In the first post of this series I argued that we should not build colormaps for azimuth (or phase) data by interpolating linearly between fully saturated hues in RGB or HSL space.
A first step towards the ideal colormap for azimuth data would be to interpolate between isoluminant colours instead. Kindlmann et al. (2002) published isoluminant RGB values for red, yellow, green, cyan, blue, and magenta based on a user study. The code in the next block show how to interpolate between those published colours to get 256-sample R, G, and B arrays (with magenta repeated at both ends), which can then be combined into a isoluminant colormap for azimuth data.
This is a good example in general of how to interpolate to a finer sampling one or more sequence of values using the interp function from the Numpy library. In line 04 we define 7 evenly spaced points between 0 and 255; this will be the sample coordinate for the r, g, and b colours created in lines 01-03. In line 05 we create the new coordinates we will be interpolating r, g, and b values at in lines 06-08 (all integers between 0 and 256). The full code will come in the Notebook accompanying the last post in this series.
This new colormap is used in the bottom map of the figure below, whereas in the top map we used a conventional HSV azimuth colormap (both maps show the dip azimuth calculated on the Penobscot horizon). The differences are subtle, but with the isoluminant colormap we are guaranteed there are no perceptual artifacts due to the random variations in lightness of the fully saturated HSV colors.
Another possible strategy to create a perceptual colormap for azimuth data would be to set lightness and chroma to constant values in LCH space and interpolate between hues. This is the Matlab colormap I previously created, and shown in Figure 4 of New Matlab isoluminant colormap for azimuth data. In the next post, I will show how to do this in Python.
In New Matlab isoluminant colormap for azimuth data I showcased a Matlab colormap that I believe is perceptually superior to the conventional, HSV-based colormaps for azimuth data, in that it does not superimposes on the data the color artifacts that plague all rainbows. However, it still has a limitation, which is that the main colours do not correspond exactly to the four compass directions N, E, W, and S.
My intention with this series is to go back to square one, deconstruct the conventional colormaps for azimuth, and build a new one that has all the desired properties of both perceptual linearity, and correct location of the main colors. All reproducible in Python.
If we wanted to build from scratch a colormap for azimuth (or phase) data the main tasks would be to generate a sequence of distinguishable colours at opposite quadrants, or compass directions (like 0 and 180 degrees, or N and S), and to wrap around the sequence with the same colour at the two ends.
But to do that, we should avoid interpolating linearly between fully saturated hues in RGB or HSL space.
To illustrate why, it is useful to look at the figure below. On the left is a hue circle with primary, secondary, and tertiary colours in a counter-clockwise sequence: red, rose, magenta, violet, blue, azure, cyan, aquamarine, electric green, chartreuse, yellow, and orange. The colour chips are placed at evenly spaced angular distances according to their hue (in radians).
Left, primary, secondary, and tertiary colour chips arranged using hue for angular distance; right, the same colour chips arranged using intensity for angular distance.
This looks familiar and seems like a natural ordering of colors, so we may be tempted in building a colormap, to just take that sequence, wrap it around at the red (or the magenta) and linearly interpolate to 256 colours to get a continuous colormap [1], and use it for azimuth data, which is how usually the conventional azimuth colormaps are built.
On the right side in the figure the chips have been rearranged according to their intensity on a counter-clockwise sequence from 0 to 255 with 0 at three hours; so, for example blue, which is the darkest colour with an intensity of 29, is close to the beginning of the sequence, and yellow, the brightest with an intensity of 225, is close to the end. Notice that the chips are no longer equidistant.
The most striking is that the blue and the yellow chips are more separated than the other chips, and for this reason blue and yellow features seem to stand out a lot more in a map when using this color sequence, which can be both distracting and confusing. A good example is Figure 3 in New Matlab isoluminant colormap for azimuth data.
