Trig pillars today

The humble trig pillar is an iconic part of the Great British landscape. They’re part of the vast original retriangulation network of over 31000 points that included lighthouses, turrets, and beacons. With substantial foundations, 6500+ of these weathered pillars can still be found standing today as loved elements of our landscape.

Hexagonal cartogram, illustrating both trig pillar occurrence and elevation across Great Britain

To celebrate the 87th anniversary of the Trig Pillar, our Geospatial Graduate Alice Cunningham describes how she used a bivariate cartogram to analyse trig pillars today, to show us where trig pillars are concentrated, and how does their elevation vary across the landscape.

Step 1: Finding and filtering the data

To find some data to work with, I took a dive into the OS Complete Trig Archive.

This is a legacy dataset that is no longer maintained, but a good open-source representation of trig pillars in Great Britain.

When creating any visual representation of geospatial data, it’s important to take a look at the data first. My preferred programme of choice is Safe Software’s FME – but you can use the attribute table within your chosen GIS, or various programming languages to achieve the same result. So, let’s take a look at what we’ve got:

FME Workbench, used to filter the OS Complete Trig Archive

First, I used a Statistics Calculator to summarise 3 variables in the archive:

· TYPE OF MARK (including trig pillars)

· DESTROYED MARK INDICATOR (a binary value assigned for the status of the trig point)

· HEIGHT (the height of the trig pillar above sea level)

This told me that there were roughly 6500 pillars still standing across the UK at the time of archived data publication, and that their height varied between -1 m (Little Ouse, near Ely) and 1345 m (Ben Nevis) (accounting for the trig pillar height itself).

After using a simple Test Filter to extract non-destroyed pillars only, I wrote this out to a shapefile.

Placing the data in ArcGIS Pro illustrates the density of the data to work with

Step 2: Creating a hexagonal grid

I popped the shapefile into ArcGIS Pro, using the ONS Countries (Dec 2021) GB BFC outline for reference. There were far too many trig points for the reader to comprehend on a single scale.

I could create a heatmap, but to keep things even and neat I decided to create a grid of hexagons to represent data. Using an equal-area projection, I ran the “Generate Tessellation” tool for transverse hexagons.

It is often best to create multiple grid sizes before settling on your final visual. Using the formula for hexagon area on the ESRI toolset, I decided that hexagons with a side of 20 km was detailed enough to represent the data without cluttering the view

Using the Generate Tessellation tool in ArcGIS Pro to create a hexagonal grid across Great Britain

Step 3: Performing a spatial join

At the moment we have a grid of hexagons, but we don’t have any data assigned to them. Often we manage data using joins and relates, perhaps joining data on the basis of a common attribute.

Here, the link between the hexagon and trig archive is purely spatial – i.e., how many trig pillars occur within the area of a given hexagon? By assigning a spatial join on the hexagonal grid to our trig pillar shapefile, we can count the number of trig pillars inside that hexagon.

By altering the legend, we can communicate that height actually means elevation above sea level in this context

Step 4: Playing with bivariate symbology

Taking a step back from our map so far, we’ve got not one but two variables to play with here: the count of trig pillars in each hexagon, and the average height of trig pillars in that area.

We could create two separate maps with a single hue colour ramp, one for occurrence, and one for elevation. But thinking back to their purpose, trig pillar occurrence and their elevation are intrinsically linked.

There are several ways to visualise two linked variables on one cartogram, but in this instance we can make use of colour to blend these variables. Not only does this declutter the visualisation, the blending of two colours (called a bivariate colour scheme) purposefully indicates the fundamental link between the two variables. This creates what’s known as a “bivariate” cartogram. I’m calling this a cartogram rather than a map as the hexagonal grid is a statistical diagram of the UK, not intended as a cartographically accurate map.

By using ArcGIS Pro, I amended the automated legend from “HEIGHT” to “Elevation”, to better communicate that it’s the height above sea level, not the height of the pillar itself.

ArcGIS Pro has some helpful CVD filters, that should be used to check your colour scheme before sharing your design

Step 5: The all-important colour scheme

We need to choose a colour scheme that accommodates for varying Colour Vision Deficiency perception. Helpfully, ArcGIS Pro has a CVD filter that helped me choose a blue-green colour scheme that improves CVD accessibility for the cartogram.

Adding an outline to the map

Step 6: The finishing touches

Taking the layout out of ArcGIS Pro, I used Adobe Illustrator to add the finishing touches to this map. Adobe Illustrator isn’t strictly necessary here, and on other projects I have opted for Inkscape as an open-source alternative, or solely used layouts in GIS.

Some of our hexagons were left empty from the spatial join, so these need to be filled in to represent “holes” in our data. Coastal hexagons with no data were omitted due to the spatial join process, meaning this map is a generalised representation of Great Britain. We tend not to use solid black for these hexagons on a GeoDataViz layout, as this creates unwanted focal points and dominates the visualisation.

The final touch was a grey outline and gentle blue halo, used to symbolise the ocean without using a solid fill, creating the impression of the cartogram “popping” off the screen.