the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Mar 2026
The rise of stratification – Climate-induced changes in the thermal structure of a shallow polymictic lake until 2100
Abstract. Climate change exerts a significant influence on lake ecosystems by altering stratification and thermal dynamics. These changes are commonly quantified using climate model projections to assess future conditions. However, climate model simulations generally have a daily temporal resolution, which is inadequate for resolving the fast, sub-daily processes governing shallow lakes. Consequently, long-term impacts of climate change on shallow lakes remain underrepresented in literature. This study presents a modeling workflow for simulating and quantifying future climatic conditions in shallow environments, where sub-daily resolution is necessary. The proposed workflow comprises a weather generator for the temporal downscaling meteorological forcing, as well as a physics-based one-dimensional model for simulating lake thermal dynamics. The workflow is applied to Lake Blaton, a large polymictic lake, to assess the effects of the projected climatic changes on the lake through the end of the century. Changes are analyzed as a function of time and water depth, under the RCP4.5 and RCP8.5 climate scenarios, using an ensemble of 14 climate model simulations. Our findings indicate that stratification is likely to intensify throughout the century, in a complex interplay with water depth, in which mutually enhancing effects are accompanied by progressive dampening. The number and duration of stratified events show a slight increase, which does not reflect the magnitude of the projected intensification in stratification, suggesting that wind forcing remains a dominant factor in regulating stratification dynamics. Lastly, the evaluation of lake heatwaves using a temporally varying baseline indicates no significant changes in their characteristics.
- Preprint
(1524 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 08 May 2026)
- RC1: 'Comment on egusphere-2026-708', Anonymous Referee #1, 01 Apr 2026 reply
-
RC2: 'Comment on egusphere-2026-708', Anonymous Referee #2, 10 Apr 2026
reply
The manuscript addresses climate-driven changes in thermal stratification in shallow polymictic lakes. However, the central research questions, methodology, and conclusions lack sufficient novelty.
1) Over the past decade, numerous studies have extensively examined the response of lake thermal structure and stratification dynamics to climate change, including in shallow and polymictic systems. For example, Woolway et al. (2022) clearly demonstrated that climate warming leads to earlier onset, prolonged duration of stratification, and altered mixing processes, with important ecological consequences. More recently, Sullivan et al. (2025) further provided mechanistic insights into how climate change regulates lake thermal regimes and ecosystem functioning.
R Iestyn Woolway, Sapna Sharma, John P Smol, Lakes in Hot Water: The Impacts of a Changing Climate on Aquatic Ecosystems, BioScience, Volume 72, Issue 11, November 2022, Pages 1050–1061, https://doi.org/10.1093/biosci/biac052
Sullivan, C.J., Read, J.S. and Hansen, G.J.A. (2025), Climate-driven alterations of lake thermal regimes. Limnol Oceanogr, 70: 2348-2364. https://doi.org/10.1002/lno.70128
In shallow and polymictic systems specifically, similar questions have already been explored in depth, see (Török et al., 2025; Zhang et al., 2025). The study seems to emphasize sub-daily processes as a key innovation. However, this aspect is also not new. Previous studies have already investigated sub-daily dynamics in shallow lakes. For example, Frassl et al. (2018) examined sub-daily variability in lake temperature dynamics, and Piccioni et al. (2021) demonstrated that thermal stratification and mixing can alternate on hourly timescales.
Török S D, Torma P. Long-term changes in summer stratification of a shallow polymictic lake by climate change and anthropogenic water level regulation[J]. Science of The Total Environment, 2025, 1009: 181025.
Zhang, M., Leppäranta, M., Heikkilä, M., Weckström, K., Korhola, A., Kirchner, N., … Weckström, J. (2025). The thermal structure of small and shallow Arctic Fennoscandian lakes. Arctic, Antarctic, and Alpine Research, 57(1). https://doi.org/10.1080/15230430.2024.2433829
2) While the authors devote substantial effort to describing the weather generation and overall modeling workflow, the actual lake model setup, calibration, and validation are insufficiently documented. These aspects are critical for any numerical simulation study, yet the authors only briefly states that “model setup, including the calibration and validation procedures and their results, are provided in Török and Torma (2025, 2024).” This is not adequate. Moreover, In Section 3.1, the validation focuses primarily on the temporal downscaling procedure. However, it should also include validation of the lake model itself, particularly its ability to reproduce thermal dynamics, including both stratified and mixed conditions. Without such validation, the reliability of the projected results remains uncertain.
