-REA_IT shows widespread underestimation and the weakest temporal correspondence with IT-SNOW: this may be due to its fully coupled atmosphere–land architecture with no assimilation of meteorological observations. Overall, the results suggest that forcing biases explain a substantial part of the observed SWE discrepancies, although not all of them. This study provides useful insights for the use of SWE estimates from large-scale land-surface products in long-term hydro-climatological assessments over Italy and helps to understand whether the products can be used to evaluate snow dynamics in mountainous regions lacking high-quality benchmark estimates.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Hydrology and Earth System Sciences.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1832 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2484', Anonymous Referee #1, 17 Jun 2026
-
AC1: 'Reply on RC1', Mattia Neri, 30 Jun 2026
We thank the reviewer for the careful reading of the manuscript, the appreciations towards our work and the very constructive comments. Below we provide a point-by-point response.
General comment
For the evaluation of the land surface datasets against IT-SNOW (for SWE and snow cover duration) or SCIA (for 2-m temperature and precipitation), the authors look into spatial biases between the different products, but do not evaluate the statistical significance of these biases, which make me wonder whether the biases presented are statistically significant or not. As the authors work with a relatively short period (2010-2024) and a large number of grid points, I would suggest assessing the statistical significance of SWE, snow cover duration, temperature and precipitation biases using a two-tailed students T-test combined with a False Discovery Rate (FDR) approach following Wilks (2016).
Reply: We would like to thank the Reviewer for this highly pertinent and constructive observation, which significantly enhances the statistical robustness of our findings. We fully agree that evaluating the statistical significance of the spatial biases is crucial, particularly given the relatively short evaluation period (2010–2024) and the large number of grid points involved in the analysis. Following your valuable suggestion, we have already carried out the analysis assessing the statistical significance of the biases for all the variables using the proposed methodology. In the revised version of the manuscript, we will update the results section accordingly. Specifically, rather than overlaying stippling directly onto the main manuscript maps—which would severely compromise their readability given the spatial resolution (9 km), the small pixel size, and the compact format of the figures—we plan to provide dedicated additional materials. These will likely be included in a new Supplement and will probably consist of (i) separate maps explicitly showing the spatial distribution of the statistical significance, and (ii) additional graphical outputs illustrating the relationship between the magnitude of the bias and the significance of the estimate. However, the results of this new analysis confirm our initial findings. The statistical significance is generally consistent with the interpretations provided in the original draft; therefore, the core message and the overall conclusions of the Technical Note will remain unchanged, but they will now be supported by a much more rigorous statistical framework. We sincerely thank the Reviewer again for pointing this out and for helping us improve the quality of our manuscript.
Specific comments (SC)
SC1: L124 and Figure 1: Could the authors add more geographical information to the map of Italy (other than “Alps” and “Apennines”) to help readers locate places more easily? For example, could they indicate the location of the Ligurian sector, Sardinia, Sicily, the Strait of Messina, etc.? I know where these places are, but I imagine that not all readers do, so it would be helpful to include a little more geographical background information.
Reply to SC1: We agree that adding more geographical references to Figure 1 will help the international readership better orient themselves and follow the regional differences discussed in the text. In the revised version of the manuscript, we will update Figure 1 to include additional geographical background information.
SC2: L146-147: Could the authors be a little bit more specific on how the precipitation-phase partitioning is implemented in IT-SNOW?
Reply to SC2: We thank the Reviewer for pointing this out. Phase partitioning in S3M relies on both air temperature and relative humidity, which generally provides more reliable results than using air temperature alone (Zhang et al., 2017). Rain and snow proportions are derived based on the simple method by Froidurot et al. (2014). The separation is performed at an hourly time scale. In the revised manuscript, we will include these additional details.
SC3: L148: Is the hybrid temperature-index and radiation-driven melt approach in IT-SNOW similar to the enhanced temperature-index approach of Pellicciotti et al., 2005 (https://doi.org/10.3189/172756505781829124)?
Reply to SC3: Yes, the approach in S3M was inspired by the approach of Pellicciotti et al. (2005). The only, significant deviation is the addition of two modulation parameters that reduce snowmelt based on antecedent temperature. This was done to include in the approach some forms of cold content, that is, thermal inertia of snow. While a comprehensive description is available in Avanzi et al. (2023), we agree that briefly clarifying this connection will benefit the readers. Accordingly, we will include these details and the reference to Pellicciotti et al. (2005) in the revised text.
SC4: Appendix Figures B1-B3: For the middle row figures (representing the temperature biases) I recommend reversing the color bar. In the current state, the colors representing the Tbias are somewhat counterintuitive in my opinion, i.e. warmer is purple and colder is red whereas a positive SWE bias is blue (usually associated with colder temperature conditions) and a negative SWE bias is red (usually associated with warmer conditions).
