the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2026
Integrating propagation and recovery dynamics into groundwater drought vulnerability assessment through exposure, pressure, and aquifer system response
Abstract. Groundwater drought is influenced more by system-specific response dynamics than by meteorological forcing alone. We introduce a multi-scale framework that combines exposure, pressure, and sensitivity with process-based metrics of drought propagation and recovery to assess groundwater drought vulnerability. The Drought Impact Potential Index (DIPI) is developed and tested across a regional aquifer system. Propagation probability, median recovery time, and resilience metrics are examined across temporal scales and in groundwater systems at different depths. The findings reveal that spatial vulnerability patterns are driven by variations in system memory and response time. Deeper aquifer systems tend to have higher propagation probability, longer recovery periods, and increased vulnerability, indicating delayed responses and persistent drought signals. Conversely, shallower systems respond more quickly and recover faster, leading to lower drought persistence. The spatial distribution of DIPI remains consistent whether using weighted or unweighted versions, confirming that the identified patterns are robust and reflect fundamental hydrogeological controls. These results demonstrate that groundwater drought vulnerability arises from interactions between external forcing and internal system dynamics and cannot be understood solely through static indicators. An area-based analysis of exposure–pressure contrast shows that 60.4 % of the study area is dominated by the intrinsic system response, compared to 21.8 % driven primarily by human pressure. The proposed framework offers a process-based approach for groundwater drought assessment and can be applied to other diverse aquifer systems.
- Preprint
(1999 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-1904', Anonymous Referee #1, 29 Jun 2026
-
AC2: 'Reply on RC1', Katarzyna Sawicka, 08 Jul 2026
We sincerely thank Referee #1 (RC1) for the careful and constructive assessment of our manuscript. We appreciate the reviewer’s recognition of the relevance of the topic and of the potential value of the SPI-SGI analysis for understanding groundwater drought propagation across different aquifer systems.
We agree that the hydrometeorological interpretation of drought propagation should be developed more explicitly and that the DIPI framework should be presented more cautiously. In the revised manuscript, we will place greater emphasis on the SPI-SGI relationship, lagged response, attenuation, storage memory, and recovery asymmetry as the central hydrometeorological components of the study. We will also clarify that SPI is used as an indicator of precipitation-deficit propagation into groundwater levels, not as a complete representation of all meteorological and recharge-related controls.
At the same time, we would like to clarify the manuscript's intended scope. Our aim is not to develop a comprehensive recharge model or a complete hydrometeorological water-balance attribution framework. Rather, the study quantifies how precipitation deficits translate into groundwater drought signals across different hydrogeological conditions and examines how these observed response metrics can support a spatial, index-based assessment of groundwater drought impact potential.
In response to the reviewer’s comments, we will revise the manuscript to: (1) expand the interpretation of the SPI-SGI coupling process; (2) justify the use of SPI and explicitly discuss the absence of PET, evapotranspiration, soil moisture, and recharge indicators as a limitation; (3) soften conclusions based on the two deep confined wells; and (iv) present DIPI as a diagnostic screening framework rather than a fully causal or predictive model. We have addressed the reviewer’s comments point by point below.
1. We appreciate the reviewer's insightful comment. We concur that the relationship between SPI and SGI, along with the multi-timescale analysis of drought propagation and recovery, constitutes the primary hydrometeorological contribution of this study. Additionally, we acknowledge that DIPI, including components such as Exposure-Pressure-Sensitivity, GDEs, groundwater quality, abstraction, and mining pressure, is more directly linked to assessing groundwater drought vulnerability and water resource screening than to hydrometeorological process analysis alone.
In the revised manuscript, we will rebalance the structure and interpretation. The SPI-SGI analysis will be presented as the core process-oriented component of the study, whereas DIPI will be described as an index-based extension that translates observed groundwater drought-response metrics into a spatial assessment of drought-impact potential.
