the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Feb 2025
Drought Propagation and Ecosystem Resilience in a Peri-Urban Catchment of Berlin-Brandenburg
Abstract. This study investigates the drought dynamics and their effects on surface water, groundwater, and vegetation across the Tegeler Fließ catchment in Berlin/Brandenburg from November 2008 to April 2021. Calculating drought indices for atmospheric, hydrological and groundwater drought, namely the Standardised Precipitation Index (SPI), the Standardised Surface Water Level Index (SSWLI) and the Standardised Groundwater Level Index (SGLI), respectively, the analysis identifies station-specific drought events and their propagation across three locations: Schildow, Luebars, and Tegel. The three indices allow us to take a closer look into the differences and the propagation of drought processes over different parts of the hydrological system. The study also assesses the impact of drought on vegetation health using the Normalized Difference Vegetation Index (NDVI). Our results strongly differ at different locations: the peri-urban area (Tegel) experienced the most severe and prolonged groundwater droughts, while the groundwater in the nature reserve and fen meadow area (Schildow) remained more resilient but faced significant surface water stress. Agricultural land (Luebars) displayed variability in both surface and groundwater responses, with surface water systems being more resilient. NDVI analysis revealed that vegetation remained largely within moderate to dense classes throughout the study period, showing resilience despite severe drought conditions from 2018 to 2020. Spearman correlation tests did not show any significant relationship between NDVI and drought indices, while Granger causality tests revealed that SPI, and for some stations also SSWLI, significantly Granger-caused NDVI with a lag of one month. These findings highlight the need for localized drought management strategies tailored to both surface and groundwater resources, alongside enhanced vegetation monitoring that goes beyond traditional indices like NDVI.
- Preprint
(1891 KB) - Metadata XML
-
Supplement
(1549 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-471', Anonymous Referee #1, 31 Mar 2025
The paper titled "Drought Propagation and Ecosystem Resilience in a Peri-Urban Catchment of Berlin-Brandenburg" by Polina Franke, Aryan Goswami and Mark Somogyvari explores the dependencies between hydro-meteorological drought and the resilience of vegetation by analysing monthly time series of drought indices and vegetation indices at a local catchment north-west of Berlin.
I find the toolchain and approach presented in this study to be highly practical for large-scale investigations of drought and vegetation resilience. However, the current methodology is not well-suited for a localized study. The authors have overlooked significant opportunities provided by the specific site, such as ground-truth monitoring, which could have enhanced the robustness of their findings. Given these limitations, I recommend rejection but encourage resubmission following a substantial revision that incorporates site-specific data and methodologies
Major comments
My primary concern is that the authors did not take advantage of the opportunity to visit the field site and use field data to better understand and support the observed differences between the chosen indices and locations. Instead, the manuscript offers speculative explanations without providing the depth of analysis that could have been achieved through direct field observations and data.
For instance,
- l375: “In particular, Tegel groundwater experienced a severe 17-month drought from 2019 to 2020, while Schildow showed more resilience in groundwater levels but showed stress in surface water systems”. How are the aquifer properties in the different areas which may lead to this resilience? For instance, If the porosity is larger, water level declines are less for the same amount of water extracted.
- 309: “ In-situ measures of vegetation health could provide more precise insights and supplement satellite-derived indices in these analyses.” While this is a valid suggestion, the study would benefit from a more comprehensive inclusion of land cover distribution for each MODIS pixel analyzed. The variation in NDVI changes is highly dependent on vegetation type. For instance, if the NDVI signal predominantly reflects trees with deep root systems, a summer drought may have little impact on leaf greenness. In contrast, grasslands are likely to show more noticeable NDVI changes under similar drought conditions. Additionally, although the manuscript briefly mentions the influence of agriculture, it lacks specific and quantitative details. It would be beneficial to clarify the proportion of agricultural land within the MODIS pixels and to specify which types of crops are present in these areas. For a study focused on local conditions, this information should be readily obtainable and incorporated into the analysis. For the reader, at least one picture of each study site would be helpful
Another critique is the usage of the SSWLI index. As the authors state mention by themselve, that “that this approach has a lower accuracy than traditional discharge based indices, as river levels (especially in smaller streams) are more susceptible to nonlinear behavior due to the river profile”, but they did not provide any estimate of how less accurate the chosen approach is. Considering that all gauges are along the same river, different resilience to drought can only be estimated by comparing discharges due to the impact of cross-section and streambed roughnesses variability at each station (see Manning-Strickler Formula). I suggest conducting a number of discharge measurements at each station and deriving a rating curve (https://en.wikipedia.org/wiki/Rating_curve) to convert the water level estimates to discharges.
