the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Feb 2026
Confidence-Aware Framework for Mapping Satellite-Derived River Reaches to Gridded Routing Networks
Abstract. The Surface Water and Ocean Topography (SWOT) mission delivers reach-scale observations of river water surface elevation, contextualized by the vector-based SWORD database. Assimilating these observations into gridded routing models such as CTRIP is hindered by structural mismatches between object-based river geometries and pixel-based flow networks. We present a global, confidence-aware pipeline that assigns SWORD reaches to CTRIP pixels by combining geometric and hydrological criteria such as intersection, proximity, upstream-area consistency, reach length, and flow-direction alignment into a composite score. Each assignment receives a confidence tier (Tier 1: single; Tier 2: scored; Tier 3: fallback; Tier 4: unassigned), and Tier-2 cases are further refined by a confidence score (high/medium/low). Applied globally at 1/12°, the framework assigns >99 % of CTRIP pixels; the vast majority are resolved either unambiguously (Tier 1) or as high-quality scored matches (Tier 2–High), with no fallback assignments and <0.5 % unassigned. Independent diagnostics based on basin-hash continuity confirm hydrological integrity. Code and outputs (CSV, NetCDF, shapefiles) are openly available and directly usable for assimilation into CTRIP or can be applied to any other gridded river network, providing a reproducible foundation for bridging SWOT observations with global river routing models.
- Preprint
(4769 KB) - Metadata XML
-
Supplement
(75 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
CEC1: 'Comment on egusphere-2026-509', Astrid Kerkweg, 13 Mar 2026
-
AC3: 'Reply on CEC1', Kaushlendra Verma, 16 Jun 2026
Comment 1: The main paper must give the model name and version number (or other unique identifier) in the title.
Response: We thank the Executive Editor for highlighting this requirement. In accordance with GMD editorial guidelines, we have updated the manuscript title to include the name and version of the mapping pipeline.
We introduce the acronym SWORD-CTRIP Mapping (SCM v1.0) as a unique identifier for the methodology presented in this study. This modification ensures clear identification and traceability of the development in line with GMD standards.
Manuscript Change:
The revised title is now as: “SWORD-CTRIP Mapping (SCM v1.0): A Confidence-Aware Framework for Assigning Vector River Reaches to Gridded Routing Networks"
Code and Data Section: The version SCM v1.0 corresponds to the exact implementation archived in this study.
Citation: https://doi.org/10.5194/egusphere-2026-509-AC3
-
AC3: 'Reply on CEC1', Kaushlendra Verma, 16 Jun 2026
-
RC1: 'Comment on egusphere-2026-509', Anonymous Referee #1, 18 Apr 2026
Review of the paper entitled : “Confidence-Aware Framework for Mapping Satellite-Derived River Reaches to Gridded Routing Networks”, by Kaushlendra Verma and Simon Munier
Dear authors,
Please find below my review and suggestions of minor to moderate revisions regarding this interesting technical note.
Best regards,
General
This technical note presents a pipeline whose objective is “to identify the single most hydrologically consistent SWORD reach for each valid CTRIP pixel.” The topic is relevant and timely, addressing an important challenge in linking vector-based river network datasets with gridded hydrological models.
Overall, the manuscript is well developed, and both the methodology and figures are interesting. The comments below mainly aim at improving clarity and reinforcing key aspects.
Some clarifications would be beneficial, particularly regarding topological configurations, sensitivity to weighting, the use of zoomed examples, and the justification of how hydrological connectivity and drainage consistency are preserved. In addition to drainage/mass considerations, a brief discussion of hydrodynamic implications, especially in the context of data assimilation, would further strengthen the manuscript.
In Section 3, the methodology is sound but could be made easier to follow with minor improvements in notation and a clearer description of the weighting optimization and selection process. Including a few zoomed examples of typical reach configurations within pixels would also help. It would be useful to clarify whether the algorithm ensures consistent basin drainage without duplication, which is important for mass conservation.
Providing the spatial resolution of the CTRIP grid and discussing sensitivity to resolution would help better position the approach. From a hydrodynamic perspective, a more detailed discussion of whether the reach attribution preserves network structure and supports consistent flow propagation would be valuable.