We will also expand the discussion of how precipitation anomalies are transformed into groundwater drought signals through recharge connectivity, storage memory, lag, and attenuation. The explanatory power of SPI will be discussed as scale- and aquifer-dependent: stronger, faster coupling is expected in shallow unconfined systems, whereas weaker, delayed, and more persistent responses are expected in confined systems with greater storage and longer memory.
Concerning the difference between meteorological and human-induced signals, we acknowledge that this distinction cannot be strictly causal when relying solely on SPI-SGI correlations and SGI-based metrics. Groundwater levels likely reflect the combined influence of climate, extraction, and mining-related dewatering. Therefore, we will not suggest that the E-P contrast directly separates intrinsic aquifer behavior from human activity.
2. We agree that the SPI-SGI analysis should be interpreted more explicitly in terms of hydrometeorological processes. The manuscript already includes a lagged SPI-SGI correlation analysis across 3-, 6-, and 12-month aggregation windows, with SGI shifted relative to SPI over lags of 0-12 months. The analysis was also repeated using detrended indices to reduce the risk that correlations primarily reflect common long-term trends. However, we agree that these results are not yet sufficiently discussed from a process perspective.
In the revised version, we will interpret the SPI-SGI relationships through four linked mechanisms: (1) attenuation of short-term precipitation variability, (2) lagged groundwater response, (3) increased system memory with aquifer depth and confinement, and (4) asymmetric recovery following groundwater drought events. We will frame the SPI-SGI analysis as a diagnostic tool for identifying dominant response timescales and groundwater memory, rather than as a complete recharge process model.
3. We thank the reviewer for raising this point. We agree that groundwater drought development and recovery may be influenced by evapotranspiration, temperature, soil moisture, recharge timing, snow processes, and groundwater abstraction. We also agree that a complete hydrometeorological attribution would require additional variables, such as PET, actual evapotranspiration, soil moisture, or independently estimated recharge.
The choice to focus on SPI and SGI was a deliberate methodological decision. This study aims to quantify how precipitation deficits influence groundwater-level anomalies across aquifer systems with varying storage and connectivity, rather than developing a comprehensive water-balance or recharge model. SPI was chosen because it is a straightforward, widely adopted, and reliable indicator of precipitation deficit, available consistently throughout the entire study period and across all meteorological stations involved.
We therefore do not intend to add SPEI or PET-based indicators as a new analytical component to the revised manuscript. Including them would broaden the paper's scope toward climatic water-balance modeling and would require additional assumptions about PET estimation, spatial representativeness, soil-water storage, and recharge partitioning. Instead, we will clarify the scope: SGI is used as the integrated groundwater-system response, whereas SPI represents precipitation-deficit forcing.
We will explicitly acknowledge this limitation in the revised manuscript. It will clarify that the SPI-SGI analysis indicates the propagation of precipitation-deficit signals into groundwater levels, rather than fully attributing groundwater drought to all hydrometeorological factors. Additionally, suggestions for future research will include incorporating SPEI-based analysis, PET/recharge modeling, and soil-moisture data.
4. We agree with the reviewer. The limited number of monitoring wells, especially in the deep confined aquifer, does not allow us to draw strong statistical or spatially representative conclusions about the entire aquifer system. The two deep confined piezometers provide useful evidence from the available monitoring network, but they should not be interpreted as a statistically robust sample of all deep confined groundwater conditions in GWB 43.
In the revised manuscript, we will temper the language used when discussing conclusions about aquifer depth and confinement. We will steer clear of claims suggesting that deeper aquifers typically show higher propagation probability, longer recovery times, or increased vulnerability as statistically confirmed patterns. Instead, we will present these findings as observations derived from the monitoring network that align with the conceptual hydrogeological model.
For example, statements such as “deeper aquifer systems have higher propagation probability, longer recovery periods, and increased vulnerability” will be revised to “within the available monitoring network, the deep confined wells showed longer drought persistence and slower recovery, consistent with the expected influence of storage, confinement, and delayed recharge.” We will also strengthen the limitation statement regarding the expert-based aquifer grouping and the limited number of confined and deep confined wells.
