the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2026
Climate change intensifies hydrological seasonality in Denmark: Insights from an integrated model assessment
Abstract. Temperate regions across Europe, such as Denmark, are projected to be subjected to substantial changes in the hydrological cycle due to climate change. Changes in climate can materialize as general changes in long term means, extremes, or in seasonal patterns, e.g., dampening or intensification of the seasonal contrasts. Changes in seasonal patterns can affect the hydrological cycle in various ways, due to the interlinkage between hydrological compartments. To detect, track and quantify the impact of changes in climate and seasonal patterns, integrated hydrological modelling is needed. This makes Denmark an ideal test case due to the established integrated and physically based National Hydrological Model of Denmark (DK-model). Utilizing climate projections from 17 RCP8.5 climate models, downscaled and bias-corrected for Denmark, we calculate climate change impacts on both overall values and seasonality for the variables soil moisture, streamflow, shallow and deeper groundwater to the end of the century. Moreover, standardized hydrological drought indices are calculated for the same variables. Climate change projections point towards a future with higher annual precipitation, mainly due to wetter winters, while climatic water balance deficits increase during summer; thus, intensifying the seasonal contrast. The increased contrast is reappearing in the fast-responding hydrological variable, soil moisture; while streamflow and shallow groundwater clearly reproduce increase during the wetter winter, the summer signal differs. The deep groundwater systems experience higher future groundwater heads across the entire year. Common for all variables is a larger seasonality, defined as contrast between intra-annual low and highs. Notably, the ensemble of hydrological projections is more in agreement regarding the seasonal contrast than on the direction of absolute change, with results agreeing for 85 % to 99 % of the area of Denmark on increased seasonality, whereas only agreeing for 50 % to 98 % on the absolute direction of that change. The drought indices exhibit a similar seasonal change, with more droughts during summer and more wet anomalies during winter for soil moisture, while summer droughts for streamflow and shallow groundwater partially are buffered by wetter winters and the related recharge increase. In summary, the results indicate that despite considerable increases in precipitation, projected climate change for Denmark is expected to enhance hydrological seasonality instead of producing a uniform transition to a wetter regime, potentially impacting the climate adaptation and mitigation effort, agricultural yields, and water supply.
- Preprint
(4314 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-1531', Benjamin Poschlod, 17 Jun 2026
-
AC1: 'Reply on RC1', Raphael Schneider, 22 Jun 2026
Our most honest gratitude to Benjamin Poschold for his constructive and thorough review. Concerning the reviewer’s major comments:
We understand the critique on RCP8.5 not being considered a plausible trajectory anymore. And it is a good point to be brought up in the review. Indeed, the most clean solution would the suggested option (b) to re-structure everything In terms of global warming levels. Based on our current setup this, however, will require too much extra work. It is, however, something we already discussed and will consider in upcoming hydrologic projections. We will therefore follow the reviewer’s suggestion (a), i.e. centre the analysis around RCP8.5 mid century (2041-2070) results, and reformulate so that the expressed concerns regarding plausibility of RCP8.5 are clearly stated. So this means changes to Fig. 3, 4, 6, 9 (and respective figures in the appendix) and respective parts of the Results section, as well as, of course the introduction of the climate models in Sect. 3.2 and relevant sections in the Discussion. But, indeed, the main conclusions remain the same, regardless of the chosen period (mid or end century).
Thanks a lot also for pointing out the sign inconsistency when presenting depth to phreatic and its changes. Internally, we use depth to phreatic with negative signs indicating values below ground. Then, an increase of the value means the groundwater table moves up / closer to the surface. However, we present results in Fig. 5 etc without that negative sign. So, thanks for spotting this. In the revised version, we’ll make sure to be consistent, and use the more common terms “shallow groundwater” or “uppermost groundwater” table.
Thanks a lot also for minor comments in the attached annotated pdf, appreciated input to clear up some things and correct minor mistakes. To name a few explicitly:
We appreciate the idea of modifying Fig. 1 by adding an overview map of Denmark, and will do so for the revisions.
We will also improve the description of the performed bias correction, and improve clarity on the calculation method and bias correction of PET (and its implications).
All in all, we foresee no issues in addressing the raised concerns in this review, and look forward to improving our manuscript accordingly in the revisions.Citation: https://doi.org/10.5194/egusphere-2026-1531-AC1
-
AC1: 'Reply on RC1', Raphael Schneider, 22 Jun 2026
-
RC2: 'Comment on egusphere-2026-1531', Anonymous Referee #2, 18 Aug 2026
Schneider et al. 2026 present a timely and well-presented study. The integrated national model over Denmakr presents an interesting view of how climate signals propagate through SM, streamflow, and GW. Its emphasis on seasonal contrasts is useful for both, the science and water management. I have read RC1 and the authors' response, so I do not repeat the comments on RCP8.5 or shallow-groundwater terminology. My main concern is, however, that the claim that seasonal amplification is more robust than changes in mean conditions is not yet entirely sufficiently demonstrated. I therefore recommend a major revision, as explained further below.
