the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Feb 2025
Groundwater recharge in Brandenburg is declining – but why?
Abstract. Brandenburg is among the driest federal states in Germany, featuring low rates of ground water recharge (GWR) across large parts of the state. This GWR is fundamental to both water supply and the support of natural ecosystems. There is strong observational evidence, however, that GWR has been declining since 1980: first, river discharge (which is almost exclusively fed via GWR) has been significantly decreasing in many catchments (by around 40 % since 1980). Second, ground water levels in the groundwater recharge areas show a significant long-term decline. In this study, we search for potential reasons behind this decline, by investigating five catchments across Brandenburg that we consider as largely unaffected by direct anthropogenic interference with the water balance. Using the Soil-Water-Atmosphere-Plant model (SWAP) to simulate long-term trends in GWR, we found that significant increases of air temperature, solar irradiation and leaf area index (LAI) since 1980 acted towards a decrease in GWR in the order of -21 to -4 mm a-1 per decade from 1980 to 2023. The Brandenburg-wide LAI trend of +0.1 m2 m-2 per decade was inferred from a recently published, spatio-temporally consistent LAI reconstruction. The contribution of this LAI trend to the decrease of GWR amounted to -5 to -3 mm a-1 per decade. Based on our results, we consider it as very likely that the decrease in discharge since 1980 can be explained by a decrease in GWR which, in turn, was caused by climate change in combination with an increasing LAI. However, we also found that precipitation trends can be highly incoherent at the catchment scale. Even though these precipitation trends are not significant, they can have a fundamental impact on the significance, the magnitude and even on the sign of simulated GWR trends. Given the uncertainty of the precipitation trend, four out of five catchments still appear to exhibit a gap between negative simulated GWR trends and more negative observed discharge trends. We provide a comprehensive discussion of possible reasons and uncertainties to explain this gap, including the effects of the limited length and the inhomogeneity of climate and discharge records, the role of land cover and vegetation change, irrigation water consumption, latent anthropogenic interventions in the catchments water balance, uncertainties in ground water table depth, as well as model-related uncertainties. Addressing these uncertainties should be a prime subject for prospective research with regard to the effects of environmental change on GWR in Brandenburg. Water resources management and planning in Brandenburg should, however, already take into account the possibility of GWR to decrease further. Given the fundamental importance of precipitation trends and their large uncertainty in future projections, we strongly advise against putting our hopes in a future increase of GWR as projected mainly on the basis of expected future increases in winter precipitation.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(13288 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(13288 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-222', Anonymous Referee #1, 06 Mar 2025
This is a very interesting paper that analyses the groundwater recharge behavior in 5 catchments in Brandenburg. The paper fits very well within the scope of the special issue.
I think the paper has great potential, but it is somewhat hindered by missing information and lacking methodology descriptions. This makes the paper somewhat difficult to follow, I often had to backtrack quite a lot in the text to see if I missed some detail or information. Therefore I am sending below my recommendations to improve the manuscript.
My main question is about the model setup. Are the models set up on a per catchment basis, or over a grid or other spatial sub-units? This was very confusing to me – sometimes I was sure everything is aggregated over catchments, but in some parts further inhomogeneities were resolved. It would be important to include more details about this in the text. Please find my specific comments below.
Specific comments:
L33: These few paragraphs could use some references to support the text.
L36: groundwater
L42: Does the term “combination” here means a gauging station after multiple rivers join together?
L50: can you provide a reference?
L52: what about cross-flow with lower aquifer layers? The recent study of Tsypin et al., 2024 showed that deeper layers interact with the uppermost at certain geological settings, mainly where the Rupelian clay layer is eroded. These flows could act as a water source or sink as they flow upwards or downwards at different locations.
Tsypin, Mikhail, et al. "Modeling the influence of climate on groundwater flow and heat regime in Brandenburg (Germany)." Frontiers in Water 6 (2024): 1353394.
Fig.1: Which wells were used to create this map? Is the data coverage homogeneous, or are there any areas with fewer information?
This figure would be a better place to show the used gauge locations than fig. 5.
L67: “any long term trend in discharge can thus be interpreted as a long term trend in GWR.”
Can you elaborate on this a bit further: what is the long-term criteria, why short term changes cannot be used in such way? Is there a model that can show this behavior/or a past study where this was investigated?
L108: reference to the dataset
L125: reference needed
L129: Can you support this statement with a reference, or by referring back to the introduction
L136: What does partly mean here?
