Downstream export dominates the fate of groundwater-derived CO<sub>2</sub> in a boreal stream

Olid, Carolina; Hauptmann, Demian; Karlsson, Jan; Klaus, Marcus

doi:10.5194/egusphere-2025-5892

Preprints

https://doi.org/10.5194/egusphere-2025-5892

Preprints

06 Feb 2026

| 06 Feb 2026

Downstream export dominates the fate of groundwater-derived CO₂ in a boreal stream

Carolina Olid, Demian Hauptmann, Jan Karlsson, and Marcus Klaus

Abstract. Groundwater inflow is increasingly recognized as a major source of carbon dioxide (CO₂) to streams. Yet, its fate – whether it is emitted to the atmosphere or exported downstream – remains poorly characterized, partly due to the challenges of quantifying groundwater inflow rates at high spatial (meter) and temporal (days) resolutions. In this study, we assessed the fate of groundwater-derived CO₂ in a 400 m boreal headwater stream reach by combining fine-scale measurements of groundwater inputs, emissions and downstream export of CO₂. Spatial patterns in groundwater-derived CO₂ inputs were primarily driven by the magnitude of groundwater inflows, which were controlled by catchment characteristics, such as stream slope and localized aquifer properties. Temporally, peaks in groundwater CO₂ inputs during snowmelt were primarily driven by increased groundwater discharge rather than elevated CO₂ concentrations in the groundwater, whereas peaks during summer and early autumn were associated with rainfall events and higher CO₂ concentrations in groundwater, likely resulting from enhanced soil respiration. Overall, groundwater CO₂ inputs exceeded CO₂ emissions by up to fourfold, with 40–60 % of terrestrial CO₂ transported downstream. This indicates that a substantial portion bypasses immediate atmospheric emission and may contribute to CO₂ emission further along the stream network or be cycled through in–stream processes downstream. Our results demonstrate how and to what extent groundwater inflows contribute to the variability of CO₂ fluxes from headwater streams. These findings highlight the importance of integrative assessments of CO₂ fluxes (i.e. groundwater inputs, emissions, and downstream export), which consider both in-stream processes and catchment-scale dynamics. This is particularly important in the context of climate-driven changes in hydrology and terrestrial carbon cycling.

Received: 27 Nov 2025 – Discussion started: 06 Feb 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 1244 KB)

Supplement (475 KB)

Download & links

Carolina Olid, Demian Hauptmann, Jan Karlsson, and Marcus Klaus

Status: final response (author comments only)

RC1:
'Comment on egusphere-2025-5892', Anonymous Referee #1, 10 Mar 2026

This study quantified the influence of groundwater inflow on CO2 emissions and export from a headwater stream in northern Sweden, with a cold, humid boreal climate, over 13 sampling campaigns in 2020. For this, the authors used radon 222 as a hydrologic tracer to obtain direct estimates of groundwater flow and groundwater CO2 fluxes. In general this is a well written manuscript and could represent an important contribution to our understanding of the C cycle in fluvial systems; however, there is room for improvement. Not enough care and attention was put towards the methods section which I found confusing at times. In addition, I recommend the authors expand the sensitivity/uncertainty analysis for the groundwater CO2 inputs to the other fluxes too (downstream export and emissions to the atmosphere), this will help better understand the magnitude of the sources of error when building mass balances for CO2. Also, I’d like to see a comparison between the direct estimations of CO2 inputs via Rn222 and by mass balance of CO2. Does the budget add up? This is particularly concerning because the authors observed that their estimates of groundwater flow based on Rn222 were only a fraction of the groundwater inputs when estimated based on upstream/downstream differences in stream discharge. More detailed comments as follows:
Either in the intro or somewhere in the discussion I’d like to see some discussion on the role of CO2 vs other GHGs in the emissions of GHGs from boreal systems, just to have an idea of why the focus on CO2 and what we are missing by not measuring CH4 or N2O emissions.
METHODS
Mind the organization in the methods, make sure you have consistency in the order of CO2, and Rn222 sampling/processing for all the water sources. Be consistent in the use of the equation terms throughout the method descriptions to facilitate understanding exactly which pool or flux the authors are referring to. Consider creating a simple diagram that shows the equation terms and simple version of how they were measured.
L143—What is the reasoning behind acidifying the sample, if the interest is only in the CO2 concentration? Why not to measure the CO2 in the original sample? Is it to fix the sample (avoid microbial metabolism) I’d be a bit concerned about the kind of error introduced from the back-calculation of the original CO2 concentration from the CO2 concentration of an acidified sample and pH, simply because pH is so sensitive and prone to error. Perhaps the error is small because of the low pH…?
L168 – This is confusing. The previous paragraph explains how Rn222 is measured in the water. But this line then refers to the concentration in the air, and how the concentrations in the water are determined based on the concentrations in the air…?
L186 – Why were concentrations of Rn222 in the groundwater measured using a different technique/equipment than stream water (or air, I am not sure, see comment for L168 above)?
L189 – Do the authors mean “to account for radioactive decay” as in L172? If so make sure wording is consistent.
L189 – Why sampling soil air instead of the actual water for groundwater CO2 concentrations?
L201—Was discharge measured during every sampling campaign?
L202 – There are other more general citations for salt pulse that could be used here instead of a paper in revision.
L204-207 – This section about gas transfer should be moved below to the section on CO2 emissions. Was k measured during every sampling campaign?
L221 – Are these incubation experiments what is explained in the following paragraph? Is it soils or is it sediments? Is the diffusing radon input Fdiff mentioned below?
L222 – Did the authors mean sediment instead or soil?
L222, L232 – Why/how concentration of Rn222 in the groundwater again? How are these different from the concentration of Rn222 in the groundwater described in L186? Is one advective and one diffusive? If so, make that very clear.
L232—It is unclear to me where the Rlab and Rfield come from. If Rlab (bottle) refer to the bottle experiments described in L223? If so, it would really help to refer to the Rlab and (Rfield) terms in their corresponding sections.
L246—Ci-1, I assume concentrations in the stream water? Add “streamwater” as a qualifier.\
L257 – Equation 2?
L258 – The description of the uncertainty estimations should be in a different section, perhaps under “statistics and sensitivity analysis”
L297 – How good was this model? This is important to know, to understand the accuracy of the CO2 emissions.
STATISTICS
Why is the sensitivity analysis only conducted for Fgw? This same approach could be used for most other measurements to obtain associated errors for CO2 emissions and export.
Sound is not a well-known method, perhaps expand on the error associated with this type of measurement? This is in part why I suggest a sensitivity analysis for CO2 emissions.
RESULTS
Some of the figures could be joined to reduce the large number of figures and to better allow for comparisons of concentrations and fluxes of the different pathways. For example, consider moving Figure S3a and S3b to the main document and combine it with Figure 2, so that stream and gw concentrations are plotted side by side. Same applies to Figure S3c and S3d, combine with Figure 3 but use the same type of graph, either box plot or line, but consistent.
There is lots of inconsistencies with the number of significant digits used within the different rates. Make sure the significant digits are also consistent between text and figures too.
L329 – Constant is a strong word for such variability. I do not see anything constant in the longitudinal trends.
L419 – What drives the somewhat less variable CO2 emissions? How did gas exchange vary longitudinally and temporally?
DISCUSSION
A few things I’d like to see discussed: how does the “conservative” (L375) nature of your methodology affects your results and conclusions?; the study is based on a relatively limited number of wells, how does this affect the overall results and conclusions? (besides not capturing potentially large spatial variability); how do gas exchange rates and CO2 emissions compare to other studies/methods in similar systems and how does the uncertainty on those values affect the relative importance of downstream export vs emissions to the atmosphere?; does the budget close? Or could the authors compare their gw input results with values obtained from the difference between upstream inputs and (downstream export+emissions)?
L496—It is interesting that there is less variability in gw CO2 concentrations than in gw fluxes. I’d imagine the hydrologic patterns would have some effect on the biological factors controlling CO2 concentrations…?
L557—focus on published work that supports your findings.
L572—Residence time?