5. We thank the reviewer for this detailed and helpful comment. We agree that DIPI is more closely aligned with groundwater drought vulnerability screening and spatial assessment than with hydrometeorological process analysis alone. In the revised manuscript, we will present DIPI as an index-based diagnostic framework rather than a fully causal or predictive model.
We also agree that the weighting scheme is subjective. The comparison between the weighted and equal-weighted DIPI variants will be presented only as a preliminary sensitivity check, not as a systematic uncertainty analysis. We will soften statements such as “the hotspot configuration is robust” and instead note that the broad spatial pattern was similar under the two tested aggregation schemes, while a more complete sensitivity analysis would be required to support stronger conclusions.
The reviewer is also correct that Exposure and Pressure are not fully independent. Because Exposure is derived from SGI-based drought metrics, it may already reflect the influence of abstraction, mining-related dewatering, or other long-term human modifications to groundwater levels. Therefore, the E-P contrast will no longer be interpreted as a causal separation between “intrinsic aquifer response” and “human pressure.” We will present it as a diagnostic contrast between SGI-derived drought response and mapped anthropogenic pressure.
We also acknowledge that the spatial interpolation of DIPI is limited by the sparse monitoring network. Consequently, the DIPI maps should be viewed as preliminary screening tools highlighting general regional trends rather than precise high-resolution predictions. We will include a clearer statement emphasizing that thorough spatial uncertainty analysis, interpolation cross-validation, and error mapping are dependent on a denser, more evenly distributed monitoring network, particularly for confined and deep confined aquifers.
Minor comments
We thank the reviewer for identifying these inconsistencies. We will use the name Drought Impact Potential Index (DIPI) consistently throughout the manuscript. The incorrect phrase “Drought Impact and Pressure Index” will be replaced.
We will also correct the figure reference in Sect. 2.3. The pressure component is not shown in Fig. 6, which presents propagation-recovery trajectories, but appears in the DIPI spatial outputs. We will revise the sentence to refer to Appendix A, Fig. A1, and, where appropriate, to Fig. 7.
Summary of planned manuscript changes
- Revise the Introduction to foreground the SPI-SGI propagation and recovery analysis as the primary hydrometeorological component and to present DIPI as a vulnerability-oriented extension.
- Revise the Methods to justify the use of SPI as a precipitation-deficit indicator and to clarify that SPI does not represent PET, actual evapotranspiration, soil moisture, snow processes, or recharge timing.
- Expand Sect. 5.4 and the Discussion to interpret lagged SPI-SGI relationships in terms of attenuation, lag, storage memory, recharge connectivity, and recovery asymmetry.
- Soften statements in the Abstract, Results, Discussion, and Conclusions regarding the deep confined aquifer, emphasizing that the two deep confined piezometers provide indicative evidence from the available monitoring network rather than population-wide statistical support.
- Revise DIPI-related sections to present the index as an exploratory screening and diagnostic framework. Replace strong claims about robustness with cautious language and present the weighted/equal-weighted comparison as a preliminary sensitivity check.
- Clarify that Exposure and Pressure are not fully independent because SGI-derived metrics may already incorporate the effects of abstraction and mining-related dewatering. Interpret the E-P contrast as a diagnostic index contrast, not as a causal separation between natural and anthropogenic controls.
- Strengthen limitations related to spatial interpolation, monitoring density, and the lack of full uncertainty analysis, cross-validation, and interpolation error maps.
- Use the name Drought Impact Potential Index (DIPI) consistently and correct the erroneous reference to Fig. 6 in Sect. 2.3.
