Last time we had an introduction to colour in scientific graphs, where we had two categorical data series. What happens if we have continuous or sequential data, plotted over an area?
Here are two X-ray maps using the classic ‘rainbow’ colour scheme, the first of Ti in allanite:
Looks good, no? We can see the background that has very little Ti compared to the allanite. The allanite has a Ti rich “core”, intermediate “mantle”, and a Ti-poor “crust”. It is also possible to see some “hotspots” of Ti in the mantle.
Let’s see another example, this time of Mn in epidote:
Not as good as before, but we can see the important things. The epidote is mostly in the ~40 area (green), with some Mn-rich zones (blue and violet). The other mineral in the top is low on Mn (yellow).
If you’ve read my last post, then you can already see the problems in this. First of all, this will go bad in black and white printing. Let’s simulate that:
See how the colour scale makes absolutely no sense? The second obvious problem is the accessibility: the use of both red and green makes it difficult for people with colour deficiency to properly understand the figure.
In plots such as this, there is a third problem – the way our eye perceives colour. The computer “sees” the data as continuous. However, when plotted with different colours (hues) our brain interprets what it sees as discrete colours. You see green, red, blue, violet, etc. The problem is exacerbated because not all colours have the same numerical range. If you look at the scale, you can see that you have only one orange whereas you have several greens. Therefore, differences within different shades of the same colour are lost.
In order to make better colour schemes, luminosity should be used instead of hue. The human eye and brain can detect subtle differences in luminosity, especially when it is the same colour. Let’s see how it works for our two X-ray maps:
Oh my, now that’s beautiful. And clear. I used two different schemes – white to dark colour and black to bright colour. Both work fantastically. You can now see that the allanite is actually sector zoned in respect to Ti, and it has a weird Ti-poor ring surrounding its core. You can also discern the delicate details of Mn zoning in the epidote. Notice that I used only one colour! This also works great in grey scale printing.
What happens when you have a certain middle value (such as 0) where the values are diverging from it? It is a good idea to use two different colours and a neutral middle colour:
Even though it doesn’t make much sense to use it in mineral chemistry X-ray map, the result is surprisingly beautiful. It is, however, quite useful in topography (below and above water), geophysics (magnetic anomalies), meteorology (temperature), etc. Just be sure that you’re using colour blind friendly colours.