L143: I am just curious here: what about not interpolated, but dynamically downscaled datasets, such as CERv2 (https://www.tu.berlin/klima/forschung/regionalklimatologie/mitteleuropa/cer) . Would they not be a better choice for gridded data?
L161: parenthesis missing
Were there any weighting used when multiple gauges were included? I think using a buffer area as a criteria here could easily introduce further uncertainties to the data (as precipitation in Brandenburg could be very heterogeneous, especially during extreme rainfall events) – does including them make a big difference? (does it worth it to use them instead of just the climate station)
L196: The used dataset is a categorized dataset for groundwater distance. Does its non-continuous nature create any issues for your analysis?
General question for methods: Was the analysis done over a grid, over catchment or any other spatial unit? It is not clear for me from the text
Also, for a better comprehensibility, this section could really use a flowchart on how the different parameters interact, or even a conceptual figure on how the modeling concept looks like.
L214: I really like this concept for filling in missing data.
Can you give more details on how this package was used (as it is a specific module for photovoltaic modelling, it is not straightforward how to use it in this setting). Which function was used for modelling with what parameters?
L228: same question: what random forest model was used and how?
L266: How did you check that this spin up period was enough for the model?
L270: was the modelling run on a grid? – I am really confused about this at this point
L282: why was this setup needed?
Table3: this table is a bit complicated and difficult to understand. Please consider revising it.
L320: maybe: “(small triangles)”
Figure 4: can you add a legend instead of explaining everything in the captions?
L318: are the blue/red lines are simply drawn by connecting the points of the corresponding LAI setup?
L327: This could also be referred to as a linear relation between precipitation and GWR
L338: Can you explain further how the significance was calculated?
Figure 5: This figure is very complex, but I think the chosen visualizations are really the best way to show the findings. You could make the figure a bit less busy by removing the gauge locations from the map. Also positioning the legend next to the map is not ideal.
Could you consider adding another figure where (some of – maybe just the main configuration) the modelled GWR timeseries are shown against the discharge timeseries for the catchments? It would help the text in my opinion…
L351: rephrase “While this is plausible…”
L361: what are the blue and red lines, are they trendlines or just connectors between the points?
L365: which offset? Could you be more exact?
Can one state here that there is an unknown water loss from the system as the discharge trend is steeper?
L370: Does it mean more humid conditions or a faster change towards humid conditions?
L373: point out whiskers on the legend
L380: where was this pointed out?
How could the model have GWR estimations for different depths – was it over a grid?
5 limitation and uncertainties: I really like this section, but it could have a more exact title like “Explaining the gap between trends”
L437: a similar behavior was pointed out locally by Somogyvari et al., 2024 for a lake system in the region.
Somogyvári, Márk, et al. "A hybrid data-driven approach to analyze the drivers of lake level dynamics." Hydrology and Earth System Sciences 28.18 (2024): 4331-4348.
They also used a combination of different factors as an explanation for system water loss, together with an environmental tipping point. Did you consider such explanations?
L511: Could you rank the different explanations based on plausibility?
Citation: https://doi.org/10.5194/egusphere-2025-222-RC1 - AC1: 'Reply on RC1', Maik Heistermann, 12 May 2025
-
RC2: 'Comment on egusphere-2025-222', Anonymous Referee #2, 28 Apr 2025
The manuscript by Francke and Heistermann is giving an approach to explain dropping groundwater recharge in Brandenburg. Overall, the study is impressive and very interesting and will doubtless find its place in the journal, where it fits well into.
The manuscript is appropriately designed and consistent, figures are of high quality and in principle well prepared. Few minor things should be modified, which I marked below.
More specifically I was wondering, why the authors mainly base their arguments on discharge measurements only rather than including direct groundwater observations to justify dropping groundwater recharge.
During reading, I also had the problem, that I often had to jump back in the text to be sure I understand what is meant and don’t miss things. Sometimes expressions are difficult to read and to follow, but when authors carefully inspect the manuscript again, I’m confident it will be improved and become published in N-HESS.
Below, more specific comments and suggestions are given.
L30 f. I would rather use deep and shallow instead of distant and close water tables.
L38. In an intensely cultivated and urbanized region as Brandenburg, there might be increased abstraction by man different actors (farmers, water suppliers, private wells, etc.). How comes you generally state: “there is no basis to assume such a widespread abstraction of groundwater”? That statement is a fundamental prerequisite of the coming explanation for dropping gwr, but there is no evidence presented.
L39. How is it possible, an aquifer gets confined by a river? Please think about wording.