Furthermore, Lake Balaton is a large (surface area ~596 km²) and elongated lake, where spatial heterogeneity, especially differential heating driven by meteorological forcing, is likely to play an important role in thermal dynamics. The current modeling approach appears to neglect the spatial variability of meteorological inputs, which may limit its ability to accurately represent lake-wide processes.
In addition, although the authors considers several meteorological variables (e.g., shortwave radiation, air temperature, and relative humidity), the treatment of wind forcing is unclear and appears insufficient despite some discussion of the limitation of wind in Discussion. Wind is a primary driver of mixing and thermal structure in open-water systems, and its long-term variability is well documented. For example, surface wind speeds have increased by 10–20% in some regions (e.g., over Lake Superior; Desai et al., 2009), while decreasing by similar magnitudes elsewhere (Vautard et al., 2010). Neglecting or inadequately representing wind forcing raises concerns about the physical realism of the model and the robustness of the conclusions.
Desai A R, Austin J A, Bennington V, et al. Stronger winds over a large lake in response to weakening air-to-lake temperature gradient[J]. Nature Geoscience, 2009, 2(12): 855-858.
Vautard R, Cattiaux J, Yiou P, et al. Northern Hemisphere atmospheric stilling partly attributed to an increase in surface roughness[J]. Nature geoscience, 2010, 3(11): 756-761.
3) The Results section appears to lack direct presentation of the simulated thermal structure. In particular, there is no clear visualization of depth–time thermal dynamics showing the evolution of stratification and mixing. Instead, the manuscript immediately proceeds to climate change projections and derived metrics. This makes it difficult to assess the physical realism of the model simulations. For a modeling study of this type, it is essential to first demonstrate that the model can reproduce realistic thermal behavior, including both stratified and mixed conditions, before moving on to climate-driven changes. The absence of such direct simulation outputs (e.g., temperature contour plots over depth and time) significantly limits the reader’s ability to evaluate the robustness of the results.
In addition, regarding the climate forcing, the analysis appears to focus primarily on precipitation and air temperature. It is unclear why other key meteorological drivers—particularly wind—are not explicitly considered in terms of their long-term trends. Wind forcing plays a dominant role in regulating mixing and stratification dynamics, especially in shallow polymictic systems. Neglecting its projected changes raises concerns about the completeness and physical consistency of the forcing framework.
4) The Discussion section remains largely descriptive and does not sufficiently interpret the results within a mechanistic or conceptual framework. Wind forcing is a key driver of thermal stratification and mixing dynamics, especially in shallow polymictic lakes. However, the modeling framework appears to lack an explicit treatment of wind variability, particularly its long-term trends under climate change. This omission raises significant concerns regarding the reliability of the projections, as changes in wind forcing can fundamentally alter stratification and mixing regimes. Furthermore, the discussion of ecological implications is largely generic and reflects well-established textbook knowledge (e.g., increasing temperature leads to reduced dissolved oxygen, and stronger stratification promotes hypoxia). To enhance the scientific significance of the work, the authors should provide quantitative estimates—for example, how much dissolved oxygen is expected to decline under different climate scenarios?
5) There are also many concerns about Figs.
Figure 1 is overly simplified and lacks essential information. The figure should be structured into clearly defined panels (e.g., (a), (b), and (c)), each with explicit descriptions in the caption. It is also unclear whether the single measurement site, located in the western corner of the lake, is representative of the entire Lake Balaton system. Given the elongated morphology and known spatial heterogeneity of the lake, one site may not adequately capture basin-wide thermal dynamics.
In addition, the figure lacks geographic coordinates. The colored lines should also be explicitly defined in the caption (e.g., as bathymetric contours or isobaths).
Figure 2 is overly complex and difficult to follow. The figure contains too many elements, including multiple GOTM simulations (1 h and 6 h), different meteorological datasets, and repeated bias correction steps, which together reduce clarity. In particular, the logic within the “Lake modeling” section is unclear. The relationship between the 1-hour and 6-hour simulations is not well explained, and it is not evident why both are required. The use of dashed lines further adds ambiguity, as their meaning is not clearly defined.
Moreover, the caption is overly descriptive but does not clearly communicate the purpose of the figure. It would benefit from an opening sentence that concisely states the figure’s objective, for example: “Figure 2. Schematic overview of the modeling framework used to simulate lake thermal dynamics under present and future climate conditions.”
