Pelletier depth-to-bedrock

[Victor Oladoja]

Hi all, 

I noticed that in the watershed workflow, the default depth-to-bedrock layer derived from Pelletier is set to average_soil_and_sedimentary-deposit_thickness.tif, which does not appear to include the regolith (specifically the saprolite layer). Pelletier also provides upland_hill-slope_regolith_thickness.tif, which seems more consistent with a depth-to-bedrock definition that includes saprolite, although it is more for upland areas. I was wondering if there is a specific reason why average_soil_and_sedimentary-deposit_thickness.tif was selected as the default.

Thanks,

Victor

[Coon, Ethan]

Hi Victor,

Precursor:  IANAG (I am not a geologist)! If I am interpreting this incorrectly, I’m happy to change our defaults!

“Bedrock” is a qualitative term here.  Remember how that layering works — effectively we use a 3-layer model:

Layer 1 — NRCS soils data.  This is based on measurements and cover at most the top 2m of “soil”

Layer 2 — GLHYMPS media.  This is “unconsolidated sediments” and is used to describe anywhere there is no NRCS data and any sediments from 2m down.

Layer 3 — “bedrock"

So the important questions are:

1. To what depth are GLHYMPS values valid?  
2. What values are you providing as “bedrock”?  Pristine, unfractured values?  Or fractured bedrock effective values (saprolite)?
3.  And what are you thinking of as the “bottom” of your domain (which is almost always no flux, and therefore should be thought of as effectively 0 permeability rock)?

I read the GLHYMPS data description as more likely to describe the “average soil and sedimentary deposit” layer than the saprolite layer.

So if there are sediments on top of saprolite on top of pristine bedrock, it is likely that the saprolite would be better described by spatially uniform properties than by GLHYMPS properties.  So I believe there are a few potential strategies:

1. Assume no saprolite, using bedrock properties as “pristine, unfractured bedrock"
2. Use “bedrock” properties to describe fractured bedrock.  Then one should think of the bottom of their domain as the top of the “pristine, unfractured bedrock”.
3. Or even better, one could take their knowledge of their site and use it to introduce a 4-layer model, where saprolite is explicitly included and given valid properties for porosity, permeability, and WRM curves.

Effectively we are doing the 1st by default.  I’m not sure if we have sufficient valid geologic data across the entire US to do the third programmatically.  

This is why it is what it is — I am very open to other suggestions or implementations of alternative strategies to define this layering.

Ethan

From: ats-...@googlegroups.com <ats-...@googlegroups.com> on behalf of Victor Oladoja <oladoj...@gmail.com>
Date: Monday, January 5, 2026 at 11:06 AM
To: Amanzi-ATS Users <ats-...@googlegroups.com>
Subject: [EXTERNAL] Pelletier depth-to-bedrock

This Message Is From an Untrusted Sender

You have not previously corresponded with this sender. If suspicious, use the Report Phish button in Outlook.

 

--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ats-users/6e4bb2f4-d595-40ba-b083-b9b90a03c8a0n%40googlegroups.com.

[Victor Oladoja]

Hi Ethan,

Thank you very much for the detailed explanation of the rationale behind adopting the average soil and sedimentary deposit as the default. I also appreciate you outlining the key questions about GLHYMPS and bedrock definitions. I hadn’t thought of GLHYMPS values in that context before.

Regarding DTB, I see that SoilGrids2017 defines bedrock as “the consolidated solid rock underlying unconsolidated surface materials, such as soil or other regolith,” which is broadly consistent with Pelletier’s definition of an average soil and sedimentary deposit. One difference, however, is that SoilGrids2017 may implicitly include saprolite within its regolith representation, whereas Pelletier's average soil and sedimentary deposit (transported materials) may not.

At my study site, we observe little to no soil or unconsolidated sediments (<2 m) and a spatially heterogeneous saprolite layer in places where we have seismic refraction data. So, DTB maps from Pelletier and SoilGrids2017 differ substantially. If it is safe to assume that GLHYMPS values (representing unconsolidated sediments, i.e., transported materials) are more consistent with the Pelletier DTB definition than with SoilGrids2017 (at least in my study site), then I will proceed using the second or third strategy you suggested, changing some of the default TIFF files. Thanks again for the clarification; this has been very helpful.

Victor

[Coon, Ethan]

Yeah, I don’t have a good answer here.  You should probably try both.

In my experience, the Pelletier product has been better in places with shallow DTB and/or existing geologic observations.  The SoilGrids 2017 DTB product is a machine learning correlation, and doesn’t seem as faithful to observations as the Pelletier data.  They actually dropped it for the newer SoilGrids 2.0 (I don’t know why).

But it’s hard to know if differences are due to definitions or just plain error/differing methodologies.

I would suggest that, if you have data, consider using Pelletier as a suggestion, and look at both possible options in Pelletier.   Then also look at the GLHYMPs properties at your site and decide if they seem plausible as saprolite properties or not.  If so, then a deeper DTB + GLHYMPS as saprolite may be reasonable.  If not, then you may have to come up with your own saprolite properties.

Ethan

To view this discussion visit https://groups.google.com/d/msgid/ats-users/6a64c433-e4e9-4411-a22f-5c5826cfc26an%40googlegroups.com.

[Victor Oladoja]

I will do exactly that. Thank you very much for your time and for the many helpful suggestions. I really appreciate!

-Victor