Andy Woodruff once wrote an article on free cartography, making the point that our profession is undervalued and issuing the rallying call, ‘Maps are easy. Good maps are not. Your skills are valuable.’
And yet free cartography has rapidly become a necessity in the era of open geodata. Applying good cartography to geographic data immediately makes it more useful and more usable, as is explained in more detail in my guest post on the GeoVation blog.
At Ordnance Survey, we have been spending an unbelievable amount of time on creating stylesheets for our data products. It has become CartoDesign team’s biggest project of 2014 and arguably our most important too. And we are not the only ones doing it. For example, there has been a lot of chatter on Google+ with the guys on the QGIS User Group happily sharing around styled layer descriptors and QML files for Phase 1 of the UK Habitat Classification data.
QML is a variant of the popular Extensible Markup Language (XML) computer file format. It is used by the open-source Geographic Information System (GIS) software Quantum GIS, now more commonly known as QGIS, which is used to create, edit, visualise, analyse and publish geospatial information, and offers an alternative to proprietary GIS software such as ArcGIS or MapInfo.
The business is keen for us to continue with our styling strategy because it unlocks so much value to our products and in turn makes them easier to both resell and market.
I have just finished working on what I believe to be our best QML stylesheets to date. This latest set are for OS VectorMap Local, the first of our new generation of stylesheets to support a licensed (paid-for) product.
Creating QML for OS VectorMap Local was a fairly difficult task. Our strategy is to style all features, even if they are turned off by default, in case the customer wishes to display them; and if possible to only load the data in once, i.e. we try to maintain a one-to-one relationship between QGIS layers and data files.
There are also some bugs with the software. It turns out many are fixed in the master build but this is little use if you are not a code geek and/or you are working on a managed version of Windows!
OS VectorMap Local is by far the most detailed Ordnance Survey product I or my colleagues have styled so far. This created many problems, for example styling different features from within the same layer based on more than one variable, e.g. multiple classifications as well as scale. This particular example was resolved with the use of rule-based styling. Similar issues rose with labelling, where some features (from within the same layer) needed to be different colours, weights or even not shown at all. Thankfully version 2.2 of QGIS allowed me to resolve many of these issues by using expressions. Alpha channel transparencies were also used to turn off features we don’t display in our default style, e.g. contours. By doing this all the user needs to do should they wish to style contours is to turn the alpha channel back to 255 (on). Another example of a decision I had to make was in choosing to use scalable vector graphics (SVG) for vegetation patterns – the first time we have included vegetation in external OS VectorMap Local stylesheets. Although the user has to go through initial hassle of copying the folder across, the results are so much crisper than using a font or even standard symbols and this process only needs to be done once.
These OS VectorMap Local stylesheets are available on GitHub here.
Away from Ordnance Survey, I have also been working on QML files for the newly released to open data, EuroGlobalMap from EuroGeographics. This nominal 1 : 1,000,000 scale map is a very complex and confusing dataset but it contains a whole load of useful data and I was happy to see it released as open. But in its raw format it is so hard to get one’s head around that some stylesheets are surely going to be useful.
My colleague Charley (Glynn) had a similar idea and so we have both been working on two very different styles for the product:
I have created a style based upon the OS Full colour style seen across our contextual mapping portfolio, including OS VectorMap (Local and District), Meridian 2 and Strategi. Just like our own products, I have tried to style as much of the data as possible, which is a lot of features. My belief is that this will be useful to anyone already working with the Ordance Survey products who wishes to be able to add further scales for Europe to their portfolio in a style which is consistent.
My ‘OS’ full colour style for EuroGlobalMap remains a work in process, nevertheless it can be downloaded from GitHub here.
Charley has been more selective in the data from EuroGlobalMap and has used the elements required to make a simple contextual or reference map. He has created a new style of his own and I think it looks quite neat. This is also available on GitHub along with other stylesheets he has played with.
So if we can produce cartography for data products for free, then why are paper maps so expensive? Well I can offer two reasons for this, but I must stress that this is all my own personal – albeit educated insider – opinion:
First of all data products bring in important revenue streams to most mapping agencies and businesses. By adding value to them we increase revenue. Even by providing stylesheets for open data, we are selling the package and so promoting our premium products.
Secondly, creating and publishing paper maps is unfortunately relatively expensive. When you take into account the amount of man hours and salary that go into making every paper map, then add on the cost of printing and selling (which alone is roughly equivalent to the sale price) then actually paper maps represent very good value to the consumer.
So taking these two points into consideration, the cartography customers receive is often free cartography; and maybe Andy Woodruff was right, maybe we remain an undervalued profession.