General Comment

This paper presents a range of models for unsaturated soil hydraulic conductivity curves (UHCCs), differing in how they combine approaches across three conceptual domains. Building on the author’s previous work (de Rooij, 2024a), the paper introduces a new model, JUV, alongside earlier models (ADV, AMV, GMV, HMV, KGV), as summarized in lines 399–404. The use of the RIAfitter and KRIAfitter programs for model comparison is commendable, and it is particularly valuable that all data, code, and results are openly provided. The full description of the fitting procedures enhances the reproducibility of the research.

For each model, five parameter fitting strategies were tested, as outlined in lines 370–374. Among these, methods 2 and 5 rely on SWRC parameters that have already been fitted from retention curve data, whereas methods 1, 3, and 4 re-fit all parameters, including those of the SWRC. This is somewhat confusing, given that line 188 states: “Before KRIAfitter 1.0 can be run to determine the values of the parameters of the chosen UHCC model for a particular soil, the parameters of the SWRC of Eqs. (1a–d) need to be fitted using RIAfitter 2.0.” If SWRC parameters are already known, it is unclear why they are re-fitted during UHCC fitting. If these parameters change during the UHCC fitting, then the SWRC curve would no longer match the originally fitted SWRC. This raises the question of whether the SWRC parameters in the SWRC equations are treated as independent from those in the UHCC equations (e.g., is the "n" in SWRC distinct from the "n" in UHCC?). From a modeling standpoint, it would seem more logical to fix the SWRC parameters and only fit the additional UHCC parameters, as is done in methods 2 and 5.

Specific Comments

1. Table 1: Based on RMSE, the most flexible parameter set (method 1) yields the best results in nine cases, while the second-most flexible (method 3) does so in one case. Since optimization aims to minimize RMSE, and method 1 likely encompasses the parameter space of method 3, it is unclear why method 3 would ever outperform method 1. If method 1 includes all of method 3’s parameters, then the optimal set found by method 3 should also be attainable by method 1. This discrepancy could be due to method 1’s parameter space not actually covering that of method 3, or because the fitting algorithm failed to locate the global optimum within the broader space of method 1.
2. As mentioned in the general comment, I believe that fixing all SWRC parameters during UHCC fitting is a reasonable approach. In this context, method 2 fixes the SWRC parameter α, but only method 5 fixes all SWRC parameters. I would be interested in seeing more variations on this fixed-parameter approach—such as versions assuming Assouline's τ = 0.0 or Mualem’s γ = 2.0—alongside method 5. The author’s model in fact resembles the model proposed by Peters (2013), which originally used γ = 2.0. The equation involving γ and τ is specifically intended to describe the capillary range. If the focus is solely on the capillary range, as in this study, it may be sufficient to fix γ = 2.0, without fitting both γ and τ. This could be tested as a simplification, though it is ultimately up to the author whether to include such comparisons.

Technical Comment

1. Equation 5a: Is the parameter “a” defined somewhere? Should this be “Dₐ”?

References

• Peters (2013): http://doi.org/10.1002/wrcr.20548
• Seki et al. (2023): https://doi.org/10.2478/johh-2022-0039

The manuscript presents a new junction model (JUV) to parameterize the unsaturated soil hydraulic conductivity curve (UHCC). The model assumes a switch from capillary flow to film flow at a critical matric head. In addition the manuscript documents a Fortran code, KRIAfitter, that estimates the parameters in the new JUV given measured values of hydraulic conductivity and matric head/water content. This code also includes several other UHCC models.

The idea of the junction model presents an interesting simplification for combining capillary flow and film flow. The provided fortran code for fitting of UHCC models ensures reproducibility and can be a useful tool for the community.

Presenting both, the new model and the code in one manuscript is challenging and unfortunately not successful. Some necessary information is missing or provided later in the manuscript. The manuscript needs to be reworked carefully to be able to review it. Below are some comments that may help improve the manuscript.

Specific comments:

Lines 86-87: “This correction needs to be applied to hd in the logarithmic branch of the SWRC as shown in Eq. (1a). No correction is needed in the UHCC.” Equations 2 and 6 show c as well. So c is used in the UHCC. Please clarify.

Lines 98-109: I agree that “Simple addition of film and capillary conductivities may therefore not be accurate.” However, the proposed simplification is clearly not accurate also. This should be discussed as well.

Equation 5a: What is a? Is this supposed to be Da? Please introduce Da then.

Equations 5b-d: units don’t match

Lines 167: is hd not fitted? Please clarify.

Lines 188-192: The split into two separate programs seems to limit the application. Could it make sense to fit alpha and n based on both the SWRC and UHCC?

Line 196: If alpha and n are already fitted as part of the SWRC, why are they not fixed? Or is this an iterative process?

Line 204-205: it is not clear what these parameters are.

Line 226 and following: Why is the SCE algorithm needed to determine the optimal parameters. This is a global algorithm to avoid local minima. Why is it not possible to use a simple gradient based method?

Line 232 (as an example): Very specific settings for the code like “but only if input variable FewComplexes is set to ‘T’ on input” should be limited to the user manual.

Line 356: What pressure plate data is referred to here?

Lines 370-374: Here the 5 different sets of parameters that were fitted are introduced. They are later referred to inconsistently. Sometimes as set 1-5, sometimes as ‘Mualem’ or ‘alpha only’. Please be consistent and introduce the naming convention here.

It is not clear what “Ks,a (when applicable means)”. Ks,a is calculated using equation 4.

In set 2: what about parameter n? Set 2 also lists hd to be used from the SWRC, the other lists don’t mention it.

Line 388: pF was not introduced

Lines388-390: This is unclear. What lower weight?

Lines 390-393: Where do these specific values for two soils come from?

Lines 399-404: Here the different methods the JUV is compared against are introduced. This should be done in the methods section. Additionally, this lacks crucial information. The methods all refer to (de Rooij, 2024a). Does this mean these are all recently introduced methods? The Kosugi model is established. What are the modifications? Please clarify. Also, why not compare to a simple established method?

The following discussion of the results is difficult to follow. Please at least refer to the Figures with their Figure number.

What is the evaluation if fits are good or not based on? Quantitative measures? Qualitative features? If so, which features?

Lines 454-455: This seems to be related to the range the observations cover. To me to does not make sense to include datasets that do not cover the dry range. Please clarify.

Line 462: Here the Akaike’s Information Criterion is introduced. It is later implied that this accounts for the number of parameters fitted. Please introduce earlier in the methods.

Line 480-484: Here the range of the underlying data is discussed. It could make sense to discuss the ranges also in terms of the calculated hj. This would allow the reader to better understand these ranges.

Figure 1: please add to the caption which method was used for each of the graphs.