Figures are generally interesting but pixelated, and some global maps are difficult to interpret in detail; localized zooms could improve readability. In Figure 2, a brief comment on sensitivity to the weighting scheme would also be helpful.
Minor
- “Observations of water surface elevation (WSE), facilitating the estimation of river discharge (e.g., Biancamaria et al., 2016; Durand et al., 2016)”. Water surface slope is also a key hydraulic observable for discharge estimation, together with dry channel geometry. Discharge estimation from water surface geometry alone is an ill-posed problem which can be better constrained by hydrological closure (https://doi.org/10.1029/2024WR038455 and refs therein), hydrology (MGB, could be cited also as large scale vectorial H&H model?) here connected to vectorial river network)
- Recent raster–vector frameworks (https://doi.org/10.5194/egusphere-2026-1557, https://doi.org/https://doi.org/10.1029/2024WR038183)
- Area and connectivity conservative are important in hydrology-hydrodynamic consistency, for gradient back propagation in emerging differentiable hydrological–hydraulic modeling approaches for basin scale inference (https://doi.org/10.5194/gmd-15-6085-2022) and double H&H regionalisation (DOI: 10.22541/au.176901862.25424328/v1)
- Another recent ref that could be relevant regarding river network capture in large scale gridded models: https://doi.org/https://doi.org/10.1029/2024WR038183
- “extant solutions”
- Fig1 grey quarter circles on borders?
- L103, clarify “per Pfafstetter hydrological zone basis”
- L106, “in a hydrologically coherent”, clarify vs connectivity and mass/area conservation
- L113, clarify “ghost or unresolved reaches”
- L144 “cantered on”
- L146, “numerical instability” of what? You mean high values of A ? also improve readability of this inline equation and variable names in paragraph
- L155 and after, homogenize notations, bold in table 1. Clarify index i. ‘The candidate with the lowest score is selected” : clarify among list of reaches within a pixel? Is it the confidence score?
- L159, wheight adjustment by pfaster zone : geomorphological regularity or else assumed?
- L198, “The confidence levels assigned”, define it clearly in method section, also what is “reach-pixel mapping” vs index i in eq 1?
- “the mapping preserves upstream–downstream structure at pixel scale”, I have difficulties to understand/see that, could be clarified in methodo and results.
- Clarify “Looking ahead, the framework is readily extensible: adaptive weighting schemes, integration with alternative routing models, and incorporation of additional hydrological metrics could further refine assignment confidence”. Which other metric could help in this (from sword or else), why not done here. Clarify routing model you mean topology, complexity?
Citation: https://doi.org/10.5194/egusphere-2026-509-RC1 -
AC1: 'Reply on RC1', Kaushlendra Verma, 16 Jun 2026
We thank the referee for the careful and constructive review of our manuscript. We are pleased that the overall methodology and relevance of the work are recognized. We have revised the manuscript to improve clarity, strengthen methodological explanations, and better position the approach within hydrological and hydrodynamic contexts.
Below we respond point-by-point to all comments. Reviewer comments are reproduced in italics followed by our responses.
General Comments
Reviewer 1 – General Comment 1: “Some clarifications would be beneficial, particularly regarding topological configurations, sensitivity to weighting, and justification of hydrological connectivity and drainage consistency are preserved.”
Response: We thank the reviewer for raising this important point regarding drainage consistency and potential duplication. The mapping framework does not explicitly enforce full network-scale mass conservation constraints during the assignment step. Instead, it ensures hydrological plausibility through a combination of local and structural criteria:
- River identity filtering based on MERIT Hydro identifiers restricts candidate pixels to those belonging to the same river system, both ensuring topological coherency and preventing cross-basin assignments.
- Upstream-area consistency (A) constrains assignments to matches with similar contributing drainage area, thereby reducing the risk of inconsistent upstream attribution.
- Geometric alignment criteria (distance and flow-direction agreement) further ensure that selected reaches are spatially and directionally consistent with the CTRIP routing structure.
Because these constraints are applied locally at the pixel level, they do not formally guarantee global mass conservation or strict enforcement of downstream connectivity. To assess whether such inconsistencies arise in practice, we introduce an independent basin-hash coherence diagnostic (Section 4), which evaluates whether assigned reaches preserve upstream connectivity patterns derived from MERIT Hydro.