- I would suggest that the definition and robustness of the seasonality metric need some more clarification. Seasonality is defined as the maximum minus minimum monthly value (Lines 90 and 358-365). However, it is not fully clear to me, whether this range is calculated for each year and then summarized, or from the 30-year monthly climatology(?). I expect that the range can be sensitive to one month, timing shifts, etc. Please clarify this. Then, the variable-specific marginal thresholds in Table 2 and the 75 % agreement criterion also need justification. As this can matter when shifting the focus to mid-century: Table 2 shows clear increases over only 55 % of the area for deep groundwater and 75 % for shallow groundwater. Therefore, the abstract's current range of 85-99 %, which describes end-century results, should be revised.
- Lines 255-258 define the factor as Q_reference / Q_future, but state that values above 1 indicate increasing future streamflow. I guess, that this interpretation is mathematically reversed: under the stated definition, a value above one means Q_future is lower than Q_reference, whereas a value below one means future streamflow is higher. Or did I overlooked something? Then, Section 4.1 and Figure 4 instead appear to use and interpret Q_future / Q_reference. This inconsistency may affect the direction and magnitude of the reported changes, not only the notation. Please verify the calculation used in the analysis code and all associated figures and tables, then define and apply one ratio consistently throughout. If the intended factor is Q_future / Q_reference, Lines 255-258 should be corrected accordingly.
- The model simulations are equally weighted although several share a driving GCM or RCM. Please list all GCM-RCM-initial-condition chains. If this is the ensemble in Denager et al. (2026, Table S5; https://doi.org/10.5194/bg-23-441-2026), five runs use HadGEM2-ES, four EC-EARTH, and three MPI-ESM-LR. Thus, 12 of 17 runs originate from only three GCMs, so the median and 13-of-17 agreement threshold may overweight their signals. Authors could consider repeating the key results with equal weight per driving GCM, or one representative run per GCM, before describing them as robust.
- Figure 2 presents an evaluation of groundwater-level amplitude, but this is based on the calibration period and shows an underestimation of approximately 0.5 m. Please discuss how this bias may affect the interpretation of projected changes in seasonality. No soil-moisture evaluation is presented here, although soil moisture provides the clearest summer-drying signal. Schneider et al. (2026) report lower ESSMI skill and identify limitations related to only three soil-moisture stations, the lumped root-zone representation, and fixed climatological vegetation development. Please summarize these limitations and clarify that agreement across the climate projections remains conditional on a single hydrological model structure and parameter set.
- The integrated model is the study's main strength, but the process explanations remain qualitative. Winter recharge is used to explain summer buffering, and irrigation to explain local deep-groundwater drought, without showing recharge, actual evapotranspiration, drainage, baseflow, or abstraction changes. A compact seasonal water balance would make the claimed propagation much more convincing.
Minor comments
- Line 355: “significant exceptions” may imply statistical significance, although no statistical test is presented. Please use “notable” or “substantial.”
- The main text refers to the winter and summer maps as Figures B3 and B4, whereas the appendix labels them B2 and B3. The caption of Figure B1 also says it is equivalent to Figure 3, although it is equivalent to Figure 4?
- The data and code statement say that scripts and results will be provided upon request. Please, double check the latest Copernicus guidelines, regarding placement of some scripts and data under persistent repository with a DOI.
Citation: https://doi.org/10.5194/egusphere-2026-1531-RC2 -
AC2: 'Reply on RC2', Raphael Schneider, 24 Aug 2026
Thanks a lot to Reviewer #2 for their constructive feedback – highly appreciated. Concerning their major comments, we outline how we plan to address them in a potential revision of our manuscript:
- Seasonality is calculated based on the max-min range of each year, irrespective of the timing of min and max. We also looked into timing shifts during the preparation of this manuscript; at monthly scale, however, no change can be detected for either variable. We will make this more clear in the revised manuscript. Moreover, correctly noted that the abstract states end-century values; we will adapt accordingly (but, as previously mentioned in the reply to Reviewer #1, the overall conclusion remains the same)
- Our apologies, thanks for spotting that mistake. The streamflow factors indeed are calculated as Q_fut / Q_ref. We will correct lines 255ff.
- Yes, this is the ensemble seen in Denager et al. 2026 Table S5, and which originally is described by Pasten-Zapata et al. 2019. We will address the raised concern on the robustness of using only a limited number of GCMs in the revisions. Currently, we cannot tell which way we’ll go, but the reviewer’s suggestions with equal weights per GCM or a representative run per GCM seem good suggestions and definitely doable.
- Those are some valid points on model performance. Concerning groundwater level amplitude: True, the error on the simulated groundwater level amplitudes is substantial. However, there is skill in the model in distinguishing small and large amplitudes. For soil moisture, no suitable observations exist beyond potentially the 3 CRN stations discussed in Schneider et al. 2026. Hence, no soil moisture data was part of the calibration; hence it is missing in Figure. We will mention this as a limitation of the study in the revised version. We also want to point out that the overall seasonal dynamics of the model are simulated robustly, which can e.g. be deducted from the (besides soil moisture) successful validation of the model against drought indices and their propagation throughout the hydrological cycle (Schneider et al. 2026); and the generally reasonable model performance (e.g. summer streamflow water balances etc; Figure 2).
