the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jan 2026
Explicit Representation and Calibration of Different Landscape Units for a Robust Catchment DOC Export Model
Abstract. Elevated dissolved organic carbon (DOC) concentrations are a major concern for ecosystems and drinking water supply. Data-driven studies revealed variable functioning of different landscape units (upland, riparian zone, and groundwater) in catchment DOC mobilization and export. However, lumped and landscape-explicit (separating upland and riparian zone) model structures are generally calibrated to stream DOC concentrations, while the internal DOC dynamics often do not receive sufficient attention. Here, we developed a flexible model with a lumped and landscape-explicit structure for four headwater catchments in the Harz Mountains, Germany. We evaluated these models under a baseline calibration (only using stream DOC concentration) and a constrained calibration (using stream DOC and internal DOC concentrations). Under the baseline calibration, both model structures can reasonably represent stream DOC dynamics in some catchments (Kling–Gupta efficiency of some behavioural simulations > 0.6), but with unreasonably high groundwater DOC. By contrast, the constrained calibration reduces the KGE for stream DOC concentrations but ensures a more realistic representation of internal DOC dynamics. Additionally, the landscape-explicit model structure is more robust than the lumped model structure under changing boundary conditions. Our study thus highlights the necessity of representing different landscape units explicitly in combination with constraining the calibration of DOC concentrations in these landscape units.
Competing interests: Rohini Kumar is a member of the editorial board of Hydrology and Earth System Sciences
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.- Preprint
(1986 KB) - Metadata XML
-
Supplement
(956 KB) - BibTeX
- EndNote
Status: open (until 27 Apr 2026)
- RC1: 'Comment on egusphere-2025-6246', Anonymous Referee #1, 14 Mar 2026 reply
-
RC2: 'Comment on egusphere-2025-6246', Anonymous Referee #2, 10 Apr 2026
reply
The present manuscript (MS) compares the performance of a lumped and a landscape-explicit version of a conceptual DOC catchment model. While both versions could satisfactorily simulate streamflow DOC, low groundwater DOC and increased C-input could only be simulated by the more complex, landscape-explicit model version. The paper is nicely written and gets to the point that a more complex model structure and additional information, i.e. low groundwater DOC, adds to the realism of model simulations.
However, these results seem expectable at first sight and so far the contributions to general knowledge which go beyond this particular model and beyond this particular application to four specific German basins seem rather limited. Eventually, a more detailed representation of DOC formation, but for sure a more detailed data analysis are warranted to justify the model structure: is the presented model really a valid compromise between required complexity and computational efficiency and therefore might be regarded as a straight-forward tool for practitioners? As follows three major concerns are listed.
1. Simplified representation of DOC origin
The conceptualization of carbon transformation follows the well-known INCA-C model by (Futter et al., 2007) but is even simplified by single net transformation between the soil organic carbon pool and DOC. While the authors state that this reduces equifinality, this very simple assumption combines various individual multicomponent reactions. An existing example in HESS shows that more physical basis is possible with reasonable computational effort (Wen et al. 2020, https://doi.org/10.5194/hess-24-945-2020). Those authors incorporated coupled elemental cycling, stoichiometry, thermodynamics and kinetics in their DOC catchment model already 6 years ago. This means that a more sophisticated representation of DOC formation is possible and might be warranted also in the present model. If the authors decide to maintain their simplified approach, this needs strong justification and more detailed checks.
2. Missing carbon mass balance
In the introduction, the authors criticize the model by Birkel et al. 2014 for not closing carbon mass balances. But in their own model application, they only study runoff and DOC concentrations separately, but do not calculate DOC loads or carbon mass fluxes. This would also be valuable for their “scenario” of forest dieback: Is their landscape-explicit model version really superior when it comes to a realistic carbon mass balance? By the way, also for drinking water reservoirs, realistic DOC loads are more important than reproducing DOC concentration patterns.
3. DOC-export during different flow conditions
The authors only test their models by overall performance during a multi-year period. They do not study DOC simulations during different flow conditions. Again the study of Wen et al., 2020 might be relevant here. Their study showed that DOC-stocks are formed during periods of high temperature, while the main DOC export occurs during following wet conditions or during high magnitude runoff events. They also showed that streamflow DOC equals groundwater DOC during dry conditions, riparian zone DOC during intermediate conditions, and hillslope DOC during real wet conditions. These results could easily be checked or challenged by the present application. This might also prove the physical soundness and robustness of the present approach. Potentially, a shortcut of riparian buffer zones during high magnitude events might also play a role which is not included in the present model so far. Especially in the “Warme Bode”, the landscape-explicit model fails to predict single DOC peaks. Those are particularly evident following the 2015 European drought which was characterized by high temperatures and below average rainfall. The different behavior between of the four basins and how an improved model adequately represents this could be another test here.
Citation: https://doi.org/10.5194/egusphere-2025-6246-RC2 -
RC3: 'Comment on egusphere-2025-6246', Anonymous Referee #3, 10 Apr 2026
reply
This manuscript evaluates if accurate DOC simulations at catchment outlets reliably indicate realistic internal DOC dynamics. By comparing lumped and landscape-explicit models across diverse headwater catchments, the study assesses how model structures and calibration strategies (baseline and constrained) perform under varying conditions. The results demonstrate that good performance at the catchment outlet does not necessarily imply realistic internal process representation, as stream-level agreement can mask substantial internal discrepancies. The analysis further shows that landscape-explicit models incorporating physically motivated internal constraints provide enhanced interpretability and are better suited for scenario analyses. Overall, the study highlights that robust DOC modelling requires an adequate representation of internal dynamics and supports the use of landscape-explicit, internally constrained model structures for catchment-scale applications.
