the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Nov 2025
Uncovering the melt: UAS and in-situ sensor synergies reveal DOC pathways in a northern peatland
Abstract. Spring snowmelt is a critical period for dissolved organic carbon (DOC) export from northern boreal peatlands, yet the spatiotemporal dynamics of this process remain poorly understood. To reveal the spatial patterns, we used a novel combination of high-resolution Unmanned Aircraft System (UAS) snow depth mapping, topographic wetness index, and high-frequency stream monitoring. Our results show that substantial DOC leaching is triggered after widespread snow cover depletion, likely due to thawing of surficial peat layers. High-resolution UAS snow surveys captured the progression of snowmelt from drier, south-facing slopes and forested areas toward wetter fen areas, with the expansion of snow-free areas in high-wetness zones initiating hydrological connectivity and rapid DOC flushing. Event-based hysteresis and flushing analyses enabled by high-frequency stream monitoring revealed transitions from deeper to more surficial flow paths towards the final peak melt. The integration of high-resolution spatial and temporal datasets enabled the detailed identification of DOC transport mechanisms during the snowmelt period. These findings underscore the sensitivity of peatland carbon dynamics to late winter processes and snow conditions, highlighting their potential vulnerability to future shifts in climate.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3730 KB)
-
Supplement
(1029 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3730 KB) - Metadata XML
-
Supplement
(1029 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4682', Anonymous Referee #1, 18 Nov 2025
- AC1: 'Reply on RC1', Petra Korhonen, 23 Jan 2026
-
RC2: 'Comment on egusphere-2025-4682', Anonymous Referee #2, 06 Dec 2025
This study presents the use of Unmanned Aircraft System (UAS) snow cover/depth measurements on a fen to complement the interpretation of how the dynamics of source areas and flow paths influence the export of dissolve organic carbon (DOC) from a small, high latitude catchment during snowmelt. Oxygen isotopes, continuous groundwater level measurements and snow surveys, as well as continuous measurement of an optical proxy for DOC are valuable complements to the data used in the analysis. Hysteresis and flushing indices are a key part of the reasoning used in ascribing the role of hydrological connectivity bringing in new source areas and dilution of source areas during events.
Major concerns:
Much emphasis is placed on interpreting the variation in DOC concentrations. However what strikes me more is the overall stability of DOC when the authors say that as much as 50% of the runoff peaks could come from snowmelt/precipitation inputs that are likely to be sources with little DOC.
The interpretation of the rich data sources in this paper would be much more convincing if the different catchment DOC sources and pathways alluded to in the discussion were quantified in an approach that keeps track of mass balances of water and carbon. Without that, I am concerned that conclusions may be drawn that are at odds with what the data show. In such a quantification, care should also be taken to explain what the inclusion of the UAS data adds to the interpretation of the other data.
Minor comments
Section 2.1
Is there any information on how well this groundwater level represents the spatial variability in GW level across the fen?
Is there information on how the water from the larger, upslope catchment area moves through the fen? Are subsurface, preferential flowpaths an important feature?
Section 2.2 – Were the daily stream isotope samples taken at the same time each day. During spring, the runoff can have strong diurnal variation, so where one samples in that diurnal variation is important.
Section 2.3 – While there is good coverage of snow depth in the open fen, is there any information about the progress of snowmelt in the larger, upslope forested area which is presumably a large source of the water measured at the flume?
How well does the 10 cm resolution in elevation measurement of each cell compare to the variation of the topography across the fen? Is this important for assessing how well TWI can be identified.
Snow water equivalent was apparently measured at 5 points. What is the variation between points? This should be of relevance for the assumption of a uniform SWE across the fen. While an average SWE may be relevant for the water balance of the catchment, if one wants to consider local inputs of water to the catchment, then that variation in SWE will be important. A relationship between degree of change in snow depth and SWE from your point measurements might be a way to improve precision on where water is being input to the fen surface. The use of a reference from alpine areas (Lopez-Moreno) to argue that SWE is relatively uniform may be problematic. It would be better to know how SWE varies in a high-latitude fen as opposed to an alpine area with much more relief and different diurnal patterns of insolation.
Section 2.4.1 – is there any test of how well the processing of UAS images was able to capture the variation of snow depth and SWI across the fen?
Line 251: Data is plural, so “data was” should read “data were”.