The results show spatially coherent basin structures with minimal fragmentation or interleaving, indicating that the mapping preserves large-scale drainage organization despite the absence of an explicit global constraint. We have clarified this distinction in the revised manuscript and moderated the wording accordingly to avoid overinterpretation.
Manuscript Change:
Section 3: After scoring paragraph: "The assignment framework operates at the level of individual CTRIP pixels and does not explicitly enforce global network constraints such as downstream continuity or strict mass conservation. Instead, hydrological consistency is approximated through local criteria, including river identity filtering, upstream-area agreement, and flow-direction alignment. The extent to which these local constraints preserve basin-scale topology is evaluated a posteriori using an independent diagnostic (Section 4)."
Section 4: Paragraph 225 and line "previous upstream-downstream structure at pixel scale” has been replaced as follows: “The basin-hash diagnostic provides an empirical assessment of whether the assignment preserves large-scale drainage organization. While the mapping does not explicitly enforce global connectivity constraints, the emergence of spatially coherent basin structures indicates that the local assignment criteria are sufficient to maintain hydrological consistency in practice. Future developments could incorporate explicit network-constrained optimization to enforce downstream continuity, although results here suggest that local criteria are already sufficient for global-scale applications.”
Reviewer 1 – General Comment 2: “In addition to drainage/mass considerations, a brief discussion of hydrodynamic implications, especially in the context of data assimilation, would further strengthen the manuscript.”
Response: We thank the reviewer for this important remark. The objective of the mapping is to ensure that observations are associated with routing pixels in a manner compatible with the model’s flow structure, rather than to enforce hydrodynamic constraints explicitly.
This is achieved through:
- Flow-direction alignment (Θ) ensures that selected reaches are consistent with the CTRIP routing direction at the pixel level
- Upstream-area consistency (A) constrains assignments to reaches with comparable contributing drainage area, which is a key control on discharge magnitude and routing continuity.
- Spatial proximity and reach representativeness reduce the likelihood of associating observations with geometrically inconsistent or disconnected river segments.
These criteria ensure that observations are injected at hydrologically meaningful locations, allowing routing models to propagate flow according to their native dynamics. While the method does not enforce mass conservation or downstream continuity explicitly, it is designed to avoid systematic inconsistencies that could disrupt routing behavior.
We have added a dedicated discussion to clarify this role of the mapping as a preprocessing step for data assimilation systems.
Manuscript Change:
Section 4: After results: "From a hydrodynamic perspective, the assignment framework is designed to ensure compatibility with the routing structure rather than to enforce dynamic flow constraints explicitly. The inclusion of flow-direction alignment and upstream-area consistency in the scoring ensures that assigned reaches are locally consistent with the CTRIP flow network. As a result, observations mapped through this framework are introduced at hydrologically coherent locations, allowing routing models to propagate information according to their native dynamics. While the method does not impose explicit constraints on mass conservation or downstream continuity, it is designed to avoid systematic inconsistencies that could disrupt flow propagation in data assimilation applications. This property is particularly important for assimilation systems such as CTRIP-HyDAS, where observation increments are propagated along the river network through model dynamics."
Reviewer 1 – General Comment 3: “In Section 3, the methodology is sound but could be made easier to follow with minor improvements in notation and a clearer description of the weighting optimization and selection process.”
Response: We thank the reviewer for this important question regarding the weighting strategy and its sensitivity. The weighting scheme is not a free tuning parameter, but a mechanism to resolve residual ambiguity after stronger hydrological constraints are applied. Candidate reaches are first restricted by river identity and upstream-area consistency, which significantly reduce the solution space. As a result, remaining candidates are often geometrically similar, and spatial proximity becomes the dominant discriminant. This explains why distance frequently dominates the optimal weighting, while other metrics may receive negligible weight.
This behavior reflects the hierarchical structure of the mapping pipeline, rather than a lack of sensitivity. The final assignments are therefore robust, with limited sensitivity to the precise weighting configuration. We have clarified this interpretation and improved notation throughout Section 3.
