RW solid strat model: using pseudo-boreholes to enforce surficial geology

[jacksonn]

Hello,

I have ~290 boreholes with stratigraphic picks; 5 strat units. Model area is 18,400 x 19,000 x 2,190 ft (XYZ, respectively); spacing is 500 x 500 x 10 ft. This is a reduced model size for quicker testing of variables and input data and will be expanded with a finer resolution once I've resolved the best approach. I also have a constraining surface of the DEM (same DEM that was used to compute the borehole elevations) and have a baseplate set to the lowermost unit.

 

I have a surficial geologic map that I've converted into a raster/grid matching the project dimensions. For each surficial model cell without a "real" borehole and stratigraphic data, I've created a pseudo-borehole to represent the surficial geologic map. I set the borehole's location to the cell's centroid and set a stratigraphic interval of 0-10 ft to match the surficial map. The pseudo-boreholes are identified by a boolean field in the Location tab. [I alternately used (1) an interval of 0-unknown ft and (2) two intervals: 0-10 ft and 10-unknown. The three options don't seem to differ much in their results.]

 

Conceptually, the inclusion of these pseudo-boreholes should result in a solid model (or layered model) that has a surface matching the surficial geologic map. Yet, I can't achieve the desired result. I've been tinkering with the modeling algorithm, modeling sequence, and other options, to no avail. Some are clearly more suitable than others, but none seem to generate a surface that truly matches expectations, even when hiding thin units in a layered model. The subsurface of my attempts still leaves something to be desired as well, but I have ideas for how to resolve that. 

 

What do you think the best approach would be for enforcing the surficial stratigraphy? I have to imagine that the inclusion of these data should be able to improve model results, but I'm struggling with the mechanism.

 

In anticipation of the request to send the data for further investigation, I've also emailed the dataset (and this prompt) to the RW team.

 

Thanks!