Citation: https://doi.org/10.5194/egusphere-2025-5892-RC1
- AC1: 'Reply on RC1', Carolina Olid, 11 May 2026
  
  Referee’s comment 1 (R1C1): This study quantified the influence of groundwater inflow on CO2 emissions and export from a headwater stream in northern Sweden, with a cold, humid boreal climate, over 13 sampling campaigns in 2020. For this, the authors used radon 222 as a hydrologic tracer to obtain direct estimates of groundwater flow and groundwater CO2 fluxes. In general this is a well written manuscript and could represent an important contribution to our understanding of the C cycle in fluvial systems; however, there is room for improvement. Not enough care and attention was put towards the methods section which I found confusing at times. In addition, I recommend the authors expand the sensitivity/uncertainty analysis for the groundwater CO2 inputs to the other fluxes too (downstream export and emissions to the atmosphere), this will help better understand the magnitude of the sources of error when building mass balances for CO2. Also, I’d like to see a comparison between the direct estimations of CO2 inputs via Rn222 and by mass balance of CO2. Does the budget add up? This is particularly concerning because the authors observed that their estimates of groundwater flow based on Rn222 were only a fraction of the groundwater inputs when estimated based on upstream/downstream differences in stream discharge.
  Author’s comment (AC_R1C1): We thank the referee for their thorough and constructive comments. We will organize the Methods section to improve clarity, introducing new subsections that more clearly describe the field measurements, laboratory analyses, and modelling approaches related to the ²²²Rn mass balance. We will also expand the uncertainty analyses to include not only groundwater CO₂ inputs, but also downstream export and emissions to the atmosphere. This will allow for a more comprehensive assessment of the main sources of uncertainty affecting CO₂ flux estimates.
  Regarding the suggestion to compare direct measurements of CO₂ inputs based on ²²²Rn with a full CO₂ mass balance, we acknowledge the value of such an approach. However, we did not quantify all relevant CO₂sources and sinks within the stream (e.g. in-stream metabolism) and therefore cannot construct a complete CO₂ mass balance. Importantly, the primary objective of this study is not to close the CO₂ budget, but rather to assess the relative importance of groundwater-derived CO₂ inputs compared to other CO₂ fluxes, and to highlight their high spatial and temporal variability. Capturing this level of variability at high temporal and spatial resolution is essential for improving and constraining future carbon mass balance approaches.
  We recognize that previous studies have successfully combined ²²²Rn with carbon mass balance frameworks and/or modelling approaches to quantify groundwater-derived CO₂ inputs and their fate (e.g.,(Duvert et al., 2019; Larsen et al., 2025; Wang et al., 2021)). However, these studies are typically based on short-term campaigns or rely on model parameterizations and therefore do not fully resolve the temporal variability of groundwater-derived CO₂ inputs. Our repeated sampling highlights that this variability is substantial and likely represents a major, yet still poorly constrained, component of stream carbon dynamics. We also acknowledge that the title and parts of the introduction may have suggested that we aimed to construct a complete CO₂ budget, which may have led to this misunderstanding. We will therefore revise both the title and the introduction to better clarify the objectives of the study.
  We also agree with the reviewer that groundwater inflow rates derived from ²²²Rn were at the lower end of the range obtained from discharge-based calculations. As noted in the manuscript, our approach was conservative, as it assumes no ²²²Rn losses other than those due to atmospheric evasion and downstream export. Additional losses of ²²²Rn may occur through water loss from the stream to the subsurface (i.e., losing reaches), as well as through hyporheic exchange processes. These processes are not explicitly accounted for in our model, but if they were, it would lead to higher inferred groundwater inflows and bring these estimates closer to those derived from discharge data. In the revised manuscript, we will clarify this point and emphasize that our estimates likely represent a lower bound, with higher fluxes expected in stream sections where water losses may occur.
  ------------------------------------------------------
  R1C2: Either in the intro or somewhere in the discussion I’d like to see some discussion on the role of CO2 vs other GHGs in the emissions of GHGs from boreal systems, just to have an idea of why the focus on CO2 and what we are missing by not measuring CH4 or N2O emissions.
  AC_R1C2: We thank the reviewer for this suggestion. We agree that considering other GHGs, such as CH₄ and N₂O, is important for a comprehensive assessment of GHG emissions from boreal stream systems. In our study, we focused on CO₂ because it typically represents the dominant component of GHG emissions in boreal headwater streams (Pilla et al., 2022; Song et al., 2024; Soued et al., 2016; Wallin et al., 2018).
  CH₄ concentrations were measured but were consistently very low, in many cases approaching or below the detection limit of our method. This reflects the low concentrations of CH₄in the studied system, which is consistent with previous observations in similar boreal environments. Given these low concentrations and the associated uncertainty, we did not include CH₄in our analysis. We will add a paragraph in the Introduction to better contextualize the role of CO₂ relative to other GHGs, and to acknowledge that our estimates do not capture the full GHG budget.
  ------------------------------------------------------
  R1C3: Mind the organization in the methods, make sure you have consistency in the order of CO2, and Rn222 sampling/processing for all the water sources. Be consistent in the use of the equation terms throughout the method descriptions to facilitate understanding exactly which pool or flux the authors are referring to. Consider creating a simple diagram that shows the equation terms and simple version of how they were measured.
  AC_R1C3: We thank the reviewer for this helpful suggestion. We will reorganize the Methods section to ensure a consistent structure in the presentation of CO₂and ²²²Rn sampling and processing across all water sources (stream water and groundwater). We will also revise the description of equation terms to ensure consistent terminology and to introduce them in a clear and logical order, facilitating the identification of each pool and flux throughout the text.
  We agree that a schematic diagram could help clarify the overall methodological framework. We will therefore consider including such a figure in the revised manuscript, although its inclusion will depend on the final number of figures and overall manuscript structure.
  ------------------------------------------------------
  R1C4: L143—What is the reasoning behind acidifying the sample, if the interest is only in the CO2 concentration? Why not to measure the CO2 in the original sample? Is it to fix the sample (avoid microbial metabolism) I’d be a bit concerned about the kind of error introduced from the back-calculation of the original CO2 concentration from the CO2 concentration of an acidified sample and pH, simply because pH is so sensitive and prone to error. Perhaps the error is small because of the low pH…?
  AC_R1C4: We thank the reviewer for this important comment. Samples were acidified to convert all dissolved inorganic carbon (DIC) species (HCO₃⁻ and CO₃²⁻) into CO₂, allowing for consistent quantification of total DIC via headspace equilibration. This approach effectively stops biological activity during storage and handling and minimises alterations in sample composition. It also avoids uncertainties related to CO₂ speciation, particularly in higher pH systems where HCO₃⁻and CO₃²⁻ represent a significant fraction of the DIC pool.
  We agree that back-calculating in situ CO₂ concentrations from acidified samples and pH measurements introduces some uncertainty, particularly given the sensitivity of pH. However, in our case, this uncertainty is expected to be relatively small. The studied system is characterized by low alkalinity and generally low pH conditions, where CO₂ dominates the DIC pool. Here, on average, CO₂made up 98.2 ± 0.7% of the DIC pool (Hauptmann et al., 2026). Under these conditions, the conversion from DIC to CO₂ is less sensitive to pH errors compared to higher alkalinity systems. Furthermore, any potential errors associated with this approach are expected to be consistent across sites and sampling campaigns, allowing for a robust comparison of spatial and temporal patterns.
  In addition, CO₂concentrations derived from grab samples were consistent with continuous in situ measurements obtained using a Vaisala CO₂sensor, as shown in (Hauptmann et al., 2026). Specifically, the distribution of CO₂ evasion estimates from grab samples did not differ significantly from those derived from continuous logger data, indicating that the grab sampling approach is representative of seasonal conditions. We will include this point in the revised manuscript to further support the robustness of our CO₂ estimates.
  ------------------------------------------------------
  R1C5: L168 – This is confusing. The previous paragraph explains how Rn222 is measured in the water. But this line then refers to the concentration in the air, and how the concentrations in the water are determined based on the concentrations in the air…?
  AC_R1C5: We thank the reviewer for pointing this out. We agree that the original description was unclear and will revise the text for clarity. In our setup, ²²²Rn is first transferred from the water sample to an air loop using a closed-circuit bubbling system connected to a RAD7 detector. The RAD7 directly measures the ²²²Rn concentration in the air phase. The corresponding ²²²Rn concentration in the water sample is then calculated from the measured air concentration, taking into account the volumes of air and water, temperature, and partitioning between phases. We will revise the manuscript to clearly distinguish between the measurement of ²²²Rn in the air phase and the subsequent calculation of its concentration in water.
  ------------------------------------------------------
  R1C6: L186 – Why were concentrations of Rn222 in the groundwater measured using a different technique/equipment than stream water (or air, I am not sure, see comment for L168 above)?
  AC_R1C6: We thank the reviewer for this comment. In this study, ²²²Rn concentrations were measured using two standard techniques: RAD7 and liquid scintillation counting (LSC), which were selected based on sample type and practical constraints. RAD7 requires relatively large water volumes to achieve low detection limits and was therefore used for stream water, where sufficient sample volume could be collected. In contrast, collecting large, gas-tight groundwater samples is often challenging. For this reason, LSC was used for groundwater samples, as it requires much smaller sample volumes and is well-suited for waters with high ²²²Rn concentrations. We will clarify this rationale in the revised manuscript.
  ------------------------------------------------------
  R1C7: L189 – Do the authors mean “to account for radioactive decay” as in L172? If so make sure wording is consistent.
  AC_R1C7: Yes, we refer to correcting ²²²Rn concentrations for radioactive decay to the time of measurement. We will revise the wording to ensure consistency throughout the manuscript.
  ------------------------------------------------------
  R1C8: L189 – Why sampling soil air instead of the actual water for groundwater CO2 concentrations?
  AC_R1C8: Measuring CO₂ concentrations in the same groundwater samples as those analysed for ²²²Rn, would have been favourable. However, groundwater volumes that could be extracted across the wells were limited and had to suffice for a range of analyses, including analyses not included in this paper, but needed for other parts of a wider research project (e.g. dissolved organic carbon and nitrogen). To assure consistent methodology across sites and sampling occasions that provides enough sampling material even at low groundwater tables, we installed soil gas probes at the same locations as the groundwater wells so that groundwater can be prioritized for ²²²Rn, and other parameters that rely on liquid samples, and soil gas probes can be used for CO₂ analysis. The soil gas probes were always below the groundwater table and formed a headspace in equilibrium with the surrounding groundwater. We are confident that air samples from these probes hence represent the same groundwater conditions as the ²²²Rn samples.
  ------------------------------------------------------
  R1C9: L201—Was discharge measured during every sampling campaign?
  AC_R1C9: Discharge was manually measured a total of 23 times between 2020 and 2022 using salt slugs (Hauptmann et al., 2026). With these measurements, we established a rating curve that allowed us to calculate continuous discharge estimates for the occasions where we did not manually estimate discharge. Between each sampling station with these continuous records, we assumed a linear increase in discharge. This point will be clarified in the revised version of the manuscript.
  ------------------------------------------------------
  R1C10: L202 – There are other more general citations for salt pulse could be used here instead of a paper in revision.
  AC_R1C10: We agree that more general references could be used to describe the salt slug method for discharge estimation. The study previously cited as “in revision” is now published, and we will update the reference accordingly and ensure that appropriate general references are also included (e.g. (Moore, 2005)).
  ------------------------------------------------------
  R1C11: L204-207 – This section about gas transfer should be moved below to the section on CO2 emissions. Was k measured during every sampling campaign?
  AC_R1C11: As suggested by the reviewer, we will move this section right below the section on CO₂emissions to improve the clarity of the methods. We will also clarify how k was estimated and that we have manual k measurements in 10 occasions. Similar to discharge, we used assumptions of linear increase obtained using an ordinary least squares regression between discharge and k₆₀₀ for the sampling occasions where we did not record ambient sound.
  ------------------------------------------------------
  R1C12: L221 – Are these incubation experiments what is explained in the following paragraph? Is it soils or is it sediments? Is the diffusing radon input Fdiff mentioned below?
  AC_R1C12: We thank the reviewer for this comment. Yes, the incubation experiments referred to those described in the following paragraph. These experiments were conducted using streambed sediments (not soils). Sediment incubation experiments were performed in the laboratory to quantify both the diffusive ²²²Rn flux from sediments (F_diff) and the ²²²Rn concentration in groundwater, following (Chanyotha et al., 2014). The initial phase of the incubation (approximately the first 14 hours) was used to estimate the diffusive flux (F_diff), while the later stages (after >100 hours) were used to determine the equilibrium ²²²Rn concentration in groundwater. We will revise the manuscript to clarify these points and explicitly link the incubation experiments to the estimation of these parameters.
  ------------------------------------------------------
  R1C13: L222 – Did the authors mean sediment instead or soil?
  AC_R1C13: Yes, we meant 'sediment' and not 'soil'. We will amend the sentence accordingly.
  ------------------------------------------------------
  R1C14: L222, L232 – Why/how concentration of Rn222 in the groundwater again? How are these different from the concentration of Rn222 in the groundwater described in L186? Is one advective and one diffusive? If so, make that very clear.
  AC_R1C14: ²²²Rn concentrations in groundwater were first estimated from water samples collected in wells in the field. However, this approach is subject to uncertainties, particularly due to the need to minimize gas exchange with the atmosphere during sampling. To address this, we also performed laboratory incubation experiments under controlled conditions (see response to AC12). These experiments provide an independent estimate of ²²²Rn concentrations in groundwater, helping to reduce uncertainties associated with field sampling. We will clarify the rationale for using both methods in the revised manuscript.
  ------------------------------------------------------
  R1C15: L232—It is unclear to me where the Rlab and Rfield come from. If Rlab (bottle) refer to the bottle experiments described in L223? If so, it would really help to refer to the Rlab and (Rfield) terms in their corresponding sections.
  AC_R1C15: We thank the reviewer for bringing this to our attention. We agree that the origin of the R_lab and R_fieldterms was not sufficiently clear. R_lab and R_field refer to ratios of volume of water to sediment in the incubation chamber (lab) and in the field (which is a function of the porosity), respectively. We used the notation used in (Stieglitz et al., 2013), to which we refer for the method. We will revise the notation and define both terms explicitly in their respective sections. We will also ensure consistent use of the terms throughout the manuscript to improve clarity.
  ------------------------------------------------------
  R1C16: L246—Ci-1, I assume concentrations in the stream water? Add “streamwater” as a qualifier.\
  AC_R1C16: We will add “streamwater” to clarify the text.
  ------------------------------------------------------
  R1C17: L257 – Equation 2?
  AC_R1C17: Yes, we will modify the text and cite here Equation 2 instead.
  ------------------------------------------------------
  R1C18: L258 – The description of the uncertainty estimations should be in a different section, perhaps under “statistics and sensitivity analysis”
  AC_R1C18: We agree with the reviewer. This description will be included under the caption 'Statistics and sensitivity analyses' at the end of the 'Methods' section.
  ------------------------------------------------------
  R1C19: L297 – How good was this model? This is important to know, to understand the accuracy of the CO₂ emissions.
  AC_R1C19: We thank the reviewer for this comment. The k₆₀₀ estimates are based on the sound spectral analysis approach described in (Klaus et al., 2019), which links stream sound spectral properties to gas exchange velocity, reflecting their shared physical basis in bubble entrainment and emission. This method has been validated using independent CO₂ flux chamber measurements under both controlled lab and field conditions similar to our study stream. In our study, continuous k₆₀₀ values were derived from site-specific relationships between discharge (Q) and k₆₀₀, following (Hauptmann et al., 2026) (now published), where these relationships were established based on concurrent measurements and showed strong agreement. To address the reviewer’s concern, we will include additional information in the revised manuscript describing the performance of these relationships and discussing the associated uncertainties, to better constrain the accuracy of the CO₂ emission estimates.
  ------------------------------------------------------
  R1C20: Why is the sensitivity analysis only conducted for Fgw? This same approach could be used for most other measurements to obtain associated errors for CO2 emissions and export.
  AC_R1C20: We thank the reviewer for this valuable suggestion. We agree that uncertainty and sensitivity analyses are important to assess the robustness of CO₂ emission and export estimates. A detailed sensitivity analysis of these fluxes has been previously addressed in (Hauptmann et al., 2026), where the main sources of uncertainty associated with CO₂ emissions and downstream export are discussed in detail. Rather than duplicating this analysis, we will expand the current manuscript to better describe the key parameters controlling these fluxes (e.g., k₆₀₀, discharge, and CO₂ concentrations) and discuss their associated uncertainties. This will allow us to provide a clearer assessment of the robustness of our estimates while keeping the manuscript focused on the main objectives of the study.
  ------------------------------------------------------
  R1C21: Sound is not a well-known method, perhaps expand on the error associated with this type of measurement? This is in part why I suggest a sensitivity analysis for CO2 emissions.
  AC_R1C21: As mentioned in AC_R1C19, we will include additional information in the revised manuscript describing the performance of the sound method to estimate k₆₀₀ and discussing the associated uncertainties. This will help to better constrain the accuracy of the CO₂ emissions.
  ------------------------------------------------------
  R1C22: Some of the figures could be joined to reduce the large number of figures and to better allow for comparisons of concentrations and fluxes of the different pathways. For example, consider moving Figure S3a and S3b to the main document and combine it with Figure 2, so that stream and gw concentrations are plotted side by side. Same applies to Figure S3c and S3d, combine with Figure 3 but use the same type of graph, either box plot or line, but consistent.
  AC_R1C22: We thank the reviewer for this helpful suggestion. We agree that combining figures will improve clarity and facilitate comparison between groundwater and stream water concentrations. Following this suggestion, we will merge Figures S3a and S3b with Figure 2, and Figures S3c and S3d with Figure 3, to present groundwater and stream data side by side. We will also ensure consistency in the type of graphical representation used (e.g., box plots or line plots) across figures. These changes will reduce the number of figures and improve the overall readability of the manuscript.
  ------------------------------------------------------
  R1C23: There is lots of inconsistencies with the number of significant digits used within the different rates. Make sure the significant digits are also consistent between text and figures too.
  AC_R1C23: We thank the reviewer for pointing this out. We will revise the manuscript to ensure consistent use of significant digits across all reported values, both in the text and in the figures.
  ------------------------------------------------------
  R1C24: L329 – Constant is a strong word for such variability. I do not see anything constant in the longitudinal trends.
  AC_R1C24: We agree that the term “constant” is too strong given the observed variability. We will revise the text to use more appropriate wording (e.g., “relatively stable” or “within a narrow range”) to better reflect the longitudinal trends.
  ------------------------------------------------------
  R1C25: L419 – What drives the somewhat less variable CO2 emissions? How did gas exchange vary longitudinally and temporally?
  AC_R1C25: We thank the reviewer for this insightful comment. We agree that understanding the drivers of the relatively low variability in CO₂emissions is important. CO₂emissions are primarily controlled by gas exchange velocity (k₆₀₀), which is in turn driven by turbulence and bubble-mediated exchange processes (Klaus et al., 2019). In low-slope streams such as the one studied here, recent work has shown that k₆₀₀ does not necessarily increase strongly with discharge but is instead largely controlled by stream morphology (Aho et al., 2025). This likely contributes to the relatively low temporal variability observed in CO₂ emissions in our study.
  The variability in CO₂ emissions and its controls have been examined in detail in (Hauptmann et al., 2026), based on the same study system. Therefore, in the revised manuscript, we will include a summary of these controls, refer to Hauptmann et al. for a more comprehensive discussion, and explicitly link these processes to the temporal and spatial patterns observed in our dataset.
  ------------------------------------------------------
  R1C26: A few things I’d like to see discussed: how does the “conservative” (L375) nature of your methodology affects your results and conclusions?; the study is based on a relatively limited number of wells, how does this affect the overall results and conclusions? (besides not capturing potentially large spatial variability); how do gas exchange rates and CO2 emissions compare to other studies/methods in similar systems and how does the uncertainty on those values affect the relative importance of downstream export vs emissions to the atmosphere?; does the budget close? Or could the authors compare their gw input results with values obtained from the difference between upstream inputs and (downstream export+emissions)?
  AC_R1C26: We thank the reviewer for this comprehensive and insightful comment. We agree that these aspects are important to better contextualize our results and their implications.
  Regarding the conservative nature of our approach, as discussed in the manuscript, we assume no additional ²²²Rn losses beyond atmospheric evasion and downstream export. This likely leads to an underestimation of groundwater inflow rates and associated CO₂ inputs (see response to R1C1). We will further clarify in the Discussion that our estimates should therefore be interpreted as conservative, lower-bound values. We will also include a more detailed uncertainty analysis to assess how these assumptions may influence our estimates of groundwater inflow and associated CO₂ fluxes.
  We also acknowledge that the relatively limited number of groundwater wells may not fully capture the spatial heterogeneity of groundwater CO₂ concentrations. However, the wells were strategically placed to cover the expected range of wetter and drier sections of the riparian zone, as riparian wetness is a key control on CO₂ dynamics in boreal headwater streams (Leith et al., 2015). We will clarify this sampling strategy in the revised manuscript and discuss its implications for representing groundwater endmembers.
  With respect to gas exchange rates and CO₂ emissions, we will include additional discussion comparing our estimates with previous studies in similar boreal systems and highlight how turbulence and bubble-mediated processes control gas exchange dynamics (Klaus et al., 2019). We will also clarify that our sound spectral properties and derived k₆₀₀ values fall within the range of the calibration dataset, supporting the robustness of our estimates. In addition, we will place our k₆₀₀ and CO₂ evasion estimates in a broader context by comparing them with global observations (Rocher-Ros et al., 2019). We will also discuss how uncertainties in k₆₀₀ may influence the relative importance of groundwater discharge versus atmospheric emission.
  Finally, we recognize the interest in evaluating whether the CO₂ budget closes. As mentioned above (see R1C1), we did not quantify all relevant sources and sinks (e.g., in-stream metabolism), and therefore our results should not be interpreted as a closed CO₂ mass balance. Instead, we will clarify that our approach focuses on constraining groundwater-derived CO₂inputs and their variability, which represents a key uncertainty in carbon budgets.
  ------------------------------------------------------
  R1C27: L496—Interestingly, there is less variability in gw CO2 concentrations than in gw fluxes. I’d imagine the hydrologic patterns would have some effect on the biological factors controlling CO2 concentrations…?
  AC_R1C27: We thank the reviewer for this interesting comment. We agree that hydrological patterns may influence the biological processes controlling groundwater CO₂ concentrations, for example through variations in residence time, connectivity, or organic matter inputs. While we did not directly assess these mechanisms in this study, our results showing lower variability in groundwater CO₂ concentrations compared to fluxes suggest that hydrological controls may indeed play a more important role. In addition, CO₂ concentrations likely reflect the complex interplay between production and consumption processes. It is therefore possible that hydrological variability affects both inputs and outputs similarly, resulting in relatively limited variation in net CO₂ concentrations. We will expand the Discussion to acknowledge these potential links and highlight them as an area for future research.
  ------------------------------------------------------
  R1C28: L557—focus on published work that supports your findings.
  AC_R1C28: The study previously cited is now published, and we will update the reference accordingly. We will also ensure that all supporting statements are backed by appropriate published literature in the revised manuscript.
  ------------------------------------------------------
  R1C29: L572—Residence time?
  AC_R1C29: We agree that “residence time” is a more appropriate term in this context. We will revise the text accordingly.
  ------------------------------------------------------
  (The reference list can be found at the end of the response to R2)
  