Manuscript Change:
Section 3: Line 163-169 has been refined as: "The optimal weights exhibit marked regional variability, but a consistent pattern emerges in which distance frequently dominates the scoring function, while reach length and angular alignment may receive negligible or zero weight. This behavior reflects the hierarchical structure of the mapping pipeline: candidate reaches are first constrained by river identity and upstream-area consistency, which significantly reduce ambiguity. The remaining candidates within a pixel are therefore often geometrically similar, such that spatial proximity alone is sufficient to discriminate between them.
In this context, the weighting step should be interpreted as a mechanism for resolving local geometric ambiguity rather than as a physically calibrated parameterization. Sensitivity of the final assignments to the weighting scheme is consequently limited, as the primary hydrological consistency is enforced prior to scoring."
Addition in the caption of Figure 2: "The limited spread of high-confidence assignments across weighting configurations indicates low sensitivity of the final mapping to the precise choice of weights."
Reviewer 1 – General Comment 4: "Including a few zoomed examples of typical reach configurations within pixels would also help. It would be useful to clarify whether the algorithm ensures consistent basin drainage without duplication, which is important for mass conservation."
Response: We agree with the reviewer that global maps alone can obscure local behavior. To address this, we have:
- added zoomed examples illustrating representative configurations (single-reach, multi-reach, complex systems),
- increased figure resolution to improve readability.
These additions provide explicit visualization of the assignment process at pixel scale and improve interpretability.
Manuscript Change:
Addition in the section 4: “To improve interpretability of the assignment process at pixel scale, we include zoomed examples of representative configurations (Figure S1). These examples illustrate how the scoring framework resolves ambiguity in multi-reach pixels and how assignments behave in structurally complex regions such as braided or deltaic systems.”
Addition of figure caption: "Figure S1: Local examples of SWORD–CTRIP reach-to-pixel assignment. CTRIP routing pixels are shown as grid-cell rectangles, SWORD reach centerlines as polylines, non-selected candidate reaches in light blue, and the selected reach in red. Panels show (a) a Tier 1 single-reach assignment, (b) a Tier 2 multi-reach case resolved by scoring, and (c) a complex braided or deltaic configuration."
Reviewer 1 – General Comment 5: “Providing the spatial resolution of the CTRIP grid and discussing sensitivity to resolution would help better position the approach. From a hydrodynamic perspective, a more detailed discussion of whether the reach attribution preserves network structure and supports consistent flow propagation would be valuable.”
Response: We thank the reviewer for this suggestion.The mapping is performed on the CTRIP routing grid at a spatial resolution of 1/12° (~10 km at the equator), which is now explicitly stated in the revised manuscript. This resolution determines the typical number of SWORD reaches intersecting a pixel and therefore influences the frequency of single- versus multi-reach cases.
The methodology itself is not specific to this resolution and can be applied to other grid configurations. However, the degree of ambiguity in the assignment is expected to vary with resolution: coarser grids increase the likelihood of multiple reaches per pixel, making the scoring step more critical, whereas finer grids tend to reduce ambiguity and lead to more direct (Tier 1) assignments.
We have added a short discussion to clarify this point and to better position the approach with respect to grid resolution.
Manuscript Change:
Addition in the Section 2: “The mapping is performed on the CTRIP routing grid at a spatial resolution of 1/12° (approximately 10 km at the equator). This resolution influences the degree of reach–pixel ambiguity, as it controls the number of SWORD reaches intersecting each grid cell.”
Addition in the Section 4: “Although the methodology is not tied to a specific grid resolution, its behavior depends on the relative scale of the routing grid and the underlying river network. At coarser resolutions, multiple SWORD reaches are more likely to intersect a given pixel, increasing the importance of the scoring framework. At finer resolutions, ambiguity is reduced and assignments are more frequently resolved through direct intersection (Tier 1). This scaling behavior reflects the geometric nature of the problem rather than a limitation of the method. Future work could explicitly quantify this sensitivity by comparing assignment statistics across multiple grid resolutions.”
Reviewer 1 – General Comment 6: “Figures are generally interesting but pixelated, and some global maps are difficult to interpret in detail; localized zooms could improve readability. In Figure 2, a brief comment on sensitivity to the weighting scheme would also be helpful.”
