irizarry@batcomputer.tn.cornell.edu (Gil Irizarry) (03/11/88)
I have a question for the group: Suppose I have a bunch of data for groundwater on a certain planet (say Mars). Say the data consists of two numbers for every square meter of ground, on number representing concentration of the water, and the other representing the depth at which the water is found. What is the best way to display this data? Suppose we want to project the planet's surface onto a Mercator map. Someone suggested that we can display water depth as a color (blue - near surf , red - very deep) and concentration a intensity of this color. I think this will not look too readable, but we haven't actually tried it yet. Can anyone think of a better way, or is this the best. Note that we need to see that whole planet, or at least a very large part, at one time. You can either e-mail or post the responses. Thanks in advance. -- Gil Irizarry BS CS '89 irizarry@tcgould.tn.cornell.edu.ARPA Cornell University irizarry@cnrlthry.bitnet xady@crnlvax5.bitnet Space: the dream lives on...
bobr@zeus.TEK.COM (Robert Reed) (03/12/88)
In article <4007@batcomputer.tn.cornell.edu> irizarry@tcgould.tn.cornell.edu (Gil Irizarry) writes:
Suppose I have a bunch of data for groundwater on a certain planet (say
Mars). Say the data consists of two numbers for every square meter of
ground, on number representing concentration of the water, and the other
representing the depth at which the water is found. What is the best
way to display this data?
Depends. Do you need to be able to see surface relief? Is there only one
concentration/depth pair per geographic position? What sort of continuity
exists in the data? (Are adjacent values of concentration more likely to be
the same than depth?) What is the resolution of your display medium?
The first approach which occurred to me was to use hue and saturation to
represent concentration and depth. This would leave intensity available to
show relief (by shadow casting or whatever), so that you could more easily
connect moisture features with land features. The best assignment of hue and
saturation depends on the data. Adjacent hue changes are much more visible
than adjacent saturation changes, so hue would be better for highlighting
the variance of a subtly changing parameter.
Using all three axes of the color space requires adequate resolution to
differentiate. If depth is lacking (like 8 bits per pixel rather than 24),
the subtle changes in intensity and saturation will be lost in the
quantization noise.
If your data does include multiple points per location, it will be difficult
to avoid muddiness in a two dimensional projection. Perhaps a series of
sectional plots arranged by depth might be a better solution. If you have
a graphics system of sufficient performance, providing an interactive
fly-through of the 3-D data may be advantageous. Individual depths and
locations would provide the 3-D positioning, and hue or intensity could be
used for concentration. Using semiopaque voxels would allow the
investigator to grasp the shape of the moisture concentrations.
There are lots of alternatives, some of which I've only alluded in this
note. I hope some of this helps.
--
Robert Reed, Tektronix CAE Systems Division, bobr@zeus.TEK