Reply to SC4: We completely agree that the current color combination for the temperature bias might be counterintuitive for the readers. In the revised manuscript, we will update the color scale. Specifically, we will reverse the color bar, using red to indicate temperature overestimation (positive bias).
SC5: L358-359: Another potential explanation for the overestimation of snow depth in the higher parts of the Alps could be that (according to Orsolini et al., 2019; https://doi.org/10.5194/tc-13-2221-2019) data assimilation of snow cover (using IMS – Interactive Multisensor Snow and Ice Mapping System) was discontinued in ERA5 above 1500 m a.s.l. This could explain the snow biases in ERA5 itself, but as snow biases also affect ERA5 temperature (via snow-albedo feedbacks) I can imagine that the negative temperature biases propagated from ERA5 into ERA5-Land also have affected the snow conditions in ERA5-Land, particularly at high Alpine elevations. And what about snow-albedo feedback in ERA5-Land itself? Could these feedbacks also have contributed to the SWE bias in ERA5-Land?
Reply to SC5: We thank the Reviewer for this highly insightful comment and for pointing out the relevance of Orsolini et al. (2019). We agree that these represent very plausible hypotheses that could help explain the observed biases. First, it is reasonable to hypothesize that the discontinue IMS snow cover assimilation above 1500 m a.s.l. in ERA5 introduces snow and temperature biases in the atmospheric reanalysis. These biases may then propagate into ERA5-Land through the atmospheric forcing. Specifically, the temperature at the “lowest model level” (10 m), which drives the offline land surface model after being adjusted for elevation differences between the two grids via a standard lapse rate. Second, regarding the snow-albedo feedback, we agree that this could also play a role. Specifically, while ERA5-Land runs offline (meaning there is no two-way land-atmosphere coupling), such a mechanism could be driven by the internal thermodynamics of its underlying land surface model CHTESSEL. An initial overestimation of SWE could lead to an anomalously high surface albedo, which would reduce the absorbed shortwave radiation. This process might lower the computed skin temperature and reduce the sensible heat flux, thereby potentially delaying snowmelt. Consequently, this internal thermodynamic loop may act to exacerbate and prolong the positive SWE biases, particularly at higher Alpine elevations. In the revised manuscript, we will expand the discussion (around L358-359) to explicitly cite Orsolini et al. (2019) and to discuss these two intertwined, potential mechanisms—the propagation of biases from the ERA5 forcing and the possible internal thermodynamic loop within CHTESSEL—as plausible drivers of the overestimation in the Alpine sector.
SC6: L371-374 and beyond: I guess the negative SWE biases in the Apennines or some regions of the Alps, like the Trentino-Alto Adige region could also be related to differences in how ERA5-Land (via precipitation forcing from ERA5) or CERRA-Land/VR-REA-IT and the S3M model (the basis for IT-SNOW) deal with the precipitation phase, particularly for temperature regimes around 0°C. I don’t know how the precipitation-phase partitioning in S3M works, but could it be that precipitation falls as snow in S3M, whereas in ERA5-Land or the other 2 models precipitation falls as rain? To address this question, I recommend adding two more rows to Figures B1-B3 that visualize the seasonal biases in rainfall and snowfall.
Reply to SC6: We fully agree that differences in the precipitation-phase partitioning (solid vs. liquid) around 0°C between the evaluated products and the S3M model are highly likely to be a key driver of the SWE biases observed in such regions. We agree with the rationale behind your suggestion to visualize the seasonal biases in rainfall and snowfall. However, we are unfortunately unable to perform this specific evaluation due to data availability constraints. To ensure a consistent and "symmetrical" evaluation, our analysis relies strictly on the variables that are publicly and openly distributed for all datasets. While ERA5-Land does provide partitioned snowfall and rainfall outputs, these specific fluxes are not available in the public data releases for CERRA-Land and VHR-REA_IT. Also, the reference dataset (IT-SNOW) includes only the final snow variables (SWE, snow depth, bulk snow density and liquid water yield), but not the snowfall and rainfall fields internally computed by the S3M model. Consequently, computing a direct, comparative spatial bias for snowfall against the reference dataset is not feasible. While a quantitative evaluation is not feasible, during the revision process we will evaluate how best to acknowledge this aspect in the text. We plan to briefly mention the possibility that differences in the precipitation-phase partitioning approaches could potentially influence the SWE discrepancies, particularly in regions where temperatures frequently fluctuate around the freezing point.