Based on your map, there appear to be surface tributaries (e.g., Kindelfließ) and a lake (Hermsdorfer Lake) within the study area, which could influence the SSWLI index at the downstream gauges. How might these features impact the observed values? This could help explain the variations in drought conditions observed at each station (Figure 3).
One assumption of your approach is that the dynamics of groundwater and surface water levels can be directly attributed to drought conditions. However, groundwater levels, for example, may also be influenced by anthropogenic factors, such as the lowering of local groundwater levels due to construction activities or changes in the extent of sealed surfaces. Did you consider these potential impacts in your analysis?
Although the authors reference the drought index from the Helmholtz Centre, they did not analyze soil drought conditions, which are conceptually the link between meteorological drought and groundwater drought. If the authors choose not to include this analysis, they should provide a clear justification for this decision.
The study would benefit from a discussion on the applicable scale of your analysis and I feel there is a lack of mentioned studies which did similar works, please provide some more references. Are there limitations to using individual MODIS cells for the indices, considering the sensitivity of values to neighboring cells? Addressing this potential issue could be valuable and could be incorporated into the introduction section.
Minor comments
l18. The second part of the sentence is an empty phrase, as the planet naturally includes the Berlin/Brandenburg region.
L44. SenMVKU reference (and some others) have no year, please update
L46. It is called Helmholtz Centre for Environmental Research and not Helmholtz Institut
L50 – L 68. I believe there is an overlap with Section 2.3. In the introduction, you should list different indices for precipitation, hydrometry, and groundwater resources, discussing their advantages and disadvantages one by one. The specific details of the technique used in this study can be covered in Section 2.3."
L73. Provide reference to Granger Causility test
Figure1: Surface Water instead of Surfacewater
Section 2.2. Here you mentioned the Software “R” first. So here you should reference R and mention, which version did you use
L190: Why did time lag analysis fail? Perhaps you should not mention it here but using it in the discussion section
Figure3: The plot of absolute surface water levels and groundwater levels for comparing different stations provides limited insight. In Surface Water Hydrology, with a focus on water budgets rather than water levels, it is more informative to compare changes in discharge rather than stage, as stage is dependent on discharge and cross-sectional area. Since discharge data is not available, I recommend normalizing the stages and presenting them in a boxplot or histogram to visualize the distribution across all stations. For groundwater levels, the absolute values provide limited context for the reader. Instead, it is more meaningful to assess the distance from the surface, as this helps evaluate how accessible the groundwater is for vegetation, which is a central aspect of your study.Therefore, I suggest revising Section 3.1 to reflect these points, too.
Fig4: All plots should share the X Axis to make temporal comparison more easy.
L285-295: The authors provide several potential explanations for the low NDVI values during the winter months between 2009 and 2011, such as reduced vegetation cover and lower growth rates. I believe the answer is very simple: There was a long lasting snow layer during this years. I recommend checking the DWD snow data to verify this and rephrasing the discussion to provide a more explicit and clear explanation, avoiding any ambiguity.
L313: “:. v”, improve language
L379: If you say 13 years is not enough to capute longer-term climate cycles, why did you start in 2008 and end in 2021?
Citation: https://doi.org/10.5194/egusphere-2025-471-RC1 - AC1: 'Reply on RC1', Márk Somogyvári, 11 Jul 2025
-
RC2: 'Comment on egusphere-2025-471', Anonymous Referee #2, 20 May 2025
Review on: Drought Propagation and Ecosystem Resilience in a Peri-Urban Catchment of Berlin-Brandenburg
By Polina Franke et al.
This study aims to investigate the regional implications of droughts, aligning with the objectives of the NHESS special issue. While the research question is relevant, the study lacks a comprehensive conceptual framework for the field settings. This absence leaves the connection between groundwater and surface water droughts and vegetation responses vague, despite the application of various statistical methods. A more robust conceptual model is necessary to illustrate the relationships among the land surface, groundwater, and surface water at the study sites. Understanding these hydrological connections is critical to understand or verify how drought propagates through this system and affects vegetation. The writing and language are clear, with a logical structure and appropriate text length. The balance of figures and text is satisfactory, although improvements to some figures are recommended (as discussed in specific comments).
Please consider my comments as suggestions to enhance the manuscript. I hope to support the improvement of the study and apologize if any remarks come across as too direct.
