[GRASSLIST:3538] Re: Geophysics modules for GRASS

[Michael Barton]

Craig,

That was an amazing 3D analysis you did.

Because I haven't done more than just look briefly at the G3D modules, 
you might have the best answers from folks working with it the most 
(e.g., Helena Mitasova) via the grass list. I will tell you what I know 
so far.

>
> I have not looked at all into the 3-D capabilities of GRASS - I am a 
> little out of touch with these features. In 5.7 do the functions like 
> v.buffer & v.distance work in 3-D?

These do not have 3D counterparts. This is where new programming is 
needed.

>  If they do than I suppose that 3-D vector geometry could be used to 
> create binary query volumes through a 3-D version of v.to.rast. 
> Modules to generate vector iso-surfaces from 3-D raster volumes should 
> be pretty straight forward to develop. Volume visualization (density 
> clouds) is also pretty easy as long as the hardward supports 3-D 
> texture maps.

I *think* that isosurfaces can be generated via NVIZ (at least they 
were possible previously in the mkdsp module that the updated NVIZ 
replaced). Given this, constructing isosurfaces should be doable 
without a great deal of effort.

>
> I suppose Vis5D could be used to for these purposes also.

There is a Vis5D export module

>
> Another issue that was very important in the South African case was 
> that the 3-D volumes needed to be orientated along the trend of the 
> data. These volumes used a fairly small sample interval (high-res). So 
> to conserve memory it was critical that the grid volumes where 
> orientated along some bearing and plunge. In other words, having the 
> volume defined along a northing, easting, and depth was not adequate 
> because a lot of the grid points in the volume where never used (up to 
> 90%).  A 3-D raster volume was positioned in space through a 3-D 
> transformation matrix which contains scaling, rotation and translation 
> components. I assume this is probably not an issue for you? But for 
> others it can be important. Can GRASS handle this?

Again I *think* the updated NVIZ can do this. Note that a feature 
missing in GRASS's extensive analysis routines is a trend surface 
module. My understanding is that the data model or framework for 3D (or 
perhaps even nD) is in place and fairly robust. However, there are few 
tools yet written to work with the data.

Michael

____________________
C. Michael Barton, Professor
School of Human Origins, Cultures, & Societies
PO Box 872402
Arizona State University
Tempe, AZ  85287-2402
USA

Phone: 480-965-6262
Fax: 480-965-7671
www: <www.public.asu.edu/~cmbarton>
On May 28, 2004, at 7:53 AM, Craig Funk wrote: