Mapping Internal Administrative Boundaries (part 1)
Several weeks ago, Marissa and I attended an event at Columbia University’s Lamont-Doherty Earth Observatory on mapping internal administrative boundaries. The purpose of this conference was to bring together professionals from the public and private sectors and from all different backgrounds (GIS, human geography, cartography, etc) to have a discussion about the practices and problems of mapping internal boundaries. There are a variety of freely available data sources for boundaries, such as GADM and Natural Earth, however there are also a multitude of commercial providers who maintain and license their own internal boundary data. And depending on how the data is being used, one of these options might be better than the other, but it is not always a straight forward process to make this decision.
Although our primary focus here at International Mapping has been on international boundaries (often referred to as “admin 0” or “ADM0”) and disputes, there are obvious overlaps in dealing with the sub-national boundaries nested hierarchically within national boundaries. We very often wrestle with internal boundaries when making maps for our clients, trying to ensure the data is up to date not only in the geometry/shape of each administrative unit, but also in the naming and attribution.
In this post I hope to summarize some takeaways from the event as well as some of the primary difficulties we face when mapping and maintaining admin boundary data. I refer to this post as part 1 because in an upcoming post I will go over some ideas for moving forward.
Background
To kick off the event, we heard from a panel of several speakers about their use cases and challenges with internal boundaries, and on the importance of mapping sub-national administrative units. One longer term effort has been led by the UN with its Second Administrative Level Boundaries (SALB) project, which was launched in 2001. In short, the goal of SALB is to aggregate internal boundary data directly from UN members states, collected from authoritative sources such as individual National Mapping Agency or relevant divisions of government. Nearly two decades later, the project’s goal has not fully come to fruition, primarily due to a lack of participation from member states. At the time of this writing, the SALB site lists datasets contributed by only 45 of the 197 states listed.
The event also featured a number of lightning talks from professionals in both the public and private sector, discussing their interaction with internal boundaries or showcasing products they’ve built around them. It was quite insightful to hear common (and familiar) threads on the difficulties so many are facing when dealing with boundary data.
Difficulties in Mapping and Maintaining Administrative Boundary Datasets
The list below comes from way too many hours spent dealing with administrative boundary data, but as mentioned above these frustrations are shared by many:
- Finding “authoritative” data
In some cases, internal boundary data is distributed by countries via the web and maintained by some designated agency. In these rare cases we would consider the data authoritative, as who else but a country themselves can decide on their internal organization? However more often than not data isn’t available in this way, and it is difficult to tell whether data from other sources can be considered reliable, well-sourced and/or up to date.
Sometimes, a list of administrative units might only be found in a primary source or authoritative piece of legislation. However with no spatial data or map accompanying the legislation, how are we to handle these instances? I recently found this to be the case while attempting to update Mali’s cercles (ADM2 divisions). A set of 2012 decrees established a number of new ADM1 regions and defined the cercles making up each. After some time spent researching, I came to discover that the legislation was way ahead of the reality on the ground, and that we likely won’t see the newer divisions come to fruition for years to come.
- Licensing, permissions, copyright and use restrictions
Associated permissions, licenses or use cases attached to a data source can restrict a user from any commercial use, or the modifying and combining of data with other sources. This often leads to datasets being updated and utilized on an internal basis, and not shared back to the broader community for fear of violating permitted uses. In many cases permitted use can also be vague or confusing, with even the “authoritative” datasets mentioned above having tight restrictions on use. - Assigning of codes or unique identifiers
Internal administrative units are often used for mapping tabular or statistical data by creating choropleth or other maps for data visualization. A major component of this is tracking unique identifiers for each unit in order to join the tabular data to the spatial. There are a number of methodologies and standards for these unique identifiers, and it is up for debate which is best, however it is often likely that the datasets you download will require some additions or “QC” to ensure an accurate join to your spatial data. - Resolution or intended scale for data
This is how detailed the data is. We describe Sovereign Limits resolution as holding up at scales of 1:25,000 and larger. More often, Sovereign Limits boundaries have been created at even larger scales than this. For admin units that we source, the resolution of data will rarely hold to those standards. We have learned to expect huge variation when it comes to resolution in administrative data from different sources. It will always be ideal to locate and make use of the highest resolution and up to date data available. - Maintaining topology
This is the “connectivity” of the spatial data. Unintended gaps or overlaps between admin unit data can be seen as problematic, as this type of data is intended to align like puzzle pieces. So, a certain amount of processing and data validation is often required to make sure the topology is sound. While nice and tidy topology isn’t required in spatial datasets, it is often a sign of a sloppy data quality when polygons crisscross one another or aren’t aligned seamlessly. Editing or processing data can introduce its own errors, so it is often a time consuming task to ensure that the topology is sound. - Coastline
Coastline data comes in various resolutions, and in many cases can be quite cumbersome to deal with. Admin unit data sourced for coastal states will nearly always include a coastline in the dataset, and the coastline for one country dataset is extremely unlikely to match the source or resolution used in the coast of another. This means that, for a cohesive dataset, it is necessary to process admin unit data to either 1) a single coastline source or 2) to ignore coastline and extend admin units to the edge of a territorial sea, avoiding the difficult task of maintaining complex and cumbersome coastline data. - Disputed areas
The way one views the world’s boundaries will have an effect on the treatment of administrative units in many places. For example, if I am creating a map of Georgia (the country), should municipalities within the breakaway republics of Abkhazia or South Ossetia be included? Most often the answer depends on your or your organization/client’s geopolitical views, adding a further complexity in dealing with this data.
There are plenty of other fun issues that come up when dealing with this data, but I believe the above are the most common and pertinent. As said earlier, in a future post I’ll plan to share some thoughts for strategies moving forward, but in the meantime I need to check into the status of Mali’s cercles again.