HabitableZone

Flame » in reply to I would probably fail a Turing test.

Something else, though,
podrock — (archived)

My handle here, as you might recall, is from a rare and enigmatic metamorphic rock, nodular sillimanite gneiss. I've been studying this rock for decades.

In the field, it outcrops with these nodules of quartz, sillimanite, and other minerals. They are harder than the gneisic matrix so they tend to poke out of the surface. And they come in all sorts of varieties. Oblate, prolate, rods... We have a pretty good idea now about how the mineralogy comes about, but it is the structures that haunt me.

You'll go up to one of these outcrops while mapping and observe the shapes and orientations of the nodules. Sometimes they are fairly consistent. Other times, not. But there is a pattern.

In order to document that pattern, I'd have to use my Brunton compass to take readings of their orientation. I would measure their principal dimensions, then do some 3D graphical statistics. And still be confused.

The problem was the depth of the data set. It takes time to take these measurements out in the wild. And boring.

What if, I could just have a drone with LiDAR, have it circle the outcrop a couple of times, also taking pictures, and feed that data into the AI, and have it measure not a dozen orientations, but all of them? A 3D model I can study while sitting on my butt in my grotto office?

A thousand data points. Plotted on a steronet. Numbers crunched with great statistics.

Now that would be fun.

Log in or register to post.

The whole thread (42 posts)

Related discussions