  Citation: https://doi.org/10.5194/egusphere-2025-5892-AC1
RC2:
'Comment on egusphere-2025-5892', Anonymous Referee #2, 02 Apr 2026

General comments
The manuscript by Olid et al. addresses an important question regarding the role of groundwater inputs in delivering CO2 to headwater streams, and the fate of this CO2. The use of 222Rn as a tracer of groundwater inflows is a clear strength of the study. However, I have several concerns that need to be addressed before the paper can be considered for publication in HESS.
First, I have concerns regarding the conceptual framework. The authors quantify one input (groundwater CO2) and two outputs (emission and downstream export), but it is unclear whether these terms capture all sources and sinks of CO2 in their system. Without showing that the CO2 budget closes, interpreting the fraction of downstream CO2 export attributed to groundwater inputs (as in Fig 7) is potentially misleading. In-stream metabolism could represent another significant CO2 source – would it be possible to estimate the contribution of in-stream metabolism from the dataset? For example, could any mismatch between the input and the sum of the two outputs be attributed to metabolism? If not, how do the authors interpret any mismatch between inputs and outputs? I feel like the dataset has some potential, but it needs to be examined more thoroughly.
Second, the paper lacks clarity in the Methods section. Several aspects are confusing and require more justification. For example, multiple approaches are presented for obtaining the same variable (e.g. 222Rn in groundwater), but the rationale for using multiple methods is not explained.
Third, the literature review is somewhat outdated and incomplete, which gives a false impression of novelty in the introduction. In particular, there is an increasing number of studies that combined the use of 222Rn and CO2 to trace groundwater contributions, which have been overlooked. A more thorough review of this literature would improve the impact and relevance of this ms.
Lastly, the paper appears to have a high degree of overlap with a companion study cited as ‘in revision’, which I assume is now published (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024JG008671). That paper addresses similar research questions at the same site using comparable methods, yet it is not even mentioned in the introduction. The authors need to clarify the added value and novelty of their study, how the two papers complement each other, and the extent of overlap between the two.

Specific comments
Introduction
L71-75. While this statement is broadly accurate, the authors have overlooked several relevant studies. Just a few examples below:
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019JG005047
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2022JG006954
https://www.sciencedirect.com/science/article/pii/S0043135424015835
L87-89. Once again this statement does not reflect the recent literature on the combined use of 222Rn and CO2 to trace groundwater CO2 inputs. I am not simply suggesting that the authors cite the below studies – instead I would invite them to engage with these studies and identify any remaining knowledge gaps. I believe doing so would increase the impact (and relevance) of this work.
https://www.sciencedirect.com/science/article/pii/S0048969721012985
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018JG004912
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024JG008592
L93-99. It might be useful to rephrase the hypotheses for clarity – for example adding one hypothesis specifically on the partitioning between emission vs downstream export.

