the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Jan 2026
Impact of seasonal snow on the recharge of a mountain karst aquifer under climate change: the Dévoluy case study (Southern Alps, France)
Abstract. Seasonal snow strongly influences groundwater recharge in mountain aquifers, yet its role in mid-altitude karst systems under climate warming remains poorly quantified. We investigated the Dévoluy karst aquifer (Southern French Alps) to assess how snow controls recharge and how spring discharge may respond to rising temperatures. Using the KarstMod platform, we developed a rainfall–snow–discharge model incorporating a degree-day snow routine to partition precipitation between rainfall and snow, and simulate the snowmelt. The model was calibrated and validated over four contrasting years (two low-snow, one high-snow, and one very high-snow year). Results show that accounting for snow processes is essential to reproduce the observed discharge dynamics, highlighting the dominant role of snow accumulation and melt in controlling both flow timing and magnitude. Under +2 °C and +4 °C warming scenarios, simulated winter flows increase while snowmelt peaks occur earlier, resulting in earlier and more severe summer low-flow periods. August discharge decreases by 28 % to 44 %, respectively, compared to present conditions. These findings demonstrate the critical role of seasonal snow in regulating recharge in mid-altitude karst aquifers and highlight that ongoing warming will substantially reduce summer water availability in mountain regions.
- Preprint
(1893 KB) - Metadata XML
-
Supplement
(120 KB) - BibTeX
- EndNote
Status: open (until 20 Feb 2026)
- RC1: 'Comment on egusphere-2025-5654', Anonymous Referee #1, 22 Jan 2026 reply
-
RC2: 'Comment on egusphere-2025-5654', Anonymous Referee #2, 27 Jan 2026
reply
Overall, I found the models and methods were well described. The authors did a good job justifying the inclusion of snow in the karst discharge simulation under multiple climate-change temperature scenarios. Overall, I found the manuscript well written and clear. Thank you for letting me review this research.
Overall, there are a few major requests I have:
- Greatest recommendation: the research needs to include RMSE, MAE, bias and/or other metrics between the simulations and in-situ data to calculate and quantify the amount of model error. These will also need associated visualization for the reader to ensure that the model statistically performed well.
- I recommend aggregating the time series (e.g. figure 4-6) into bar charts and/or box and whisker, etc. so the reader can easily compare the simulation results.
Please see my comments below:
Abstract:
Line 18: At the end of this sentence, “Results show that accounting for snow processes is essential to reproduce the observed discharge dynamics” - I think it would be clearer to add “in karst environments”
Introduction:
Line 57: The paragraph on hydrologic modelling as an efficient tool and the models that incorporate snow as a parameter come before the next paragraph that explains the importance of incorporating snow into karst models. I think these two paragraphs should be swapped – e.g. paragraph at line 57 should go at line 46.
Methods:
Data
137: It is unclear how “precipitation (mm/day) and mean temperature (°C)” factor into the sentence. Are you trying to say that precipitation (mm/day) and mean temperature (°C) also have the four grid meshes with 8x8 spatial resolution, etc.? If so – then this needs its own sentence – the format of this sentence leaves the reader confused. Please clarify.
Accuracy Assessment:
Please include an RMSE, MAE, bias for evaluating the accuracy of the simulated results compared to the in-situ results. The KGEnp mentioned in Line 411 is one of the few mentions of model efficiency. I would like to see more on model accuracy and bias. Please include figures to visually demonstrate these results.
Results:
Figure 4: the pink/blue in Fig 4A and yellow/gray are hard to discern (e.g. the colors overlap). Can you try more high contrast line colors?
The graphics don’t line up with the text. For example, line 470 mentions Figure 8e, then the next figure is Figure 6. Please put the figures after the sections where they are first mentioned.
I find the time series difficult to interpret. I think it would be easier if you included aggregated final results for the readers such as:
Figure 4: bar chart showing the average discharge for the simulated and discharge for the warmup, calibration, etc. to easily compare if the simulated had a similar amount, or more, or less predicted snow.
Figure 5: I recommend a similar aggregated visual that makes it easy for the readers to compare. Also, make the caption clearer to define was E-S, M-S, and C-S mean (e.g. explain it – the reader needs more than a copy/paste from the visual).
Figure 6 and 8 – same comment – aggerate so the reader can easily compare.
For example, Figure 7 was easily digestible to me because I could make side-by-side comparisons.
I think you should reference Figure 4 in the Figure 8 caption so readers can easily reference the different model simulations.
Figure 6A: I find this unimportant – a reader could just as easily find that in the text – or you could put it in the caption, but the bar at the top seems out of place and unnecessary.
