the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Aug 2025
Hotspots and hot moments of metal mobilization: dynamic connectivity in legacy mine waters
Abstract. Monitoring and treatment of contaminated mine water conventionally focuses on end-of-pipe assessment and remediation techniques, at the downstream outlet of mining sites after closure. Conversely, the initial stages of pollutant release and their pathways within abandoned mines have been largely overlooked. This study examines subsurface mining-affected anthropogenic structures and the dynamic hydrogeochemical loadings and drainage pathways within them, revealing how variable subsurface flow activation impacts metal(loid) mobilization and opens novel direct mitigation options. We identified complex hydrological patterns through the mine (Reiche Zeche, Ore Mountains, Germany) in which percolation paths were dynamically connected to the drainage based on flow conditions. Using in-situ sensors, hydrogeochemical monitoring and stable water isotopes, we reveal a hydrodynamic regime in which episodic shifts in subsurface connectivity govern metal(loid) mobilization from localized storage zones, ultimately controlling solute export to surface waters. We use concentration–discharge (C–Q) relationships, hysteresis indices, and the Pollution Load Index (PLI) to evaluate metal transport during the annual pattern of flow regimes. Our analyses of event-scale C–Q hysteresis patterns reveal site- and element-specific shifts in flow path activation in a very short time. Despite low flow periods traditionally considered low risk for contaminant mobilization, contaminant hotspots within poorly connected hydrological zones can emerge during these times, with high pollution potential and solute accumulation governed by the sequence and timing of crossing or exceeding a connectivity or flow threshold, as described by the hydrological fill-and-spill and geochemical lotic-lentic cycle concepts. Notably, Zn loads during low flow, pre-flush periods reached levels up to six times higher than median values. Preceding the flushing events, geochemical and microbial-mediated metal leaching create the spatially distributed contaminant stock, remobilized during reconnection events. With a large proportion of heavy metal loads occur during low flow and especially just before the high flow (flush) period, source-related mitigation with decentralized water treatment structures becomes much more feasible than end-of-pipe solutions that require higher throughput volumes and multi-element filtering. This work also highlights the need for event-sensitive monitoring and treatment strategy options that prioritize internal system behavior to mitigate pollution risk in abandoned mines and other caverned hydrological systems.
- Preprint
(3204 KB) - Metadata XML
-
Supplement
(2087 KB) - BibTeX
- EndNote
Status: open (until 15 Oct 2025)
-
RC1: 'Comment on egusphere-2025-4092', Anonymous Referee #1, 26 Sep 2025
reply
-
AC1: 'Reply on RC1', Anita Sanchez, 01 Oct 2025
reply
We thank the referee for reviewing our manuscript “Hotspots and hot moments of metal mobilization: dynamic connectivity in legacy mine waters”. We are very grateful for the important clarifications the referee requests for, which we will address when revising the manuscript. Briefly, the major changes that we are planning are:
- Revise hysteresis interpretation – We will substantially revise our hysteresis analysis and discussion to clearly distinguish between enrichment and dilution contexts. We will remove generalizations such as “clockwise loops = proximal sources” and instead interpret hysteresis loops relative to concentration-discharge regime at each site. In addition, we will update the current conceptual representation for C-Q relationships by amending Fig. 1b. To improve our illustration of how hysteresis interacts with enrichment and dilution dynamics, we will explore replacing this with a time-series based schematic.
- Clarify terminology (“load” and “loading”) – To avoid confusion with “load” in terms of flux (C*Q), we will explicitly define our use of “loading” in the manuscript. We will refer to “loading” as temporal concentration accumulation when describing temporal increases in solute concentrations, while reserving “load” for flux-based metrics (C*Q). This will be stated early on in the abstract of the manuscript when these terms are first used.
- Improve phase definitions and visualization – We will revise and clarify in the methods how hydrogeochemical phases were identified and separated (based on C-Q slopes and Hysteresis index values during rising/falling discharge). In the figures, phases with similar C-Q relationships will be grouped with more consistent color schemes. We will also revise the text to make clear that while “loading” and “dilution” both describe a negative relationship between C and Q, these phases are distinguished by their relative timing of concentration response compared to discharge. For example, loading refers to concentration increases during low or rising flows (C responds faster than Q), whereas dilution refers to delayed concentration decreases relative to rising discharge (C responds slower than Q). Similarily, “flushing” and “recession” are differentiated by whether concentration lags or leads discharge changes. We will also adjust the figure colors so that the phases with the same C-Q slope direction (positive vs. negative) are grouped in similar hues.
