I am wondering if anyone has tried Kriging mathemathical interpolation methods in Grasshopper yet? There are numurous software that can run this from point clouds, but I haven’t found it in GH yet.
Appreciate any leads for this one.
Kind regards,
James
I remember that technique from my days as a mining engineer. Used to estimate an average grade for a massive ore deposit, although I am sure there are many other applications (water reservoirs, biomass, epidemiology, population density, etc). A value, i.e., an assay sample, is assigned to an area or a volume, then the total value, i.e., assay grade, is calculated.
A 2d or 3d veronai will give you areas or volumes from your sample point distribution; value times area, totaled, divided by total area, will give your average number. you can also divide an area by meshing, if you prefer classical Kriging vs the veronai method. All of this is easily done in grasshopper. Try it, a few points, then make a veronai
Hi Markz,
Thank you for your quick response. I’ve been working with borehole point clouds using simpler Delaunay meshing and manual booleans, but they don’t work well with complex stratification I don’t think.
I think I understand your explanation. I only need to factor in the weight of the points to make this work. However, If you have any simple examples I could refer to, that would be really helpful to ensure I’m on the right track!
Kind regards,
Hi James,
well, basic theory should apply, I am thinking that you want a 2-d, or 2-1/2 d volume tro deal with strata; linear if you have a vein type deposit, or coal seam, perhaps spherical shells if a porphyry, perhaps cubic if you are in the Democratic Republic of Congo after high grade copper/cobalt.
what I mean by 2-1/2 d, start with a veronai, or delaney if you prefer, then extrude the cells by the indicated thickness of your orebody, you could do it by strata thickness, or each pay zone for each borehole; more accurate if you have core for metallics, or gamma/neutron density, or even raman spectroscopy, for sidewall in a borehole. well, however you are sampling, for each volumetric element, ore value = grade * volume, then the sum the elements, divide by total volume and you will have total contained grade. standard kriging.
I’ve never coded this up, but i suspect a veronai will be better than a delaney, but try both.
Any of peter’s architectural shell examples, elsewhere on this discussion, are basically the visual model of what you are after, except your end product is an estimated grade, instead of a physical structure.
The attached is a not bad intro on that matter.
AU_L_KRIGING.pdf (6.5 MB)
also
kriging_course_mnmuq2014_hal.pdf (677.1 KB)
BTW: If you are not familiar with coding (mid to advanced) avoid trying that kind of stuff.
https://www.tandfonline.com/doi/full/10.1080/15481603.2014.1002379
BTW: I’m, far and away from base (summer windsurfing period started) but I found some Krieging snaps: Get some pts (like in a Patch thing) and do the interpolation (a zillion times faster than Patch) for some more “elaborated” result.
Moral:
