A Colour Ramp You Will Love or Hate

A Colour Ramp You Will Love or Hate

You are going to love this, or hate it.

Take a look at this colour ramp below (that’s the colour palette we use to display data on a map).

To the undiscerning eye, this looks pretty innocuous. It appears to meet good standards (https://earthobservatory.nasa.gov/blogs/elegantfigures/2013/08/06/subtleties-of-color-part-2-of-6/) for a colour ramp so that:

  • It is mostly driven by gradual brightness change from one end to the other; ?
  • The change is gradual and even from low to high values; ? and
  • It works for all types of visual colour blindness. ?

So why does it cause such a stir whenever I use it on a map?

Here’s why…

Let’s take a map such as the WHRC Pantropical National Level Carbon Stock Dataset (available on Earth Engine). Their default visualisation looks like this (as shown in the header image too):


There are serious problems with this colour palette. Notice how there is a sharp perceptual boundary when it goes from green to orange, even though this boundary has no physical meaning – it doesn't mean the vegetation cover has stopped. And the lowest values are white, which means that if you have a white basemap (such as a roadmap), you can’t tell the difference between low values and absent data.

An alternative might be “Greens”, a colour ramp that, for example, Meta/WRI recently used to represent their 1m canopy map. But that also suffers from low values tending to white, making low values difficult to distinguish from “no data”.

At Earth Blox , we like to use “Speed” (from cmocean) because the progressive “greenness” still makes sense for forest properties, but the lowest values go to a pale yellow instead of white, making it easy to distinguish low values and “no data”. (We take our time at Earth Blox to have default colour ramps that make it easy for users to just crack on with their analysis without having to think too hard).

This is what the same data looks like using Speed:



There is a lot to commend this map, such as the low range of hues that keeps it clear and simple, and the progressively darker green being an intuitive indicator of denser forest.?

But now, look at the same data using the new colour ramp introduced at the top of the page and pay attention to your immediate response.?



Some of you will look at this and wonder what all the fuss is about – it looks like a satellite image. Isn't that an effective way to communicate the content of the data in a way that is immediately recognisable? For many people, this makes sense as there is very little cognitive translation needed to understand this map. It instantly makes sense.?

But I’m sure that, like many of my colleagues, some of you will feel slightly uncomfortable. You will squirm in your seat a little as you look at this map and try to understand why you don’t like it.? It is data… but it looks like an image… that’s not normal…?

If you are feeling queasy about this map, I’m guessing you are probably a data analyst – someone who is used to colour palettes that represent data in unnatural renderings. Unnatural-looking maps reassure you that what you are looking at is data, not simulation.?

But it's not a simulation, it's just a different colour ramp, and a pretty simple one at that.?

We all know that beauty is in the eye of the beholder, but what about colour ramps? Let me know what you think. (And if you want the colour ramp, get in touch). ??

William Roper

Freelance GIS Developer

1 个月

Have you run it through https://github.com/matplotlib/viscm, or something similar? I like it visually, but I feel like there's less distinction than the speed example, which makes it harder (for me) to parse at a quick glance. viscm might help identify why that's the case.

Christine Ratcliffe

Researcher | Writer | Communicator - freelance

1 个月

I read 'How to lie with maps' when I was doing my GIS master's. I seem to remember a section of the book that spoke about how you can change people's interpretation of how bad crime is in their city by changing the colours you use in a crime map. One of the many things I learnt in my master's is there is that there is so much nuance in effective communication.

Dimitrios Michelakis, PhD

SaaS | Product | Analytics | Data | Technical Leadership | ESG | Climate Change | Sustainability

2 个月

I believe design is heavily dependent on the target audience and the problem at hand. If you were to present two maps to existing or prospective customers, would their decisions differ based on what they see? If not, the choice of color-coding is primarily an aesthetic one—it might help with marketing if it’s visually appealing. However, if their decisions do differ, was the decision better or worse? In the former case, the color-coding becomes critical and could significantly differentiate your map. Personally, I prefer using a single color with varying shades to represent one attribute (e.g., carbon), rather than multiple colors.

That is a very interesting post. I have always been against the multi colour ramps for the very reasons you laid out (see- I was paying attention in the data visualization class). Speed is probably my preferred option, especially for ndvi maps as tending to yellow as you approach 0 makes intuitive sense. But for data like lai I don't like how it suggests low values which are still greenish from above look like they are bare. I'd like to try the new ramp (please fire it over). I do wonder how good the contrast is if printed in B&W. Then again, who bothers printing nowadays... I think that shows my age!

Nathan Thomas

Senior Lecturer in GIS and Remote Sensing at Edge Hill University. Specialist in mangrove and blue carbon extent, change and structure.

2 个月

There is only viridis ??

要查看或添加评论,请登录

社区洞察

其他会员也浏览了