the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A reference basin based framework for identifying and transferring permafrost hydrological coupling regimes across China
Abstract. Permafrost degradation is altering runoff generation, subsurface storage, and hydrological connectivity across cold region river basins. Scaling process understanding from well-studied basins to broader permafrost regions remains challenging because the subsurface hydrothermal controls that organize permafrost hydrological coupling are rarely observed at large spatial scales. This study develops a reference basin approach for identifying and transferring permafrost hydrological coupling regimes across China using simplified, broadly available, and process relevant proxy variables. Using the Source Region of the Yellow River as a process constrained reference basin, we construct a two-dimensional process space with two coordinates, dominant control depth DCD and coupling strength C. DCD describes the relative vertical position of hydrological control, while C represents the integrated magnitude of permafrost hydrological linkage. Temporal trajectories in this space are clustered into four coupling regime archetypes. These archetypes are then transferred to broader permafrost regions using a parsimonious proxy set that represents climatic background, topographic setting, and vertically integrated soil thermal organization. To define the applicability boundary, out-of-distribution screening identifies regions where target environmental conditions depart from the reference basin support. The results show coherent regional patterns of coupling regimes across the Qinghai Tibetan Plateau and northeastern permafrost regions, while out-of-distribution areas indicate where the transferred classification is less directly interpretable. The framework provides a bounded and physically interpretable pathway for scaling locally constrained coupling archetypes to broader data sparse regions.
- Preprint
(1184 KB) - Metadata XML
-
Supplement
(640 KB) - BibTeX
- EndNote
Status: open (until 12 Aug 2026)
-
RC1: 'Comment on egusphere-2026-1989', Anonymous Referee #1, 12 Jul 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-1989/egusphere-2026-1989-RC1-supplement.pdfReplyCitation: https://doi.org/
10.5194/egusphere-2026-1989-RC1 -
CC1: 'Comment on egusphere-2026-1989', Nima Zafarmomen, 13 Jul 2026
reply
The manuscript develops a reference-basin framework for identifying and transferring permafrost–hydrological coupling regimes across China. Using the Source Region of the Yellow River as the reference basin, the authors define a two-dimensional process space based on dominant control depth and integrated coupling strength. Temporal trajectories in this space are clustered into four coupling archetypes, which are subsequently transferred to other Chinese permafrost regions using a generalized additive model and environmental predictors such as precipitation, latitude, elevation, and ERA5-Land-derived soil thermal indicators. Out-of-distribution screening is used to identify locations where environmental conditions are insufficiently represented by the reference basin. The study provides an interesting and potentially useful framework for regionalizing process-based hydrological information in data-sparse permafrost environments.
- The manuscript should clarify the novelty of the proposed framework relative to conventional hydrological regionalization, environmental similarity classification, and machine-learning transfer approaches. The authors should explain more explicitly how the DCD–C trajectory framework provides information that cannot be obtained from static permafrost or hydroclimatic classifications.
- The selection of a fixed three-month lag for calculating depth-dependent coupling requires stronger justification. Although a sensitivity analysis is mentioned, hydrological response times may vary spatially, seasonally, and among permafrost regimes. The authors should discuss whether using one lag for all grid cells could influence the estimated dominant control depth and coupling strength.
- The choice of four clusters appears to rely partly on physical interpretability. The manuscript would benefit from a clearer quantitative justification using cluster-validity statistics, stability analysis, or resampling. The authors should also indicate whether the four archetypes remain stable under reasonable changes in window length, time step, or clustering initialization.
- The authors are strongly recommended to cite recent work demonstrating the value of assimilating remotely sensed environmental information into integrated hydrological models. In particular, Zafarmomen, N., Alizadeh, H., Bayat, M., Ehtiat, M., and Moradkhani, H; is relevant to the manuscript’s broader discussion of transferring and constraining hydrological process information using spatially distributed observations.
- The national transfer model achieves an overall accuracy of 58.15% and a Kappa coefficient of 0.41. These values indicate only moderate predictive agreement. The limitations of this performance should be emphasized more clearly, especially when interpreting detailed regional patterns and temporal transitions outside the reference basin.