Line 270-280 –MAJOR It appears that different source areas are being assumed AND that the concentration of the source areas is changing. How can one distinguish between changes in source area concentration and changes in source areas? Please be more clear about the assumptions being used in the hysteresis and flushing analyses.
Section 3.1, Line 299 – should RMSE have the unit of cm?
Fig. 3. Please include uncertainty in this. It is worry ing that such a large portion of the predicted data is excluded as unreasonable. What does the non-excluded data actually say about the snow depth, especially when the ambition is to map the spatial variability of the snow depth.
Lines 330-335. The relation of SWI and snowdepth is discussed. Does the SWI get influences by snowdepth?. I presume SWI is calculated during the snow-free period, but it would be good to clarify this in the methods.
Line 384-388 The change in groundwater levels from being above the soil surface until the final runoff peak, after which the water table is below the soil surface seems to indicated a very important shift in flow paths and hydrological connectivity in the fen. Yet this major change gets little mention and does not seem to have a major impact on the DOC concentration. This seems to deserve more attention in the discussion where groundwater does not get much mention.
Figure 7 – The “snow” value of the oxygen isotopes: is that sampled in the snow pack I assume? Please make that clear. If it is the isotope signature remaining in the snowpack, this means the snowpack is contributing meltwater with a less depleted isotope ratio. This complicates the interpretation. A quantitative hydrograph separation would be a much more satisfactory basis for the interpretation of the runoff sources. I would recommend an explicit hydrograph separation. There will of course need to be assumptions, but then the assumptions will be explicit.
Results:
Much is made about the variation in the DOC concentration, but what strikes me is how stable the DOC is, especially when the precipitation and snowmelt entering the catchment on top of frozen fens soils are potential sources of runoff with almost no DOC until they interact with the subsurface. But that interpretation depends on quantifying the inputs of “recent precipitation” along flow paths that stay out of the soil/peat. The fact that the groundwater well is frozen, with water tables above the peat surface during most of the study period (Fig. 7e) would indicate that the input of precipitation/snowmelt on the mire that makes it to the stream would be a source without much DOC at all.
Figure 8 – please find a way to help the reader better compare panes (a) and (b) of the figure. Stacking one on top of the other, rather than side by side would be one easy way, even if it would take more space.
Line 480. The role of ice is mentioned, with the UAS snow depth analysis interpreting surficial ice as snow. Please say more about “formation of ice in low lying areas”. My experience is that the upper centimetres of mires are often frozen during the winter, whereas forested upland areas often do not have much ice in the soil profile. Is the entire fen frozen to a certain depth (some centimetres), or is it a crust of ice on the snow and or soil surface that is referred to here. This is good to clarify since the depth of soil frost has a bearing on the interpretation of flow paths and where the DOC sources are that keep DOC from diluting more as snow melts on the fen, creating large inputs of low-DOC water close to the flume.
Speaking of the flume, does it capture all the water running off from the upslope areas? With water tables above the ground surface on a relatively flat landscape, there might be a possibility of water bypassing the flume.
Citation: https://doi.org/10.5194/egusphere-2025-4682-RC2 - AC2: 'Reply on RC2', Petra Korhonen, 23 Jan 2026
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4682', Anonymous Referee #1, 18 Nov 2025
General comments
This is an interesting setup touching a weak spot in hydrology and solute mobilization: While connectivity is used to explain solute export dynamics, it is hard to actually map and measure. This study can go a step in this direction by a drone-based assessment of snow cover changes during a major snowmelt event combined with concentration dynamics in the receiving stream. This topic is of great interest to readers of HESS.
While the data and effort are impressive, I am not totally convinced about the way results are described and interpreted. From my point of view the result section can be more concise and the discussion section much more integrative. I see a tendency to overinterpret patterns. My suggestion is a careful revision following the points raised below. I moreover encourage the authors to look for a measure of connectivity in the snow cover data that can be used as an explanatory variable in the event C-Q analysis. Just fraction of area covered by snow or snow depths is not telling a story of connectivity of source zones to the stream. However, measures of spatial connectivity of snow-free area changing over time may provide that.
Specific comments
Title
For me UAS is not self-explaining and I would even avoid DOC as an abbreviation in the title.
Abstract
L9: I miss a reference to the location and size of the catchment/area to help the reader understanding what dimension you are talking about.