SC7: Also, I am wondering whether there are differences in the density of ultrasonic sensors (used for assimilation of IT-SNOW) among the different regions? Could it be that the density of sensors is higher in the central Apennines and some regions of the Alps, such as Trentino Alto-Adige? I am asking this because the negative SWE biases presented in Figure 3 seem to appear consistently over the same regions, independent from the comparison with the different land surface datasets. Even the SWE biases between ERA5 Land and IT-SNOW show patches of negative SWE biases in the Trentino Alto-Adige region, which seem to appear in the lower-elevated areas of the respective region.
Reply to SC7: Yes, there are indeed regional differences: the bulk of the snow data are in the northern regions, particularly within the Po river basin in the northwest. While a lower sensor density might indeed affect the performance of the model, previous comprehensive assessments of IT-SNOW against a spatially distributed Sentinel-1 snow depth dataset did not reveal any clear spatial bias patterns linked to the sensor network density (Avanzi et al., 2023). So we are inclined to attribute most of these patterns to the evaluated datasets, rather than to IT-SNOW. We appreciate the Reviewer for prompting this clarification. In the revised manuscript, we will briefly expand upon this aspect in the text.
References
Avanzi, F., Gabellani, S., Delogu, F., Silvestro, F., Pignone, F., Bruno, G., Pulvirenti, L., Squicciarino, G., Fiori, E., Rossi, L., Puca, S., Toniazzo, A., Giordano, P., Falzacappa, M., Ratto, S., Stevenin, H., Cardillo, A., Fioletti, M., Cazzuli, O., Cremonese, E., Morra di Cella, U., Ferraris, L., 2023. IT-SNOW: a snow reanalysis for Italy blending modeling, in situ data, and satellite observations (2010–2021). Earth Syst. Sci. Data 15, 639–660. https://doi.org/10.5194/essd-15-639-2023
Froidurot, S., Zin, I., Hingray, B., Gautheron, A., 2014. Sensitivity of Precipitation Phase over the Swiss Alps to Different Meteorological Variables. J. Hydrometeorol. 15, 685–696. https://doi.org/10.1175/JHM-D-13-073.1
Orsolini, Y., Wegmann, M., Dutra, E., Liu, B., Balsamo, G., Yang, K., De Rosnay, P., Zhu, C., Wang, W., Senan, R., Arduini, G., 2019. Evaluation of snow depth and snow cover over the Tibetan Plateau in global reanalyses using in situ and satellite remote sensing observations. The Cryosphere 13, 2221–2239. https://doi.org/10.5194/tc-13-2221-2019
Pellicciotti, F., Brock, B., Strasser, U., Burlando, P., Funk, M., Corripio, J., 2005. An enhanced temperature-index glacier melt model including the shortwave radiation balance: development and testing for Haut Glacier d’Arolla, Switzerland. J. Glaciol. 51, 573–587. https://doi.org/10.3189/172756505781829124
Zhang, Z., Glaser, S., Bales, R., Conklin, M., Rice, R., Marks, D., 2017. Insights into mountain precipitation and snowpack from a basin‐scale wireless‐sensor network. Water Resour. Res. 53, 6626–6641. https://doi.org/10.1002/2016WR018825
Citation: https://doi.org/10.5194/egusphere-2026-2484-AC1
-
AC1: 'Reply on RC1', Mattia Neri, 30 Jun 2026
-
RC2: 'Comment on egusphere-2026-2484', Steven Margulis, 08 Jul 2026
General comments: The manuscript provides a useful comparison of three primarily model-based snow products over Italy relative to a more observationally constrained snow product. The contribution provides useful insight into how model-based estimates perform with respect to snow water equivalent estimation over mountainous environments, and how errors in these estimates may be attributed to errors in model inputs (meteorological forcings). The specific comments below are suggestions meant to improve the manuscript before publication.
Specific comments:
- Line 59-67. In discussing other observationally-constrained snow datasets, you should consider the mention of Cortés and Margulis (2017), Liu et al. (2021), and Fang et al. (2022), which demonstrate multi-decadal SWE estimation over High Mountain Asia, the Andes, and the Western U.S. respectively.
- Figure 2, panel (b). It may make it easier for the reader to interpret the SWE if you centered the mean annual temperature colorbar on zero degrees Celsius. It would provide visual evidence of regions that are below freezing for a significant fraction of the year and thus more likely to have snowfall than rainfall.
- Section 3.1. Given that the IT-SNOW is the reference product, a bit more detail on how it works would help the reader. For example, how is the assimilation implemented? Is it in near-real-time? Retrospective? What is the assimilation method used? While these details are in the provided references, having the basic summary would be helpful.
