Quantifying key parameter sensitivities for water table depth in hydrological schemes of CoLM-PSUADE

Wu, Tingting; Zhang, Shupeng; Yang, Xiaofan; Dai, Yongjiu

doi:10.5194/egusphere-2026-2275

Preprints

https://doi.org/10.5194/egusphere-2026-2275

Preprints

23 Apr 2026

| 23 Apr 2026

Quantifying key parameter sensitivities for water table depth in hydrological schemes of CoLM-PSUADE

Tingting Wu, Shupeng Zhang, Xiaofan Yang, and Yongjiu Dai

Abstract. Accurately representing groundwater dynamics in land surface models (LSMs) is crucial for understanding water-energy cycles and assessing water resources. However, most LSMs lack systematic sensitivity analyses of parameters regulating water table depth (WTD). This study couples the Common Land Model (CoLM) with Problem Solving environment for Uncertainty Analysis and Design Exploration (PSUADE) in a single-point framework to facilitate systematic parameter analysis and calibration aimed at improving WTD simulation. The CoLM-PSUADE framework was then applied to evaluate groundwater-related parameters using WTD observations from the Gongga Mountain site. A comprehensive analysis integrating qualitative sensitivity analysis, quantitative sensitivity analysis, and parameter optimization techniques was conducted to evaluate the sensitivity of 56 parameters associated with key hydrological processes and to determine their optimal ranges. The results indicate that eight parameters can be identified as robustly sensitive, including those controlling unsaturated soil water movement (56-soil_alpha, 53-soil_n), subsurface runoff (40-rsubmax), plant hydraulic processes (49-beta, 45-krmax, 46-ck0), and net surface water infiltration (4-alpha_rain, 10-rhol_nir). Among them, the subsurface runoff parameter r_sub,max exhibits a well-defined optimal range (on the order of 10⁻⁴) and can regulate both the magnitude of subsurface runoff and its decay with increasing WTD when combined with another empirical parameter in the SIMTOP (Simple TOPMODEL-based) scheme, f_drai, thereby exerting strong control on WTD. The soil hydraulic parameter α shows the highest sensitivity. It regulates unsaturated hydraulic conductivity and soil water retention, thereby exerting a dominant influence on the variability and lagged response of WTD. Based on these findings, a stepwise calibration strategy is recommended, in which the subsurface runoff parameters (r_sub,max and f_drai) are first adjusted to constrain the mean WTD, followed by optimization of other key parameters, such as α, to improve the temporal dynamics of WTD. It is demonstrated that CoLM-PSUADE provides a useful tool for sensitivity-guided parameter optimization in high-dimensional LSMs and hydrological models.

Received: 22 Apr 2026 – Discussion started: 23 Apr 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, 5151 KB)

Supplement (3830 KB)

Download & links

Tingting Wu, Shupeng Zhang, Xiaofan Yang, and Yongjiu Dai

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-2275', Anonymous Referee #1, 01 Jun 2026
The authors have developed and applied a CoLM2024–PSUADE coupling framework to examine 56 hydrologically relevant parameters using multiple sensitivity-analysis methods. They further integrate parameter screening, quantitative sensitivity analysis, and parameter optimization to evaluate and improve the simulation of water table depth at a mountain forest site. Overall, I appreciate the considerable effort invested in model coupling, experimental design, and the subsequent analyses. The study addresses an important topic and presents a potentially useful workflow for sensitivity-guided calibration in land surface models. However, several aspects of the experimental design and the interpretation of the results would benefit from further clarification and strengthening before publication.
Major comments:
The manuscript emphasizes two main contributions: the development of a CoLM2024–PSUADE coupling tool and the robust sensitivity analysis of water table depth-related parameters. The coupling framework is useful from a technical and practical perspective. However, the sensitivity-analysis results for key parameters are derived from a single-point simulation at one mountain forest site. Although this site-specific analysis may provide useful insights into WTD simulation under the local hydroclimatic and soil conditions, it is difficult to infer that the identified sensitive parameters or their optimal ranges are generally applicable to other regions, land-cover types, climate regimes, or groundwater conditions. Therefore, I recommend that the authors either expand the analysis to additional sites or grid cells. If additional multi-site experiments are not feasible, the conclusions should be refocused on the current watershed/site. Thus, the current title of the manuscript is inappropriate.