Citation: https://doi.org/10.5194/egusphere-2026-1904-AC2
-
AC2: 'Reply on RC1', Katarzyna Sawicka, 08 Jul 2026
-
CC1: 'Comment on egusphere-2026-1904', Justyna Kubicz, 03 Jul 2026
This paper presents an interesting approach to groundwater drought by introducing the Drought Impact Potential Index. The overall writing is clear, the study area is well described, and the general idea of moving from static maps to dynamic time-series metrics is a step in the right direction.
However, there are serious concerns regarding how the methods and results are interpreted from a hydrogeological perspective. The primary issue lies in designating this framework as a process-based approach. In groundwater science, a process-based approach implies solving physical equations of water flow using specific parameters such as hydraulic conductivity and storativity. This study relies entirely on statistical indices like the Standardized Groundwater Index. While using time lags and propagation metrics represents an advancement over simple static maps, it remains a strictly statistical approach. Labeling it as process-based might mislead readers into believing that actual physical flow dynamics were calculated.
This terminology issue leads directly to the most critical misinterpretation in the paper. The authors note that the deep aquifer system shows no recovery and exhibits very long drought durations, attributing this to a long system memory. However, the study area is heavily influenced by open-pit mines. When a mine continuously pumps water to keep the pit dry, the water table drops and remains depressed. This phenomenon constitutes human-made drawdown rather than a natural drought. The groundwater index will naturally remain negative because the system is artificially drained, not because the climate exhibits a long memory. Concluding that deep aquifers have poor drought recovery based on wells affected by mine dewatering is physically incorrect. The paper conflates natural climate deficits with permanent human-induced hydraulic changes.
Furthermore, the authors divide the aquifers into three clusters, yet the deep cluster relies on only two monitoring wells. In a variable glacial geological setting characterized by interbedded layers of clay and sand, two points cannot adequately represent the behavior of an entire regional aquifer system covering over 3600 km2. Interpolating data from just two wells to generate smooth continuous maps for the whole region creates a false sense of precision. The conclusions regarding deep aquifer behavior should be strictly limited to these two specific locations and not generalized across the entire region.
The methodology used to quantify human pressure also requires improvement. The authors employ water well density as a proxy for pressure, which does not reflect hydrogeological reality. A single household well pumping minimal volumes and a large mine dewatering system are treated identically as single points on a map. True hydraulic pressure depends on the actual volume of water being extracted, rather than the mere number of boreholes. Utilizing well density creates an output that appears precise but fails to reflect the actual physical stress exerted on the hydrodynamic system.
There is also a significant missed opportunity regarding vertical flow between layers, and addressing this is crucial for the paper to be scientifically sound. In a multi-layered system, intensive pumping from the deep aquifer for mining purposes induces a downward hydraulic gradient. Consequently, water from the shallow aquifer leaks downward into the deeper zone. The authors might be observing a decline in shallow groundwater levels and misinterpreting it as meteorological drought propagation, whereas it may actually represent water being physically drawn down into the mining drainage system. The authors do not necessarily need to run complex new calculations to fix this, but they must address this mechanism directly in their text to show they understand the physical limits of their data.
First, they should add a dedicated paragraph in the Discussion section explaining that intensive pumping reverses natural vertical gradients, creating a downward pull. They must explicitly state that this physical drainage could be recorded by shallow monitoring wells as a drought signal, admitting that what looks like drought propagation might actually be water drawn down into the mining system. Second, the authors need to address their data limitations by explaining that separating natural propagation from mining-induced leakage requires calculating vertical hydraulic gradients using pairs of shallow and deep piezometers at the exact same site. Since their network consists of individual scattered wells rather than purpose-built multi-level nests, they cannot calculate these exact gradients, and being honest about this limitation is much better than ignoring the physics. Third, the authors should adjust the methodology section to acknowledge this upfront by noting that standardized indices aggregate all water level declines into a single signal and cannot independently distinguish between a drop caused by a lack of rainfall and a drop caused by vertical leakage. Finally, the authors must soften their conclusions regarding the shallow aquifer. Right now, the paper strongly links shallow aquifer declines to meteorological propagation. The relevant sentences should be revised to state that the observed declines are likely a combined effect of meteorological drought propagation and potential vertical leakage driven by the artificial drawdown of the deep aquifers. Making this change shows scientific maturity and proves that while their tool is statistical, their understanding of the actual hydrogeological system is grounded in physical reality.
