the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Jan 2025
Multi-variable process-based calibration of a behavioural hydrological model
Abstract. Behavioural hydrological modelling aims not only at predicting the discharge of an area within a model, but also at understanding and correctly depicting the underlying hydrological processes. Here, we present a new approach for the calibration and evaluation of water balance models, exemplarily applied to the Riveris catchment in Rhineland-Palatinate, Germany. For our approach, we used the behavioural model WaSiM. The first calibration step is the adjustment of the evapotranspiration (ETa) parameters based on MODIS evaporation data. This aims at providing correct evaporation behaviour of the model and at closing the water balance at the gauging station. In a second step, geometry and transmissivity of the aquifer are determined using the Characteristic Delay Curve (CDC). The portion of groundwater recharge was calibrated using the Delayed Flow Index (DFI). In a third step, inappropriate pedotransfer functions (PTFs) could be filtered out by comparing dominant runoff process patterns under a synthetic precipitation event with a soil hydrological reference map, Then, the discharge peaks were adjusted based on so-called signature indices. This ensured a correct depiction of high-flow volume in the model. Finally, the overall model performance was determined using signature indices and efficiency measures. The results show a very good model fit with values for the NSE of 0.88 and 0.9 for the KGE in the calibration period and an NSE of 0.81 and a KGE of 0.89 for the validation period. Simultaneously, our calibration approach ensured a correct depiction of the underlying processes (groundwater behaviour, runoff patterns). This means that our calibration approach allows selecting a behaviourally faithful one from many possible parameterisation variants.
- Preprint
(3367 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 05 Mar 2025)
-
RC1: 'Comment on hess-2024-369', Dan Myers, 23 Jan 2025
reply
Reviewer comments for hess-2024-369
This study calibrated a hydrologic model not just for discharge, but including other water balance components such as evapotranspiration and groundwater flow. I found it to be an interesting approach, and suitable for HESS. The modeling approach seems solid to me, but I think more work is needed to fit it within the state of existing literature and really piece out the novelties of the work. My comments below aim to find ways to address this concern, and thus are mostly for the introduction and discussion.
Major Comments:
Page 2, lines 27-28. I think it would be helpful for the readers to describe the “why” other water balance components besides discharge are neglected. For instance, you could add a second clause to the sentence explaining that it takes additional resources and data, the researchers incorrectly deem them as nonessential to their research questions, or whichever reasons there are.
Page 2, lines 34-40. I think this paragraph about behavioural modelling should be expanded. In its current form, it is very light about literature for previous behavioural and multi-variable calibration approaches. For instance, hydrologic models have previously been calibrated for discharge and evapotranspiration, or discharge and groundwater, or various water balance combinations together, as you mention in the following paragraphs. There are existing calibration programs, such as R-SWAT (Nguyen et al., 2022) or AMALGAM (Vrugt and Robinson, 2007), that can incorporate multiple parts of the water balance into calibrations. Similarly, Dangol et al. (2023) is one paper about the importance of calibrating multiple variables. What is the state of literature regarding these approaches? This will help later set up the gap in knowledge or modeling capabilities that this research will fill.
Page 3, lines 61-62. To me, this is a very important sentence in the paper, identifying the research gap and why it is important. However, in its current form it is not very convincing to me, and I suggest re-conceptualizing it. The current sentence basically states that while other studies have calibrated discharge with ET, or with groundwater or other processes, this study also does that but in a different combination that hadn’t been matched exactly yet in the literature. Instead, I suggest explaining the importance here of doing the combination of calibrations used in this work, compared to those previous literature has tested, as making a significant advancement beyond the current state of knowledge – say, for simulating a land use or climate change correctly. Really highlight the bigger novelty in this sentence and why it is important for HESS readers.
Page 3, lines 64-65. The intro mentions some challenges to getting groundwater flow and evapotranspiration right. However, I think there should be more information, perhaps, a paragraph, about why this is important for modelers. To clarify, I’m not saying it isn’t important, just that it should be explained more. For example, say someone calibrated a model on discharge alone and deemed it to have satisfactory performance (NSE > 0.7, etc.). Why would they then go through the additional steps of this approach, particularly if they’re short on budget or time? To help alleviate this concern, I think you could explain more about how models calibrated on discharge alone can behave differently when simulating land or climate changes if they don’t get the other water balance components right.
