Continuing our series to introduce some friendly faces from the people working at OS and showcase the wide variety of work we do, meet Joe Harrison. Joe joined OS on our graduate scheme in 2017 as a data scientist and has worked in a few areas of the business so far. If you’re a follower of the blog, you may recognise some of the project Joe’s worked on…
Hi, I’m Joe, one of the data science grads from the most recent graduate scheme. My first placement was an eight-month stay in the Consultancy & Technical Services team from September until the end of March. I then moved to the GeoDataViz team with Charley and Paul for a two-month placement.
Scottish Boundary Commission
During my first week in the team, Charley was contacted by the Scottish Boundary Commission who wanted advice on how to visualise the new Scottish constituency boundaries and the changes. I created a couple of examples using QGIS. I improved the clarity of the maps by reducing the number of colours, and by adding shadows to emphasise areas of interest.
This was my favourite piece of work during my time in GeoDataViz. I contacted one of my old lecturers, Alasdair Rae, from the University of Sheffield to see if there were any opportunities to collaborate. He suggested we take a look at David Garcia’s “The Philippines: 100 of 7641 Islands”. We decided to create a similar map for Great Britain, mapping the 82 islands greater than 5km2 in size. This used a variety of data sources. I used a coastline shapefile derived from the Mean High Water Line in Boundary-Line and OS Terrain 5 for the visualisation, and AddressBase Plus and OS Open Names to add some extra attribution such as residential address counts and island names. After identifying the islands joined by bridges which required manual removal, OS Terrain 5 could be clipped to the island outlines and visualised in QGIS. From here, each island was exported and combined in a poster layout using Adobe Photoshop. A final filter was applied to make the colours pop. Alasdair also created a gif to explore the islands in more detail. Alasdair’s gif and a quiz is available in this blog post, and the poster itself is available from A0 to A4 on OS’s Flickr page. The statistics I produced found their way into a BBC article and a Telegraph article, and the visuals ended up in Country Walking Magazine and the I newspaper.
Having seen Paul’s Premier League Grounds poster, I thought it would be a nice idea to have a crack at doing a few more. I was tired of the focus on the Premier League (possibly due to being Sheffield Wednesday fan), so I set about putting all 92 English football league grounds for the 2018-19 season on one poster. While I was at it I thought I’d get all the major trophies won by each team by scraping a Wikipedia page.
I pulled each stadium footprint out of OS Open Map-Local and tidied up some of the generalised shapes of the other stadiums too. Once retrieved, I styled each ground in QGIS with the club colours and exported using the Atlas Feature. From here, all the stadiums were lined up in Adobe Photoshop for the final poster layout. The poster is available to view here in all its A0 glory, alongside individual images of each stadium.
As part of Geovation’s outreach programme, Charley, Paul and Luke have been running GeoDataViz Masterclasses in different parts of the country. I’ve helped out with four; in Birmingham, Manchester, Bristol and Cardiff and I’m looking to add Brighton and Newcastle to the list. It was great to get out and about and engage with start-ups interested in geospatial data and data visualisation to help solve geographic problems.
I also helped the team put together the A to Z of Cartography which ran on Twitter through June and July.
Flight path visualisation
My final project in the team involved animation. I hadn’t used data with a time element before, so this was something pretty new to me. The data I decided to use contained coordinates dropped along the flight paths of the planes used to capture aerial photography across Great Britain between 2011 and 2017. This dataset was pretty big as it contained 7,678,681 points.
A GIF wouldn’t be able to contain all the frames needed to visualise all the data at once, so I decided a video would be the final output. I made the frames for this using a QGIS plugin called Time Manager. Obviously, I didn’t want 7,678,681 frames, so to make the data more usable I only kept every 10th point. I then created a new time field to make all the years run consecutively. So instead of it running January – December of each year separately, all January flights ran at once, all February ran at once etc. This produced 59,471 frames which took Time Manager a rather painful 9 days to produce. FFmpeg was then used to stitch them all together. You can see the final video below:
In all, I loved my time in GeoDataViz. I learned some new tricks in Adobe Photoshop and Illustrator, tested out some Python and SQL and I got access to a lot of data visualisation literature which has changed my outlook and approach. Thanks to Charley and Paul for having me!
Find out more about career opportunities at OS on our website.