Response: We agree with the reviewer. In the revised manuscript:
- we have added zoomed regional examples (Figure S1) illustrating representative configurations (single-reach, multi-reach, and complex systems),
- increased figure resolution for improved readability.
These additions improve interpretability of the assignment process at pixel scale for Figure 2 and Figure 3.
Minor Comments
Reviewer 1 – Minor Comment 1: “Observations of water surface elevation (WSE), facilitating the estimation of river discharge (e.g., Biancamaria et al., 2016; Durand et al., 2016)”. Water surface slope is also a key hydraulic observable for discharge estimation, together with dry channel geometry. Discharge estimation from water surface geometry alone is an ill-posed problem which can be better constrained by hydrological closure (https://doi.org/10.1029/2024WR038455 and refs therein), hydrology (MGB, could be cited also as large-scale vectorial H&H model?) here connected to vectorial river network)”
Response: We thank the reviewer for this important clarification. We agree that water surface elevation alone is insufficient to constrain discharge, and that additional information such as water surface slope, channel geometry, and hydrological context are required. The Introduction has been revised to reflect this more accurately and to position routing models as a key component in providing hydrological closure.
Manuscript Change:
Introduction Line 31: “facilitating the estimation of discharge” has been replaced as: “facilitating the estimation of river discharge when combined with additional constraints such as water surface slope, channel geometry, and hydrological context”
Reviewer 1 – Minor Comment 2: Missing recent references: “Recent raster–vector frameworks (https://doi.org/10.5194/egusphere-2026-1557, https://doi.org/https://doi.org/10.1029/2024WR038183.). Another recent ref that could be relevant regarding river network capture in large scale gridded models:https://doi.org/https://doi.org/10.1029/2024WR038183"
Response: We thank the reviewer for these suggestions. We have incorporated the recommended references to better position our work within recent developments in:
- Raster–vector integration frameworks,
- Large-scale hydrological and hydrodynamic modeling,
- River network representation in gridded systems.
These additions strengthen the context and relevance of the proposed framework.
Manuscript Change:
Introduction Line 53: "Recent efforts have also explored hybrid raster–vector representations and their implications for large-scale hydrological and hydrodynamic modeling (e.g., Berkaoui et al., 2026; Shrestha et al., 2024). These approaches highlight the importance of maintaining consistency between network structure and model representation, particularly in emerging frameworks that couple hydrological and hydraulic processes.”
Reviewer 1 – Minor Comment 3: “Area and connectivity conservative are important in hydrology-hydrodynamic consistency, for gradient back propagation in emerging differentiable hydrological–hydraulic modeling approaches for basin scale inference (https://doi.org/10.5194/gmd-15-6085-2022) and double H&H regionalisation (DOI: 10.22541/au.176901862.25424328/v1)”
Response: We agree with the reviewer that area and connectivity consistency are fundamental for hydrological–hydrodynamic modeling, particularly in emerging differentiable frameworks. In this study, these aspects are not enforced explicitly at the network level but are approximated through local constraints (upstream area consistency, river identity filtering, and flow-direction alignment). We have clarified this positioning and added references to related modeling approaches.
- Raster–vector integration frameworks,
- Large-scale hydrological and hydrodynamic modeling,
- River network representation in gridded systems.
These additions strengthen the context and relevance of the proposed framework.
Manuscript Change:
Introduction / Discussion: clarified relationship to connectivity-constrained modeling.
Reviewer 1 – Minor Comment 4: “Terminology and wording: extant solutions | L103: per Pfafstetter hydrological zone basis | L106: hydrologically coherent | L113: ghost or unresolved reaches.”
Response: We thank the reviewer for these suggestions. We have revised the manuscript to improve clarity and readability by replacing ambiguous or overly technical expressions and clarifying terminology throughout.
Manuscript Change:
Terminology replaced as follows:
- “extant solutions” → “existing approaches”
- “per Pfafstetter hydrological zone basis” → “independently for each Pfafstetter hydrological zone”
- “hydrologically coherent” → “consistent with river connectivity and drainage structure”
- “ghost or unresolved reaches” → “topologically undefined or invalid reaches”
Reviewer 1 – Minor Comment 5: “Typographical and clarity: L144: cantered → centered | unclear phrasing in upstream-area equation | L146: numerical instability.”