- Line 151: How informative is the snow covered area in updating SWE? It typically has limited instantaneous relationship with SWE (i.e. often saturates near 100% for much of the accumulation season). Does the assimilation more heavily rely on the snow depth observations as a constraint?
- Line 58: It would be helpful to the reader to provide some of the bulk verification statistics from previous work (i.e. define what is meant by “consistent” and “low mean bias” more quantitatively).
- Figures 3 and 4. I would suggest adding a third row at the bottom showing the bias as a percent of the reference. Given the large difference in mean annual SWE or SCD values, it would be complementary to show the percentage bias in addition to the absolute bias.
- Line 359: Follow-on studies to those cited above (i.e., Liu et al. (2022) and Fang et al. (2023) looked at relationships between SWE errors and meteorological forcing (including ERA5 and ERA5-Land). These might be useful for context and supportive of the findings in this manuscript.
- I would suggest the authors consider whether the Alps vs. Apennines results should be discussed in terms of seasonal vs. ephemeral snow. Liu et al. (2022) partitioned their analysis of HMA snow into seasonal and ephemeral and found varying errors for the different regimes. From the results shown, it appears the Apennines may be more ephemeral and that may explain some of the errors seen.
- Line 393: The fact that the highest resolution (and coupled) model is presumably most computational expensive and yet worst in some error metrics is perhaps worth emphasizing more. Given that meteorological forcings are so important in SWE evolution, but typically relatively coarse, does not guarantee that results will be more accurate at higher resolution. This emphasizes the importance of observations in high-resolution applications.
- Section 5.2: Can you provide a bit more detail on how the models parameterize snow covered area? Is it a prognostic variable or is it parameterized as a function of snow depth? Is it binary or fractional?
- Figure 5: This is a more general comment relevant to some of the other figures too, but Figure 5 provides a useful example. The authors should consider whether using a more discretized colorbar will help communicate the message more clearly. In this case, having a continuous colorbar to demonstrate correlations is not as visually useful as perhaps discretizing the colorbar into a few bins where “poor”, “moderate”, and “high” correlation would be easier to infer.
- Figures A1-B3. Would using consistent colorbars for bias in these plots be easier to interpret or do the authors think using different colorbars for different variables are helpful?
Minor suggested edits/corrections:
- Line 112: Is “SCIA” an acronym? If so, define in first instance.
- Line 352: Use “SWE” instead of “snow water equivalent”.
References:
Cortés, G., and S. Margulis (2017), Impacts of El Niño and La Niña on interannual snow accumulation in the Andes: Results from a high-resolution 31 year reanalysis, Geophys. Res. Lett., 44, 6859–6867, doi:10.1002/2017GL073826.
Fang, Y., Y. Liu, and S.A. Margulis, 2022. A western United States snow reanalysis dataset over the Landsat era from water years 1985 to 2021, Scientific Data. https://10.1038/s41597-022-01768-7.
Fang, Y., Y. Liu, D. Li, H. Sun, and S.A. Margulis, 2023. Spatiotemporal snow water storage uncertainty in the midlatitude American Cordillera, The Cryosphere, 17, 5175–5195, https://doi.org/10.5194/tc-17-5175-2023.
Liu, Y., Fang, Y., and S.A. Margulis, 2021. Spatiotemporal distribution of seasonal snow water equivalent in High-Mountain Asia from an 18-year Landsat-MODIS era snow reanalysis dataset, The Cryosphere, 15, 5261–5280, 2021. https://doi.org/10.5194/tc-15-5261-2021.
Liu, Y., Y. Fang, D. Li, and S.A. Margulis, 2022. How well do global snow products characterize snow storage in High Mountain Asia? Geophysical Research Letters, 49, e2022GL100082. https://doi.org/10.1029/2022GL100082.
Citation: https://doi.org/10.5194/egusphere-2026-2484-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 268 | 80 | 20 | 368 | 13 | 13 |
- HTML: 268
- PDF: 80
- XML: 20
- Total: 368
- BibTeX: 13
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of egusphere-2026-2484: “Technical note: Evaluation of snow water equivalent from large-scale land-surface products over Italy” by Sarigil et al.
This manuscript evaluates the snow water equivalent (SWE) estimates of three large-scale land surface products - ERA5-Land, CERRA-Land, and the Italian VHR-REA_IT - against the national reference dataset IT-SNOW in Italy. To this end, a (spatial) bias assessment is performed on the mean annual SWE and snow cover duration. This is combined with temporal correlation analyses of daily SWE series and complemented by an evaluation of temperature and precipitation biases in the land surface products. The latter provides insight into how each product represent atmospheric conditions and how this could contribute to SWE discrepancies.