Beyond this general plan, here is the detailed response to the reviewers comments:
Referee #1: In their manuscript entitled „Hotspots and Hot Moments of Metal Mobilization: Dynamic Connectivity in Legacy Mine Waters“, Sanchez et al., (2025) analyzed metal(loid) pollution dynamics in a mining-affected area. Based on an extensive monitoring program and using C-Q relationships, the authors reveal the high spatial and temporal variability of hydrological metal(loid) transport. The results of this study are quite interesting and therefore likely to be a highly valuable contribution for the readers of HESS. Furthermore, the study is well-written and comprises clear and appealing figures. I therefore recommend considering publishing this study in HESS. However, I do not entirely agree with some of the interpretations of the results, especially regarding the hysteresis analysis, but also with the “loading” term and the definition of phases. Therefore, additional consistency checks and potential revisions to certain parts of the discussion may be necessary, as detailed below.
Thank you very much indeed for your thoughtful evaluation of our manuscript. We see that we have cause some haze with insufficiently crisp wordings piling up to slight inconsistencies. This will be addressed along the lines presented above.
Hysteresis analysis and interpretation
A clockwise hysteresis coupled to an enrichment pattern can imply something very different than a clockwise hysteresis coupled to a dilution pattern, and vice versa. Together with an enrichment pattern, clockwise hysteresis indicates a faster reaction of concentrations (C) as compared to discharge (Q), in the form of a faster increase of C, a faster recovery after the peak, or both. Consequently, counterclockwise and enrichment indicate a slower reaction of C as compared to Q. Instead, dilution and clockwise hysteresis imply a slower reaction of C as compared to Q, in the form of a delayed decrease in C, a slower recovery/increase after the peak, or both. Counterclockwise hysteresis and dilution imply a faster response of C compared to Q. As this may not be entirely intuitive, I have added a hand-drawn sketch to illustrate this for dilution patterns (as these were the primary patterns reported). I hope it is readable.
Given these different C responses for the same hysteresis pattern, one must be very careful and avoid interpretations based on enrichment patterns from other studies. So one cannot generally say that “Clockwise loops suggest rapid mobilization from proximal sources”, as stated in the manuscript and later on built upon that statement for further interpretation. This needs to be carefully checked throughout the manuscript, and interpretations based on this need to be re-evaluated.
We appreciate for pointing out this important distinction and for providing a hand-drawn sketch to better understand how to properly differentiate these relationships. We agree that hysteresis behavior must be interpreted in the context of enrichment versus dilution regimes. Our attempt to unify some of the common approaches (e.g., Pohle et al., 2021; Speir et al., 2024; Rose et al., 2018; Lloyd et al., 2015; Zuecco et al., 2015; Vaughan et al., 2017; Knapp et al., 2020; Musolff et al., 2021) obviously has left this distinction open. We agree and will revise the manuscript to reflect this. Specifically, we will remove generalized statements linking clockwise or counterclockwise loops to proximal or distant sources. In the revised text we will interpret hysteresis loops relative to the underlying concentration-discharge pattern at each site by making it clear that clockwise loops under dilution reflect delayed solute recovery compared to discharge and clockwise loops under enrichment reflect a faster response of concentration than of discharge. In addition, we will update Fig. 1b accordingly. Specifically, we are reconsidering the current conceptual representation and will explore replacing it with a time-series based schematic that better illustrates how hysteresis interacts with enrichment and dilution dynamics. We believe these revisions will improve the rigor and clarity of our hysteresis interpretation.
Additionally, while I agree that hot moments observed for 3A are striking, the dynamics for 2 and 3B are less pronounced. They appear to show slow and steady buildup of concentrations, for example, due to an increasing water age during low flow, and a rapid dilution with incoming (younger and less exposed) event water (as also described in lines 465-466). I do not see much of the mobilization pattern or threshold behavior here, which is constantly referred to in the manuscript. Overall, I recommend a clearer differentiation between the potentially driving processes at different sites and a reevaluation of the potential underlying processes.