The manuscript states that the study was conducted at a long-term integrated observation site, which is an important advantage of the dataset. However, the sensitivity analysis appears to be based on three years of data (2005-2007). The length and representativeness of the analysis period may be critical because WTD dynamics and parameter sensitivity may vary substantially. I recommend that the authors extend the sensitivity analysis to cover a longer observation period, or at least justify why the selected three-year period is representative of the long-term hydrological conditions at the site.

The qualitative sensitivity-analysis results show some variability across different methods, sampling techniques, sample sizes, and objective functions. In particular, Figures 7 and 8 suggest that 56-soil_alpha is the most consistently identified sensitive parameter, whereas the rankings of several other parameters vary among experimental configurations and with increasing sample size. This variability is understandable given the high dimensionality of the parameter space and the stochastic nature of the sampling procedures, but it would be helpful to further clarify the robustness of the identified sensitive parameters. I suggest that the authors consider adding repeated stochastic sensitivity-analysis experiments under the same settings, for example by using multiple random seeds for selected method–sampling technique–sample-size–objective-function combinations. The robustness of the sensitive parameters could then be evaluated using statistical indicators such as the mean and standard deviation of sensitivity indices, confidence intervals, coefficient of variation, selection frequency across repeated realizations.

The role and notation of the subsurface runoff decay factor require further clarification. In Table 2, the 56 adjustable parameters include 38-fsatdcfa, which is described as a runoff decay parameter, but the later discussion refers to the decay factor as f_drai. It is therefore unclear whether f_draiis identical to 38-fsatdcfa or represents a separate parameter. I suggest that the authors clarify the relationship between 38-fsatdcfa and f_drai.

The optimization experiment design and the interpretation of the optimized WTD simulations would benefit from further clarification. The manuscript compares SCE, PSO, and BOBYQA using a fixed budget of 500 evaluations, but it is not entirely clear whether this number is sufficient to assess optimizer convergence and robustness in a multi-parameter calibration problem. This is particularly relevant because the manuscript reports that PSO does not show a clear convergence trend, while BOBYQA exhibits stronger fluctuations after about 300 evaluations. I suggest that the authors provide a clearer justification for the 500-evaluation limit and, if computationally feasible, consider additional optimization experiments with larger evaluation budgets, for example 1000–2000 evaluations, or repeated runs using different random seeds. At minimum, the authors should discuss the potential limitations of comparing optimization methods under this relatively limited evaluation budget and report the convergence criteria more explicitly.

In addition, Figure 16 suggests that although the SCE-optimized simulation improves the overall WTD level, it does not fully reproduce the two pronounced WTD declines observed in 2007. This point deserves further discussion, because capturing such event-scale or seasonal drawdown dynamics is important for evaluating whether the optimized parameters improve not only the mean WTD but also the temporal variability. The authors should clarify whether these discrepancies are related to limitations in the runoff/groundwater parameterization, freeze-thaw processes, meteorological forcing, observational uncertainty, or the optimization objective function. It would also be helpful to add event-based or seasonal performance metrics to quantify model performance during these pronounced drawdown periods.

Minor comments:
L16: In the Abstract, the phrase “sensitivity analysis” is repeated several times in the sentence

L281: Please double-check the formula labelled as as it appears to correspond to the Nash–Sutcliffe efficiency (NSE)
Citation: https://doi.org/10.5194/egusphere-2026-2275-RC1
- AC1: 'Reply on RC1', Tingting Wu, 15 Jun 2026
  