- The out-of-distribution screening is an important component of the framework, but its implementation is not described in sufficient detail. The authors should specify the statistical distance, threshold, or predictor-space criterion used to identify OOD grid cells and provide a sensitivity analysis showing how the mapped applicability boundary changes with the selected threshold.
- Some physical interpretations of regime transitions appear stronger than the available evidence supports. For example, transitions among the four statistical archetypes are interpreted as changes in active-layer storage, groundwater connectivity, and thaw-driven flow paths. These interpretations should be presented more cautiously unless they can be validated using independent observations of active-layer thickness, groundwater, soil moisture, or baseflow.
Citation: https://doi.org/10.5194/egusphere-2026-1989-CC1 -
CC2: 'Reply on CC1', Nima Zafarmomen, 13 Jul 2026
reply
The manuscript develops a reference-basin framework for identifying and transferring permafrost–hydrological coupling regimes across China. Using the Source Region of the Yellow River as the reference basin, the authors define a two-dimensional process space based on dominant control depth and integrated coupling strength. Temporal trajectories in this space are clustered into four coupling archetypes, which are subsequently transferred to other Chinese permafrost regions using a generalized additive model and environmental predictors such as precipitation, latitude, elevation, and ERA5-Land-derived soil thermal indicators. Out-of-distribution screening is used to identify locations where environmental conditions are insufficiently represented by the reference basin. The study provides an interesting and potentially useful framework for regionalizing process-based hydrological information in data-sparse permafrost environments.
- The manuscript should clarify the novelty of the proposed framework relative to conventional hydrological regionalization, environmental similarity classification, and machine-learning transfer approaches. The authors should explain more explicitly how the DCD–C trajectory framework provides information that cannot be obtained from static permafrost or hydroclimatic classifications.
- The selection of a fixed three-month lag for calculating depth-dependent coupling requires stronger justification. Although a sensitivity analysis is mentioned, hydrological response times may vary spatially, seasonally, and among permafrost regimes. The authors should discuss whether using one lag for all grid cells could influence the estimated dominant control depth and coupling strength.
- The choice of four clusters appears to rely partly on physical interpretability. The manuscript would benefit from a clearer quantitative justification using cluster-validity statistics, stability analysis, or resampling. The authors should also indicate whether the four archetypes remain stable under reasonable changes in window length, time step, or clustering initialization.
- The authors are strongly recommended to cite recent work demonstrating the value of assimilating remotely sensed environmental information into integrated hydrological models. In particular,Assimilation of Sentinel-based leaf area index for modeling surface–ground water interactions in irrigation districts, Water Resources Research Zafarmomen, N., Alizadeh, H., Bayat, M., Ehtiat, M., and Moradkhani, H; is relevant to the manuscript’s broader discussion of transferring and constraining hydrological process information using spatially distributed observations.
- The national transfer model achieves an overall accuracy of 58.15% and a Kappa coefficient of 0.41. These values indicate only moderate predictive agreement. The limitations of this performance should be emphasized more clearly, especially when interpreting detailed regional patterns and temporal transitions outside the reference basin.
- The out-of-distribution screening is an important component of the framework, but its implementation is not described in sufficient detail. The authors should specify the statistical distance, threshold, or predictor-space criterion used to identify OOD grid cells and provide a sensitivity analysis showing how the mapped applicability boundary changes with the selected threshold.
- Some physical interpretations of regime transitions appear stronger than the available evidence supports. For example, transitions among the four statistical archetypes are interpreted as changes in active-layer storage, groundwater connectivity, and thaw-driven flow paths. These interpretations should be presented more cautiously unless they can be validated using independent observations of active-layer thickness, groundwater, soil moisture, or baseflow.
Citation: https://doi.org/10.5194/egusphere-2026-1989-CC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 8 | 4 | 1 | 13 | 0 | 1 | 0 |
- HTML: 8
- PDF: 4
- XML: 1
- Total: 13
- Supplement: 0
- BibTeX: 1
- EndNote: 0
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1