The justification for the expert-based clustering and the chosen index weights remains somewhat weak, although testing the unweighted version was a commendable step to verify the robustness of the spatial patterns. In summary, the authors do not need to construct a complex numerical groundwater model to address these shortcomings. They simply need to isolate the wells affected by mining from those reflecting natural conditions in their drought analysis, avoid interpolating two deep wells over an expansive area, and remove the claim that this constitutes a process-based physical approach. If they refine their conclusions to explicitly acknowledge the limitations of statistical indices in explaining physical groundwater processes, the paper will represent a valuable contribution to the journal.
Citation: https://doi.org/10.5194/egusphere-2026-1904-CC1 -
AC1: 'Reply on CC1', Katarzyna Sawicka, 03 Jul 2026
We thank the commenter for the careful and constructive assessment of our manuscript. The comment raises important hydrogeological issues regarding the terminology used to describe the framework, the interpretation of SGI-based drought indicators in a mining-affected groundwater body, and the limitations of the monitoring network. We agree that these points should be clarified more explicitly in the revised manuscript.
Process-based terminology
We agree that describing our approach as “process-based” may be too strong if interpreted strictly in the groundwater modeling sense, i.e., as an approach based on solving groundwater-flow equations using parameters such as hydraulic conductivity, storage coefficient, recharge, and boundary conditions. Our approach is not a numerical groundwater-flow model. It is an indicator-based, statistical framework that integrates hydrogeologically relevant drought-response metrics, including propagation probability, recovery time, vulnerability, and resilience.
In the revised manuscript, we will replace or qualify the term “process-based” throughout the text. We will instead use terms such as “process-informed”, “hydrogeologically informed statistical framework”, or “indicator-based framework incorporating propagation and recovery metrics”. We will also revise statements that imply a “mechanistic” or “physically grounded” interpretation to better distinguish observed drought-response patterns from physically simulated groundwater-flow processes.
Mining-induced drawdown and SGI interpretation
We agree that SGI-based analyses cannot, by themselves, distinguish groundwater-level decline caused by meteorological drought from decline caused by anthropogenic drawdown. This is particularly important in GWB 43, where open-pit lignite mining and associated dewatering alter hydraulic gradients and groundwater storage conditions.
Our aim was not to classify all negative SGI values as purely natural drought. The manuscript already includes mining-related pressure as a separate component of the DIPI framework, but we agree that the interpretation of drought duration and recovery, especially in deep aquifers, should be more cautious. In the revised Discussion, we will clarify that prolonged or non-recovering SGI drought episodes in deeper aquifers may reflect a combined signal of climatic drought, aquifer memory, and persistent mining-induced drawdown. We will avoid attributing the absence of recovery solely to natural system memory.
While we do not fully agree that wells affected by mining should be excluded, it's important to note that this study aims not to reconstruct a purely natural groundwater drought signal but to evaluate the potential impact of groundwater drought in a human-modified groundwater system. In GWB 43, mining-related dewatering is a key aspect of the hydrogeological setting and a major factor influencing groundwater conditions. Omitting these piezometers would eliminate the areas where drought effects and human activities interact most heavily.
However, we agree that SGI signals from mining-affected wells must be interpreted cautiously. In the revised manuscript, we will explicitly state that negative SGI values in such wells may reflect a combined signal of meteorological drought, aquifer memory, abstraction, mining-induced drawdown, and possible vertical leakage. Therefore, these results will be interpreted as groundwater-level drought signals under combined climatic and anthropogenic forcing, rather than as purely natural drought responses.