  We sincerely thank the reviewer for the time and effort dedicated to reviewing our manuscript and for providing constructive comments. We are encouraged by the reviewer’s positive remarks regarding the potential usefulness of our coupling framework for sensitivity-guided calibration in land surface models. In addition, the reviewer rightly points out key challenges faced by our study, concerning the generalizability of the identified parameters to other conditions, the robustness of sensitivity analysis, the sufficiency and convergence of optimization experiments, and the clarity of conclusions and parameter definitions. We agree with this diagnosis and will systematically address these issues.
  In the revised manuscript, we will implement comprehensive improvements targeting the problems highlighted by the reviewer:
  1. Improving alignment between content and conclusions and clarifying limitations
  (1) Title and scope updates: To align with the current site-specific content, we will revise the manuscript title to clearly indicate that the results are related to the selected mountain site.
  (2) Conclusions justification: We will modify the conclusions to focus on the current site and emphasize the limitation that the ranking of sensitive parameters is specific to the local hydrological and climatic conditions. The intended scope and limitations of the study will be clarified accordingly.
  (3) Water year selection clarification: Considering the relative stability of measured water table depth dynamics across the chosen years and the computational demands of multiple experiments, we have chosen data in the 2006-2007 period for analysis. We will expand Section 2.1.2 to clarify the representativeness of this period with respect to long-term hydrological conditions at the site.
  2. Strengthening robustness of interpretations
  (1) Clarifying the robustness of the identified sensitive parameters: We will try to perform two additional sensitivity-analysis experiments using different random seeds. Together with the existing experiment, this will yield three realizations in total. Based on these, we will compute the suggested statistical indicators to evaluate parameter robustness. In addition, we will attempt, if computationally feasible, to expand the qualitative analysis to an additional site on the Tibetan Plateau, which is covered by grassland, another dominant land-cover type besides forest (Wang et al., 2022), to discuss the robustness of the identified sensitive parameters.
  (2) Addressing insufficiencies in optimization evaluations: We will expand optimization experiments to 1000 evaluations in Section 3.3, discuss the potential limitations of comparing optimization methods under relatively limited evaluation budget, and report the convergence criteria more explicitly in the Appendix. In addition, we will also perform a clear assessment about convergence of selected optimizers from the perspective of the principle of the methods. In addition, we acknowledge that the SCE-optimized simulation does not fully capture the two pronounced WTD declines observed in 2007, and we agree that this deserves further discussion. In the revised manuscript, we will add a dedicated subsection in the Discussion section to further analyze the performance of the parameter optimization.
  3. Enhancing content and equation clarity
  (1) Abstract revision: We will revise the sentence to avoid the repetition of “sensitivity analysis”.
  (2) Equation correction: We thank the reviewer for catching this inconsistency. The formula previously labeled as R² is indeed the Nash–Sutcliffe efficiency (NSE), where the denominator is normalized by the variance of the observed values (Duc and Sawada, 2023). We will correct the label throughout the manuscript, update all related descriptions accordingly, and add an explicit clarifying comment in the publicly released code to avoid further confusion.
  (3) Parameter clarification: We will clarify the relationship between 38-fsatdcfa and f_drai in Section 4.2.
  We thank the reviewer for these helpful comments again. We believe these planned systematic revisions will directly resolve the issues raised and significantly enhance the scientific rigor and transparency of our study. A detailed point-by-point response will be provided in the revised manuscript.
  References
  
  Duc, L. and Sawada, Y.: A signal-processing-based interpretation of the Nash–Sutcliffe efficiency, Hydrol. Earth Syst. Sci., 27, 1827-1839, https://doi.org/10.5194/hess-27-1827-2023, 2023.
  
  Wang, Y., Lv, W., Xue, K., Wang, S., Zhang, L., Hu, R., Zeng, H., Xu, X., Li, Y., Jiang, L., Hao, Y., Du, J., Sun, J., Dorji, T., Piao, S., Wang, C., Luo, C., Zhang, Z., Chang, X., Zhang, M., Hu, Y., Wu, T., Wang, J., Li, B., Liu, P., Zhou, Y., Wang, A., Dong, S., Zhang, X., Gao, Q., Zhou, H., Shen, M., Wilkes, A., Miehe, G., Zhao, X., and Niu, H.: Grassland changes and adaptive management on the Qinghai–Tibetan Plateau, Nat. Rev. Earth Environ., 3, 668-683, https://doi.org/10.1038/s43017-022-00330-8, 2022.
  
  Citation: https://doi.org/10.5194/egusphere-2026-2275-AC1
RC2:
'Comment on egusphere-2026-2275', Anonymous Referee #2, 04 Jun 2026

This study employed the Common Land Model (CoLM) to find the most sensitive parameters for water table depth (WTD) in the Gongga Mountain site, and to calibration the most sensitive parameters aimed at improving WTD simulation. This study suggested that CoLM-PSUADE was a useful method to find the most sensitive parameters, and the most sensitive parameters could be used to improve the temporal dynamics of WTD.
(1）The language in the text needs to be revised, particularly with regard to verb tense.
（2）The methodology section of the paper needs to be expanded. Further details are required regarding the parameter sensitivity analysis methods.
（3）Classifying the physical parameters in the model will aid the reader’s understanding. For example, parameters could be categorised into soil parameters and vegetation parameters, and so on.
（4）Please provide a more detailed explanation of the physical significance of the sensitive parameters.
（5）A full explanation is required as to why the parameter sensitivity analysis and parameter optimisation are being carried out concurrently.

