This is an archival copy of the Visualization Group's web page 1998 to 2017. For current information, please vist our group's new web page.
Mining for Copper
Karsten Pruess (Earth Sciences Division, LBL) and George Brimhall (Geophysics, UC Berkeley) have an LDRD to study and model geophysical and geochemical processes which have resulted in an ore body of particular interest in the desert of the Andes Mountains in the El Salvador district of Chile. The images presented on this page show a collaboration with the Visualiztion Group. The work done to create these images lays the groundwork for further study, as the focus of these images is the display of a large amount of field data. The future work will involve coupling the computational resources of the NERSC machines which will model ore movement over time with visualization tools.
(Feb 2004 - the following URIs have gone stale, but have been left in the document as plain text for historical reference.) More information about the Pruess and Brimhall LDRD can be found http://www.lbl.gov/Publications/LDRD/1996/Pruess.html - at LBL or at http://socrates.berkeley.edu/~earthres/ - UCB.
The amount of field data is extensive. There are over 11,000 boreholes and more than 1 million assay records. These records indicate the amount of copper present in rock samples taken from the boreholes. There is also topographic data. The first image shows a far-away view of just the head of each borehole, along with an interpolated mesh representing surface topography.
|
If we then add colorized representations of the assay records, and change the view so that we don't see all of this data at once, we still have a mess.
|
Conceptually, what we really want to do is to show only a part of the millon-plus assay records. Further, we want to let the user choose which of the records to display. A very natural way to do this is to "draw a box" around some of the boreholes, then display the assay records associated with only those holes.
In order to accomplish this, we had to extend our existing VRModules library to create what we'll call a lasso box. The way this module works is that it first creates a box which encompasses an entire dataset. Then, the box's size and position in space can be controlled with one of the three- or six-dimensional input devices in the Visualization Laboratory.
A second module was written that will take an array of data (which can be thought of as a one-dimensional list of items, each of which contains an x,y,z coordinate and "some data"), along with the extents of the lasso box, and "filter out" data which lies outside the range of the lasso box.
This image shows what happens when we choose to show assay data from only a subset of the entire model.
|
We can take this idea one step further, and do some more general visualization of the assay data which we have selected. In the following image, we have first used a locally-written three-dimensional kriging tool to compute a new set of assay values which lie on a structured grid, then we have taken a slice or two from this data and computed an isoconcentration surface. The kriging operation on this subset of points takes about fifteen or twenty seconds on a reasonably powerful microprocessor. Kriging the entire million-plus assay records takes on the order of three hours using the same hardware.
|
If we combine the visualization of assay, topology with a kriged geological model, we get something like this.
|
A more recent visualization shows the difference between the currently-observed top-of-ore and a theoretical top-of-ore in the distant past. The theoretical model is based upon knowledge of the geologic structure in the area, of geophysical processes and other information.
(Pictures not ready right now)
Immediate plans for the future include:
- Development of a simulation tool for modeling ore movement using a mass-balance equation on the NERSC T3E, and visualizing the results.
- Bethel, Pruess and Brimhall were awarded an LDRD for FY1998 to tackle problems pertaining to the visualization of large datasets. The Pruess-Brimhall project will be a source for such large datasets.