Forcing the SURFEX/Crocus snow model with combined hourly meteorological forecasts and gridded observations in southern Norway

Luijting, Hanneke; Vikhamar-Schuler, Dagrun; Aspelien, Trygve; Bakketun, Åsmund; Homleid, Mariken

doi:https://doi.org/10.5194/tc-12-2123-2018

Articles | Volume 12, issue 6

https://doi.org/10.5194/tc-12-2123-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/tc-12-2123-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 12, issue 6

Research article

|

21 Jun 2018

Research article |

| 21 Jun 2018

Forcing the SURFEX/Crocus snow model with combined hourly meteorological forecasts and gridded observations in southern Norway

Hanneke Luijting, Dagrun Vikhamar-Schuler, Trygve Aspelien, Åsmund Bakketun, and Mariken Homleid

Download

Final revised paper (published on 21 Jun 2018)
Preprint (discussion started on 03 Nov 2017)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Referee review for the article entitled "Forcing the SURFACE/Crocus snow model with continued hourly meteorological forecasts and gridded observations in southern Norway"', Anonymous Referee #1, 30 Nov 2017
- AC1: 'Response to anonymous referee #1', Hanneke Luijting, 22 Jan 2018
RC2: 'Review of the paper “Forcing the SURFEX/Crocus snow model with combined hourly meteorological forecast and gridded observations in southern Norway” by H. Luijting et al. submitted to The Cryosphere', Anonymous Referee #2, 18 Dec 2017
- AC2: 'Response to anonymous referee #2', Hanneke Luijting, 22 Jan 2018
EC1: 'Editor comment', Ross Brown, 27 Mar 2018

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Hanneke Luijting on behalf of the Authors (25 Feb 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (01 Mar 2018) by Ross Brown

RR by Anonymous Referee #2 (22 Mar 2018)

Suggestions for revision or reasons for rejection

Review of the revised version of the paper “Forcing the SURFEX/Crocus snow model with combined hourly meteorological forecast and gridded observations in southern Norway” by H. Luijting et al. submitted to The Cryosphere

General comments

During the revision process, the authors have made several changes to the initial manuscript including:
• An extended introduction that better presents the context of the paper with respect to previous studies and that also gives an overview of the current status of snowpack modeling in Norway for various applications.
• A better description of the study area and the meteorological forcing provided to Crocus.
• 1-D simulations to test the impact of blowing snow sublimation on snowpack evolution in Crocus
• A quantification of the agreement between modeled snow cover and snow cover observed retrieved from MODIS.

These changes have improved the quality of the manuscript. On the other hand, the authors have chosen not to include any additional 2D simulations as suggested by the two reviewers. This would have potentially increased the quality of the paper and provided interesting results for the community. The computational cost cannot be the only argument considered when deciding to rerun a simulation or not. In particular, my main concern regarding the simulations presented in the revised version of the paper is related to the simulation driven by AROME-MetCoop. In this simulation, air temperature is taken from a post-processed gridded temperature at 500 m whereas precipitation (rainfall and snowfall) are obtained from AROME at 2.5 km grid spacing. Is the precipitation phase modified based on the high-resolution post-processed temperature data? This is never mentioned in the paper. It can have a impact in the western side of the domain where temperatures can be closed to zero during precipitation events occurring in wintertime. An error in precipitation phase can have a strong impact on the snowpack evolution. If this correction of precipitation phase has not been carried out, I strongly recommend the authors to rerun the Crocus simulation driven by AROME-MetCoop with consistent precipitation fields.

The quality of the discussion should also be improved prior to publication. In the conclusion and at the beginning of the discussion section, the authors explain that AROME-MetCoop-Crocus provides the best representation of the spatial distribution of snow cover during the melting season whereas it gives worse results than GridObs Crocus when evaluated at stations measuring snow depth. I recommend the authors to better discuss the reason for these differences when estimating the performances of AROME-MetCoop-Crocus. Is it related to different performance of the gridded precipitation as a function of elevation? The challenge of estimated accurate precipitation at high-elevation in mountainous terrain when very few stations are available is never clearly discussed in the paper whereas it is a key element when modeling snowpack in these regions.

Specific comments

P 3 L6 : what is the meaning of “internal structure”? FSM and JULES do not include a description of the snow metamorphism contrary to Crocus and SNOWPACK.

P 3 L 23: the snow maps generated by Vionnet et al (2016) and Quéno et al (2016) cannot be qualified of “snow maps of high quality”. Indeed, these two studies have shown the potential of high-resolution NWP system to drive snowpack model in alpine regions but the snow depth bias in their experiments are still large. These studies have also shown that a distributed precipitation analysis system using the guess from the high-resolution NWP system is required.

P 5 L 1-2: the description of the configuration of the soil model is not clear. Do the author mean that the number of soil layers used to solve the evolution of temperature and the soil moisture is different?

P 5 L 11-13: a spin up of the land-surface scheme is generally required to generate the initial soil conditions (temperature and moisture). Here, the authors uses the default values in SURFEX. They should include a comment on this topic and mention the potential impact at the beginning of the snow season when the snow cover is potentially sensitive to the heat flux from the ground.