Introduction
The introduction reads very well – not much to revise from my point of view. However, I found the isotope analysis as not at all motivated in the introduction. You should mention at some point here and / or in the methods what this analysis is used for.
L56: “water table” or better “water levels” is enough.
L60: These studies use the TWI at catchment scale or in riparian zones but not in peatlands. Are there also examples from peatlands pointing a dominant topographic control of discharge generation?
Methods
L109: To what is the abbreviation “FMI” pointing to?
L115: Try to be more precise here. Is this the pH in the surface water or in the soil water?
Fig. 1: I find it rather unusual to use a copy of a topographic map as a background. All shown features are useless unless explained in a legend. But are all shown features necessary to know? Isolines lack numbers! Potentially use the right map to show the location of the research station/ precipitation sampler.
L177: Explain GCPs.
Fig. S2: This relationship looks rather weak. Can you report on the R2 and bias and have you used the confidence interval later on as shown (but not described) here?
Results
I find the description of snowmelt with 5 figures and 2 tables as too extensive. Information can be transferred in a more condensed way focusing on the information that is really needed in the subsequent analysis.
L291: What caused the rapid snowmelt? Temperature? Rainfall on snow?
L292: To what statistic measure “variation” is referring to? Can you be more quantitative here?
L331: For me it is a surprise that the SWI is dynamic. The description in the method does not point to SWI being used as a dynamic surface feature. I interpreted it as a static topography feature. So, for me this is hard to understand. Do you describe snow melt in different classes of the snow-free topography or temporally dynamic SWI of the snow surface?
L334f: Can you state if difference were statistically significant?
Table 3: For me it would be a better option to show the content of that table in Fig. 6 as a third panel.
Chapter 3.2: The title implies a description of DOC only but the chapter contains much more.
Fig. 7: Use TSSeq concentrations in the axis as well. Is this mg/L as a unit or rather unitless? What is the data source for snow depths here? Is this the same data as described above (UAS)? I am not sure if there is a reason to display load of DOC and TSS concentrations in the same plot. Same for WTD and water temperature.
L355: Consider a different wording as “followed by” implies that first discharge increased and then air temperature and rainfall increased (that actually triggered the discharge?).
L362-374: Concentration of DOC hardly change over time so that the discharge dynamics are the overwhelmingly dominant driver of the load, right? This stark differences in the variation of both could be mentioned here.
L398: This statement puzzles me as the relative position of DOC and Q in the plot (Fig. 8) is matter of the scaling of the two different Y-axes.
Fig. 8: Why is cumulative Q and SWE loss with the same unit referring to different Y-axes. This should be on the same axis. I have troubles understanding the SWE loss in the figure. Majority happens at the 9th May – I see that this is due to a gap in the data. However, the way to display this is not helpful. Consider to leave out the vertical line starting from 0.
L433-435: Some of the information are redundant here as anticlockwise and HI<0 is the same thing.
Fig. 9: Use TSSeq on the axis.
Discussion
I have issues with the cut between chapter 4.1, 4.2 and 4.3. For me the separation is not clear but redundancies are large. Discussion circles around the same processes that are explained by different data in the different chapters. The idea of a discussion should be more integrative and less along the steps of the result section, especially when the same processes are discussed.
L462: Again, I have issues to make the link of snowmelt and SWI. A high SWI marks areas in the landscape that tends to be wetter as flow paths converge (large upstream area, low slope) while low SWI values mark areas that are steeper and have smaller upstream area. How does that come together with the snow melt? In a direct causative way? Or because both are a function of topography? Steeper hillslopes do not allow for snowpack accumulation and are more exposed to radiation… So how does that link to connectivity in the landscape?
L478f: The ice cover is a new result brought up here. For me it does not really explain why ice is forming here especially.
L503: This relative increase was not convincingly shown nor quantified in the result. So, it is a bit hard to follow that argument here.
L509f: However, consider that the dilution effect is very small with concentration hardly changing during the event. This speaks rather for a transport and not a source limitation of DOC. So, from my observation I see very mild dilution effects only and therefore nearly every flow path loaded with DOC and therefore no major changes in sources of flowpath. This is basically a chemostatic system.
L538-553: For me this discussion repeats former statement but add some TSS data. I suggest to strongly reduce redundancies and combine with the discussion above.