The main findings suggest that VHR-REA_IT generally performs worse than ERA5-Land and CERRA-Land in simulating SWE and snow cover duration, and that CERRA-Land performs better than ERA5 in the Italian Alps. Much of the bias can be attributed to forcing biases in each product, or to the absence of the assimilation of meteorological observations in VHR-REA_IT’s coupled atmosphere-land modelling framework.
The manuscript is well and clearly written and pleasant to read. It is an interesting note and provides valuable insights. I have added several comments that need to be addressed prior to publication.
General Comments
For the evaluation of the land surface datasets against IT-SNOW (for SWE and snow cover duration) or SCIA (for 2-m temperature and precipitation), the authors look into spatial biases between the different products, but do not evaluate the statistical significance of these biases, which make me wonder whether the biases presented are statistically significant or not. As the authors work with a relatively short period (2010-2024) and a large number of grid points, I would suggest assessing the statistical significance of SWE, snow cover duration, temperature and precipitation biases using a two-tailed students T-test combined with a False Discovery Rate (FDR) approach following Wilks (2016):
Wilks, D. S., 2016: “The Stippling Shows Statistically Significant Grid Points”: How Research Results are Routinely Overstated and Overinterpreted, and What to Do about It. Bull. Amer. Meteor. Soc., 97, 2263–2273, https://doi.org/10.1175/BAMSD-15-00267.1.
Differences are then marked as statistically significant when p-values are lower than or equal to 0.05, while also considering the increased likelihood of false positives that can appear when statistical testing is performed over a large number of grid points.
Specific Comments
L124 and Figure 1: Could the authors add more geographical information to the map of Italy (other than “Alps” and “Apennines”) to help readers locate places more easily? For example, could they indicate the location of the Ligurian sector, Sardinia, Sicily, the Strait of Messina, etc.? I know where these places are, but I imagine that not all readers do, so it would be helpful to include a little more geographical background information.
L146-147: Could the authors be a little bit more specific on how the precipitation-phase partitioning is implemented in IT-SNOW?
L148: Is the hybrid temperature-index and radiation-driven melt approach in IT-SNOW similar to the enhanced temperature-index approach of Pellicciotti et al., 2005 (https://doi.org/10.3189/172756505781829124)?
Appendix Figures B1-B3: For the middle row figures (representing the temperature biases) I recommend reversing the color bar. In the current state, the colors representing the Tbias are somewhat counterintuitive in my opinion, i.e. warmer is purple and colder is red whereas a positive SWE bias is blue (usually associated with colder temperature conditions) and a negative SWE bias is red (usually associated with warmer conditions).
L358-359: Another potential explanation for the overestimation of snow depth in the higher parts of the Alps could be that (according to Orsolini et al., 2019; https://doi.org/10.5194/tc-13-2221-2019) data assimilation of snow cover (using IMS – Interactive Multisensor Snow and Ice Mapping System) was discontinued in ERA5 above 1500 m a.s.l. This could explain the snow biases in ERA5 itself, but as snow biases also affect ERA5 temperature (via snow-albedo feedbacks) I can imagine that the negative temperature biases propagated from ERA5 into ERA5-Land also have affected the snow conditions in ERA5-Land, particularly at high Alpine elevations. And what about snow-albedo feedback in ERA5-Land itself? Could these feedbacks also have contributed to the SWE bias in ERA5-Land?
L371-374 and beyond: I guess the negative SWE biases in the Apennines or some regions of the Alps, like the Trentino-Alto Adige region could also be related to differences in how ERA5-Land (via precipitation forcing from ERA5) or CERRA-Land/VR-REA-IT and the S3M model (the basis for IT-SNOW) deal with the precipitation phase, particularly for temperature regimes around 0°C. I don’t know how the precipitation-phase partitioning in S3M works, but could it be that precipitation falls as snow in S3M, whereas in ERA5-Land or the other 2 models precipitation falls as rain? To address this question, I recommend adding two more rows to Figures B1-B3 that visualize the seasonal biases in rainfall and snowfall.
Also, I am wondering whether there are differences in the density of ultrasonic sensors (used for assimilation of IT-SNOW) among the different regions? Could it be that the density of sensors is higher in the central Apennines and some regions of the Alps, such as Trentino Alto-Adige? I am asking this because the negative SWE biases presented in Figure 3 seem to appear consistently over the same regions, independent from the comparison with the different land surface datasets. Even the SWE biases between ERA5 Land and IT-SNOW show patches of negative SWE biases in the Trentino Alto-Adige region, which seem to appear in the lower-elevated areas of the respective region.