- Thanks for that inspiration. We agree, changes to ETa, recharge and similar fluxes could be a relevant addition. Whilst we would like to keep the climate change impacts more qualitative in this study (and focus on the changes to seasonality, instead of, e.g., climate change impacts on the amount of the available groundwater resource or similar), we see the point of including some effects, beyond the four major states and fluxes in Figure 4 (which are soil moisture, streamflow, shallow groundwater, deep groundwater). Potentially by adding future values to Figure 1, or in a separate, new figure or table (in the appendix?)
All in all, we foresee no issues in addressing the raised concerns in this review, and look forward to improving our manuscript accordingly in the revisions.
Citation: https://doi.org/10.5194/egusphere-2026-1531-AC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 194 | 85 | 17 | 296 | 27 | 14 |
- HTML: 194
- PDF: 85
- XML: 17
- Total: 296
- BibTeX: 27
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors investigate how projected climate change propagates through the interconnected hydrological system of Denmark, using the physically-based and integrated National Hydrological Model of Denmark (DK-model, MIKE SHE based, 500 m resolution, 24h time steps). They force the model with 17 EURO-CORDEX RCM projections (CMIP5-driven, RCP8.5), downscaled and bias-corrected for Denmark, and compare a mid-century (2041-2070) and an end-century (2071-2100) period against a 1991-2020 reference. The analysis covers four compartments of the hydrological cycle: root-zone soil moisture, streamflow, depth to the phreatic surface (shallow groundwater), and the deeper aquifer groundwater head. In addition to changes in monthly mean conditions, the authors compute standardized hydrological indices to quantify changes in the frequency and severity of dry and wet anomalies, and explicitly quantify changes in seasonality.
This is a careful, well-written, and policy-relevant study with an appealing visual presentation of results. The integrated modelling approach, and the multi-compartment treatment of seasonality is a nice contribution that distinguishes it from the more common single-compartment assessments. However, there are still some points, where I see the need for improvement. My major comments are below, while I provide an annotated version of the PDF attached for the minor stuff. In sum I'd still consider it a "minor revision".
Major comments:
1. RCP8.5 framing and the emphasis on end-of-century
The study relies exclusively on RCP8.5 ("business-as-usual storyline", L197-199). RCP8.5 is now widely regarded as an implausible rather than a business-as-usual trajectory (https://doi.org/10.5194/gmd-19-2627-2026), and this is especially consequential for the end-of-century period, where RCP8.5 diverges most strongly from more plausible pathways. As written, the abstract, the headline figures (Figs. 4, 5, 6, 7, 9), and most of the narrative are anchored on the 2071-2100 results, which risks overstating the magnitude of the projected changes relative to a realistic emissions future. I recommend one of the following two options:
(a) Shift the framing, figures, and discussion towards the mid-century (2041-2070) period as the primary result, relegating end-century to a secondary/illustrative role. The mid-century results already exist in the appendix (Fig. B1, Fig. B4) and could be moved to the main text. The "business-as-usual" wording should be removed or corrected regardless.
or preferably
(b) Re-structure the analysis in terms of global warming levels (GWLs, e.g. +1.5, +2, +3 °C above 1850-1900) rather than fixed time slices under a single RCP. This decouples the results from the RCP8.5 trajectory and also reduces the climate model uncertainty, where the different equilibrium climate sensitivity (ECS) of the driving GCM is a major uncertainty source for the diverging 17 simulations. GWLs further make the results transferable to other scenarios and to the GWL framing used in IPCC AR6. The years at which each driving GCM reaches a given GWL are e.g. provided by Hauser et al. 2022 and are openly available: https://doi.org/10.5281/zenodo.3591807
The authors could still use 30-year periods instead of Hauser's standard 20 years.
At minimum (option a), the RCP8.5 limitation and its implication for the end-century numbers should be stated explicitly (at best briefly in the methods when describing the data and later in detail in Sect. 5.2.). The main conclusion (increased seasonality is a more robust signal than the mean change) will very likely hold in either case.
2. "Depth to phreatic" label vs. the sign of the reported change
There is a sign/labelling inconsistency around the shallow groundwater variable throughout the paper. "Depth to phreatic surface" is to my understanding defined as depth in metres below ground surface (L263, L339-340: "depth to phreatic (below surface [m])"). By that definition, a larger value means the water table is deeper, i.e. shallow groundwater is lower / further from the surface. However, the results consistently treat positive changes in "depth to phreatic" as more/higher groundwater (e.g. L324-326: "the groundwater levels generally rise. For the depth to phreatic, climate change impacts show a clear seasonal signal, with highest increases during late winter (up to +0.27 m)"; Fig. 4 and Fig. 6 colour positive changes blue = increase = wetter).
I'd suggest to fix this by re-naming the label to "shallow groundwater level" consistently.