the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Feb 2025
Evaluating Unsaturated Hydraulic Conductivity Models for Diverse Soils and Climates: A Functional Comparison of Additive, Junction, and Kosugi Parameterizations
Abstract. Soil water moves as capillary flow, film flow, and vapour diffusion. The additive model for the unsaturated soil hydraulic conductivity curve adds up conductivities of capillary water, adsorbed water, and an equivalent vapour conductivity. A recently introduced junction model with liquid water in either films or capillaries has one parameter less. We compared calculated water fluxes based on both models and Kosugi’s model (which only considers capillary water) by fitting the RIA soil water retention curve and the three conductivity models to data for three soils. Five subsets of the model parameters were calibrated by fitting, with the other parameters fixed. For all 135 resulting cases, we ran the Hydrus-1D numerical model for uniform columns of these soils subjected to generated weather records for three climates. Hydrus-1D crashed 14 times for the additive model, twice for Kosugi’s model and once for the junction model. The conductivity models and fitting parameter sets only significantly affected the drainage flux at 2 m depth (and, in one case, the transpiration), but the effect was only large in two cases. If the conductivity models disagreed, the additivity model was usually the outlier. An analysis of the water balance terms revealed the impact of soil type to be limited on transpiration or infiltration, but stronger on groundwater recharge. The conductivity models had only a minor effect. The fluxes were insensitive to differences in the dry-range conductivity. Fewer fitted parameters rarely altered the results significantly. This favours the more parsimonious and robust junction and Kosugi models.
Competing interests: At least one of the (co-)authors 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 preprint. The responsibility to include appropriate place names lies with the authors.- Preprint
(9216 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-412', Anonymous Referee #1, 31 Mar 2025
The proposed study compares three different models for the unsaturated hydraulic conductivity curve (UHCC), namely the Kosogi model, the additive and junction model within a numerical exercise. Each of these models is fitted to data obtained for three different soils, using 5 different variations of free fitting parameters, and these are used to simulated the water balance of a 2 m homogeneous soil column for a temperate, Monsoon and semi-arid climate using weather data from sites in Ukkel in Belgium, Tamale in Ghana and Colombo in Sri Lanka. The comparison reveals that most simulated fluxes show a stronger sensitivity to climate, while the drainage flux in 2m shows the strongest sensitivity to changes in soil hydraulic properties.
I fully agree with the authosr, that the ultimate purpose of UHCC is to be used in simulations, and I would highly value a comparison of those different models against independent data obtained in different real world settings using a model of appropriate complexity.
Unfortunately, I cannot confirm that the presented pure model comparison has much scientific value. Both the choice of the soils and the climate setting appear very much ad-hoc, while other findings appear constraint by using a simple 1d model approach.
I also regret to tell, that the presentation quality is neither up to the standards required for a top journal nor does it stick to the HESS specific standards. All should parameters named, clearly defined with their units.
The abstract and the introduction are very technical and appear to be written for an “insider”.
“A recently introduced junction model with liquid water in either films or capillaries has one parameter less.” One parameter less than what?
“We compared calculated water fluxes based on both models and Kosugi’s model (which only considers capillary water) by fitting the RIA soil water retention curve and the three conductivity models to data for three soils”. What is the RIA soil water retention curve?
The manuscript is also not self-contained. The soil hydraulic model should at least be briefly explained. In this context, I wonder how the authors account of water vapor fluxes. The latter needs proper accounting for the heat balance, as saturated water vapor pressure in soil is a function of temperature. And the water vapor diffusion coefficient is a product of the average free path length and the average thermal velocity of vapor molecules. According to kinetic gas theory, both are functions of temperature as well. Does this implies that the model can only be used in numerical models solving the coupled water and heat balance?
Using a 1d model to quantify overland flow and removing water after a ponding height of 1cm appear arbitrary, why not 0.5 cm. This approach is simply too simple!
I also wonder about the authors expectation/ hypothesis when comparing the three soils. Which differences would you expect? A comparative plot of the three retention functions would be helpful to formulate a kind of hypothesis.
I don’t get the point of compare different parameter fitting scenarios. You can and should reject fits that drop below a performance certain threshold. Otherwise optional differences in simulated water balance components are contaminated by deficiency of the fits.
The difference between many water balance components are very small, What about defining on simulation/soil type as reference and looking at differences. This will easier reveal differences.
Citation: https://doi.org/10.5194/egusphere-2025-412-RC1 - AC1: 'Reply on RC1', Gerrit H. de Rooij, 28 May 2025
-
RC2: 'Comment on egusphere-2025-412', Anonymous Referee #2, 17 Apr 2025
General comments
The manuscript provides a functional evaluation of three unsaturated hydraulic conductivity models, namely the Kosugi (KGV), additive (ADV), and junction model (JUV), in simulating the water balance under different soils and climate conditions. To this end, multi-year simulations are conducted with the HYDRUS 1D model, and the results are compared in terms of aggregated mean fluxes at both the upper and bottom boundaries of the soil profile.
The selected functional approach is noteworthy as it allows to compare soil hydraulic models with different characteristics (i.e., considering vapor diffusion) directly in terms of their effects on simulated outputs. However, it should be pointed out that despite being fairly conducted, the study is basically a numerical exercise with a limited link with real data either in terms of the selected soils or simulated water fluxes. In fact, the soil hydraulic conductivity data are gathered from literature while the simulated fluxes are not compared with measured ones, and, therefore, their validity can only be assessed in relative terms. Thus, this study adds a limited contribution beyond what was already investigated in previous studies conducted by the same author(s).
Specific comments
The presentation of the selected hydraulic conductivity models is poor and their properties are difficult to understand if the readers do not refer to the previous studies. I think the manuscript's readability could be improved if the models were presented, even showing the related equations. There are several symbols in text and tables (e.g., hj, Ksc, Ksa, τ, γ) that are not defined when they are introduced for the first time or not defined at all. Furthermore, the choice of using the Kosugi model for comparison should be motivated as well as the advantage of ADV and JUV over classical models should be highlighted.
I hardly understand the rationale for considering the different combinations of fixed and fitted parameters in model simulations. Indeed, if the soil hydraulic conductivity data are measured, the best choice is the one that yields the lowest values of fitting statistics (RMSE, R2, AIC, MBE….), and comparing alternative strategies in terms of model outputs is trivial without independent reference data. However, it should be recognized that the selection of the parameter fitting strategy is a crucial step once the experimental data are obtained and guidelines helping practitioners in this choice are probably lacking but, in my opinion, this point should be addressed with larger soil databases including soil with different origins and characteristics.
Minor comments
L.11: What is RIA?
- 49-64: Is the discussion conducted here strictly necessary for the study? I checked that these points were already discussed in the papers where the models were developed.
L.71: On parameter less than what?
L.84: “all of them including diffusive movement of water vapor”: is it true also for KSG model? If yes, how the water diffusion was accounted for in KSG model?
- 153: What is the KRIAfitter code?
L.155-163: Most of this information was already given in L.90-103. I suggest to unify these sections.
L.214-215: Is this conclusion of general validity? I don’t think so given only three soils were considered.
Citation: https://doi.org/10.5194/egusphere-2025-412-RC2 - AC2: 'Reply on RC2', Gerrit H. de Rooij, 28 May 2025
Data sets
Simulation Results for Hydraulic Conductivity Models Across Three Soils and Three Climates Asha Nambiar and Gerrit H. de Rooij https://doi.org/10.5281/zenodo.14753321
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 191 | 85 | 18 | 294 | 12 | 26 |
- HTML: 191
- PDF: 85
- XML: 18
- Total: 294
- BibTeX: 12
- EndNote: 26
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1