Response: We have corrected typographical errors and improved clarity of the upstream-area consistency metric. In particular, we now explicitly state that instability refers to undefined or inflated values of the mismatch metric when drainage area approaches zero.
Manuscript Change:
- Corrected typos
- Improved explanation of upstream-area mismatch equation
- Clarified handling of near-zero drainage areas
Reviewer 1 – Minor Comment 6: “Equation and notion clarity: L155: clarify index | clarify score definition | numerical instability | distinguish score vs confidence | homogenize notion.”
Response: We thank the reviewer for this important point. Section 3 has been revised to clearly define all variables and indices. The candidate index i is now explicitly introduced, and the distinction between the selection score and the final confidence score has been clarified. Notation has been homogenized throughout the section and Table 1.
Manuscript Change:
Before equation the lines have been added as follows: “For each CTRIP pixel, let i denote the index of candidate SWORD reaches intersecting that pixel. The composite score Si is computed for each candidate i and used to rank the candidates.”
For the clarify score vs confidence: “The composite score used for candidate selection is distinct from the confidence score reported in the final mapping, which is derived from rescaled score values and used for classification and diagnostics.”
The changes being done as:
- Added definition of candidate index i
- Clarified that score is used for ranking candidates within a pixel
- Distinguished score vs confidence metric
- Harmonized notation across text and tables
Reviewer 1 – Minor Comment 7: “L159, wheight adjustment by pfaster zone : geomorphological regularity or else assumed?”
Response: We clarify that weight optimization is empirical and not prescribed based on geomorphological assumptions. Instead, weights reflect the relative importance of metrics after hydrological constraints are applied. The resulting variability across zones reflects differences in network geometry rather than predefined assumptions.
Manuscript Change:
Section 3: clarified interpretation of weight optimization
Reviewer 1 – Minor Comment 8: “L198: The confidence levels assigned”, define it clearly in method section, also what is “reach-pixel mapping” vs index i in eq 1?”
Response: We thank the reviewer for this suggestion. We have clarified the definition of confidence levels and the terminology related to reach-pixel mapping in the Methods section, ensuring that all classification criteria and variable definitions are explicitly stated and consistently used throughout the manuscript.
Manuscript Change:
Section 3: added explicit definitions of confidence levels and mapping terminology.
Reviewer 1 – Minor Comment 9: “The mapping preserves upstream–downstream structure at pixel scale”, I have difficulties to understand/see that, could be clarified in methodo and results.”
Response: We thank the reviewer for this remark. We have clarified that preservation of upstream–downstream structure is evaluated empirically using the basin-hash diagnostic, rather than being explicitly enforced. The revised text emphasizes this diagnostic interpretation and avoids overinterpretation.
Manuscript Change:
Section 4: revised interpretation of basin-hash results.
Reviewer 1 – Minor Comment 10: “Clarify “Looking ahead, the framework is readily extensible: adaptive weighting schemes, integration with alternative routing models, and incorporation of additional hydrological metrics could further refine assignment confidence”. Which other metric could help in this (from sword or else), why not done here. Clarify routing model you mean topology, complexity?”
Response: We thank the reviewer for this comment. We have clarified the discussion of potential extensions by providing examples of additional metrics that could be incorporated, such as stream-order constraints, reach-level observability metrics from SWORD, or alternative hydrological indicators. These were not included in the present study to maintain a minimal and globally applicable framework but represent natural extensions for future work.
Manuscript Change:
Section 5: Line 249: “Potential extensions include the incorporation of additional metrics such as stream-order constraints, reach-level observability indicators, or alternative hydrological descriptors. These were not included here in order to maintain a minimal and globally consistent framework.”
Reviewer 1 – Minor Comment 11: “Fig1 grey quarter circles on borders?”
Response: We thank the reviewer for pointing this out. The figure has been corrected to remove unintended graphical artifacts.
Manuscript Change:
Figure 1: It has been updated.