Vertical leakage and hydraulic gradients
We agree that mining-induced drawdown may create or intensify downward hydraulic gradients, potentially inducing leakage from shallow aquifers into deeper drained zones. We will add a dedicated paragraph to the Discussion explaining that dewatering of deeper horizons in a multilayered aquifer system can alter vertical gradients and cause downward leakage. Consequently, groundwater-level declines recorded in shallow monitoring wells may reflect both the propagation of meteorological drought and leakage toward the mining drainage system.
We also note that separating these mechanisms would require paired shallow and deep piezometers at the same sites to calculate vertical hydraulic gradients. Because our dataset consists of spatially distributed monitoring wells rather than nested multilevel piezometers, this separation cannot be performed robustly with the available data.
Limitations of SGI
We agree that this limitation should be stated earlier in the manuscript. We will add a sentence to the Methods section noting that standardized groundwater indices aggregate all groundwater-level anomalies into a single standardized signal. Therefore, SGI identifies deviations from normal groundwater-level conditions but does not independently diagnose their cause. In human-modified systems, SGI may capture signals from meteorological drought, groundwater abstraction, mining dewatering, and induced vertical leakage.
Deep aquifer cluster and spatial interpolation
We concur that the deep-confined group, represented by only two wells, cannot be used to draw broad regional generalizations. Although the manuscript mentions that confined groups have few wells and should not be regarded as population-wide data, we believe this limitation should be highlighted more prominently.
We will revise the text and figure captions to clarify that interpolated surfaces based on monitoring points are exploratory visualizations intended to support comparison with pressure and sensitivity layers, not precise regional maps of deep-aquifer conditions. Conclusions about the deep confined aquifer will be limited to the monitored locations and will not be generalized to the entire regional aquifer system.
Pressure component and well density
We agree that well density is an imperfect proxy for anthropogenic pressure because it does not reflect actual abstraction volumes. A household well and a high-capacity dewatering system do not impose the same hydraulic stress. In our framework, mining-related pressure was included as a separate layer, but we agree that the abstraction-density component should be described more cautiously.
We will revise the Methods and Limitations sections to clarify that well density reflects the spatial occurrence of abstraction pressure rather than actual withdrawal intensity. Where abstraction rate or dewatering volume data are unavailable or incomplete, this proxy provides only an approximate indicator of pressure. We will also avoid wording implying that borehole density directly quantifies hydraulic stress.
Shallow aquifer interpretation
We concur that the interpretation of the shallow aquifer should be more tentative. Therefore, we will revise the relevant sentences to indicate that observed declines in shallow groundwater levels likely result from a combination of meteorological drought effects, local water extraction, hydraulic alterations from mining, and potential vertical leakage into deeper drained zones. This update will align the interpretation more closely with GWB 43's hydrogeological context.
Weights and expert-based grouping
We acknowledge that both the expert-based grouping and the selected DIPI weights introduce uncertainty. The revised manuscript will better justify these choices and explicitly refer to the unweighted DIPI variant as a sensitivity test. We will also clarify that the purpose of the weighting scheme is not to provide a unique physical solution but to construct a transparent diagnostic index that integrates exposure, pressure, and sensitivity.
Summary of planned revisions
In response to this comment, we will revise the manuscript by:
- replacing or qualifying “process-based” terminology;
- clarifying that DIPI is a hydrogeologically informed statistical/index-based framework;
- adding a limitation that SGI cannot distinguish climatic drought from anthropogenic drawdown;
- expanding the Discussion on mining-induced drawdown, vertical gradients, and possible downward leakage;
- softening the interpretation of deep-aquifer memory and recovery;
- limiting conclusions from the two deep wells to monitored conditions rather than the whole regional aquifer;
- clarifying that interpolated maps are exploratory visualizations;
- describing well density as a proxy for pressure, not a direct measure of abstraction intensity.
Overall, we believe these revisions will improve the hydrogeological precision of the manuscript while preserving the study’s main contribution: a transferable diagnostic framework for identifying where groundwater drought signals, anthropogenic pressure, and aquifer sensitivity coincide.