Methods
Figure 1. “Experimental” (typo). Groundwater wells G1, G3, G5, and G6 do not appear on the figure. Need to remove blue triangles in a) (unless they relate to something relevant to this study).
L110-126. Please specify catchment area.
L143-159. It seems like DIC was measured and then CO2 was inferred based on DIC and pH? Why not measuring CO2 directly instead? Please clarify.
L176. I do not think that such shallow piezometers (max depth 1.3 m) can be considered to intersect “regional aquifers”. They likely only capture the very near-surface portion of the saturated zone (unless the site is located within a regional groundwater discharge area).
L185-189. Unclear why the groundwater samples were not analysed on the RAD7, like for the surface water samples.
L189-194. This paragraph is also confusing – why was the same method not applied to CO2 measurements in groundwater samples as was used for streamwater samples?
L210-213. Why was pH not measured in the field? Your indirect CO2 estimates require accurate pH measurements, so this procedure needs to be better justified.
L222. It is unclear why 222Rn in groundwater is measured here, as this was already described earlier. If the intention is to compare methods, please state this explicitly.
L243. Add reference(s) before the equation, as this mass balance is derived from previous work.
L299-301. Multiplying a discharge by a concentration should give absolute fluxes in g/d, not area-normalised fluxes in g/m2/d. Please explain how you obtained area-normalised fluxes.

Results
L366. Figure 3. It would be very helpful to show the discharge series here as well for context.
L390-391. Please mention if the relationship is positive or negative.
L393-398. CV calculated on what?
L423-429. For downstream fluxes, how were values normalised, i.e. with which surface area? It makes little sense to me to normalise a dissolved CO2 flux by stream surface area. While I understand that normalisation can help compare fluxes across sites, I would recommend presenting absolute downstream CO2 fluxes and comparing these with absolute CO2 emission fluxes and absolute groundwater CO2 inputs. At the very least, this could be done for the catchment outlet (e.g. site S18).
L437. Remove “being”
L444. Clarify what is meant by “stream CO2 export”, as there are two possible export pathways.
L444-449. I struggle to understand the usefulness of presenting the contribution of groundwater CO2 inputs to downstream CO2 export. This ratio does not account for concurrent CO2 emission and therefore may be difficult to interpret in terms of actual “contribution”. Why not present the contribution of groundwater inputs to emissions as well? Alternatively, expressing groundwater inputs relative to total CO2 outputs (i.e. emission + downstream export) would be more consistent with a mass balance perspective.

Discussion
L456. This subsection provides very limited new insight, as spatial and temporal variability in groundwater inflows to streams is already well established. I suggest reducing it significantly.
L470-478. I would be more careful with this interpretation – as per Fig S4, there are several sites with high slope and no groundwater inflow. Also, an incised valley (i.e. a high hillslope gradient) and a high channel slope are not equivalent.
L517-520. References are needed to support this statement.
L529-537. This section is not supported by references, is very generic, and adds limited value to the discussion. Please consider improving or removing.
L556-557. This statement and reference are confusing – isn’t the finding that a substantial portion of groundwater CO2 inputs is transported downstream a main finding of the present study? Or is it taken from the Hauptmann study?

Conclusions
L565-584. This section is very unspecific and adds little to the paper. I would suggest removing it.

Citation: https://doi.org/10.5194/egusphere-2025-5892-RC2
- AC2: 'Reply on RC2', Carolina Olid, 11 May 2026
  
