Data Processing :
Combining data through color compositing

A GIS / remote sensing method known as color compositing can be employed to visualize multiple geophysical data sets simultaneously. One advantage is that it allows relationships, commonalities, and differences between the various sensing techniques to be recognized. In addition, different anomalies detected by the various methods can be complementary, as when resistivity shows the floor of a house due to a soil change while magnetometry reveals interior features like hearths.

Up to three geophysical data sets can be employed, by assigning each to a primary color (blue, green, or red, above). If one has more than three data sets then dimensionality might be reduced through a principal components analysis.

One must be wary that anomalies in one image do not "cancel out" those in another. For example, is both resistivity and EM conductivity surveys are employed, as above, high resistivity anomalies will be expressed as low conductivity and vice versa. Because one is the negative of the other, when composited they will effectively cancel each other out. The solution is to invert the scale on one so that the information is complementary.

The color composite is derived by combining the three color primaries into a single image. The various combinations and colors that result point to similarities and differences between the various input data sets. Assuming 8-bit data for each of the input images (256 display values), then it is possible to achieve 2563 = 16,777,216 color combinations in the resulting image!


Color composite of the above data sets. Each square is 20 m.

Interpretation

Interpretation is straightforward, but requires familiarity with the RGB color model and a little practice. Black represents low measurements in all three inputs while white points to parallel high measurements on all three. Dominance of a primary color means that only that input contributes. An equal contribution of all three primaries produces a gray with brightness determined by their magnitude; unequal contributions result in other colors. The following key presents significant combinations.

Click here to see an RGB animation

(Data source: Whistling Elk Village, SD. Each data set has been pre-processed.)

Back to Main Topic Page

Contribution by: Kenneth L. Kvamme, Archeo-Imaging Lab, University of Arkansas