Citation: https://doi.org/10.5194/egusphere-2026-1904-AC1 -
RC2: 'Reply on AC1', Anonymous Referee #2, 08 Jul 2026
I thank the authors for their comprehensive response. After incorporating the corrections into the manuscript, it will be suitable for publication.
Citation: https://doi.org/10.5194/egusphere-2026-1904-RC2 -
AC4: 'Reply on RC2', Katarzyna Sawicka, 24 Jul 2026
We thank the referee for this positive assessment. All corrections described in our response will be incorporated into the revised manuscript if we are invited to submit a revised version.
Citation: https://doi.org/10.5194/egusphere-2026-1904-AC4
-
AC4: 'Reply on RC2', Katarzyna Sawicka, 24 Jul 2026
-
AC1: 'Reply on CC1', Katarzyna Sawicka, 03 Jul 2026
-
RC3: 'Comment on egusphere-2026-1904', Anonymous Referee #3, 17 Jul 2026
This manuscript examines groundwater drought vulnerability by integrating drought propagation, recovery, and spatial indicators. The topic is relevant, and the proposed open workflow is potentially useful. Several important conceptual issues have already been raised in RC1 and CC1. My comments are as follows.
1. Table 2 defines propagation probability as the fraction of SGI-3 events developing into SGI-12 events, yet Table 4 reports a probability for every SGI scale. Some values are impossible under the stated formula: when `N = 1`, `P` cannot be 0.17, and when `N = 2`, it cannot be 0.14. The authors should provide the implemented equation and a worked example, and then correct the affected results.
2. The reported CRT values cannot be reconciled with the equation in Table 2. Values of 0.90–0.98 would imply that negative SGI occurs in only about 2–9% of all months, which conflicts with Fig. 4. The authors should verify the formula, indicator definitions, and reported values.
3. Table 3 and Fig. A1 define exposure as `E = 0.4VUL′ + 0.3CRS_inv′ + 0.3MaxDur′`, whereas Fig. 7a uses `E = 0.6VUL′ + 0.4CRS_inv′`. The authors should identify the equation actually used and regenerate all dependent maps and statistics.
4. Sect. 5.3 assigns values of 11 months, 23 months, and −17.07 to Cluster 3 at SGI-3, whereas Table 4 reports 6 months, 6 months, and −8.2. The values in the text instead match Cluster 1 at SGI-12. This section should be corrected against the analysis output.
5. The spatial interpolation inputs are unclear. The Methods state that drought metrics were derived from aquifer-group mean SGI series, but the maps are described as being interpolated from piezometers. If group-level values were assigned to individual wells, the interpolation contains only three distinct drought-metric values. If well-level metrics were used, they should be reported and made available.
6. The monitoring network contains only 16 wells, including two deep wells. Without cross-validation and spatial uncertainty estimates, this network provides insufficient support for precise regional area percentages. The maps should therefore be presented as exploratory screening products.
7. The manuscript states that all series were truncated to 2005–2024, whereas Table 1 gives 2005–2021 for the shallow group. The authors should report the actual period used for each well and explain how missing observations were handled.
8. The SGI procedure is insufficiently documented. The manuscript should clarify whether the normal-score transformation was performed by calendar month, whether temporal aggregation occurred before or after standardization, and how incomplete windows were treated.
9. The lag analysis uses overlapping and autocorrelated series and selects the lag with the maximum correlation. A time-series-aware uncertainty analysis or stability test is needed. The current correlations show temporal association but do not establish a propagation mechanism.
10. The framework is described as process-based and mechanistic, although it consists of standardized indices, event statistics, and weighted spatial layers. No groundwater-flow or water-balance calculation is included, and hydraulic parameters do not enter the index. “Hydrogeologically informed statistical framework” would be a more accurate description.