We appreciate your observation. We agree that the strength of threshold-driven mobilization differs across sites, with site 3A showing the clearest evidence. However, we emphasize that site 2, exhibiting gradual solute buildup during low flow, still revealed threshold-like dynamics when viewed in detail. These features include stratification and density-driven storage, abrupt concentration drops at the onset of flushing, and phase-specific hysteresis behavior. These features indicate that threshold exceedance processes were also operating at site 2, though modulated by site-specific hydrodynamics. In the revised manuscript, we will clarify that threshold-driven behavior was most pronounced at site 3A, while at site 2 and 3B additional process (e.g., dilution by younger recharge waters, stratified storage) also played a role as well. Since obviously, a more detailed monitoring at site 3A (using a spectrolyser there too) would have been advantageous but simply has not been done, we will add this aspect in the discussion, too.
Loading
Finally, I got a little confused by the terms 'load' or 'loading'. My initial interpretation was 'load' in terms of C * Q. Only when it came to explaining the PLI did I understand that it is about concentrations, which makes a difference in how to interpret the results. To avoid this confusion, I suggest choosing a different term or explaining this difference in terminology early on in the manuscript.
Thank your for drawing our attention to this ambiguity. In the revised manuscript, we will explicitly define “load” as flux (C*Q) and state that the term “loading” is used to describe temporal concentration accumulation. This distinction will be introduced early on in the manuscript when these terms are first presented and applied consistently throughout the manuscript.
Hydrogeochemical phases
I find the naming and visualization of ‘phases’ somewhat confusing. In the sense of the directional C-Q relationship, what is here defined as ‘Loading’ and ‘Dilution’ describe the same negative relationship between C and Q. The same applies to what is called ‘flushing’ and ‘recession’, which both describe a positive relationship between C and Q. While it might make sense to divide these phases also into those of rising and falling discharge, their common C-Q relationship should be made clearer in the text and in the figures, for example by coloring the same C-Q relationships in a more similar way. Also, I did not entirely understand how that was calculated. If it is based on the C-Q relationships, was it calculated using a moving average, or how were the phases separated? I might have missed it, but the manuscript would benefit from stating that clearer in the methods.
You are completely right for pointing out that our wording could be clearer. The phases in our study were defined from point-to-point C-Q slopes and hysteresis index (HI) calculations, as outlined in section 2.6.5 of the methods. Each segment of the C-Q trajectory between consecutive sampling points was classified based on the sign of the slope (positive/negative, i.e., enrichment vs. dilution) and the hysteresis direction (clockwise/counterclockwise, HI > 0 or HI < 0). This procedure allowed us to identify whether concentrations responded faster or slower than discharge during the rising and falling limbs. To clarify, the four phases mentioned are differentiated as follows:
- loading – negative slope (C increases relative to Q), typically during low or rising flow; concentrations respond faster than discharge
- dilution – negative slope (C decreases relative to Q), typically during rising flow; concentrations respond more slowly than discharge
- flushing – positive slope (C increases with Q), usually on the rising limb; concentrations lag behind discharge but increase with renewed connectivity
- recession – positive slope (C decreases with Q), typically on the falling limb; concentrations decline in parallel or faster than discharge.
To avoid further confusion, we will revise section 2.6.5 to explicitly state this stepwise classification procedure and will clarify that no moving averages were used since the phases are derived directly from sequential data points. For the figures, we will also adjust the figure colors so that the phases with the same C-Q slope direction (positive vs. negative) are grouped in similar hues.
Line-to-line comments
- L21-24: This sentence is hard to understand. I suggest splitting and rewriting it.
Thank you for highlighting this. We see how the phrasing could be confusing. We will split and rewrite the sentence in a more straightforward and easy to understand way.
- L39: I am not sure ‘discrete’ is the right term here. As the measurement points in the study are discrete as well.
Thanks for noting this. We agree that discrete could be misleading in this context. We will replace this term with “low-resolution” to better convey that standard monitoring provides infrequent measurements and limited temporal coverage.
- L61: Readers would benefit if this was explained a little more in detail.
We’re grateful for this observation. We will expand the sentence to clarify what is meant by “lagged”, “low-pass filtered”, and “threshold-dependent”. The revised text will explain that near surface signals are delayed by slow percolation, short-term variations are dampened by storage effects, and responses often occur only after thresholds of connectivity or storage are exceeded.
- L65: adding ‘e.g.’? Here and in other exemplary citations as well?
Thanks for this suggestion. We agree that the citations presented serve as examples rather than an exhaustive list. Accordingly, we will revise the sentence to include “e.g.” before the citations to clarify this point.