  General comments
  ------------------------------------------------------
  R2C1: The manuscript by Olid et al. addresses an important question regarding the role of groundwater inputs in delivering CO2 to headwater streams, and the fate of this CO2. The use of 222Rn as a tracer of groundwater inflows is a clear strength of the study. However, I have several concerns that need to be addressed before the paper can be considered for publication in HESS.
  First, I have concerns regarding the conceptual framework. The authors quantify one input (groundwater CO2) and two outputs (emission and downstream export), but it is unclear whether these terms capture all sources and sinks of CO2 in their system. Without showing that the CO2 budget closes, interpreting the fraction of downstream CO2 export attributed to groundwater inputs (as in Fig 7) is potentially misleading. In-stream metabolism could represent another significant CO2 source – would it be possible to estimate the contribution of in-stream metabolism from the dataset? For example, could any mismatch between the input and the sum of the two outputs be attributed to metabolism? If not, how do the authors interpret any mismatch between inputs and outputs? I feel like the dataset has some potential, but it needs to be examined more thoroughly.
  AC_R2C1: We thank the reviewer for this thoughtful and important comment. We agree that constraining all relevant sources and sinks of CO₂ is essential for a complete mass balance perspective.
  In this study, we quantified groundwater-derived CO₂ inputs, as well as two major outputs pathways (atmospheric evasion and downstream export). However, we acknowledge that additional processes, such as stream metabolism, may also contribute to CO₂ dynamics and could potentially explain any mismatch between inputs and outputs. Due to the lack data (e.g. dissolved oxygen time series), we are not able to robustly quantify in-stream metabolism from our dataset and therefore cannot explicitly attribute any mismatch to metabolic processes. Instead, we will interpret potential discrepancies as the result of a combination of factors, including uncertainties in groundwater CO₂ end-member, gas exchange estimates (k6₀₀), and unquantified in-stream processes.
  We also agree that expressing groundwater inputs as a fraction of downstream export may be misleading in the absence of a closed mass balance. To address this, we will revise the manuscript to avoid presenting proportional contribution and instead directly compare the magnitude of groundwater CO₂ inputs, downstream export, and atmospheric emissions.
  Finally, we will revise the Introduction and expand the Discussion to better acknowledge these limitations and clarify that the aim of this study is not to close the CO₂ budget, but to constrain the magnitude and variability of groundwater-derived CO₂ inputs.
  ------------------------------------------------------
  R2C2: Second, the paper lacks clarity in the Methods section. Several aspects are confusing and require more justification. For example, multiple approaches are presented for obtaining the same variable (e.g. ²²²Rn in groundwater), but the rationale for using multiple methods is not explained.
  AC_R2C2: We thank the reviewer for this comment. We agree that the rationale for using multiple approaches to estimate certain variables (e.g., ²²²Rn concentration in groundwater) was not sufficiently clear in the original manuscript. Different methods were used to account for practical constraints and to better constrain key parameters (see response to R1C6). For example, direct field measurements of ²²²Rn in groundwater are subject to uncertainties related to sampling and gas exchange, whereas laboratory-based incubation experiments provide more controlled conditions and independent estimates. The use of complementary approaches therefore, helps reduce uncertainty and better capture variability. In the revised manuscript, we will clarify the rationale for using multiple methods and explicitly describe how each approach contributes to the overall estimation. As also noted in our response to Referee #1, we will revise and organize the Methods section to improve clarity.
  ------------------------------------------------------
  R2C3: Third, the literature review is somewhat outdated and incomplete, which gives a false impression of novelty in the introduction. In particular, there is an increasing number of studies that combined the use of 222Rn and CO2 to trace groundwater contributions, which have been overlooked. A more thorough review of this literature would improve the impact and relevance of this ms.
  AC_R2C3: We agree that the literature review in the original manuscript did not fully reflect recent advances in the combined use of ²²²Rn and CO₂ to trace groundwater contributions. In the revised manuscript, we will expand the introduction to include relevant recent studies (e.g. (Duvert et al., 2019; Larsen et al., 2025; Wang et al., 2021) and better position our work within this growing body of literature. Rather than only adding citations, we will more explicitly discuss how previous studies have applied combined ²²²Rn–CO₂ approaches and identify the remaining knowledge gaps. This will allow us to more clearly define the novelty of our study, particularly in terms of the high spatial and temporal resolution of groundwater-derived CO₂ inputs and their variability in boreal headwater streams.
  ------------------------------------------------------
  R2C4: Lastly, the paper appears to have a high degree of overlap with a companion study cited as ‘in revision’, which I assume is now published (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024JG008671). That paper addresses similar research questions at the same site using comparable methods, yet it is not even mentioned in the introduction. The authors need to clarify the added value and novelty of their study, how the two papers complement each other, and the extent of overlap between the two.
  AC_R2C4: We agree that the relationship between this manuscript and the companion study (now published as (Hauptmann et al., 2026)) was not sufficiently clarified in the original version. In the revised manuscript, we will explicitly introduce and cite this study in the Introduction and clearly explain how the two papers complement each other.
  While both studies were conducted at the same site and share some methodological approaches, they address different components of the carbon cycle in headwater streams, as well as different research questions. Hauptmann et al. (2026) focus on quantifying CO₂ evasion and downstream carbon export, and on understanding the controls on their relative partitioning (i.e., the evasion:export ratio). In contrast, the present study focuses on quantifying groundwater-derived CO₂ inputs using ²²²Rn as a tracer and evaluating their contribution relative to these fluxes. Importantly, our study addresses a key gap in current understanding. The fate of CO₂ in headwater streams depends on its partitioning between atmospheric evasion and downstream export, yet this partitioning cannot be fully understood without constraining upstream inputs. While recent studies have focused on the balance between these output pathways, the magnitude and variability of groundwater-derived CO₂ inputs remain poorly quantified (see response to R1C1). Our study, therefore, provides a detailed assessment of the spatial and temporal variability of groundwater CO₂ inputs, offering new insights into a key but often poorly constrained component of stream carbon budgets. We will revise the manuscript to better highlight these distinctions, clarify the complementary nature of both studies, and ensure that the novelty and added value of the present work are clearly shown. In addition, we will revise the title of the manuscript to better distinguish it from (Gao et al., 2025; Jasechko et al., 2014) and avoid potential confusion (preliminary title: “Major and spatially variable groundwater CO₂ inputs fuel downstream export in a boreal stream”)
  Specific comments
  Introduction
  ------------------------------------------------------
  R2C5: L71-75. While this statement is broadly accurate, the authors have overlooked several relevant studies. Just a few examples below:
  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019JG005047
  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2022JG006954
  https://www.sciencedirect.com/science/article/pii/S0043135424015835
  AC_R2C5: We acknowledge that some relevant studies were overlooked in the original manuscript. In the revised version, we will incorporate the suggested references and more thoroughly review the recent literature on groundwater contributions to stream CO₂ dynamics. We will also revise the corresponding section of the Introduction to reflect current knowledge better and to clearly position our study within this context.
  ------------------------------------------------------
  R2C6: L87-89. Once again this statement does not reflect the recent literature on the combined use of 222Rn and CO2 to trace groundwater CO2 inputs. I am not simply suggesting that the authors cite the below studies – instead I would invite them to engage with these studies and identify any remaining knowledge gaps. I believe doing so would increase the impact (and relevance) of this work.
  https://www.sciencedirect.com/science/article/pii/S0048969721012985
  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018JG004912
  https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024JG008592
  AC_R2C6: We thank the reviewer for making us aware of these recent studies. We agree that the original text did not sufficiently reflect recent advances in the combined use of ²²²Rn and CO₂ to trace groundwater-derived CO₂ inputs. In the revised manuscript, we will incorporate the suggested studies and engage more thoroughly with this body of literature. Rather than only adding citations, we will explicitly discuss how these studies have applied combined ²²²Rn–CO₂ approaches and identify the remaining knowledge gaps. In particular, we will highlight that, while previous studies have successfully used these tracers to quantify groundwater contributions, there is still limited understanding of the spatial and temporal variability of groundwater-derived CO₂ inputs at high resolution, especially in boreal headwater systems (see response to R1C1). We will revise the Introduction accordingly to better position our study within this context and to clarify its novelty and contribution.
  ------------------------------------------------------
  R2C7: L93-99. It might be useful to rephrase the hypotheses for clarity – for example adding one hypothesis specifically on the partitioning between emission vs downstream export.
  AC_R2C7: In the revised manuscript, we will rephrase the hypotheses and explicitly include one addressing the partitioning between atmospheric emissions and downstream export of CO₂. Specifically, we will include the following hypothesis: ”we hypothesize that groundwater-derived CO₂ inputs represent a major source of CO₂ to the stream, and that their relative importance compared to atmospheric emissions and downstream export varies in space and time”. This will help to better frame the study objectives and improve the clarity of the conceptual framework.
  Methods
  ------------------------------------------------------
  R2C8: Figure 1. “Experimental” (typo). Groundwater wells G1, G3, G5, and G6 do not appear on the figure. Need to remove blue triangles in a) (unless they relate to something relevant to this study).
  AC_R2C8: We thank the reviewer for noticing this comment. We will correct the typo in the figure caption in the revised manuscript. We also acknowledge that the labelling of groundwater wells was inconsistent between the text and the figure. In the revised version, we will ensure that the correct well codes (N3A, N3B, W3A and W3B) are used consistently throughout the manuscript. Regarding the blue triangles in panel a), we will clarify in the figure caption that they indicate the direction of stream flow.
  ------------------------------------------------------
  R2C9: L110-126. Please specify catchment area.
  AC_R2C9: The catchment area increases from 24.6 ha at the most upstream site to 34.0 ha at the most downstream site. This information will be included in the revised version of the manuscript.
  ------------------------------------------------------
  R2C10: L143-159. It seems like DIC was measured and then CO2 was inferred based on DIC and pH? Why not measuring CO2 directly instead? Please clarify.
  AC_R2C10: We thank the reviewer for this comment. CO₂ concentrations were derived from measurements of dissolved inorganic carbon (DIC), with samples acidified to convert all inorganic carbon species to CO₂prior to analysis. We acknowledge that direct measurements of dissolved CO₂ in the field are possible. However, these approaches (e.g., headspace equilibration or in situ CO₂ sensors) are generally more time-consuming, as they require several minutes per measurement for equilibration. In contrast, our approach allows for rapid sampling in the field (within seconds), enabling high spatial resolution measurements along the stream reach. Furthermore, sample acidification prevents biological activity and minimizes changes in carbonate chemistry during storage and handling, ensuring consistent and comparable measurements across samples (see response to R1C4). We will clarify this methodological choice in the revised manuscript.
  ------------------------------------------------------
  R2C11: L176. I do not think that such shallow piezometers (max depth 1.3 m) can be considered to intersect “regional aquifers”. They likely only capture the very near-surface portion of the saturated zone (unless the site is located within a regional groundwater discharge area).
  AC_R2C11: We thank the reviewer for this important comment. We agree that the shallow piezometers used in this study likely do not intersect regional aquifers but rather sample near-surface groundwater within the shallow saturated zone. In this study, we use the term “groundwater” in a broader sense to refer to subsurface water contributing to streamflow, including shallow flow paths. However, we acknowledge that our approach does not allow us to distinguish between different groundwater sources or flow paths, as we rely on a single tracer (i.e., ²²²Rn). To avoid confusion, we will revise the manuscript to clarify this definition and ensure consistent terminology throughout.
  ------------------------------------------------------
  R2C12: L185-189. Unclear why the groundwater samples were not analysed on the RAD7, like for the surface water samples.
  AC_R2C12: We thank the reviewer for this comment. As explained in our response to Referee 1 (see R1C6), different methods were used to measure 222Rn depending on sample type and practical constraints. In brief, RAD7 measurements require relatively large, gas-tight water volumes, which are difficult to obtain for groundwater samples. Therefore, liquid scintillation counting (LSC) was used for groundwater, as it requires smaller sample volumes and is better suited for high ²²²Rn concentrations. We will clarify this rationale in the revised manuscript.
  ------------------------------------------------------
  R2C13: L189-194. This paragraph is also confusing – why was the same method not applied to CO2 measurements in groundwater samples as was used for streamwater samples?
  AC_R2C13: We thank the reviewer for this comment. As outlined in our response to Referee 1 (see R1C8), measuring CO₂ concentrations in the same groundwater samples used for ²²²Rn analysis was not feasible due to limited groundwater volumes, which had to be allocated across multiple analyses for different projects. To ensure consistent sampling across sites and conditions, we installed soil gas probes at the same locations as the groundwater wells. This allowed groundwater samples to be prioritized for ²²²Rn and other analyses requiring liquid samples, while CO₂ was measured using gas samples from the probes. These probes were consistently located below the groundwater table and created a headspace in equilibrium with the surrounding groundwater. We are therefore confident that the CO₂ concentrations derived from these probes are representative of the same groundwater conditions as the ²²²Rn samples.
  ------------------------------------------------------
  R2C14: L210-213. Why was pH not measured in the field? Your indirect CO2 estimates require accurate pH measurements, so this procedure needs to be better justified.