Discussion:
Line 485: Please include your accuracy assessment values behind this sentence to justify and support that snow improved modelling results.
Citation: https://doi.org/10.5194/egusphere-2025-5654-RC2 -
RC3: 'Comment on egusphere-2025-5654', Anonymous Referee #3, 03 Feb 2026
reply
General comments
The manuscript addresses an important topic, but I recommend that the authors clarify and nuance their novelty claim regarding the effects of temperature increase on karst spring discharge. There is now a small but growing body of work that has already used modelling approaches to quantify climate‑change or warming impacts on karst recharge and spring behaviour (e.g. Doummar et al., 2018; Fan et al., 2023; and related numerical‑modelling studies on karst groundwater under climate scenarios). Rather than presenting the study as the first to model the impact of temperature increase on karst discharge, it would be more accurate and convincing to position the contribution in terms of its specific combination of features: (i) a mid‑altitude Alpine karst system with a strongly seasonal snowpack, (ii) the use of a parsimonious rainfall–snow–discharge conceptual model (KarstMod with a degree‑day snow routine), (iii) a scenario design based on simple temperature perturbations consistent with TRACC‑type warming trajectories, and (iv) a quantitative focus on seasonal metrics, particularly changes in summer/August low flows. Framing the contribution in this way would acknowledge prior work while clearly highlighting what is genuinely new and distinctive in the present case study and modelling set‑up.
I suggest adding a short clarification on the climate scenarios and the choice of perturbations. At present, the manuscript may give the impression that the French TRACC storyline is essentially a temperature‑only scenario, whereas in reality regional climate projections provide concurrent changes in temperature, precipitation and other variables. In the modelling set‑up, you chose to hold precipitation constant and perturb only temperature (+2 °C, +4 °C), which effectively turns your experiments into temperature‑sensitivity tests rather than full climate‑scenario runs. This is a reasonable and useful first‑order approach to isolate the role of warming on snow processes, recharge timing and evapotranspiration, but it should be clearly framed as such. I recommend that you (i) explicitly acknowledge that realistic future scenarios involve both temperature and precipitation changes, (ii) justify why precipitation is kept unchanged here (e.g. to avoid adding precipitation‑scenario uncertainty and to focus on snow‑related mechanisms), and (iii) state that using full regional climate projections (with modified precipitation) would be a logical next step for more comprehensive impact assessments.
In the discussion of evapotranspiration responses to warming, the manuscript rightly mentions the role of increasing temperature (via the Oudin PET formulation) and briefly of rising atmospheric CO₂, but it would be important to acknowledge that changes in phenology (longer growing seasons, earlier leaf‑out, shifts in vegetation activity) are also expected to significantly affect both potential and actual ET. Earlier onset and extended duration of transpiration can partly offset or amplify temperature-driven PET changes and will influence the seasonal partitioning between ET and recharge. I recommend adding a short paragraph noting that future ET trajectories will not depend on temperature and CO₂ alone, but also on phenological shifts and vegetation dynamics, which are not explicitly represented in the present modelling framework. This will better situate your experiments within the wider set of ecohydrological feedbacks expected under climate change.
Specific comments
Section 3.1 is missing information about the spatial distribution and discretization of model parameters and the subcatchments . The subcatchments should be additionally delineated in Fig 1.
L174 Groundwater recharge: It should be mentioned how this was estimated
L256f This information should be provided earlier in the model description (Section 3.1), where it is currently missing.
L257 compartment E – add reference to Figure or explain what this is
L331 either a notation table should be provided, or the parameters should be explained everytime when used. Same applies for Table 1- I’d suggest either adding a column explaining the variables or referring to a notation table
Technical corrections
L171 precipitation is – there is no plural form
L187 add reference or link to the used version of KarstmodCitation: https://doi.org/10.5194/egusphere-2025-5654-RC3
Data sets
Gillardes Spring discharge Bruno Arfib et al. https://data.oreme.org/snokarst/snokarst_map#zoom=9&lat=44.814&lon=5.196&layer=OpenStreetMap&overlays=TTT&selected_layer=snokarst_station&selected_station=44.76032
SAFRAN dataset Météo France https://geosas.fr/edr-viewer/
Snow depth measurement dataset Météo France https://donneespubliques.meteofrance.fr/?fond=produit&id_produit=94&id_rubrique=32
Model code and software
KarstMod modelling platform Naomi Mazzilli et al. https://hal.science/hal-02071006
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 185 | 128 | 18 | 331 | 27 | 9 | 11 |
- HTML: 185
- PDF: 128
- XML: 18
- Total: 331
- Supplement: 27
- BibTeX: 9
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Please find attached my comments.