Page 26, lines 499-501. I think here, it would be important to further discuss how your approach works in relation to previous literature about multivariable calibration approaches. For instance, is it more effective performance-wise? More user friendly? More broadly applicable to a diverse group of models? Further discussing the benefits and potential limitations of the new calibration approach would be helpful here, as well as referencing more previous literature.
Page 27, line 507 to page 28, line 516. To help convince readers, I think it would be helpful to expand this transferability section to include more specifics about how this approach is transferable to different models and catchments. You could demonstrate this in a few examples of different models or modeling approaches. For instance, would the scripts have to be adapted to other models like SWAT or CLM? What about hydrologic models like used in the airGR package for R, which can be fit using an automated approach? Or, catchments that are dominated by snowpack and snowmelt processes? I think by giving more specific insights here, you can help readers imagine replicating the approach with the models and catchments they use.
Also in the transferability and outlook section, or somewhere similar, I think you need to discuss very thoroughly why someone would go through the extra time and effort to use this approach, compared to a discharge-only calibration. Particularly for studies that are low on budget, computational resources, or time, what are the costs and benefits of going through these extra steps? Are there software that could make the additional steps quick and easy? Or, is the problem with the traditional approach substantial enough that the additional steps are necessary for accurate hydrological simulations? For clarification, I think the new approach you developed is very important, and you just need to communicate the value further. One idea would be mentioning the importance of the additional steps for getting model parameters correctly to simulate a land use or climate change. You could mention studies that have simulated land use or climate changes (e.g., Botero-Acosta et al., 2022) and how they could benefit from the parameterization steps you developed.
Page 28, line 525 to page 29, line 540. I suggest rewriting the conclusions to be more about the specific novelties of this work. In its current form, the conclusion is that the multi-criteria calibration approach is better for capturing underlying hydrological processes for ETa, groundwater, etc. But…that was already stated in the introduction to the paper while reviewing previous literature. The conclusion section should focus more on the advancements of this work beyond what was previously known, and the implications for that. I think a good piece for this to expand upon would be the last sentence about having models that can more reliably simulate varied hydrological conditions. That seems like it could be an important take-away to elaborate on, potentially relating back to land and climate change simulation.
Minor Comments:
Page 2, lines 29-30. In addition to not being suitable for investigating underlying processes, you could add that the models could also behave in unintended ways when a land or climate change is simulated.
Lines 107-108. How does the model and this calibration approach incorporate snowmelt? I suggest adding a sentence to help readers understand its applicability in snowmelt dominated catchments.
Page 17, line 333 to page 18, line 342. I think the KGE, NSE and R2 metrics are used commonly enough that you don’t need to define the equations for them, and could remove these lines.
I wish the authors best of luck with this paper and their future endeavors.
References:
Botero-Acosta, A., Ficklin, D. L., Ehsani, N., & Knouft, J. H. (2022). Climate induced changes in streamflow and water temperature in basins across the Atlantic Coast of the United States: An opportunity for nature-based regional management. Journal of Hydrology: Regional Studies, 44, 101202. https://doi.org/10.1016/j.ejrh.2022.101202
Dangol, S.; Zhang, X.; Liang, X.-Z.; Anderson, M.; Crow, W.; Lee, S.; Moglen, G.E.; McCarty, G.W. Multivariate Calibration of the SWAT Model Using Remotely Sensed Datasets. Remote Sens. 2023, 15, 2417. https://doi.org/10.3390/rs15092417
Nguyen, T. V., Dietrich, J., Dang, D. T., Tran, D. A., Doan, B. V., Sarrazin, F. J., Abbaspour, K., Srinivasan, R. (2022). An interactive graphical interface tool for parameter calibration, sensitivity analysis, uncertainty analysis, and visualization for the Soil and Water Assessment Tool. Environmental Modelling & Software, 156, 105497. https://doi.org/10.1016/j.envsoft.2022.105497.
Vrugt, Jasper A., and Bruce A. Robinson. "Improved evolutionary optimization from genetically adaptive multimethod search." Proceedings of the National Academy of Sciences 104.3 (2007): 708-711. https://doi.org/10.1073/pnas.0610471104
Citation: https://doi.org/10.5194/hess-2024-369-RC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 56 | 8 | 2 | 66 | 0 | 0 |
- HTML: 56
- PDF: 8
- XML: 2
- Total: 66
- BibTeX: 0
- EndNote: 0
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1