Using Bivariate Symbology on a map of housing prices

We’ve been looking at different mapping strategies lately and this time we’ll look at an uncommon one, uncommon enough that people are likely to be curious about how it works. I will admit that I, too, have been curious, having never really used it before. The method? Bivariate Colors. In their own words, ESRI says this method “shows the quantitative relationship between two variables in a feature layer,” whatever that means. The basic idea is that you’re trying to map two different quantities at one time using solid colours to show the differences.

In theory, this allows you to show how a feature varies across two attributes. It’s two mints in one! And a potential doubling of displayed information – or possibly doubling the confusion… This came to mind when working with postgrad Reina who was wanting to map study site locations in terms of two variables: whether a farm was a monoculture or a diverse operation and whether the study was done at the farm or catchment scale. Initially we thought she would need two separate maps, but perhaps Bivariate Colors would work here? Let’s find out.

We started with a layer of 185 study sites within Aotearoa spread across the motu:

In the table I’ve got an attribute that tells me at what scale the study was done (Farm or Catchment) and another about the operation type (Monoculture or Diverse):

To get started with Bivariate Colors, we can choose that as the Primary symbology:

This opens the settings window:

Field1 and Field2 are where I can set the attributes to be mapped, but one of the first things I notice is that it doesn’t honour text fields, meaning I can only use numerical types for this. So I added two new attributes, ScaleNum and DiverseNum and populated them so that,

  • Scale: Farm = 1, Catchment = 2
  • Diverse: Monoculture = 3, Diverse = 4

I don’t think it’s essential that the numbers are all unique; these could just as easily be 0 and 1, methinks.

Now that I’ve got numbers I can set this up and choose a colour scheme:

And here’s how that looks on the map:

Interesting, but does it work? Can my reader easily know what each point means? I stared at this for a while and ultimately thought, nah. But perhaps a bit of relabelling would help clarify what Low and High mean here:

Setting the Labels clarifies a bit, but is it working yet? I didn’t think so and was about to give up when I thought I might try one more option, again with the labels aiming for something more clear:

I’m a lot happier with this – this clarifies things a great deal for me but I had to double-check myself to ensure I had things right (i.e. clicking on sites to make sure I’ve got the right combinations for the right colours). And here would be my draft of a final map for public consumption:

I’ll say “draft” as the legend needs a bit of tweaking to get all the labels displaying but I think it sort of works? What do you think?

I have to ask myself after this little jaunt down a less-oft travelled Symbology path…was it worth it? Arguably, I could have also done this with Unique Values by assigning a unique value for each combination and ended up at the same place perhaps a simpler to understand legend:

Same information, just symoblised in a different way. Which is better? I think I would lean towards the second one, as it’s just simpler to interpret. In this case, Bivariate Colors doesn’t really add anything to our cause; it may, in fact, complicate it. But there are cases where it would work well.

I guess this gets down to the GIS craftsperson knowing their tools and which one to use for the right purpose. I suspect this one might spend most of its time in the toolbox, but good to have a better sense of what it can do.