- L71: I find it a little hard to imagine what exactly the measurement points look like, if it is not a surface catchment. Maybe some photos, even if only in the SI, would help the readers to imagine the right thing here.
Thanks for raising this point. We understand that it may be difficult to imagine how the underground sampling points look like. We will add photos of sites 1, 2, 3A, and 3B with the dates noted in which they were taken in the SI and link to them in the manuscript.
- L87-90: I suggest adding Musolff et al. (2015).
Thanks for this suggestion. Given that C-Q relationships have been applied in a wide range of studies, we have missed to add this important and clarifying reference. Since some of your conceptual suggestions appear to be well-aligned with this study, we revise the hysteresis analysis accordingly and add the citation.
- L91-92: Only for enrichment patterns. See my comment above.
Thank you for highlighting this important nuance. We agree that our original statement (“clockwise loops suggest rapid mobilization from proximal sources…”) was an oversimplification. We will revise the manuscript to clarify that hysteresis direction must be interpreted in the context of the underlying concentration-discharge regime. Specifically, we will include that clockwise and counterclockwise loops can imply different dynamics depending on whether concentrations are increasing (enrichment) or decreasing (dilution) with discharge. We will also revise the conceptual representation of C-Q relationships, specifically in Fig. 1b, to avoid any misinterpretations.
- Figure 1) Overall, the figures in the manuscript have a very high quality. Still, I have some suggestions for improvement:
c) Sorry to be a little peaky, but conceptually this is not entirely correct. For example, the lower dark blue dot (‘flush’) describes the same Q level as the upper green dot that describes ‘declining flow’. The same applies to the upper yellow and lower green dots.
We appreciate this being highlighted. We will revise Fig. 1c to avoid overlapping discharge values across phases. In the updated version, each dot (phase) will be placed along more distinct discharge ranges, ensuring a consistent trajectory from flush to declining flow to low flow. We believe this will improve the conceptual clarity of the figure.
d) I suggest not using numbers a-d twice in the figure, but instead using other numbering for the subfigures within 1d).
Yes. We will use other numbering for the subfigures in Fig. 1d to avoid confusion.
- L121: Does ‘high-resolution’ refer to spatial, temporal, or both?
Thanks for pointing this out. “High-resolution” refers to both spatial and temporal. We will clarify this in the revised manuscript.
- L270-276: I suggest adding the equation here. I assume it is Log10(C) = a + log10(Q) ^b ?
We agree. The equation is Log10(C) = a + b*Log10(Q) and we will add it to this section, which makes sense since we include the equation for PLI.
- L273-274: b=0 does not necessarily imply that concentrations are stable. While this is often the case, it could also be that concentrations vary independently of Q. For this reason, Musolff et al. (2015) combined slope b with the CVC/CVQto test whether concentrations are stable. Hence chemostatic means b ~ 0 and CVC/CVQ<<0.5. This is not only true for that figure, but for the general interpretation of results. I recommend applying the CVC/CVQ to all metal(loids) to assure that b=0 can actually be described as chemostatic behavior.
Thank you for highlighting this important nuance. Following this suggestion, we will calculate CVc/CVq for all metal(loid)s at the fours sites. In the revised manuscript, we will clarify that while some point-to-point segments with b ~ 0 also show CVc/CVq << 0.5 (consistent with chemostatic behavior), others may exhibit CVc/CVq > 0.5, indicating variability in concentrations independent of discharge. We will further state that cases will only be described as chemostatic when both criteria are fulfilled. These revisions will improve the transparency of our interpretation and classifications.
- L276: I would argue that a positive slope does not indicate ‘increased hydrological connectivity’, but an increased mobilization of solutes during increased hydrological connectivity.
We agree with this statement and will amend the text accordingly.
- L278-290: I suggest adding the equations here as well, so that readers can understand how this index was calculated.
We agree that the equations used to calculate the hysteresis index values from the methods of Lloyd et al. (2015) and Zuecco et al. (2015) should be included. We will definitely add these.
- L306: I suggest citing the python packages used.
Yes, this is fair. We will revise the methods section to include citations for the Python packages (pandas and plotly libraries) used.
- L323-325: Couldn’t it also be that the water just needs a little longer to percolate down to these deeper layers?