  AC_R2C14: We acknowledge that pH measurements are critical for the accurate estimation of CO₂ concentrations. We chose to measure pH under controlled laboratory conditions rather than in situ because field-based pH measurements in low-ionic-strength boreal waters are often associated with substantial uncertainty due to electrode instability, temperature fluctuations, and calibration drift. To ensure accuracy, we collected and handled water samples to minimize gas exchange (through airtight storage, minimal headspace, rapid analysis upon return to the laboratory). We measured pH in the laboratory using a well-calibrated electrode under stable temperature conditions. This approach reduces measurement noise and improves reproducibility compared to field measurements.
  We acknowledge the potential for pH shifts during sample handling. However, we expect these effects to have small implications for calculated CO₂ concentrations. To evaluate this, we conducted a sensitivity analysis showing that even a ± 0.1 pH unit deviation would alter the proportion of the DIC pool made up by CO₂ by only 0.82%. We calculated CO₂ from DIC using stream water pH and temperature following standard carbonate equilibrium equations (Stumm & Morgan, 1995) and found that CO₂ on average made up 98.2 ± 0.8% of the DIC pool (Hauptmann et al., 2026), ranging from 97.7 to 98.6% for a ± 0.1 pH unit deviation. This variation is minor relative to the spatial and temporal variation of CO₂ concentrations we observed in our study stream. We will clarify this procedure in the revised manuscript to better justify our approach.
  ------------------------------------------------------
  R2C15: L222. It is unclear why 222Rn in groundwater is measured here, as this was already described earlier. If the intention is to compare methods, please state this explicitly.
  AC_R2C15: We agree that this point was not clearly explained in the original manuscript. As explained in our response to Referee 1 (see R1C14), the additional measurements of ²²²Rn in groundwater were conducted to compare field-based measurements with laboratory incubation estimates. This comparison allows us to better constrain groundwater ²²²Rn concentrations and assess potential uncertainties associated with each method. We will revise the manuscript to explicitly state this objective and clarify the distinction between the two approaches.
  ------------------------------------------------------
  R2C16: L243. Add reference(s) before the equation, as this mass balance is derived from previous work.
  AC_R1C16: We agree that appropriate references should be included to support the mass balance equation. In the revised manuscript, we will add relevant references to previous studies where similar approaches have been applied.
  ------------------------------------------------------
  R2C17: L299-301. Multiplying a discharge by a concentration should give absolute fluxes in g/d, not area-normalised fluxes in g/m2/d. Please explain how you obtained area-normalised fluxes.
  AC_R2C17: We thank the reviewer for noticing this mistake in the text. We agree that multiplying discharge by concentration yields absolute fluxes (g d^-1). In our study, these fluxes were subsequently normalized by the stream surface area to allow comparison across sites and between fluxes. The stream surface area was estimated based on measurements of stream width and reach length, with width measured at multiple transects along each reach. We will revise the manuscript to clarify how area-normalised fluxes (g m^-2 d^-1) were calculated.
  Results
  ------------------------------------------------------
  R2C18: L366. Figure 3. It would be very helpful to show the discharge series here as well for context.
  AC_R2C18: In our study, discharge time series are already included in the Supporting Information (Figure S1) together with precipitation and groundwater level data. Presenting these variables together allows for a more comprehensive interpretation of hydrological dynamics, including recharge periods and the coupling between groundwater levels and stream discharge. Given that discharge data are also presented in the main text of Hauptmann et al. (2026), we believe that maintaining this information in the Supporting Information provides sufficient context while avoiding redundancy in the main figures. We will, however, ensure that the availability of discharge data is clearly referenced in the main text and figure captions to guide the reader.
  ------------------------------------------------------
  R2C19: L390-391. Please mention if the relationship is positive or negative.
  AC_R2C19: We will revise the manuscript to explicitly indicate the direction of the relationship, clarifying that a positive correlation was observed between slope and groundwater inflows.
  ------------------------------------------------------
  R2C20: L393-398. CV calculated on what?
  AC_R2C20: We agree that this sentence was unclear. The coefficient of variation (CV) refers to groundwater inflow rates, and we will revise the text to clarify this.
  ------------------------------------------------------
  R2C21: L423-429. For downstream fluxes, how were values normalised, i.e. with which surface area? It makes little sense to me to normalise a dissolved CO2 flux by stream surface area. While I understand that normalisation can help compare fluxes across sites, I would recommend presenting absolute downstream CO2 fluxes and comparing these with absolute CO2 emission fluxes and absolute groundwater CO2 inputs. At the very least, this could be done for the catchment outlet (e.g. site S18).
  AC_R2C21: We thank the reviewer for this important comment. We agree that expressing downstream CO₂ fluxes as area-normalised values may not always be the most intuitive approach. In our study, fluxes were normalised by stream surface area to allow comparison across stream reaches with different widths, as narrower sections may otherwise show lower absolute fluxes simply due to their smaller size. However, we acknowledge the value of presenting absolute fluxes for a more direct comparison between groundwater inputs, downstream export, and atmospheric emissions. To address this, we will include absolute downstream CO₂ fluxes in the supported information of the revised manuscript.
  ------------------------------------------------------
  R2C22: L437. Remove “being”
  AC_R2C22: “being” will be removed.
  ------------------------------------------------------
  R2C23: L444. Clarify what is meant by “stream CO2 export”, as there are two possible export pathways.
  AC_R2C23: We agree that the term “stream CO₂ export” may be ambiguous. In this context, we specifically refer to downstream transport of dissolved CO₂. We will revise the text to clarify this.
  ------------------------------------------------------
  R2C24: L444-449. I struggle to understand the usefulness of presenting the contribution of groundwater CO2 inputs to downstream CO2 export. This ratio does not account for concurrent CO2 emission and therefore may be difficult to interpret in terms of actual “contribution”. Why not present the contribution of groundwater inputs to emissions as well? Alternatively, expressing groundwater inputs relative to total CO2 outputs (i.e. emission + downstream export) would be more consistent with a mass balance perspective.
  AC_R2C24: We thank the reviewer for this important comment. We agree that expressing groundwater CO₂ inputs as a contribution to downstream export alone may be misleading, as it does not account for concurrent CO₂ emissions to the atmosphere. To address this, we will revise the manuscript to avoid interpreting this ratio as a direct contribution. Instead, we will focus on comparing the magnitude of groundwater inputs, downstream export, and atmospheric emissions. Where appropriate, we will also consider presenting groundwater inputs relative to total CO₂ outputs (i.e., downstream export plus emissions) to provide a more consistent framework. Accordingly, we will revise the figures to reflect this updated approach.
  Discussion
  ------------------------------------------------------
  R2C25: L456. This subsection provides very limited new insight, as spatial and temporal variability in groundwater inflows to streams is already well established. I suggest reducing it significantly.
  AC_R2C25: In the revised manuscript, we will reduce this subsection and refocus it to better highlight the specific insights provided by our dataset, particularly the high-resolution characterization of this variability and its implications for groundwater-derived CO₂ inputs. We will also rephrase the text to emphasize that the patterns observed in our stream are consistent with and confirm well-established knowledge on the spatial and temporal variability of groundwater inflows.
  ------------------------------------------------------
  R2C26: L470-478. I would be more careful with this interpretation – as per Fig S4, there are several sites with high slope and no groundwater inflow. Also, an incised valley (i.e. a high hillslope gradient) and a high channel slope are not equivalent.
  AC_R2C26: We thank the reviewer for this insightful comment. We agree that our interpretation of the relationship between slope and groundwater inflow may have been too strong. As pointed out, there are sites with high channel slope that do not show significant groundwater inflow, indicating that slope alone does not control groundwater inputs. We will revise the text to reflect this more nuanced interpretation. We also acknowledge that hillslope gradient (related to valley incision) and channel slope are not equivalent, and we will clarify this distinction in the revised manuscript.
  ------------------------------------------------------
  R2C27: L517-520. References are needed to support this statement.
  AC_R2C27: We will include relevant references to support that intensive rainfall events recharge the groundwater system and enhance interaction between groundwater and surface water (e.g. (Hauptmann et al., 2026).
  ------------------------------------------------------
  R2C28: L529-537. This section is not supported by references, is very generic, and adds limited value to the discussion. Please consider improving or removing.
  AC_ R2C28: In the revised manuscript, we will either reduce or remove this section, and where retained, we will strengthen it by including relevant references and more clearly linking the discussion to our results.
  ------------------------------------------------------
  R2C29: L556-557. This statement and reference are confusing – isn’t the finding that a substantial portion of groundwater CO2 inputs is transported downstream a main finding of the present study? Or is it taken from the Hauptmann study?
  AC_R2C29: We thank the reviewer for this comment. We agree that the statement was unclear and may have led to confusion regarding the origin of this finding. While (Hauptmann et al., 2026) demonstrated that a substantial portion of CO₂ is transported downstream rather than emitted to the atmosphere, that study did not explicitly distinguish the contribution of groundwater-derived CO₂. In contrast, our study specifically quantifies groundwater-derived CO₂inputs and shows how these inputs contribute to both CO₂ evasion and downstream transport. In the revised manuscript, we will clarify this distinction and ensure that the respective contributions of both studies are clearly separated.
  Conclusions
  ------------------------------------------------------
  R2C30: L565-584. This section is very unspecific and adds little to the paper. I would suggest removing it.
  AC_R2C30: We agree that this section was too general and did not add sufficient value to the manuscript. In the revised version, we will remove this section and instead include a concise concluding paragraph summarising the main findings of the study.
  ------------------------------------------------------
  References
  Aho, K. S., Cawley, K. M., Hall, R. O., Hensley, R. T., Dodds, W. K., Harrison, N., & Goodman, K. J. (2025). Gas transfer velocity (k600) increases with discharge in steep streams but not in low‐slope streams. Limnology and Oceanography Letters, 10(3), 308–317. https://doi.org/10.1002/lol2.70003
  Chanyotha, S., Kranrod, C., & Burnett, W. C. (2014). Assessing diffusive fluxes and pore water radon activities via a single automated experiment. Journal of Radioanalytical and Nuclear Chemistry, 301, 581–588. https://doi.org/10.1007/s10967-014-3157-3
  Duvert, C., Bossa, M., Tyler, K. J., Wynn, J. G., Munksgaard, N. C., Bird, M. I., Setterfield, S. A., & Hutley, L. B. (2019). Groundwater‐Derived DIC and Carbonate Buffering Enhance Fluvial CO ₂ Evasion in Two Australian Tropical Rivers. Journal of Geophysical Research: Biogeosciences, 124(2), 312–327. https://doi.org/10.1029/2018JG004912
  Gao, H., Ju, Q., Zhang, D., Wang, Z., Hao, Z., & Kirchner, J. W. (2025). Quantifying Dynamic Linkages Between Precipitation, Groundwater Recharge, and Streamflow Using Ensemble Rainfall‐Runoff Analysis. Water Resources Research, 61(1). https://doi.org/10.1029/2024WR037821
  Hauptmann, D., Klaus, M., Sponseller, R. A., Olid, C., Laudon, H., & Karlsson, J. (2026). Discharge Modulates the Dominance of Downstream Carbon Export Over Evasion in a Boreal Headwater Stream. Journal of Geophysical Research: Biogeosciences, 131(2). https://doi.org/10.1029/2024JG008671
  Hauptmann, D., Klaus, M., Sponseller, R. A., Olid, C., Laudon, J., & Karlsson, J. (n.d.). Discharge modulates the dominance of downstream carbon export over evasion in a boreal headwater stream. Submitted to Journal of Geophysical Research - Biogeosciences.
  Jasechko, S., Birks, S. J., Gleeson, T., Wada, Y., Fawcett, P. J., Sharp, Z. D., McDonnell, J. J., & Welker, J. M. (2014). The pronounced seasonality of global groundwater recharge. Water Resources Research, 50(11), 8845–8867. https://doi.org/10.1002/2014WR015809
  Klaus, M., Geibrink, E., Hotchkiss, E. R., & Karlsson, J. (2019). Listening to air-water gas exchange in running waters. Limnology and Oceanography Methods, 17(7), 395–414.
  Larsen, W. J., Torres, M. A., Ramos, E. J., Muñoz, S., Hou, Y., Sun, T., Ibarra, D. E., Gammerman, M., Bernstein‐Schalet, J., Suquino, K. D., & Kemeny, P. C. (2025). Limited Headwater CO ₂ Emissions Relative to Downstream C Fluxes: Insights From a Tracer‐Enabled Reactive Transport Model. Journal of Geophysical Research: Biogeosciences, 130(10). https://doi.org/10.1029/2024JG008592
  Leith, F. I., Dinsmore, K. J., Wallin, M. B., Billett, M. F., Heal, K. V., Laudon, H., Öquist, M. G., & Bishop, K. (2015). Carbon dioxide transport across the hillslope–riparian–stream continuum in a boreal headwater catchment. Biogeosciences, 12(6), 1881–1892. https://doi.org/10.5194/bg-12-1881-2015
  Moore, R. D. (2005). Slug injection using salt in solution. Watershed Magage Bulletin, 8, 1–6.
  Pilla, R. M., Griffiths, N. A., Gu, L., Kao, S., McManamay, R., Ricciuto, D. M., & Shi, X. (2022). Anthropogenically driven climate and landscape change effects on inland water carbon dynamics: What have we learned and where are we going? Global Change Biology, 28(19), 5601–5629. https://doi.org/10.1111/gcb.16324
  Rocher-Ros, G., Sponseller, R. A., Lidberg, W., Mörth, C. M., & Giesler, R. (2019). Landscape process domains drive patterns of CO2 evasion from river networks. In Limnology And Oceanography Letters (Vol. 4, Number 4, pp. 87–95). John Wiley and Sons Inc. https://doi.org/10.1002/lol2.10108
  Song, C., Liu, S., Wang, G., Zhang, L., Rosentreter, J. A., Zhao, G., Sun, X., Yao, Y., Mu, C., Sun, S., Hu, Z., Lin, S., Sun, J., Li, Y., Wang, Y., Li, Y., Raymond, P. A., & Karlsson, J. (2024). Inland water greenhouse gas emissions offset the terrestrial carbon sink in the northern cryosphere. Science Advances, 10(39). https://doi.org/10.1126/sciadv.adp0024
  Soued, C., del Giorgio, P. A., & Maranger, R. (2016). Nitrous oxide sinks and emissions in boreal aquatic networks in Québec. Nature Geoscience, 9(2), 116–120. https://doi.org/10.1038/ngeo2611
  Stieglitz, T. C., Beek, P., Souhaut, M., & Cook, P. G. (2013). Karstic groundwater discharge and seawater recirculation through sediments in shallow coastal Mediterranean laggons, determined from water, salt and radon budgets. Marine Chemistry, 156, 73–84.
  Stumm, W., & Morgan, J. J. (1995). Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters: 10: 0471511854. Wiley.
  Wallin, M. B., Campeau, A., Audet, J., Bastviken, D., Bishop, K., Kokic, J., Laudon, H., Lundin, E., Löfgren, S., Natchimuthu, S., Sobek, S., Teutschbein, C., Weyhenmeyer, G. A., & Grabs, T. (2018). Carbon dioxide and methane emissions of Swedish low‐order streams—a national estimate and lessons learnt from more than a decade of observations. Limnology and Oceanography Letters, 3(3), 156–167. https://doi.org/10.1002/lol2.10061
  Wang, C., Xie, Y., Liu, S., McCallum, J. L., Li, Q., & Wu, J. (2021). Effects of diffuse groundwater discharge, internal metabolism and carbonate buffering on headwater stream CO2 evasion. Science of The Total Environment, 777, 146230. https://doi.org/10.1016/j.scitotenv.2021.146230
  