-
RC2: 'Comment on egusphere-2026-509', Anonymous Referee #2, 27 May 2026
This manuscript proposes a method for matching grid-based river routing with vector-based SWOT data. The scoring is based on the spatial overlap of SWOT vectors within specific grid cells.
It is currently unclear how influential this matching error is compared to inherent SWOT measurement errors and routing discretization errors. When discussing the integration of SWOT data into hydrological modeling, the most dominant components of SWOT measurement error remain unaddressed in this manuscript.
To calculate the composite score, the algorithm searches across all admissible weight combinations to maximize the number of high-confidence assignments. The theoretical or practical justification for this specific scheme is missing. It is unclear why the authors estimate the weights in this way.
In real-world applications, a simple index is often insufficient. The manuscript does not demonstrate how to integrate this scoring scheme into actual modeling workflows. For instance, if a low score means a higher matching uncertainty in a DA filter, how should that uncertainty be quantitatively assigned?
The proposed framework appears strictly limited to grid-based routing. The authors did not address how (or if) this method can be adapted for vector-based routing models, such as the Muskingum method or modern hydro-fabric frameworks.
Citation: https://doi.org/10.5194/egusphere-2026-509-RC2 -
AC2: 'Reply on RC2', Kaushlendra Verma, 16 Jun 2026
We thank the reviewer for the careful and constructive evaluation of our manuscript. We appreciate the reviewer’s focus on uncertainty characterization, weighting strategy, data-assimilation applicability, and transferability of the framework. These comments have helped us clarify the scope, limitations, and intended applications of the proposed methodology. We have revised the manuscript accordingly and provide detailed responses below.
Reviewer 2 – Comment 1: " It is currently unclear how influential this matching error is compared to inherent SWOT measurement errors and routing discretization errors. When discussing the integration of SWOT data into hydrological modeling, the most dominant components of SWOT measurement error remain unaddressed in this manuscript.”
Response: We thank the reviewer for this important observation. The objective of the present study is not to quantify the complete uncertainty budget associated with SWOT-based hydrological applications, but rather to address a specific and previously underexplored source of uncertainty: the structural mismatch between vector-based SWORD reaches and gridded routing networks.
SWOT measurement uncertainties (e.g., water-surface elevation retrieval errors, discharge retrieval uncertainties, observability limitations, and radar-related effects) constitute an independent class of errors that originate from the observation system itself. Similarly, routing discretization errors arise from the representation of river networks within numerical models. The framework proposed here is designed as an intermediate re-conciliation layer between observations and models. Its purpose is to minimize ambiguity in reach-to-pixel attribution prior to any assimilation or model–observation comparison.
To avoid ambiguity regarding scope, we have clarified throughout the manuscript that the proposed confidence score quantifies assignment uncertainty only and should not be interpreted as a complete measure of SWOT observation uncertainty.
Manuscript Change:
Section 5: “The present study addresses structural assignment uncertainty arising from the reconciliation of vector-based and gridded river representations. Other uncertainty sources, including SWOT measurement errors, discharge retrieval uncertainty, and routing discretization errors, are independent components of the overall uncertainty budget and remain outside the scope of this technical note.”
Reviewer 2 – Comment 2: “To calculate the composite score, the algorithm searches across all admissible weight combinations to maximize the number of high-confidence assignments. The theoretical or practical justification for this specific scheme is missing. It is unclear why the authors estimate the weights in this way”
Response: We thank the reviewer for raising this important point. The weighting procedure was not intended as a calibration exercise against a reference dataset, since no globally available benchmark exists that provides a known “true” reach-to-pixel correspondence for all river systems. Instead, the weighting strategy was designed as a pragmatic mechanism to resolve residual ambiguity after stronger hydrological constraints had already been applied. Candidate reaches are first restricted through river-identity filtering and upstream-area consistency criteria, substantially reducing the solution space. The subsequent weighting step is therefore used only to discriminate among a small number of geometrically plausible candidates.
For each Pfafstetter zone, we evaluate all admissible weight combinations and select the configuration that maximizes the number of assignments satisfying the highest confidence criteria. The objective is not to optimize physical parameters, but rather to identify the weighting scheme that produces the largest number of internally consistent assignments.