Figure 3 provides an important validation of the weather generator; however, the comparison remains purely qualitative. The agreement between datasets is assessed visually without any quantitative metrics (e.g., RMSE, MAE, or statistical tests), which limits the strength of the validation. In addition, the use of probability density functions may not be the most clear way to present the comparison. More direct and interpretable plots (e.g., scatter plots, time series comparisons, or bias plots) could improve clarity and make differences between datasets more transparent.
Figures 5–10 appear to be repetitive and provide limited additional insight. While multiple figures are presented, they largely convey similar information (e.g., temperature trends, distributions, or stratification metrics) without offering new perspectives. For example, Figure 5 shows temperature trends, Figure 6 presents distributions, and Figures 7–9 repeat similar analyses for related variables. This redundancy reduces the overall impact of the results. The authors are encouraged to consolidate these figures and focus on presenting the most informative and non-redundant results, potentially by combining variables or highlighting key relationships rather than repeating similar plots.
Given this context, the scientific question addressed here and the some simple results are not particularly novel. Overall, the study appears to be incremental rather than providing a substantial conceptual advance, and may be more suitable for a regional or more specialized journal.
Citation: https://doi.org/10.5194/egusphere-2026-708-RC2 -
RC3: 'Comment on egusphere-2026-708', Anonymous Referee #3, 26 Apr 2026
reply
Dear Authors,
First of all, I appreciate your great important work that significantly contribute to the lake investigation field. Studying the long-term climate-induced changes in Lake Balaton’s thermal structure is definitely giving the future researchers and policy makers a good applicable/scalable approach that can be used for further development. The comments below are just for improving the current work and making it interesting for other readers as well.
My opinion is: Accepted for publication after addressing the comments attached.
Abstract:
Although the general information about the workflow has been mentioned, but still, the reader does not know what kind of modeling will be seen in the text. The text is clear and well explained in some aspects but there is a lack of modeling and some quantitative key results if there’s any.
Introduction :
Line 32: Not just in “extreme cases”, it might be possible for some shallow lakes, the oxygen depletion happens depending on the stratification strength, the bathymetry of the lake, the lake surface water temperature etc.
(Line 33): In general, when the algal blooms increases then the chance of oxygen depletion increases and may lead to water quality problems.
Line 34-35: The conclusion seems overclaimed: Climate change might generally increase thermal stability not “Even” in polymictic. The word “even” gives the a meaning of separation from the rest of the types of lakes. We should not conclude this evenness now for the climate change effect which is generally affecting all aspects of the water bodies.
Line 51: The typos in terms of writing: Either “They are more sensitive” OR “they sensitively reacting to climate change ….”
Line 50-60: While the tonality of the text before line 49,50 turned into guidance, policies, upper principals, then in these lines, we see again some introductory information that takes the reader back to the basics of the issue.
Lines 65-75 are also suffering from this problem. These lines along with the previous lines are not coherent and consistent with each other although they have been written very carefully and clearly.
Line 80: Still, we have not seen the other previous literature problems (stating the research gaps and questions that have not been covered through the previous works that can enforce/push us to continue why this is an important research that should be done. Also the model is still unclear to me. Until I go to the methodology and look at the details then I realize you are supposed to use GOTM with a new workflow that has been designed by you. If the novelty is the workflow, we have not seen in the introduction stated out clearly explicitly. If the novelty is the threshold you define or you design, it is not clear at least to me. The novelty is the time period that is being predicted, I don’t see the novelty statement selling this point.
Line 85-end: I don’t grasp the key points of these final paragraph. In the meantime that it tries to play the final paragraph role, it still shows like abstract, introductory, results discussion etc. I believe, the final paragraph is like punching the readers to proceed with the text.
In general, the introduction parts have been written good but they are not well structured and don’t follow each other however the writers might have an opposite opinion. For example, I surely know the importance of studying polymictic lakes but from the introduction I got confused whether this is study that forced to be related to polymictic because of any reason (data availability etc) or there was an urgent issue about this type of lake (that are more sensitive to the climate change and anthropogenic activity). If you see I have seen the words but I could not get this from the whole introduction which takes me to the second part of the paper (materials and methods), where you are talking about the study area.
