Dear colleagues,

Thanks for tacklig an important question in hydrogeophysics with your manuscript. I went through with pleasure. It is written straight forward, thus easy to follow. Figures are well presented.

Nevertheless I have some comments basically on the methodology. I hope they help to clarify/sharpen the main purpose of the manuscript, solve some unclear interpretation and maybe extending the research context.

To me the main purpose of the manuscript is the (more objective and formalized) cluster analyses enabled by a larger number of SNMR sounding compared to "just" manual interpretation of multiparameter geophysical data. However, I think, this part needs sharpening and clarifications.

--- Komedal:  

-Figure 3 gives the silhuette index for the clusters and the text introduces the terms "well defined", "fairly defined" and "poorly defined" that are used to decide for the 3-cluster case. First, the definitions of the terms is vague. For instance, "well defined" is defined by "most of the data ...". What does "most" mean? Give better definitions. 

-Further, to me it is unclear why the 3-cluster case is taken. That case contains a "poor defined" cluster! Taking the siluette analyses I would go for two clusters. However, it is argued with lithological knowledge one would take 3 cluster and indeed when looking at Figure 4 it very much looks like 3 cluster. But this is not the outcome of the siluette analyses. 

-Finally it is agrued that cluster 3 is interpreted differently when being shallow or deep (page 10 lines 7 to 10). To me this "additional" interpretation to the cluster analyses is really weakening the more objective and automatic approach. 

-It is argued that the cluster analyses is somehow a "brutal" approach for a smooth inversion. But then, why not using a layered inversion?

-In particular I am also suprised by the missing contrast of resistivities between saturated and unsaturated. Water saturation is a driver of the bulk resistivity. If saturation changes, resistivity changes! Is that maybe because of missing sensitivity of "inductive electromagnetic methods" to changes in "higher" resistivities, thus parameter uncertainties for rather resistive layers? It is surely not because of inability in the physical parameter as written for instance at page 12 line 10 or page 21 line 18-19. Electrical methods are used to monitor the vadose zone and water infiltration. Please be more clear. And not at least, TEM is used to monitor groundwater table! (Zamora-Luria et al 2024, Long-term monitoring of water table and saltwater intrusion ..., Near Surface Geophysics, Vol. 22). However, the cluster analyses solves that "problem" nicely into 3 clusters (because of the SNMR sensitvity to water content), thus, this seems to be the advantage of the cluster analyses. Please elaborate on this.

--- Endelave:

As for Komedal, please show the siluette plot for different numbers of cluster.

I found the parameters and hydrological interpretation in the clusters somewhat suprising. Figure 8 indicates 4 cluster, no doubt,  but especially the T2* times of sandy aquifer are quite low and the range between "clay", "till" and "sand" similar while for the saltwater saturated sand (saline sand is not a good term) T2* is higher? Why is that? Is that related to "impacts" (iron) on the T2' relaxation times? Explain! However taking a look at the spatial distribution the clusters are reasonable, so why is that? Apparently the cluster analyses works well and help identifying units. So this could be a real benefit. Please elaborate on this!

Introduction/Methodology -> Context of research:

- Clearly there are other papers already dealing with joint interpretation of electrical methods and SNMR to solve the ambiguities in hydrogeophysical tasks (saltwater - clay, or vadose zone - clay). I think those should be mentioned. For instance to name just one - there are others as well:

Guenther and Müller-Petke, 2012, Borkum ...

- Furthermore it is also necessary to point out joint inversion approaches that are in particular used to get joint layer boundaries for the interpretation. This is also necessary to point out that SNMR demands a resistivity distribution. There are different approaches around, for instance you may refer to:

Behroozmand

Skibbe

- SNMR may also be able to detect water in partly saturated sands with fast relaxation. The is research of SNMR in the vadose zone and soils. For instance:

Flinchum, B. A., Holbrook, W. S., Parsekian, A. D., & Carr, B. J. (2019). Characterizing the critical zone using borehole and surface nuclear magnetic resonance. Vadose Zone Journal, 18(1), 1-18.

Walsh, D., & Grunewald, E. (2012, January). Application of surface NMR measurements to characterize vadose zone hydrology. In Symposium on the Application of Geophysics to Engineering and Environmental Problems 2012 (pp. 229-229). Society of Exploration Geophysicists.

Hiller, T., Costabel, S., Radić, T., Dlugosch, R., & Müller-Petke, M. (2021). Feasibility study on prepolarized surface nuclear magnetic resonance for soil moisture measurements. Vadose Zone Journal, 20(5), e20138.

Discussion:

Appears more a like a repetition of already mentioned statements.

From a methodlogical point of view I would expect discussion on pro/cons compared to existing approaches of joint inversion/interpretation especially having the comments above on clustering in mind.

Also discuss why not or how to combine a joint inversion with clustering, for instance using a layered joint inversion.

It seems as only T2' is used here and might results in similar values even for different units. Why not using T2 or T1 as possible with the APSU device?

Some minor text comments:

page 2 line 23: replace inconsistent by ambigious

page 3 line 1: cannot distinguish: this is a bit harsh. there are papers around who deal with partly saturated NMR stuff (see above) 

page 3, second paragraph and figure 1: this is exactly what is described in (Günther, T., & Müller-Petke, M. (2012). Hydraulic properties at the North Sea island of Borkum derived from joint inversion of magnetic resonance and electrical resistivity soundings. Hydrology and earth system sciences, 16(9), 3279-3291.). Please cite or refer to it. It is a kind of perfect paper that lays ground for your paper as you go beyond this manual interpretation by your cluster analyses.

page 5 line 13: give numbers to deadtime

page 5 line 14: define partly or fully decay

page 5 paragraph 2.3: how is resistivity for MRS inversion handled. Any coupling there to the TEM?

page 11 table 1: as above, 1000 ohmm for saturated sand appears too high (cluster 1) and the overlap to cluser 3 in terms of resistivities is also high. this needs to be explained.

page 16, table 2: t2* times of the saltwater sand and the freshwater sand are quite different. why? 

page 16 line 17: saline sand -> replace by sand saturated with saline water (or something similar but saline sand is not correct)  

Please find my comments in the attached file.