L555ff: This is a long statement for a rather simple fact. Nearly invariant concentrations multiplied with highly variant discharge will result in a load that is exactly the same as the discharge.
L592ff: Again, I would be careful in interpreting the mild concentration changes too much. Yes, the described processes are meaningful but I don’t think we see a fundamental change of flow paths and sources but rather slight changes. So, phrases such as “quick depletion” or “sudden depletion” are a bit too much for me.
Citation: https://doi.org/10.5194/egusphere-2025-4682-RC1 - AC1: 'Reply on RC1', Petra Korhonen, 23 Jan 2026
-
RC2: 'Comment on egusphere-2025-4682', Anonymous Referee #2, 06 Dec 2025
This study presents the use of Unmanned Aircraft System (UAS) snow cover/depth measurements on a fen to complement the interpretation of how the dynamics of source areas and flow paths influence the export of dissolve organic carbon (DOC) from a small, high latitude catchment during snowmelt. Oxygen isotopes, continuous groundwater level measurements and snow surveys, as well as continuous measurement of an optical proxy for DOC are valuable complements to the data used in the analysis. Hysteresis and flushing indices are a key part of the reasoning used in ascribing the role of hydrological connectivity bringing in new source areas and dilution of source areas during events.
Major concerns:
Much emphasis is placed on interpreting the variation in DOC concentrations. However what strikes me more is the overall stability of DOC when the authors say that as much as 50% of the runoff peaks could come from snowmelt/precipitation inputs that are likely to be sources with little DOC.
The interpretation of the rich data sources in this paper would be much more convincing if the different catchment DOC sources and pathways alluded to in the discussion were quantified in an approach that keeps track of mass balances of water and carbon. Without that, I am concerned that conclusions may be drawn that are at odds with what the data show. In such a quantification, care should also be taken to explain what the inclusion of the UAS data adds to the interpretation of the other data.
Minor comments
Section 2.1
Is there any information on how well this groundwater level represents the spatial variability in GW level across the fen?
Is there information on how the water from the larger, upslope catchment area moves through the fen? Are subsurface, preferential flowpaths an important feature?
Section 2.2 – Were the daily stream isotope samples taken at the same time each day. During spring, the runoff can have strong diurnal variation, so where one samples in that diurnal variation is important.
Section 2.3 – While there is good coverage of snow depth in the open fen, is there any information about the progress of snowmelt in the larger, upslope forested area which is presumably a large source of the water measured at the flume?
How well does the 10 cm resolution in elevation measurement of each cell compare to the variation of the topography across the fen? Is this important for assessing how well TWI can be identified.
Snow water equivalent was apparently measured at 5 points. What is the variation between points? This should be of relevance for the assumption of a uniform SWE across the fen. While an average SWE may be relevant for the water balance of the catchment, if one wants to consider local inputs of water to the catchment, then that variation in SWE will be important. A relationship between degree of change in snow depth and SWE from your point measurements might be a way to improve precision on where water is being input to the fen surface. The use of a reference from alpine areas (Lopez-Moreno) to argue that SWE is relatively uniform may be problematic. It would be better to know how SWE varies in a high-latitude fen as opposed to an alpine area with much more relief and different diurnal patterns of insolation.
Section 2.4.1 – is there any test of how well the processing of UAS images was able to capture the variation of snow depth and SWI across the fen?
Line 251: Data is plural, so “data was” should read “data were”.
Line 270-280 –MAJOR It appears that different source areas are being assumed AND that the concentration of the source areas is changing. How can one distinguish between changes in source area concentration and changes in source areas? Please be more clear about the assumptions being used in the hysteresis and flushing analyses.
Section 3.1, Line 299 – should RMSE have the unit of cm?
Fig. 3. Please include uncertainty in this. It is worry ing that such a large portion of the predicted data is excluded as unreasonable. What does the non-excluded data actually say about the snow depth, especially when the ambition is to map the spatial variability of the snow depth.
Lines 330-335. The relation of SWI and snowdepth is discussed. Does the SWI get influences by snowdepth?. I presume SWI is calculated during the snow-free period, but it would be good to clarify this in the methods.