Thanks for this good perspective. While delayed percolation through the subsurface could in principle contribute to the lag between surface conditions and increased discharge, our observations suggest that deeper layers in this system generally respond rapidly, with streamflow dynamics at depth aligning closely with overall discharge. We therefore interpret the observed lag not as a simple percolation delay, but as evidence of threshold-based fill-and-spill dynamics within vertically structured storage zones. To improve clarity, we will revise the discussion to briefly acknowledge delayed percolation as a possible mechanism, but emphasize that our data support threshold-driven connectivity as the dominant explanation in this setting.
- Figure 3: Again, very nice figure. For the coloring of phases, see my comment above. Also, the light and dark blue colors are hard to distinguish, especially if the lines are dashed.
Thank you for the compliment and for the suggestion. We will address this by making loading and dilution similar hues of red and flushing and recession similar hues of blue. We will keep the chemostatic and variable sources colors as they are. We will also make the color of the dashed lines more bold so they are clearly visible and ensure that all colors presented in the figure are clearly distinguishable.
- L361: I would be interested to see the dynamics of the other metal(loids) measured as well. Maybe something for the SI?
This is a great point. We will include the load plots in the SI for the other measured metals including Fe, Pb, Cd, Cu, Al, Mn, and Ni.
- Figure 4: Please add the full legend.
We apologize for missing that. We will add the full legend with the geochemical phases for Zn.
- L407: For the combination of hysteresis and C-Q slopes, I recommend citing Winter et al. (2021) here as well.
Thanks for this great recommendation. Winter et al. (2021) provides a valuable framework for combining C-Q slopes with hysteresis analysis across multiple time scales, which is directly relevant to our study. We will include this reference in the revised manuscript and revise how to find a more clear and comprehensive phase classification.
- L409: A slope < -1 implies that not only concentrations but even the load (CxQ) decreases. Hence, it is not only a dilution of baseflow by freshly incoming water, but also the overall mass of metals declines, compared to what we had seen before the event. This is interesting. I could imagine it might be explained by the previous accumulation of metals during low flow that is not flushed out of the system, but I would appreciate the thoughts of the authors on this in the discussion.
- At the same time, I would be careful to relate a slope of -0.55 to transport-limited behavior, as this is still a
- L412: As I understand, solute concentrations do not rise, but decline. The entire paragraph needs to be carefully checked in line with my comment above.
- L449: The peak alignment is interesting and might tell us something about the underlying processes. From my interpretation, it appears to emphasize the importance of hydrological processes, specifically dilution, but it does not align well with the theory of distant versus proximal sources.
Thank you for these very thoughtful and connected comments on the interpretation of dilution slopes, load dynamics, and peak alignments. We acknowledge that in the original draft, some of these aspects may have been simplified too much or not fully elaborated, which has caused some inconsistencies. We highly appreciate that you point us to these issues. Upon revisiting our data, we recognize that the strongest patterns at sites 3A and 3B reflect not only dilution (slope < –1) but also a decline in total metal loads (after a flush event as seen in Fig. 4), pointing to depletion of previously accumulated solute pools. At site 2, by contrast, the slope of –0.55 reflects a weaker dilution signal, consistent with partial mobilization of solutes rather than purely transport-limited nor source-limited behavior. We further agree that the alignment of hysteresis peaks emphasizes hydrological processes of dilution such that whether there is a slower or faster reaction of concentration than discharge and whether there is a higher or lower loading on the falling limb. Therefore, this does not necessarily map onto a simple proximal versus distal source interpretation, as you have pointed out. In the revised manuscript, we will carefully revise Section 3.3 and associated figures to (i) explicitly separate weak vs. strong dilution patterns, (ii) clarify that slopes < –1 potentially indicate load depletion, (iii) reframe site 2 as partial mobilization rather than fully transport limitation, and (iv) present both dilution-driven and threshold-driven interpretations of peak alignment. These changes will improve the consistency, nuance, and mechanistic grounding of our discussion.
- L515-546: Overall, I find this very convincing. However, I miss a critical discussion of what might be the drawbacks or, instead, the advantages of end-of-the-pipe solutions. I am not an expert, but I could imagine end-of-the-pipe solutions are easier to implement as they only need to be implemented once, and there might be less danger of missing specific spots? When only looking at specific hot spots, there might be a risk of overlooking other hot spots, especially in hydrologically complex mining-impacted systems, right?