L50: is it indeed possible to generalize it as you do: “Given the highly permeable soils in Brandenburg and the resulting insignificance of surface runoff…“? I think the evidence coming from that statement is very serious and actually builds the foundation of the subsequent argumentation: dropping river discharge as mirror of dropping gwr, because baseflow is (in your argumentation) the only source of surface runoff.
I think that point, in combination with my concerns from line 38 (missing abstraction), should be treated with more caution and seriously justified.
Fig. 1: a) the names given therein are unclear in their meaning: are these cities, counties, catchments?
- b) in each of the figures, 2 numbers are given, but it is not clear what they represent. Also the unit is not clear to me: mm/a/dec, millimeter per year is okay, but what is “Dec”?
L69-73: I partly disagree with that statement. The analyses of gw-timeseries gives worthwhile information as for gwr and though the given issues are valid in a certain way, it is mentioned before, the catchments that are investigated are not influenced by abstraction. Of course, storativity is always an uncertain parameter, but I guess there should be few pumping tests available to at least estimate gwr in few areas, based on gw timeseries.
L75 ff. Please use consistent wording: shallow and deep gw tables and don’t switch between deep and distant.
L107: where is the dataset from? Reference?
L136: I do not understand the integration of neighbouring catchments: they do not contribute to the discharge of the gauge. And what does it mean to include them “partly”?
L162. There is a spare parenthesis at the end of the sentence.
L179. Where are the field LAI measurements were taken from? Just out of curiosity: during 2019-2024, forest dieback (as observed in other areas in Germany) did not apply to Brandenburg at all?
L196 ff. The steady state type of gw-depth information may not fit to the proposed changes in gw depth due to dropping gwr, am I wrong?
L246: where is soil moisture data taken from?
L251: is a constant head not contradicting the question, whether gwr is dropping and hence gw tables?
L265 ff.: is the concept of hydrotopes following the idea of HRUs in hydrology? How did you include the depth to groundwater, 13 classes of gw depth each mulitplied by land use, soil type etc.? How did you integrate the hydrotopes spatially: as grid or as a mesh of irregular/regular triangles?
L281: what is the unit “dec” standing for?
L291: now the explanation comes, much too late…
In general, it is hard to understand the concept of mm/a/dec. To become able to translate it into a logical unit, it needs a brief explanation at the very beginning.
L296: why was just LAI selected to be the controlling parameter of dropping gwr? Increasing ET is obviously a key-parameter as well controlling the soil moisture (fig. 3d).
L297: please explain CI
L344: where do I see it in Fig. 5? The figure is very complex but nice. However, I have difficulties to differentiate between various shades of grey in some of the graphs.
L362: in fig. 3 there is no real trend in precipitation observable, how comes it though brings the gwr to drop?
Fig. 6: how is it explainable, gwr is almost negligible in 4 out of 5 catchments fro shallow water tables? Contrastingly, the deeper groundwater receives reasonable gwr of 100-150 mm/a.
What does “area weighted” mean here?
How is areal fraction calculated?
L383 ff. I have difficulties to understand why the contribution of the shallow gw is that high, if their (however calculated) areal contribution in the total catchment is low. And why show shallow gw at all different gwr compared to deeper gw bodies? Is it a function of delay? But how was it modelled in 1D?
L403: there is a “by” too much.
L453f. Where stems the data from? A study from 2009 might not be representative to activities during the last decade.
Citation: https://doi.org/10.5194/egusphere-2025-222-RC2 - AC2: 'Reply on RC2', Maik Heistermann, 12 May 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-222', Anonymous Referee #1, 06 Mar 2025
This is a very interesting paper that analyses the groundwater recharge behavior in 5 catchments in Brandenburg. The paper fits very well within the scope of the special issue.
I think the paper has great potential, but it is somewhat hindered by missing information and lacking methodology descriptions. This makes the paper somewhat difficult to follow, I often had to backtrack quite a lot in the text to see if I missed some detail or information. Therefore I am sending below my recommendations to improve the manuscript.
My main question is about the model setup. Are the models set up on a per catchment basis, or over a grid or other spatial sub-units? This was very confusing to me – sometimes I was sure everything is aggregated over catchments, but in some parts further inhomogeneities were resolved. It would be important to include more details about this in the text. Please find my specific comments below.
Specific comments:
L33: These few paragraphs could use some references to support the text.
L36: groundwater
L42: Does the term “combination” here means a gauging station after multiple rivers join together?
L50: can you provide a reference?
L52: what about cross-flow with lower aquifer layers? The recent study of Tsypin et al., 2024 showed that deeper layers interact with the uppermost at certain geological settings, mainly where the Rupelian clay layer is eroded. These flows could act as a water source or sink as they flow upwards or downwards at different locations.