We agree that this rationale was insufficiently explained in the original manuscript and have clarified it substantially in the revised version.
Manuscript Change:
Section 3: “Because no globally validated reference dataset exists for true reach-to-pixel correspondence which actually depends on the model river network itself, weights are not calibrated against observations. Instead, an exhaustive search of admissible weight combinations is performed within each Pfafstetter zone, and the optimal combination is defined as the one maximizing the number of assignments satisfying the highest confidence criteria.”
Reviewer 2 – Comment 3: “In real-world applications, a simple index is often insufficient. The manuscript does not demonstrate how to integrate this scoring scheme into actual modeling workflows. For instance, if a low score means a higher matching uncertainty in a DA filter, how should that uncertainty be quantitatively assigned?”
Response: We thank the reviewer for highlighting this important connection between mapping uncertainty and data assimilation. We agree that assignment confidence should ultimately be translated into quantitative uncertainty estimates when observations are assimilated into hydrological models. However, the objective of the present study is limited to constructing and evaluating the mapping framework itself. The confidence metrics generated by the framework are intended to provide a quantitative indicator of assignment reliability that can subsequently be incorporated into assimilation workflows through observation screening, adaptive observation-error inflation, confidence-dependent localization, or other quality-control procedures. The calibration of a direct relationship between assignment confidence and observation-error statistics would require dedicated assimilation experiments and validation studies, which fall beyond the scope of the present technical note.
We have clarified this limitation and expanded the discussion accordingly.
Manuscript Change:
Section 5: “In data-assimilation applications, assignment confidence could be incorporated through observation-error inflation, confidence-dependent quality control, or adaptive localization strategies. Establishing quantitative relationships between assignment confidence and assimilation uncertainty requires dedicated assimilation experiments and remains the subject of future work.”
Reviewer 2 – Comment 4: “The proposed framework appears strictly limited to grid-based routing. The authors did not address how (or if) this method can be adapted for vector-based routing models, such as the Muskingum method or modern hydro-fabric frameworks.”
Response: We thank the reviewer for this valuable observation. The current implementation was developed specifically for CTRIP because the primary motivation of this work is the assimilation of SWOT observations within the CTRIP-HyDAS framework. Nevertheless, the conceptual methodology is not inherently restricted to gridded routing systems.
The proposed framework is fundamentally based on candidate identification and multi-criteria ranking. In vector-based routing frameworks, river segments or hydro-fabric elements could replace grid cells as assignment targets, while equivalent consistency metrics (e.g., network distance, drainage-area agreement, and flow-direction coherence) could be evaluated within the vector topology.
We have clarified this distinction and expanded the discussion of potential extensions beyond gridded routing models.
Manuscript Change:
Section 5: "Although demonstrated here for a gridded routing framework, the methodology is conceptually transferable to vector-based river-network representations. In such systems, routing segments rather than grid cells would serve as assignment targets, while equivalent geometric and hydrological consistency metrics could be evaluated within the vector-network topology."
Citation: https://doi.org/10.5194/egusphere-2026-509-AC2
-
AC2: 'Reply on RC2', Kaushlendra Verma, 16 Jun 2026
Model code and software
Confidence-Aware Framework for Mapping Satellite-Derived River Reaches to Gridded Routing Networks Kaushlendra Verma, and Simon Munier https://doi.org/10.5281/zenodo.18402332
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 866 | 499 | 92 | 1,457 | 208 | 190 | 325 |
- HTML: 866
- PDF: 499
- XML: 92
- Total: 1,457
- Supplement: 208
- BibTeX: 190
- EndNote: 325
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear authors,
in my role as Executive editor of GMD, I would like to bring to your attention our Editorial version 1.2: https://www.geosci-model-dev.net/12/2215/2019/
This highlights some requirements of papers published in GMD, which is also available on the GMD website in the ‘Manuscript Types’ section: http://www.geoscientific-model-development.net/submission/manuscript_types.html
In particular, please note that for your paper, the following requirement has not been met in the Discussions paper:
In order to simplify reference to your developments, please add the acronym/name of the mapping pipeline and a version number to the title of your article in your revised submission to GMD.
Yours, Astrid Kerkweg