11.E−P cannot separate natural and anthropogenic controls because exposure is derived from groundwater levels that may already contain climate, pumping, and mine-dewatering signals. Subtracting two normalized indices does not provide causal attribution, and the reported percentages should be described only as mapped E−P classes.
12. DIPI should be presented as an exploratory screening index. Its weights are not calibrated, well density is only a proxy for water use, and the mining layer does not quantify dewatering intensity. Comparing one weighted scheme with one equal-weight scheme does not establish robustness or transferability.
13. The Zenodo v1.1 package available through DOI 10.5281/zenodo.19371351 on 16 July 2026 contained placeholder material and example metadata for only PZ01–PZ03. It did not contain the complete processed data or an executable workflow. The authors should provide the actual reproducibility package or a direct link to any complete GitHub repository.
14. Use “Drought Impact Potential Index” consistently; “Drought Impact and Pressure Index” is used incorrectly in several places.
15. Sect. 2.3 incorrectly refers to Fig. 6 for the pressure map; the relevant maps are in Fig. 7 or Fig. A1.
16. Lines 144–145 list four aquifer categories, whereas the analysis uses three groups.
17. The statement that deeper aquifers generally have higher propagation probability is not consistently supported and should be reconsidered after the metric has been verified. Depth is also confounded with location and hydrogeological setting.
Citation: https://doi.org/10.5194/egusphere-2026-1904-RC3 -
AC3: 'Reply on RC3', Katarzyna Sawicka, 21 Jul 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-1904/egusphere-2026-1904-AC3-supplement.pdf
-
AC3: 'Reply on RC3', Katarzyna Sawicka, 21 Jul 2026
Data sets
Processed groundwater drought datasets for Groundwater Body no. 43: SPI, SGI, drought metrics, and DIPI components Katarzyna Sawicka https://doi.org/10.5281/zenodo.19371351
Interactive computing environment
Reproducible Jupyter notebooks for SPI–SGI processing, drought metrics, and DIPI workflow Katarzyna Sawicka https://doi.org/10.5281/zenodo.19371351
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 118 | 29 | 15 | 162 | 7 | 8 |
- HTML: 118
- PDF: 29
- XML: 15
- Total: 162
- BibTeX: 7
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript investigates groundwater drought by combining meteorological drought information, represented by SPI, with groundwater response, represented by SGI. It quantifies drought propagation, lag, recovery, and vulnerability across multiple timescales and compares these processes among shallow, intermediate, and deep aquifer systems. The study further attempts to link these temporal drought-response metrics with spatial vulnerability assessment through the composite DIPI framework.
While the topic is relevant and the SPI-SGI analysis has potential value for understanding groundwater drought propagation. The main limitation is that the hydrometeorological process mechanisms are not developed in enough depth. Instead, the paper places substantial emphasis on hydrogeological interpretation, vulnerability mapping, and the composite DIPI index.
Second, the Exposure component of DIPI is derived from SGI, but SGI itself may already be affected by groundwater abstraction and mining-related dewatering. At the same time, Pressure separately includes abstraction and mining effects. Therefore, E and P are not necessarily independent. The E−P spatial contrast should not be interpreted directly as a causal separation between “intrinsic system response” and “human pressure.”
Third, the spatial interpolation of DIPI is strongly constrained by the limited number of monitoring wells. Reviewers are likely to expect spatial uncertainty analysis, interpolation cross-validation, justification of the interpolation method, and error maps. Without these additions, statements such as “the hotspot configuration is robust” are too strong.
Minor comments
The name of DIPI is inconsistent. The abstract and methods refer to the Drought Impact Potential Index, whereas around line 135 the manuscript refers to the Drought Impact and Pressure Index. The terminology should be made consistent throughout the manuscript.
Line 115 states that the pressure component is shown in Fig. 6, but Fig. 6 actually presents the propagation–recovery trajectories. The DIPI spatial map appears to correspond to Fig. 7 or Appendix A. This figure reference should be corrected.