P 7 L 25-26 is there a reference that describes the method used to adjust AROME wind speed? Furthermore, the authors should explain how the 500-m temperature product is interpolated on the 1-km grid used for the snowpack simulation.

P 17 L 10-15: it is surprising to see that the authors have selected the simulation GridObs-Crocus at Midstova to test the impact of blowing snow sublimation. Indeed, at this station, the quality of the precipitation forcing is known to be bad for the beginning of Winter 2014/2015, leading to a underestimation of snow depth at the beginning of the season. It would have been more interesting to test the impact of the parameterization of blowing snow sublimation using the AROME forcing. Brun et al (2013) have tested the parameterization of blowing snow sublimation in a different context and have shown there may be a redundancy between using a corrected precipitation dataset and accounting for blowing snow sublimation (see Sect. 5 of Brun et al., 2013).

Brun, E., Vionnet, V., Boone, A., Decharme, B., Peings, Y., Valette, R., ... & Morin, S. (2013). Simulation of northern Eurasian local snow depth, mass, and density using a detailed snowpack model and meteorological reanalyses. Journal of Hydrometeorology, 14(1), 203-219.

P 23 L 15-19 : here the argumentation made by the authors can be discussed. Personally, I always think that model evaluation should be carried out at stations where we know that the model is supposed to represent the meteorological conditions at this station. The model can be right for wrong reasons at these stations. It does not mean that these stations must be included in the evaluation.

P 26 L 8-14: the authors explain that snow melt is underestimated in both experiments but still the snow cover melts too fast in GridObs-Crocus. As mentioned above, this suggest an underestimation of snow accumulation in GridObs-Crocus that could be better discussed in the paper.

Technical comments

P3 L 4: “SnowPack” is usually written “SNOWPACK” when referring to the Swiss snowpack model.

P 3 L 30: “??” appear at many places in the text and should be corrected.

P 6 L 3: km2 instead of km-2

P 22 L 5: do the authors mean “too late” for AROME Crocus and “too early for GridObs-Crocus?

P 25 L 32: mention explicitly “wind speed” and “air temperature” as the post-processed variables.

Hide

ED: Reconsider after major revisions (27 Mar 2018) by Ross Brown

AR by Hanneke Luijting on behalf of the Authors (16 May 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (17 May 2018) by Ross Brown

RR by Anonymous Referee #2 (31 May 2018)

Suggestions for revision or reasons for rejection

General comments

During the additional round of review, the authors have made several changes to the first revised version of the manuscript including:
• A new Crocus simulation driven by AROME-MetCoop including a modified precipitation phase based on the high-resolution post-processed temperature data. This new modification brings large improvements in terms of error statistics for the simulation of snow depth with Crocus driven by AROME-MetCoop
• Large parts of the conclusion and the discussion have been rewritten accordingly.

These changes have improved the quality of the manuscript and the new simulation nicely illustrates the necessity to include an adjustment of precipitation phase in complex terrain when carrying a simulation with a snowpack scheme running at higher resolution that the original atmospheric forcing. Based on this modification, I recommend this paper to be accepted for publication in TC. I mentioned below a few specific comments that the authors should take into account prior to publication.

Specific comments

Note: The numbering of the lines is based on the revised version of the paper in track-change mode

P4 L21: Masson et al (2013) would be a more appropriate reference for the list of different snowpack schemes in SURFEX.

P4 L30: add precipitation phase in the list of weather variables.

P 14 Figure 4: the quality of the fonts and the lines on this figure is poor compared to other figures in the paper such as Fig. 3 or Fig. 8 (which is very similar to Fig. 4). It would be good to produce a new figure easier to read.

P26–P27: the comparison between the simulations with and without the correction of precipitation phase is quite interesting. Could the authors includes the results in terms of bias and RMSE on snow depth to quantify the improvement obtained when considering the adjustment of precipitation phase? I guess these values correspond to the ones removed from Table 2 and 3.

P 30 L6: the term “significant” may not be appropriate since the authors did not the statistical significance of the differences between the 2 simulations.

Conclusion P31 L 4-9: for the results obtained with GridObs-Crocus, it would be good to make the clear distinction between (i) the underestimation of snow accumulation at high-elevation and (ii) the tendency of the model to underestimate snow melt. According to me, the sentence “during the melting season the snow cover melted away too fast” can suggest that snow melt is overestimated in GridObs-Crocus which leads to a snow cover that vanish too early. The authors should consider rewriting this sentence. The same comments apply to the Abstract (P 1, L15-20) and to the discussion (P25, L 4-14).

Hide

ED: Publish subject to minor revisions (review by editor) (31 May 2018) by Ross Brown

AR by Hanneke Luijting on behalf of the Authors (05 Jun 2018) Author's response Manuscript

ED: Publish as is (06 Jun 2018) by Ross Brown

AR by Hanneke Luijting on behalf of the Authors (07 Jun 2018)

Short summary

Knowledge of the snow reservoir is important for energy production and water resource management. In this study, a detailed snow model is run over southern Norway with two different sets of forcing data. The results show that forcing data consisting of post-processed data from a numerical weather model (observations assimilated into the raw weather predictions) are most promising for snow simulations when larger regions are evaluated.