Also, yellow and red, being two chips apart in the left circle in the figure above, are used to colour azimuths 60 degrees apart, and so do cyan and green. However, if we look at the right circle, we realize that the yellow and red chips are much further apart than the cyan and green chips [2] in the perceptual dimension of intensity; therefore, features colored in yellow and red could be perceived as much further apart (in azimuth) than cyan and green.
These differences may be subtle, but in my opinion they become important when dip azimuth is combined with other attributes, perhaps using a 3D colormap, and the resulting map is used for detailed structural interpretation. There is a really good example of this type of 3D colormap in Chopra and Marfurt (2007), where dip azimuth is rendered with hue modulation, dip magnitude with saturation modulation, and coherence with lightness modulation.
A code snippet with the main Python commands to generate the two polar scatterplots in the figure is listed, and explained below. The full code can be found in this Jupiter Notebook.
01 import matplotlib.colors as clr
02 keys=['red', '#FF007F', 'magenta', '#7F00FF', 'blue', '#0080FF','cyan', '#00FF80',
'#00FF00', '#7FFF00', 'yellow', '#FF7F00']
03 my_cmap = clr.ListedColormap(keys)
04 x = np.arange(12)
05 color = my_cmap(x)
06 n = 12
07 theta = 2*np.pi*(np.linspace(0,1,13))
08 r = np.ones(13)*2.5
09 area = 200*r**2 # size of color chips
10 c = plt.scatter(theta, r, c=color, s=area)
11 theta_i = 2*np.pi*(sorted_intensity/255.0)
12 colors = my_sorted_cmap(np.arange(12))
13 c = plt.scatter(theta_i, r, c=colors, s=area)
In line 01 we import the Colors module from the Matplotlib library, then line 02 creates the desired sequence of colours (red, rose, magenta, violet, blue, azure, cyan, aquamarine, electric green, chartreuse, yellow, and orange) using either the name or Hex code, and line 03 generates the colormap. Then we use lines 04 and 05 to assign colours to the chips in the first scatterplot (left), and lines 06, 07, and 09 to specify the number of chips, the angular distances between chips, and the area of the chips, respectively. Line 10 generates the plot. The modifications in lines 11-14 will result in the scatterplot on the right side in the figure (the sorted intensity is calculated in much the same way as in my Geophysical tutorial – How to evaluate and compare colormaps in Python).
[1] Or, perhaps, just create 12 discrete colour classes to group azimuth values in bins of pi/6 (30 degrees) each, and wrap around again at the magenta, to generate a discrete colormap.
[2] The green chip is almost completely covered by the orange chip.
This is the first post in a series on the rainbow and similar color palettes. My goal is to demonstrate it is not a good idea to use these palettes to display scientific data, and then answer these two questions: (1) is there anything we can do to “fix” the rainbow, and (2) if not, can we design a new one from scratch.
The rainbow is dead…some examples
In a previous post I showed a pseudo-3D rendering of my left hand x-ray using intensity (which is a measure of bone thickness) as the elevation. I mapped the rendering to both grayscale and rainbow color palettes, and here I reproduced the two images side by side:
I used this example to argue (briefly) that the rainbow obscures some details and confuses images by introducing artifacts. Notice that in this case it clearly reduces the effectiveness of the pseudo-3D rendering in general. It also introduces inversions in the perception of elevation. The thick part in the head of the radius bone, indicated by the arrow, looks like a depression, whereas it is clearly (and correctly) a high in the grayscale version.
This is a PA ulnar deviation x-ray of my left wrist from last month, which gives a good view of the scaphoid bone from above.
The bone is chipped in the area pointed by the arrow, due to a fall that occurred 20 or so years ago. Somewhere in there, there’s also a tiny detached fragment of cartilage that calcified (as seen in a CT scan at the time). I was lucky, because typically the result of a fall with outstretched hand for people aged 17-40 is the scaphoid fracture, which are known to have unpredictable healing. Lately, however, due to a tendonitis, the fragment too is acting out. I’m left handed so this is causing some trouble, and that’s why the recent x-rays.
