This manuscript addresses the very interesting question on the effect of along-strike strength variations on fault localisation processes. It uses state-of-the-art 3D numerical models that are complemented by observations from several examples of rifts worldwide but especially by geological observations of the Great South Basin, New Zealand, that features particularly prominent along-strike lithological variations.  The results are well discussed and linked to previous insights. I find the interaction of strain localisation processes across terrane boundaries particularly interesting and well balanced. I only have a few minor suggestions that hopefully help to improve this interesting paper even more.

*** Comments:

Fig. 1: An overview map could provide the reader with the plate tectonic context and the location of the study region relative to New Zealand. It would be useful to annotate New Zealand on the map in panel A. The cross section in panel C should be annotated in the map view of panel A. I suspect it is the black line, but I can't be certain.

It would be good to motivate the model setup a bit more with the natural example. It's absolutely fine if some of the employed values are generic, but this should be mentioned. Here are some points that could be addressed:

- Line 85: is the 5 mm/yr extension velocity based on local divergence rates? If yes this would be good to mention here.

- Line 90: Are layer thicknesses consistent with the chosen example? 

- Line 101: Why do you choose 5 km^3 blocks?

*** Minor details:

Abstract and text in general: double check capitalisation of "weak", "strong" and "initial plastic strain". 

Line 16: "extension occurs perpendicular to distinct geological terranes and parallel to terrane

boundaries" I find the description of extension "perdendicular to terranes" but "parallel to terrane boundaries" misleading. I suggest to delete "perpendicular to distinct geological terranes and".  

 

Fig. 4: Cross sections are vertically exaggerated. It would be useful to note in the caption by how much.

Section 5.1: There is a recently submitted analog/numerical modelling manuscript at Solid Earth Discussion by Schmid et al. (https://doi.org/10.5194/egusphere-2022-1203) that might be relevant to this section.

Some figures contain rather small font that could be enlarged for better readability.

 

Review of "The influence of crustal strength on rift geometry and development – Insights from 3D numerical modelling", by T. Phillips, J. Naliboff, K. McCaffrey, S. Pan, J. van Hunen, and M. Froemchen.

The manuscript by Phillips et al. presents new 3D thermomechanical models of continental rift development and propagation in heterogenous crustal domains. Such models are new and take advantages of advanced numerical developments. The manuscript is well written, well organised and the objectives are clear, i.e. to mimic surface fault patterns in rifts developed across continental regions with varying mechanical properties. The variation of the 'strength' of the crust in the models is set by imposing various range of initial frictional-plastic damage in adjacent weak regions of the continental crust layer. This work nicely illustrates the process of faults propagation and connectivity during rifting in heterogeneous regions. It is of broad interest to the tectonic community, either to those studying fault network development, or for improving our understanding of rift systems and passive margin evolution in regions with strong tectonic inheritance (i.e. pretty much everywhere).

The paper is well written and organised. I reviewed an earlier version of this work (submitted to another journal in 2021) and I am pleased to write that this version is substantially improved. I believe the manuscript is of broad interest to Solid Earth readership. It needs mostly editorial polishing, and I recommend accepting this contribution with minor revisions. 

I present below a list of comments and questions that I believe should be addressed to improve the quality of the contribution:  

- line 85: Is the extension velocity full rate or half rate? Could you please precise if the 5 mm/yr is applied on each wall or is it 2.5 mm/yr on each wall?

- There are many naming or spelling inconsistencies for the domain names. It should be “Reference domain” in line 124, 150, 157. Please make sure the case for the domain names is consistent throughout the manuscript (Strong Domain vs. Strong domain vs. strong domain, weak domain, Weak domain, etc.). I have highlighted in yellow some of the mistakes I found in the attached pdf.

- lines 135-138 (and elsewhere). A suggestion: northern and southern could be a better wording than upper and lower… upper / lower have a vertical connotation, like the upper and lower crust. Because you discuss results from 3D models, I believe it brings some confusion. Same goes with terms like “top” (e.g. line 148). 

- line 141: "increased strain", is it the total finite strain, or the cumulative frictional plastic strain, or the instantaneous strain (i.e. strain rate), or all of those? 

- line 155: I’m curious here. Is the material rheology different between the central domains and the lateral regions? As in there is no frictional plastic behaviour. Or did you just not impose any pre-existing frictional plastic strain, and yet the material could still soften?

- sub-Figures calls are lower case in the text, but upper case in the figures and captions (2b vs. 2B, 3d vs. 3D).

- line 169: I know that section is qualitative but could you provide a range for the "widely-spaced" network? Is the 10-15 km distance written line 214 the actual spacing value?

- line 177: "high bands continually migrating”: Could you provide more information on those structures? What’s that if not localisation… is it a transient localisation phenomenon? Or a feature of the model? I find it a little worrying.

- line 192-194: Because of the strong vertical exaggeration of the sections in Fig 4B) I find it difficult to assess the dip angle of the structures. It would be a nice addition in the 3D description here.

- line 196: "measure strain", again is this the frictional plastic or total strain?

- line 202: I find this method quite interesting to measure the frictional plastic strain. How do this number (18 and 22) compare to the horizontal stretching factor of the model at the same timestep?

- line 207: "Faults remain fixed": That can’t be right! As the model stretches the faults should also advect slightly at the surface

- You don’t seem to discuss the hourglass shape of the fault network with a neck in the Weak domain. Any comment on this?

- line 210: "transiently". May I suggest "progressively" instead? If it was transient after linking the faults would unlink… I doubt this is the case.

- line 221: Yet in the cross sections in figure 4 there seem to be some localisation in the northern part of the Strong domain. Which threshold do you use to decide whether strain is localised or not? In a section about quantitative analysis, I would expect to read a few more numbers.

- line 288, 289: Jourdon, spells with an “O” - it is spelled correctly in the references though.

- line 309: the term “initial timestep” bothers me a bit… it is the first output, but many calculation timesteps have taken place, right? At this resolution and with such imposed lateral velocities and thermal properties I imagine that timesteps must be of the order of 5 to 10 kyr, isn’t it?

- line 338-341: Or is it related to mantle rheology rather than the upper crust? I am not fully convinced yet of the statement that rift physiography and structural style are primarily controlled by upper crustal properties (lines 285-286) ... 

- line 366-368: There surely is an upper limit in how weaken a material can be? In your models this is the max frictional plastic value for the weakening function. After this value is reached could faults be abandoned and new one created? I suppose it will strongly depend how the fault dip evolves… as stretching accumulates the fault plane should rotate to become less steep, and thus less optimally oriented. Do you have any comment on this? Or maybe in the Myr of the models it isn’t obvious that the dip angle evolves for the faults… can’t be easily seen from the figures as the vertical exaggeration in the cross-section is very significant.

- line 369-370: If this is one timestep, how can you have multiple generation of features associated with it? It comes back to my previous question about timestepping of the code vs. of the outputs.

- Table A1: What is p exponent in the viscosity prefactor? Please edit the baseline to superscript for -1 in s^{-1}.

- Figure 1: Would be nice to add a grid or some coordinates at least for A)

Adding some orientation SW - NE to figure C) would be good too.

- Figure 3 and 4: colour scales titles could be modified slightly: + Non-initial "cumulative" Plastic Strain; + Strain rate "second" invariant; + the min scale value should for \varepsilon_{ii} should be < 1e-16 rather than 0e-16.

- Figure 7: like in the text, the case of Strong/strong, weak/Weak, Domain/domain should be cleaned-up for consistency.

Guillaume Duclaux,

Nice, 02/01/2023