We appreciate this excellent point. We agree that a critical discussion of end-of-pipe versus hotspot-focused remediation approaches strengthens the manuscript. In the revision, we will frame end-of-pipe strategies not only in terms of their ease of implementation and lower risk of overlooking active zones, but also as a potential solution that has so far received limited evaluation in the context of legacy mine systems. At the same time, hotspot-focused interventions offer the ability to intercept highly concentrated pulses but may miss other active sources in hydrologically complex settings. We will further emphasize that the very dynamic nature of contaminant mobilization, which is rarely integrated into current monitoring or planning, suggests that the most effective approach may be a combination of system-scale safeguards with targeted diagnostics.
- L540: Is that load in terms of PLI, or load in terms of C*Q?
The way load is used here is in terms of C*Q. We will clarify this by stating “load from flux-based metrics”.
Thank you very much again for your thoughtful evaluation of the manuscript and your very constructive suggestions. Your comments help us a lot to improve the paper towards coherence and clarity.
Sincerely,
Anita Alexandra Sanchez and Conrad Jackisch
(on behalf of all co-authors)
Citation: https://doi.org/10.5194/egusphere-2025-4092-AC1
-
AC1: 'Reply on RC1', Anita Sanchez, 01 Oct 2025
reply
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 2,097 | 38 | 6 | 2,141 | 15 | 22 | 16 |
- HTML: 2,097
- PDF: 38
- XML: 6
- Total: 2,141
- Supplement: 15
- BibTeX: 22
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
In their manuscript entitled „Hotspots and Hot Moments of Metal Mobilization: Dynamic Connectivity in Legacy Mine Waters“, Sanchez et al., (2025) analyzed metal(loid) pollution dynamics in a mining-affected area. Based on an extensive monitoring program and using C-Q relationships, the authors reveal the high spatial and temporal variability of hydrological metal(loid) transport. The results of this study are quite interesting and therefore likely to be a highly valuable contribution for the readers of HESS. Furthermore, the study is well-written and comprises clear and appealing figures. I therefore recommend considering publishing this study in HESS. However, I do not entirely agree with some of the interpretations of the results, especially regarding the hysteresis analysis, but also with the “loading” term and the definition of phases. Therefore, additional consistency checks and potential revisions to certain parts of the discussion may be necessary, as detailed below.
Hysteresis analysis and interpretation
A clockwise hysteresis coupled to an enrichment pattern can imply something very different than a clockwise hysteresis coupled to a dilution pattern, and vice versa. Together with an enrichment pattern, clockwise hysteresis indicates a faster reaction of concentrations (C) as compared to discharge (Q), in the form of a faster increase of C, a faster recovery after the peak, or both. Consequently, counterclockwise and enrichment indicate a slower reaction of C as compared to Q. Instead, dilution and clockwise hysteresis imply a slower reaction of C as compared to Q, in the form of a delayed decrease in C, a slower recovery/increase after the peak, or both. Counterclockwise hysteresis and dilution imply a faster response of C compared to Q. As this may not be entirely intuitive, I have added a hand-drawn sketch to illustrate this for dilution patterns (as these were the primary patterns reported). I hope it is readable.
Given these different C responses for the same hysteresis pattern, one must be very careful and avoid interpretations based on enrichment patterns from other studies. So one cannot generally say that “Clockwise loops suggest rapid mobilization from proximal sources”, as stated in the manuscript and later on built upon that statement for further interpretation. This needs to be carefully checked throughout the manuscript, and interpretations based on this need to be re-evaluated.
Additionally, while I agree that hot moments observed for 3B are striking, the dynamics for 2 and 3B are less pronounced. They appear to show slow and steady buildup of concentrations, for example, due to an increasing water age during low flow, and a rapid dilution with incoming (younger and less exposed) event water (as also described in lines 465-466). I do not see much of the mobilization pattern or threshold behavior here, which is constantly referred to in the manuscript. Overall, I recommend a clearer differentiation between the potentially driving processes at different sites and a reevaluation of the potential underlying processes.
Loading
Finally, I got a little confused by the terms 'load' or 'loading'. My initial interpretation was 'load' in terms of C * Q. Only when it came to explaining the PLI did I understand that it is about concentrations, which makes a difference in how to interpret the results. To avoid this confusion, I suggest choosing a different term or explaining this difference in terminology early on in the manuscript.