In summary, the manuscript shows clear potential; however, the outcome is somehow expectable. Several issues related to clarity, the linkage between hydrology and DOC dynamics, data uncertainty, and the interpretation of performance metrics need to be addressed:
- Although the DOC model description is detailed, a short, high-level overview would benefit readers less familiar with the framework—particularly regarding how hydrological simulation outputs are incorporated into the DOC simulations (e.g. whether and how grid-based data are aggregated).
- The introduction refers to a carbon mass balance. Please clarify whether the model applies a closed carbon mass balance and discuss any implications this assumption may have for the results.
- Hydrology plays a key role in the modelling experiments.
- The 30 selected simulations exhibit similarly strong hydrological performance, including comparable representations of individual flow components (Figure S3). Expanding the analysis to include simulations with contrasting hydrological behavior—such as improved high-flow performance—could provide additional insight and strengthen the discussion and conclusions, even if overall performance is slightly reduced but still acceptable.
- Please discuss how the underestimation of high flows affects the simulated DOC dynamics.
- Figure 4 indicates an uneven distribution of calibration and validation observations. In particular, Kalte Bode and Warme Bode have comparatively few validation observations at different times. The number of observations per period should be explicitly reported, as this imbalance may influence flow-condition representation and the interpretation of model performance.
- Please address the uncertainty associated with the DOC measurements. Could higher observation frequency or event-based (e.g. daily) DOC observations potentially improve model performance.
- The Kling–Gupta Efficiency (KGE) is used as the primary performance metric, integrating correlation, variability, and bias between simulated and observed DOC concentrations. Given that these components can contribute differently to the overall KGE—as suggested by NSE, R² and bias in the supplementary material—further discussion of their respective roles, as well as the strengths and limitations of KGE in this context, would be valuable. In particular, it would be informative to assess whether one component dominates KGE and how this relates to different flow conditions and associated DOC dynamics.
Minor Comments:
- Figures S4–S6: Please clarify in the captions whether the results refer to the calibration period, the validation period, or both.
Citation: https://doi.org/10.5194/egusphere-2025-6246-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 206 | 71 | 24 | 301 | 38 | 20 | 25 |
- HTML: 206
- PDF: 71
- XML: 24
- Total: 301
- Supplement: 38
- BibTeX: 20
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript addresses an important question in catchment-scale dissolved organic carbon (DOC) modelling: whether acceptable outlet DOC simulations are achieved for the right reasons—namely, through physically plausible internal DOC dynamics. By comparing lumped and landscape-explicit model structures across four headwater catchments, and by contrasting baseline and constrained calibration approaches, the study makes a valuable contribution to the field. The central conclusion is compelling: satisfactory stream DOC performance alone does not guarantee realistic internal behavior, and a landscape-explicit structure paired with internal constraints offers a more robust framework for interpretation and scenario analysis.
Overall, the manuscript is promising; however, several substantial issues must be addressed before it is suitable for publication.
Major Comments:
1. The constrained calibration framework requires greater transparency. While the manuscript explains that 100,000 parameter sets were generated, screened against internal constraints, and then filtered for the highest-KGE solution, it remains unclear how restrictive these constraints are and the extent to which they reduce the feasible solution space. I encourage the authors to explicitly report the number of parameter sets that satisfy each individual constraint, as well as all constraints jointly. Furthermore, it would be highly beneficial to better illustrate how the constrained calibration alters the distributions of both parameters and internal states.
2. The delineation of riparian areas necessitates stronger justification. Because the core argument of the paper is build on the explicit separation of upland and riparian units, the conclusions are highly likely to be sensitive to the definition of the riparian extent. Incorporating a brief sensitivity analysis, or at a minimum, a more comprehensive discussion regarding the uncertainties associated with the chosen delineation approach, would significantly strengthen the manuscript.
3. The scenario analysis should be interpreted with greater caution. Although the manuscript notes that the experiment focuses solely on increased upland carbon inputs without representing the hydrological alterations associated with forest dieback, this limitation must be stated more prominently. The exercise should be explicitly framed as a structural stress test rather than a comprehensive prediction of forest-dieback impacts.
4. The authors should more explicitly acknowledge that while the landscape-explicit structure represents a significant advancement, it is not yet a complete solution. The study demonstrates that baseline outlet calibration can yield unrealistic internal DOC dynamics, and that the landscape-explicit model performs better under constrained calibration and scenario testing. Nevertheless, the discussion also indicates that riparian DOC concentrations may still be underestimated relative to observations, underscoring the need for further process refinement.
Minor Comments:
1. Several formatting and technical errors must be addressed. Notably, the heading “3 Methodology” mistakenly appears at the beginning of the Results section. Additionally, the export equations should be carefully verified for consistency with the conceptual model description.
2. The reliance on Kling-Gupta Efficiency (KGE) as the primary evaluation criterion should be explicitly justified, particularly given the authors' observation that model ranking is sensitive to the choice of performance metric.
3. The manuscript would benefit from a clearer presentation of how many behavioral models satisfy the imposed internal constraints. For instance, the finding that only 53% of behavioral landscape-explicit models meet all constraints under baseline calibration is highly significant and warrants greater emphasis.
4. Several figures are visually dense and should be streamlined. Shifting some of the highly detailed material to the Supplementary Information would enhance overall readability.
5. Given that the study's novelty stems largely from its model implementation and calibration design, making the code and computational workflow openly accessible would substantially elevate the paper's impact and reproducibility.