Materials methods:
line 91: As I previously mentioned I don’t find any good reason of choosing this area as an example except the availability of the data. Is there any specific condition that excludes this lake from the other please state it otherwise, if it is just because its data was available and clean , please change the text accordingly.Lines 102: It would have been good if we had either a subheading about the data acquisition and solely talking about it or making it smoother for those who has not been familiar with the technicalities of the employed model. I see the aims of using a part of data for training the weather generator to downscale …., without prior knowledge on the details of the model unless you say it is for knowledgeable readers.
Results and discussions:
General opinion about the results discussion: Following the section is a bit hard and I suggest try to smooth it. Also, there is a lack of inconsistencies between the introduction aims and scope, and what the results are presenting which again goes back to the issues with the introduction.
Line 223: Typos, repeating “Materials and methods” instead of “Results and discussion”
Technical comments: The authors specifically designed a workflow to address the lack of sub-daily data in climate models. You validated the 6-hour downscaling against hourly ERA5 data for the 2000–2020 period. The comparison showed that while statistical consistency (mean, median, and standard deviation) was preserved, there were minor discrepancies in extreme values: water temperature was slightly overestimated (max difference of 0.3 oc), and stratification intensity () was slightly underestimated (max difference of 0.9 J m-3)
Another comment is related to the use of linear Bias correction. Although Future projections assume that changes scale linearly from the baseline, it is allowed to do that, but at the same time it introduces a potential source of error if climate responses become non-linear toward the end of the century.
Major comment on the results: One of the key issues of the predictions and projections long-term or short term is the “Confidence Intervals”, and “Uncertainty levels”. It is important to know to what extent the prediction is credible upper bound and lower bound. This is statistically significant to ensure if the predictions are correct first and more importantly safe. The figures that are showing the projections and predictions, should also show the confidence intervals and shaded area of the uncertainty upper/lower bounds.
Data sets
FORESEE Eötvös Loránd University https://meteordata.elte.hu/FORESEE/index.html
ERA5 ECMWF https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land?tab=overview
Model code and software
GOTM Lars Umlauf, Hans Burchard, Karsten Bolding https://gotm.net/portfolio/
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 232 | 82 | 15 | 329 | 7 | 12 |
- HTML: 232
- PDF: 82
- XML: 15
- Total: 329
- BibTeX: 7
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The overall approach of this study seems to be sound, but the generation of high-resolution data, the construction of the lake thermodynamic model, and the stratification simulation lack sufficiently reliable evaluation and validation. Additionally, many key parameters and settings are not clearly clarified, and the presentation of the projected results is also relatively simplistic.
Section 2.1: The lake is divided into four distinct basins, but the basins should be presented in the figure. The lake has an average depth of 3.5 meters and is driven by meteorological factors. However, it remains unclear to what extent the lake is stratified. Are there specific observational data to support this? Since the lake is divided into four basins, one monitoring point seems a bit insufficient to be comprehensive. Additionally, what indicators are being monitored at this single monitoring point in Fig. 1? Although ECMWF and FORESEE are only used for prediction analysis, and direct comparison among multiple schemes can eliminate the influence of their inherent errors, validation against ground observation data is still necessary.
Section 2.4: Regarding the lake model, it is a one-dimensional model—can it effectively carry out thermodynamic simulations of this lake? Additionally, when setting the model parameters and boundary conditions, the authors cited many references for support, yet there is a lack of observational data to underpin the model construction process. For example, how does this one-dimensional vertical lake model account for lateral inflow and outflow? The previous text mentioned that the lake is divided into four zones. Through different water depth settings, which zone’s thermodynamic stratification does this model actually represent? For example, regarding parameters and variables, water density is related to temperature, and the calculation formula should be provided. Additionally, how is the wind speed on the lake surface considered? These key aspects are not addressed.
Section 3.1: What is being validated in the figures and tables (Fig. 3 and Table 1)—water temperature or air temperature? Additionally, for the validation results, it is recommended to present them using scatter plots or similar methods, rather than just probability density curves. Since the key to this study is the generation of high-resolution input data, the reliability of the results generated by the combination of multiple models in this part needs to be thoroughly validated.
Section 3.4: Regarding lake stratification, it seems that the paper uses the potential energy anomaly index, which is calculated based on the water temperature profile. First, the specific simulation results of the water temperature profile are not presented anywhere in the text. Second, how this index actually indicates vertical stratification in the lake is not clearly explained. Although this index may be a quantitative measure, there are many well-established indicators for lake stratification that might be more convincing than the one used in this paper, such as the Schmidt Index, Lake Number.