Hydrogeochemical phases
I find the naming and visualization of ‘phases’ somewhat confusing. In the sense of the directional C-Q relationship, what is here defined as ‘Loading’ and ‘Dilution’ describe the same negative relationship between C and Q. The same applies to what is called ‘flushing’ and ‘recession’, which both describe a positive relationship between C and Q. While it might make sense to divide these phases also into those of rising and falling discharge, their common C-Q relationship should be made clearer in the text and in the figures, for example by coloring the same C-Q relationships in a more similar way. Also, I did not entirely understand how that was calculated. If it is based on the C-Q relationships, was it calculated using a moving average, or how were the phases separated? I might have missed it, but the manuscript would benefit from stating that clearer in the methods.
Line-to_line comments
L21-24: This sentence is hard to understand. I suggest splitting and rewriting it.
L39: I am not sure ‘discrete’ is the right term here. As the measurement points in the study are discrete as well.
L61: Readers would benefit if this was explained a little more in detail.
L65: adding ‘e.g.’? Here and in other exemplary citations as well?
L71: I find it a little hard to imagine what exactly the measurement points look like, if it is not a surface catchment. Maybe some photos, even if only in the SI, would help the readers to imagine the right thing here.
L87-90: I suggest adding Musolff et al. (2015).
L91-92: Only for enrichment patterns. See my comment above.
Figure 1) Overall, the figures in the manuscript have a very high quality. Still, I have some suggestions for improvement:
c) Sorry to be a little peaky, but conceptually this is not entirely correct. For example, the lower dark blue dot (‘flush’) describes the same Q level as the upper green dot that describes ‘declining flow’. The same applies to the upper yellow and lower green dots.
d) I suggest not using numbers a-d twice in the figure, but instead using other numbering for the subfigures within 1d).
L121: Does ‘high-resolution’ refer to spatial, temporal, or both?
L270-276: I suggest adding the equation here. I assume it is Log10(C) = a + log10(Q) ^b ?
L273-274: b=0 does not necessarily imply that concentrations are stable. While this is often the case, it could also be that concentrations vary independently of Q. For this reason, Musolff et al. (2015) combined slope b with the CVC/CVQ to test whether concentrations are stable. Hence chemostatic means b ~ 0 and CVC/CVQ <<0.5. This is not only true for that figure, but for the general interpretation of results. I recommend applying the CVC/CVQ to all metal(loids) to assure that b=0 can actually be described as chemostatic behavior.
L276: I would argue that a positive slope does not indicate ‘increased hydrological connectivity’, but an increased mobilization of solutes during increased hydrological connectivity.
L278-290: I suggest adding the equations here as well, so that readers can understand how this index was calculated.
L306: I suggest citing the python packages used.
L323-325: Couldn’t it also be that the water just needs a little longer to percolate down to these deeper layers?
Figure 3: Again, very nice figure. For the coloring of phases, see my comment above. Also, the light and dark blue colors are hard to distinguish, especially if the lines are dashed.
L361: I would be interested to see the dynamics of the other metal(loids) measured as well. Maybe something for the SI?
Figure 4: Please add the full legend.
L407: For the combination of hysteresis and C-Q slopes, I recommend citing Winter et al. (2021) here as well.
L409: A slope < -1 implies that not only concentrations but even the load (CxQ) decreases. Hence, it is not only a dilution of baseflow by freshly incoming water, but also the overall mass of metals declines, compared to what we had seen before the event. This is interesting. I could imagine it might be explained by the previous accumulation of metals during low flow that is not flushed out of the system, but I would appreciate the thoughts of the authors on this in the discussion.
At the same time, I would be careful to relate a slope of -0.55 to transport-limited behavior, as this is still a considerable dilution pattern.
L412: As I understand, solute concentrations do not rise, but decline. The entire paragraph needs to be carefully checked in line with my comment above.
L449: The peak alignment is interesting and might tell us something about the underlying processes. From my interpretation, it appears to emphasize the importance of hydrological processes, specifically dilution, but it does not align well with the theory of distant versus proximal sources.
L515-546: Overall, I find this very convincing. However, I miss a critical discussion of what might be the drawbacks or, instead, the advantages of end-of-the-pipe solutions. I am not an expert, but I could imagine end-of-the-pipe solutions are easier to implement as they only need to be implemented once, and there might be less danger of missing specific spots? When only looking at specific hot spots, there might be a risk of overlooking other hot spots, especially in hydrologically complex mining-impacted systems, right?
L540: Is that load in terms of PLI, or load in terms of C*Q?