Tsypin, Mikhail, et al. "Modeling the influence of climate on groundwater flow and heat regime in Brandenburg (Germany)." Frontiers in Water 6 (2024): 1353394.
Fig.1: Which wells were used to create this map? Is the data coverage homogeneous, or are there any areas with fewer information?
This figure would be a better place to show the used gauge locations than fig. 5.
L67: “any long term trend in discharge can thus be interpreted as a long term trend in GWR.”
Can you elaborate on this a bit further: what is the long-term criteria, why short term changes cannot be used in such way? Is there a model that can show this behavior/or a past study where this was investigated?
L108: reference to the dataset
L125: reference needed
L129: Can you support this statement with a reference, or by referring back to the introduction
L136: What does partly mean here?
L143: I am just curious here: what about not interpolated, but dynamically downscaled datasets, such as CERv2 (https://www.tu.berlin/klima/forschung/regionalklimatologie/mitteleuropa/cer) . Would they not be a better choice for gridded data?
L161: parenthesis missing
Were there any weighting used when multiple gauges were included? I think using a buffer area as a criteria here could easily introduce further uncertainties to the data (as precipitation in Brandenburg could be very heterogeneous, especially during extreme rainfall events) – does including them make a big difference? (does it worth it to use them instead of just the climate station)
L196: The used dataset is a categorized dataset for groundwater distance. Does its non-continuous nature create any issues for your analysis?
General question for methods: Was the analysis done over a grid, over catchment or any other spatial unit? It is not clear for me from the text
Also, for a better comprehensibility, this section could really use a flowchart on how the different parameters interact, or even a conceptual figure on how the modeling concept looks like.
L214: I really like this concept for filling in missing data.
Can you give more details on how this package was used (as it is a specific module for photovoltaic modelling, it is not straightforward how to use it in this setting). Which function was used for modelling with what parameters?
L228: same question: what random forest model was used and how?
L266: How did you check that this spin up period was enough for the model?
L270: was the modelling run on a grid? – I am really confused about this at this point
L282: why was this setup needed?
Table3: this table is a bit complicated and difficult to understand. Please consider revising it.
L320: maybe: “(small triangles)”
Figure 4: can you add a legend instead of explaining everything in the captions?
L318: are the blue/red lines are simply drawn by connecting the points of the corresponding LAI setup?
L327: This could also be referred to as a linear relation between precipitation and GWR
L338: Can you explain further how the significance was calculated?
Figure 5: This figure is very complex, but I think the chosen visualizations are really the best way to show the findings. You could make the figure a bit less busy by removing the gauge locations from the map. Also positioning the legend next to the map is not ideal.
Could you consider adding another figure where (some of – maybe just the main configuration) the modelled GWR timeseries are shown against the discharge timeseries for the catchments? It would help the text in my opinion…
L351: rephrase “While this is plausible…”
L361: what are the blue and red lines, are they trendlines or just connectors between the points?
L365: which offset? Could you be more exact?
Can one state here that there is an unknown water loss from the system as the discharge trend is steeper?
L370: Does it mean more humid conditions or a faster change towards humid conditions?
L373: point out whiskers on the legend
L380: where was this pointed out?
How could the model have GWR estimations for different depths – was it over a grid?
5 limitation and uncertainties: I really like this section, but it could have a more exact title like “Explaining the gap between trends”
L437: a similar behavior was pointed out locally by Somogyvari et al., 2024 for a lake system in the region.
Somogyvári, Márk, et al. "A hybrid data-driven approach to analyze the drivers of lake level dynamics." Hydrology and Earth System Sciences 28.18 (2024): 4331-4348.
They also used a combination of different factors as an explanation for system water loss, together with an environmental tipping point. Did you consider such explanations?
L511: Could you rank the different explanations based on plausibility?
Citation: https://doi.org/10.5194/egusphere-2025-222-RC1 - AC1: 'Reply on RC1', Maik Heistermann, 12 May 2025
-
RC2: 'Comment on egusphere-2025-222', Anonymous Referee #2, 28 Apr 2025
The manuscript by Francke and Heistermann is giving an approach to explain dropping groundwater recharge in Brandenburg. Overall, the study is impressive and very interesting and will doubtless find its place in the journal, where it fits well into.
The manuscript is appropriately designed and consistent, figures are of high quality and in principle well prepared. Few minor things should be modified, which I marked below.
More specifically I was wondering, why the authors mainly base their arguments on discharge measurements only rather than including direct groundwater observations to justify dropping groundwater recharge.
