Content page

Colours for contour plots

Many issues affect a viewer's impression of what a data set is saying. Examples include the choice of colours, the scale's minimum and maximum range, the addition of contour lines as well as colours, etc. The objective when plotting scientific or technical data is to display information with as little bias as possible. Since each data set has different characteristics it is important to experiment with the visualization in order to determine which image best presents the essential aspects of the data set.

Here are eight different ways of colouring a contour map of EM-31 apparent conductivity data gathered over a site with a complicated industrial history. The ground contains all manner of industrial waste, including wood waste, concrete - some with rebar, old oil tanks, railways lines, and other debris.

	1. Plain rainbow contouring, with the scale chosen to emphasize mid range values. Note that negative values are all one colour
	2. Rainbow colour contouring with gray bands. The scale's range is more inclusive for this data set, and the bands help provide some sense of the spatial behaviour within colour zones. The effect is similar to including contour lines on top of a colour map.
	3. This effect helps show both large scale changes, and small changes within any range.
	4. Non linear colour scales, and scales with abrupt changes in colour can help map the behaviour of specific values within the data set. However it is easy to give the wrong overall impression because the eye is drawn to the abrupt changes rather than being allowed to examine the entire data set in an un-biased way.
	5. Black and white contour plots are often necessary for reports because colour photo-copying is still more expensive and less common. Your choice of low = white high = black, or the opposite depends on the data set, and on your opinion of what works well.
	6. This is the identical scale with black and white values reversed.
	7. The same effect as number 3 can be achieved in black and white. Again, the effect is similar to, only better than, including contour lines on top of a colour map.
	8. This type of contour map is perhaps outdated. The large scale behaviour is apparent, but information about data between contour lines is completely hidden. A different choice of contour interval, and a different starting contour would provide a different impression of the data set.

As an example of considerations to bare in mind, in the first image there is reasonable detail visible between 70m and 100m north, where apparent conductivities are around 25 mS/m. However all the highest apparent conductivities are red, so details above 100 mS/m are obscured. The second figure shows these features at high values in more detail but structures around 25 mS/m are harder to discern.

All eight methods of plotting this data set are shown in thumbnail images here, for more direct comparison of the overall effect of each colouring scheme.