Line 384-388 The change in groundwater levels from being above the soil surface until the final runoff peak, after which the water table is below the soil surface seems to indicated a very important shift in flow paths and hydrological connectivity in the fen. Yet this major change gets little mention and does not seem to have a major impact on the DOC concentration. This seems to deserve more attention in the discussion where groundwater does not get much mention.
Figure 7 – The “snow” value of the oxygen isotopes: is that sampled in the snow pack I assume? Please make that clear. If it is the isotope signature remaining in the snowpack, this means the snowpack is contributing meltwater with a less depleted isotope ratio. This complicates the interpretation. A quantitative hydrograph separation would be a much more satisfactory basis for the interpretation of the runoff sources. I would recommend an explicit hydrograph separation. There will of course need to be assumptions, but then the assumptions will be explicit.
Results:
Much is made about the variation in the DOC concentration, but what strikes me is how stable the DOC is, especially when the precipitation and snowmelt entering the catchment on top of frozen fens soils are potential sources of runoff with almost no DOC until they interact with the subsurface. But that interpretation depends on quantifying the inputs of “recent precipitation” along flow paths that stay out of the soil/peat. The fact that the groundwater well is frozen, with water tables above the peat surface during most of the study period (Fig. 7e) would indicate that the input of precipitation/snowmelt on the mire that makes it to the stream would be a source without much DOC at all.
Figure 8 – please find a way to help the reader better compare panes (a) and (b) of the figure. Stacking one on top of the other, rather than side by side would be one easy way, even if it would take more space.
Line 480. The role of ice is mentioned, with the UAS snow depth analysis interpreting surficial ice as snow. Please say more about “formation of ice in low lying areas”. My experience is that the upper centimetres of mires are often frozen during the winter, whereas forested upland areas often do not have much ice in the soil profile. Is the entire fen frozen to a certain depth (some centimetres), or is it a crust of ice on the snow and or soil surface that is referred to here. This is good to clarify since the depth of soil frost has a bearing on the interpretation of flow paths and where the DOC sources are that keep DOC from diluting more as snow melts on the fen, creating large inputs of low-DOC water close to the flume.
Speaking of the flume, does it capture all the water running off from the upslope areas? With water tables above the ground surface on a relatively flat landscape, there might be a possibility of water bypassing the flume.
Citation: https://doi.org/10.5194/egusphere-2025-4682-RC2 - AC2: 'Reply on RC2', Petra Korhonen, 23 Jan 2026
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,567 | 778 | 172 | 2,517 | 307 | 115 | 116 |
- HTML: 1,567
- PDF: 778
- XML: 172
- Total: 2,517
- Supplement: 307
- BibTeX: 115
- EndNote: 116
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Pertti Ala-Aho
Bjørn Kløve
Hannu Marttila
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3730 KB) - Metadata XML
-
Supplement
(1029 KB) - BibTeX
- EndNote
- Final revised paper
General comments
This is an interesting setup touching a weak spot in hydrology and solute mobilization: While connectivity is used to explain solute export dynamics, it is hard to actually map and measure. This study can go a step in this direction by a drone-based assessment of snow cover changes during a major snowmelt event combined with concentration dynamics in the receiving stream. This topic is of great interest to readers of HESS.
While the data and effort are impressive, I am not totally convinced about the way results are described and interpreted. From my point of view the result section can be more concise and the discussion section much more integrative. I see a tendency to overinterpret patterns. My suggestion is a careful revision following the points raised below. I moreover encourage the authors to look for a measure of connectivity in the snow cover data that can be used as an explanatory variable in the event C-Q analysis. Just fraction of area covered by snow or snow depths is not telling a story of connectivity of source zones to the stream. However, measures of spatial connectivity of snow-free area changing over time may provide that.
Specific comments
Title
For me UAS is not self-explaining and I would even avoid DOC as an abbreviation in the title.
Abstract
L9: I miss a reference to the location and size of the catchment/area to help the reader understanding what dimension you are talking about.
Introduction
The introduction reads very well – not much to revise from my point of view. However, I found the isotope analysis as not at all motivated in the introduction. You should mention at some point here and / or in the methods what this analysis is used for.
L56: “water table” or better “water levels” is enough.
L60: These studies use the TWI at catchment scale or in riparian zones but not in peatlands. Are there also examples from peatlands pointing a dominant topographic control of discharge generation?