During reading, I also had the problem, that I often had to jump back in the text to be sure I understand what is meant and don’t miss things. Sometimes expressions are difficult to read and to follow, but when authors carefully inspect the manuscript again, I’m confident it will be improved and become published in N-HESS.
Below, more specific comments and suggestions are given.
L30 f. I would rather use deep and shallow instead of distant and close water tables.
L38. In an intensely cultivated and urbanized region as Brandenburg, there might be increased abstraction by man different actors (farmers, water suppliers, private wells, etc.). How comes you generally state: “there is no basis to assume such a widespread abstraction of groundwater”? That statement is a fundamental prerequisite of the coming explanation for dropping gwr, but there is no evidence presented.
L39. How is it possible, an aquifer gets confined by a river? Please think about wording.
L50: is it indeed possible to generalize it as you do: “Given the highly permeable soils in Brandenburg and the resulting insignificance of surface runoff…“? I think the evidence coming from that statement is very serious and actually builds the foundation of the subsequent argumentation: dropping river discharge as mirror of dropping gwr, because baseflow is (in your argumentation) the only source of surface runoff.
I think that point, in combination with my concerns from line 38 (missing abstraction), should be treated with more caution and seriously justified.
Fig. 1: a) the names given therein are unclear in their meaning: are these cities, counties, catchments?
- b) in each of the figures, 2 numbers are given, but it is not clear what they represent. Also the unit is not clear to me: mm/a/dec, millimeter per year is okay, but what is “Dec”?
L69-73: I partly disagree with that statement. The analyses of gw-timeseries gives worthwhile information as for gwr and though the given issues are valid in a certain way, it is mentioned before, the catchments that are investigated are not influenced by abstraction. Of course, storativity is always an uncertain parameter, but I guess there should be few pumping tests available to at least estimate gwr in few areas, based on gw timeseries.
L75 ff. Please use consistent wording: shallow and deep gw tables and don’t switch between deep and distant.
L107: where is the dataset from? Reference?
L136: I do not understand the integration of neighbouring catchments: they do not contribute to the discharge of the gauge. And what does it mean to include them “partly”?
L162. There is a spare parenthesis at the end of the sentence.
L179. Where are the field LAI measurements were taken from? Just out of curiosity: during 2019-2024, forest dieback (as observed in other areas in Germany) did not apply to Brandenburg at all?
L196 ff. The steady state type of gw-depth information may not fit to the proposed changes in gw depth due to dropping gwr, am I wrong?
L246: where is soil moisture data taken from?
L251: is a constant head not contradicting the question, whether gwr is dropping and hence gw tables?
L265 ff.: is the concept of hydrotopes following the idea of HRUs in hydrology? How did you include the depth to groundwater, 13 classes of gw depth each mulitplied by land use, soil type etc.? How did you integrate the hydrotopes spatially: as grid or as a mesh of irregular/regular triangles?
L281: what is the unit “dec” standing for?
L291: now the explanation comes, much too late…
In general, it is hard to understand the concept of mm/a/dec. To become able to translate it into a logical unit, it needs a brief explanation at the very beginning.
L296: why was just LAI selected to be the controlling parameter of dropping gwr? Increasing ET is obviously a key-parameter as well controlling the soil moisture (fig. 3d).
L297: please explain CI
L344: where do I see it in Fig. 5? The figure is very complex but nice. However, I have difficulties to differentiate between various shades of grey in some of the graphs.
L362: in fig. 3 there is no real trend in precipitation observable, how comes it though brings the gwr to drop?
Fig. 6: how is it explainable, gwr is almost negligible in 4 out of 5 catchments fro shallow water tables? Contrastingly, the deeper groundwater receives reasonable gwr of 100-150 mm/a.
What does “area weighted” mean here?
How is areal fraction calculated?
L383 ff. I have difficulties to understand why the contribution of the shallow gw is that high, if their (however calculated) areal contribution in the total catchment is low. And why show shallow gw at all different gwr compared to deeper gw bodies? Is it a function of delay? But how was it modelled in 1D?
L403: there is a “by” too much.
L453f. Where stems the data from? A study from 2009 might not be representative to activities during the last decade.
Citation: https://doi.org/10.5194/egusphere-2025-222-RC2 - AC2: 'Reply on RC2', Maik Heistermann, 12 May 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 320 | 126 | 20 | 466 | 19 | 29 |
- HTML: 320
- PDF: 126
- XML: 20
- Total: 466
- BibTeX: 19
- EndNote: 29
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Till Francke
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(13288 KB) - Metadata XML