Citation: https://doi.org/10.5194/egusphere-2026-2275-RC2
- AC2: 'Reply on RC2', Tingting Wu, 15 Jun 2026
  
  We sincerely thank the reviewer for the careful reading of our manuscript and for the suggestions. We are encouraged by the reviewer’s positive overall assessment. The detailed feedback will help us improve the presentation of the manuscript. In the revised manuscript, we will provide a point-by-point response to all comments. Below, we summarize the main aspects that we will address.
  1. Improving language consistency and scientific writing
  (1) Verb tense revision: We will thoroughly revise the manuscript to ensure proper use of tenses. Specifically, we will use the past tense for describing what we did, present tense for established facts or general truths, and present perfect tense for describing previous research or ongoing relevance.
  (2) Overall language polishing: A line-by-line check will be performed to ensure that all scientific claims are expressed accurately.
  2. Enhancing methods clarity
  (1) Detailed sensitivity analysis methods: We have already provided the principles and formulations of all sensitivity analysis methods in the Appendix A. In the revised manuscript, we will additionally expand Section 2.3.1 to summarize the characteristics of each method.
  (2) Detailed relationships between sensitivity analysis and parameter optimization: We will add additional explanations in Section 2.3 to emphasize that parameter sensitivity analysis and parameter optimization are not performed independently or concurrently, but rather sequentially. Specifically, sensitivity analysis is first conducted to identify the most influential parameters and to reduce the dimensionality of the parameter space. Subsequently, parameter optimization is performed only on the screened sensitive parameters to improve computational efficiency and to further validate the effectiveness of the identified sensitive parameters through improved simulation performance.
  3. Improving parameter interpretations
  (1) Classifying physical parameters: We will reorganize the presentation of parameters by introducing explicit categorization in Section 2.2.2.
  (2) Detailed explanation of sensitive parameters: For the sensitive parameters, we will expand Section 4.1 to explain their physical meaning and mechanistic link to water table depth dynamics.
  We thank the reviewer for these helpful comments again. We believe these planned systematic revisions will directly resolve the issues raised and significantly enhance the scientific rigor and transparency of our study. A detailed point-by-point response will be provided in the revised manuscript.
  
  Citation: https://doi.org/10.5194/egusphere-2026-2275-AC2

Tingting Wu, Shupeng Zhang, Xiaofan Yang, and Yongjiu Dai

Supplement

https://doi.org/10.5194/egusphere-2026-2275-supplement

Model code and software

Quantifying key parameter sensitivities for water table depth in hydrological schemes of CoLM-PSUADE Tingting Wu, Shupeng Zhang, Xiaofan Yang, and Yongjiu Dai https://doi.org/10.5281/zenodo.19641291

Tingting Wu, Shupeng Zhang, Xiaofan Yang, and Yongjiu Dai

Viewed

Total article views: 382 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
248	114	20	382	35	8	8

HTML: 248
PDF: 114
XML: 20
Total: 382
Supplement: 35
BibTeX: 8
EndNote: 8

Views and downloads (calculated since 23 Apr 2026)

Month	HTML	PDF	XML	Total
Apr 2026	126	40	10	176
May 2026	105	62	6	173
Jun 2026	17	12	4	33

Cumulative views and downloads (calculated since 23 Apr 2026)

Month	HTML	PDF	XML	Total
Apr 2026	126	40	10	176
May 2026	105	62	6	173
Jun 2026	17	12	4	33

Viewed (geographical distribution)

Total article views: 375 (including HTML, PDF, and XML) Thereof 375 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 15 Jun 2026

Short summary

We coupled a land surface model (CoLM) with an analysis tool (PSUADE) to evaluate processes controlling water table depth (WTD) through sensitivity analysis and parameter optimization. A small set of sensitive parameters governing WTD was identified, showing subsurface runoff controls average groundwater levels, while soil properties govern variability. A stepwise calibration was then suggested to improve accuracy. This framework offers a transferable pathway to enhance groundwater simulation.


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