Correlating Two Grid Models

RockWorks | ModOps | Grid | Statistics | Correlate

This program is used to identify spatial correlations between grid models that don't necessarily represent similar units (e.g. ppm versus occurrences per cell). Stated euphemistically: it's a way to compare apples and oranges.

This program requires that two real number RockWorks grid models (.RwGrd files) already exist and that they have the same dimensions (extents and node spacing).

Menu Options
Step-by-Step Summary

Input/Output
- Input
  - Grid Model A: Click to browse for the name of one the existing grid models (.RwGrd files) to be analyzed.
  - Grid Model B: Click to browse for the name of the second grid model to be analyzed.
- Output Grid Model: Click here to type in the name to assign to the new grid file that the program will create, which results from the correlation operation
2D Grid Map
Check this box to display the output grid as a 2D map at this time.
Click this tab to set up the 2D map layers (bitmap, symbols, labels, line contours, color-filled contours, labeled cells, map border, etc.).
3D Grid Diagram
Check this box to display the output grid as a 3D surface at this time.
Click this tab to set up the 3D map layers (surface colors, images, reference cage, etc.).
! You can request both a 2D and 3D representation of the grid model.

Select the ModOps | Grid | Statistics | Correlate menu option.
Enter the requested menu settings, described above.
Click the Continue button to proceed.
The program will load the input grid files and compute the standard deviation and mean for all of the grid nodes.

The grid nodes are then recomputed to represent the number of standard deviations from the mean in order to quantify just how anomalous a node value is relative to the entire node population. The following terminology (from mineral exploration) may help in understanding what these new values represent:
Background: Values that are less than 1 standard deviation from the mean.
Weakly Anomalous: Values that fall between 1 and 2 standard deviations from the mean.
Moderately Anomalous: Values that fall between 2 and 3 standard deviations from the mean.
Strongly Anomalous: Values that fall between 3 and 4 standard deviations from the mean.
Extremely Anomalous: Values that are greater than 4 standard deviations from the mean.
Note: During this process, negative anomalies are converted to zero. It is assumed that the user is only interested in positive anomalies.

Then, on a node-by-node basis, the absolute value of the differences between the node values are subtracted from the higher value. For example, let's say that a lead node has a value of 3.6 while the corresponding gold node has a value of 3.1. The resulting correlation coefficient would be 3.1 ( 3.6 - abs ( 3.6 - 3.1) ). That node would therefore be considered as a point where both the lead and gold are strongly correlative in a spatial sense.

The resulting node values will be stored in a new grid file in the project folder under the output file name you selected. The requested diagram(s) will be displayed in a RockPlot2D tab and/or RockPlot3D tab in the Options window.
You can adjust any of the settings in the Options window and then click the Continue button again to regenerate the diagram(s).
If you have requested a map, it will be displayed in a RockPlot window. View / save / manipulate / print / export the map in the RockPlot2D or RockPlot3D window.

This approach assumes that both data sets encompass a large enough area and sample population / distribution to establish the "background" levels for each component.
This results do not in any way, shape, or form, establish causality - just spatial correlation. Indeed, many spatial correlations are often purely coincidental.

Multivariate Anomalies for information about analyzing multiple grid files at once for anomalous regions.
Adding this program to your Playlist
Adding this program to your Favorites
Save/load/view menu settings

Back to Grid Menu Summary

RockWare home page