Review of the revised paper entitled:
From hydraulic root architecture models to macroscopic representations of root hydraulics in soil water flow and land surface models
by: Jan Vanderborght, Valentin Couvreur, Felicien Meunier, Andrea Schnepf, Harry Vereecken, Martin Bouda, and Mathieu Javaux
I carefully read the authors' response and the new manuscript.
Let me divide my comments in two parts.
First part - general impression:
What surfaces from the revision documents is a lack of care in drafting the documents. There are a few leftovers from comments apparently addressed to the co-authors that should not appear in the revised documentation.
For example, besides a large number of MS-WORD reference errors "Error! Reference source not found", the answer to the question of Rev #1 labeled as L50-65 is evidently a sentence addressed to the co-authors not meant to be in the final revision.
These mishaps are definitely not that important from the scientific point of view, but provide a general impression of carelessness, as is the large number of miprints such as expect vs except, connection vs. connecting, etc., and make the reading rather difficult in many sections.
The manuscript is in better shape in terms of readability.
Second part: Scientific.
I have to be franc here, and I apologize if I am overly "didactic" in my comments. Now that I can read the manuscript with somewhat more clarity than before, I do not find so much novelty in what is done in the paper. From the introduction (from lines 127 on) the authors state that "The objective of the paper is to derive with a bottom up approach a model that describes root water uptake considering the hydraulics of the 3D root architecture". The second objective is as follows. "This model will be scaled up to a 1D model that could be readily implemented in land surface models." After that there are numerical experiments.
Two comments are in order:
1. First objective, I have the following major reservations:
Big-root, parallel root, explicit 3D root geometry, axial flow, parallel flow, flow in series, are all processes contained in a single model framework, i.e. a stationary diffusion equation (or in mathematical terms the weighted graph Laplacian) defined on a given graph (the 3D root network). The solution of this model depends upon the topology of the graph, i.e., the (assumed or measured or modeled) 3D distribution of root sections and bifurcation nodes. Since the latter is assumed given in this manuscript, I do not understand the novelty of the contribution, besides providing a "physical-biological" interpretation of the inversion of the graph Laplacian.
I still have some difficulties in understanding the differences the authors make between the different parallel/big/general root models. To me the difference is only on the root network geometry and not at all on the processes. The weighted graph Laplacian can be interpreted as the combination of Ohm's and Kirchoff laws, i.e., force balance (Ohm's or Darcy's or Henry's law) and mass/energy balance (Kirchoff). The distinction between parallel/big/general root models is only given by the root architecture. Within the theory of diffusion equations, it is well known that we can find a re-parametrization, or more precisely in or case a redistribution of the edges of the graph (i.e., of the root architecture) and of he weights of the graph Laplacian (i.e., conductivities and edge lengths) that will connect each node of the graph directly with the sink node where the base of the plant trunk is located (the collar in the authors jargon). This would be what the authors call a "parallel root" model that is completely equivalent in terms of sap dynamics to a "big root" model. So in the relevant terms of model results, indistinguishable.
In essence, this interpretation is given in the manuscript in an imprecise way and this casts doubts on the reader on the scientific novelty/relevance of the authors' work.
This impression is reinforced by the authors' answer to comment 3 of Rev 2.
Hence, as far as the first explicit objective, there seems to be no scientific novelty, but rather a contribution to additional confusion in the "linear diffusive" root modeling frameworks.
2. Second objective: upscaling
Here the paper becomes interesting. The authors' attempt is worthwhile, but it is not explained clearly and concisely:
The objective here is well explained by the authors in their response to question 4 of rev 2 and in lines 128-129 of the introduction "This model will be scaled up to a 1D model that could be readily implemented in land surface models." However, it does not surface clearly neither from the introduction nor from the model derivations sections, which are too mixed up with the comparison between parallel/big root/3d general modeling frameworks. As a result it does not seem the main objective of the paper and thus the impression on the lack of novelty is pervasive.
Going beyond this point, it is the parametrization here that comes into play. It is impossible to come up with the complete identification of the conductivities for i) flow from the soil into the root, ii) axial flow in the root at all points (xylems are highly heterogeneous), iii) upward flow by capillarity (or whatever process or combination of processes form the upward driving force). These parametrizations act at drastically different scales, from thin (capillary) to main roots. This type of discussion is left at a "subliminal" level, but should be instead at the center of the idea that we can come up with an "upscaled" model and try to determine the "upscaled" parameters (to be used in 1D models) directly from measurements.
In conclusion I still would require major revisions to this paper.