General comments
Conceptual Model: The paper lacks a conceptual model for each of the study sites, detailing how drought propagates through the system. It's crucial to establish the distances between land surface, groundwater, and surface water. How does the groundwater well relate to stream measurements? Are groundwater flow directions understood? Moreover, soil water storage and its direct impact on vegetation should be discussed early on—a critical missing link in the observations of the system, more directly affecting drought propagation to vegetation than surface water or groundwater levels alone.
Drought Index and Data Trends: The data itself, as seen in Figure S2, already depicts long-term trends that should be discussed more thoroughly. While indices offer clearer comparisons, understanding the data and its trend provides a lot of discussion material. Moving Figure S2 to the manuscript’s main section would enhance this discussion, please consider this suggestion. Relate the findings to other regional studies. The indices primarily reflect the underlying data trends, not necessarily drought propagation effects in the system. The interpretation and discussion of the drought indices time series seems to be mainly based on visual inspection, including mainly qualitative and somewhat subjective descriptions. Thus the findings remain rather vague.
Groundwater Drought and Vegetation (NDVI): The study posits that groundwater drought impacts NDVI but does not convincingly explain this relationship. If NDVI shows minimal effects, why emphasize the need for improved vegetation monitoring over groundwater or water management strategies?
Specific comments
L28 Please rephrase this. It sounds now, as if the country would be a wet region. While Brandenburg does have many surface waters, it is overall a water limited region, which is not a contradiction but I guess needs more than this single reference and a better explanation (the tight water balance with very low precipitation for a temperate humid climate and the high atmospheric demand (evapotranspiration)).
L34 As you describe a cascade, maybe put the declining groundwater table as the last argument in the list
L45 – if there are a few studies, you should cite them here and summarize what they found
L46 –also be precise here – what significant effects did they find?
L50ff merge the two paragraphs, shorten (or move parts to the methods section, there is some overlap)
L77, not sure, if “persistent” is already an appropriate word, considering the rather short time frame of the analysis overall
L87 Add information, to which larger river the Tegeler Fließ is a tributary of.
Section 2.3.1 and 2.3.2 – why does one index gets its own section, while the others are combined in one?
Section 2.4 and Figure 2, consider to rename the section heading to “statistical tests and measures”
Your analysis process is not very difficult, so I think in such a linear process also Figure 2 is not required. Instead, structure the section with line breaks/paragraphs between the different statistical tests and measures you describe.Figure 3: Precipitation over what time frame? I don’t think the graph in its current form is very useful. The groundwater level plot show just the groundwater gradient the landscape along the Tegeler Fließ and the water level in the river actually doesn’t say a lot at all (as you also correctly mention its limitations). However, with only this data available – the critical thing to show is the relation between the groundwater level and the surface water level, as well as the distance of the groundwater level to the land surface. So either include surface water level statistics in absolute heights, or think of a better way also to include the important land surface to groundwater level information.
L196 You seem to show monthly cumulative precipitation amounts in Figure 3. Even if there is outliers in your distribution (extreme precipitation for the time frame you consider) – this is by no means to be determined as heavy rainfall from this graph! Heavy rainfall is actually large amounts over very short time frames (<day).
L197 Either you have data to support your guess, or leave it out, or discuss in more detail why you guess this (related to the missing conceptual model I) – doesn’t the Figure S2 support this claim?
L204 how much “water management” is there for the Tegeler Fließ and the surrounding groundwater? If there is a lot of influence of water management on the field sites and observed water levels, then they would not be suited to assess the study goal of observing a drought propagation from precipitation to the surface and groundwater levels.
L204ff The groundwater levels show the slope in the groundwater along the river stretch. How much is the elevation difference for the land surface? More interesting might be to discuss the variations of the water levels themselves.
L211 Not sure, what you want to say with this sentence – what the reader will see in section 3.2? Then place before the subsection heading – or consider to remove completely.
Figure 4: This is the main figure of your study – give it more space. Make sure the time series of SPI and other indices directly comparable by sharing the same x-axis! Either placing all the panels below each other (a one-page plot) or showing SPI twice, above each of the columns. The caption should give more information, especially also about the dashed line, that does not appear in the legend.
L241 Check again – generally lower is not directly what I see from Figure 4. What is with the drought event in 2016 at the Tegel site? There is none corresponding strong event in the precipitation and none at the other sites. Is this a water management related effect? Add to the following sentence: “… fluctuating between negative and positive values …”
L243 “noticeable transition…” I do not clearly see a transition.
L241ff the findings drawn in this paragraph from the visual inspection of the time series of the indices remain rather vague and seemingly subjective/only qualitative description.