  Citation: https://doi.org/10.5194/egusphere-2025-5892-AC2
- AC3: 'Reply on RC2', Carolina Olid, 11 May 2026
  
  Publisher’s note: this comment is a copy of AC2 and its content was therefore removed on 12 May 2026.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5892-AC3

Carolina Olid, Demian Hauptmann, Jan Karlsson, and Marcus Klaus

Supplement

https://doi.org/10.5194/egusphere-2025-5892-supplement

Carolina Olid, Demian Hauptmann, Jan Karlsson, and Marcus Klaus

Viewed

Total article views: 1,289 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
796	387	106	1,289	177	144	231

HTML: 796
PDF: 387
XML: 106
Total: 1,289
Supplement: 177
BibTeX: 144
EndNote: 231

Views and downloads (calculated since 06 Feb 2026)

Month	HTML	PDF	XML	Total
Feb 2026	453	213	53	719
Mar 2026	227	126	38	391
Apr 2026	107	36	11	154
May 2026	9	12	4	25

Cumulative views and downloads (calculated since 06 Feb 2026)

Month	HTML	PDF	XML	Total
Feb 2026	453	213	53	719
Mar 2026	227	126	38	391
Apr 2026	107	36	11	154
May 2026	9	12	4	25

Viewed (geographical distribution)

Total article views: 1,289 (including HTML, PDF, and XML) Thereof 1,289 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 25 May 2026

Short summary

We studied how groundwater influences carbon dioxide dynamics in a boreal forest stream by measuring water inputs, gas emissions, and downstream transport every two weeks along a 400-meter reach during the ice-free season. We found that groundwater supplies large amounts of carbon dioxide, much of which flows downstream rather than escaping to the atmosphere. These insights improve our understanding of landscape carbon cycling under changing climate and hydrological conditions.


Total:	0
HTML:	0
PDF:	0
XML:	0

Downstream export dominates the fate of groundwater-derived CO2 in a boreal stream

Supplement

Viewed

Viewed (geographical distribution)

Downstream export dominates the fate of groundwater-derived CO₂ in a boreal stream