Methods
L109: To what is the abbreviation “FMI” pointing to?
L115: Try to be more precise here. Is this the pH in the surface water or in the soil water?
Fig. 1: I find it rather unusual to use a copy of a topographic map as a background. All shown features are useless unless explained in a legend. But are all shown features necessary to know? Isolines lack numbers! Potentially use the right map to show the location of the research station/ precipitation sampler.
L177: Explain GCPs.
Fig. S2: This relationship looks rather weak. Can you report on the R2 and bias and have you used the confidence interval later on as shown (but not described) here?
Results
I find the description of snowmelt with 5 figures and 2 tables as too extensive. Information can be transferred in a more condensed way focusing on the information that is really needed in the subsequent analysis.
L291: What caused the rapid snowmelt? Temperature? Rainfall on snow?
L292: To what statistic measure “variation” is referring to? Can you be more quantitative here?
L331: For me it is a surprise that the SWI is dynamic. The description in the method does not point to SWI being used as a dynamic surface feature. I interpreted it as a static topography feature. So, for me this is hard to understand. Do you describe snow melt in different classes of the snow-free topography or temporally dynamic SWI of the snow surface?
L334f: Can you state if difference were statistically significant?
Table 3: For me it would be a better option to show the content of that table in Fig. 6 as a third panel.
Chapter 3.2: The title implies a description of DOC only but the chapter contains much more.
Fig. 7: Use TSSeq concentrations in the axis as well. Is this mg/L as a unit or rather unitless? What is the data source for snow depths here? Is this the same data as described above (UAS)? I am not sure if there is a reason to display load of DOC and TSS concentrations in the same plot. Same for WTD and water temperature.
L355: Consider a different wording as “followed by” implies that first discharge increased and then air temperature and rainfall increased (that actually triggered the discharge?).
L362-374: Concentration of DOC hardly change over time so that the discharge dynamics are the overwhelmingly dominant driver of the load, right? This stark differences in the variation of both could be mentioned here.
L398: This statement puzzles me as the relative position of DOC and Q in the plot (Fig. 8) is matter of the scaling of the two different Y-axes.
Fig. 8: Why is cumulative Q and SWE loss with the same unit referring to different Y-axes. This should be on the same axis. I have troubles understanding the SWE loss in the figure. Majority happens at the 9th May – I see that this is due to a gap in the data. However, the way to display this is not helpful. Consider to leave out the vertical line starting from 0.
L433-435: Some of the information are redundant here as anticlockwise and HI<0 is the same thing.
Fig. 9: Use TSSeq on the axis.
Discussion
I have issues with the cut between chapter 4.1, 4.2 and 4.3. For me the separation is not clear but redundancies are large. Discussion circles around the same processes that are explained by different data in the different chapters. The idea of a discussion should be more integrative and less along the steps of the result section, especially when the same processes are discussed.
L462: Again, I have issues to make the link of snowmelt and SWI. A high SWI marks areas in the landscape that tends to be wetter as flow paths converge (large upstream area, low slope) while low SWI values mark areas that are steeper and have smaller upstream area. How does that come together with the snow melt? In a direct causative way? Or because both are a function of topography? Steeper hillslopes do not allow for snowpack accumulation and are more exposed to radiation… So how does that link to connectivity in the landscape?
L478f: The ice cover is a new result brought up here. For me it does not really explain why ice is forming here especially.
L503: This relative increase was not convincingly shown nor quantified in the result. So, it is a bit hard to follow that argument here.
L509f: However, consider that the dilution effect is very small with concentration hardly changing during the event. This speaks rather for a transport and not a source limitation of DOC. So, from my observation I see very mild dilution effects only and therefore nearly every flow path loaded with DOC and therefore no major changes in sources of flowpath. This is basically a chemostatic system.
L538-553: For me this discussion repeats former statement but add some TSS data. I suggest to strongly reduce redundancies and combine with the discussion above.
L555ff: This is a long statement for a rather simple fact. Nearly invariant concentrations multiplied with highly variant discharge will result in a load that is exactly the same as the discharge.
L592ff: Again, I would be careful in interpreting the mild concentration changes too much. Yes, the described processes are meaningful but I don’t think we see a fundamental change of flow paths and sources but rather slight changes. So, phrases such as “quick depletion” or “sudden depletion” are a bit too much for me.