L247 – but the surface water index at the end of the time series shows a faster and more pronounced recovery from the drought event, while at other sites the drought persists and index values stay below 0. Stating here, a higher vulnerability of the surface waters to drought seems far fetched, also the recovery at the end of the time series suggests resilience in my opinion. Or am I missing something?
L248 “severe meteorological drought period in 2015-2016…” there is only a moderate drought (for 1 month?).
L249 I do not see a strong contrast between Tegel surface water and groundwater index – both in the same direction and mostly below 0 for the period, no?
Section 3.3: Why don’t you consider the NDVI as a measurement, as the other data you look at (precipitation and water levels) and calculate a trend for the time series. “Normalizing” the vegetation observation by deriving an index as you do on the other data might show better a drought propagation/impact on vegetation?
Section 4.1 stays vague as well, see my general comment on the missing conceptual model, I guess this section will require a revision. Includes quite some speculations, could be more precise of relying on additional data or relating to findings of other studies in the region (see maybe references within Altdorff et al. (2024): https://egusphere.copernicus.org/preprints/2024/egusphere-2024-3848/ (there are some studies cited, that deal with the 2018 -2020 drought period).
L343 There is no downward trend in your SPI3 data as shown in Figure 4 (or do you mean the indices related to surface water and groundwater?)
L350 Why should groundwater impact vegetation? This is only relevant if land surface – groundwater level distances are low and vegetation can directly use groundwater for evapotranspiration (tap into the capillary fringe) – is this the case at your sites? Otherwise, this would not be a relation to test at all.
L355 – was there irrigation? I thought there is rather little irrigation operated, so I would not expect a big effect on NDVI. Any information to support your guess?
L362 – what are “shallow-rooted plants typical of these areas”?
L363 plants rely, if at all, on shallow groundwater, never on deeper groundwater. I get what you mean though, but please stay precise and rephrase the sentence accordingly. It then again strongly depends what type of vegetation/land use you have. In this respect it would be good to show somewhere a bit more detailed what land use is covered by your NDVI data (agriculture, forest…).
L365 – How would the decline in groundwater tables (and surface waters) affect vegetation (the long term decline in the data, Fig. S2)? This is a rather vague statement – again missing is the agricultural or soil moisture drought that would be actually relevant for vegetation (see general comments).
L378 – I would like to see this point mentioned earlier in the manuscript (describing the data itself, potentially including Fig. S2 in the main manuscript). Looking at your interpolation (the Figure is good to keep in the appendix) – I do not think the interpolation introduces a lot of uncertainty here. So definitely not the interpolation/gap filling a key limitation of your study.
L382 first: all your ecosystems seem to show resilience, as you define it. Second, I think again the missing link here is the soil water storage (as you discuss just then a couple of lines below) – I would like to see this as part of the discussion chapter and not within the conclusions.
L390 I do not think assessing single plant species would make sense here – as you are interested in the effect on the ecosystem. Rather including additional data on temperature, evapotranspiration or atmospheric demand or soil moisture might close the missing gap you have so far.
Technical corrections
For all the manuscript: The figure and table captions should contain more information, to make the graph/table self-explanatory without the need for the reader to go into detail in the text, also including the explanation again for abbreviations.
Please go through the manuscript to check the units of precipitation, it should be clearly stated, over which time interval cumulative amounts are given, so not only mm but mm/day or mm/week.
L11 Rephrase the sentence for NDVI. To a reader it is not clear, what “moderate to dense classes” are. Either add “vegetation” or rather just say, that there was no, or only a moderate effect to be observed in the NDVI
L19 These studies are projections, so include “likely” or similar
L44 Not in references list, no date, please fix. I guess it is some authorities? So spelling out the abbreviation would be helpful.
Figure 1: I found the legend hard to read, usually it is first the symbol then the legend entry (left aligned close to its according symbol). As it is, the first symbol and the second column of the legend are best aligned.
Additionally: is it possible, to zoom in a little on each of the sites to better see the landuse for the NDVI area? (there is a lot of blank space in the lower right of the map)107 Two different authorities is not “various”
L108 I know that data integration is a challenge – this is always the case, I would delete the sentence
L109 Awkward sentence, rephrase
L300 line break and new paragraph following “analyses.”
Citation: https://doi.org/10.5194/egusphere-2025-471-RC2 - AC2: 'Reply on RC2', Márk Somogyvári, 11 Jul 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 195 | 51 | 15 | 261 | 14 | 12 | 20 |
- HTML: 195
- PDF: 51
- XML: 15
- Total: 261
- Supplement: 14
- BibTeX: 12
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1