Global Data Hound

When war broke out between Russia and Georgia last month, with the dispute focused on the province of South Ossetia, many of us hurried to our global maps to remind ourselves (or learn) exactly where South Ossetia is.  (Here in the geographically challenged USA, we all know already  that Georgia is the home of Coca-Cola, CNN, and Jimmy Carter).

The war was a salutary reminder that what geographers like to call administrative boundaries matter intensely.  We take for granted, because it has been the normal state of affairs for a couple of hundred years, that every single square inch of land or water on the face of the Earth is claimed by at least one nation-state.  This is not an innate property of the geo-biological system; it is not even an innate property of human civilization.  (It may, however, be an innate property of agricultural civilization, which is premised on maximizing the yield from a fixed area of land).  Up until a couple hundred years ago, vast swathes of the Earth’s surface were occupied by loosely organized hunter gatherers and nomads who had neither the inclination nor the means to create administrative boundaries. Today, the rather bland phrase “administrative boundaries” reflects a more pungent reality, which is that every square inch belongs to someone.

If you are wondering about the possible exceptions, such as Antarctica and the open ocean, note that there are in fact many national  claims to Antarctica (see the excellent Statoids site and Wikipedia); they are simply held in abeyance by a treaty regime that is specifically intended to defuse conflict. Similarly, the open oceans past the 200 (!) mile Exclusive Economic Zone are held “as the common heritage of all mankind”, but that heritage is defined by a treaty (the Law of the Sea ) which exists only because most nation-states have voluntarily agreed to a set of restrictions on their behavior.  Unfortunately, the treaty regime in the Arctic is far less clear, and there is increasing tension among the states that border on the northern ocean.

The central issue in the war that began on 08/08/08 (in the self-dramatizing Russian view) or 09/08/08 (in the self-absolving Georgian view) is, of course, whether the square inches in South Ossetia and Abkhazia belong to Russia or to Georgia.   In August, our TerraViva! product thought prewar boundaries of Russia and Georgia looked like this.  The bad news is that this was wrong; the good news is the story of how we found out and fixed it.

As soon as the war broke out, I spent some time mucking around with TerraViva! and the Web to see what more I could learn about Georgia. One thing I noticed quickly is that  Georgia and South Ossetia are in close proximity to war-torn Chechnya and Ingushetiya, which I remembered as the scene of the horrific Beslan massacre.  These fundamental geographic facts shed some light on the Russian attitude towards the Caucascus.

Another thing I noticed was that our current map of Georgia showed five first-level administrative regions: Abkhazia, Ajaria, Tbilisi, Yugo-Ossetia, Tbilisi (excluding city), and Tbilisi (city), whereas the Wikipedia entry about Georgia specifies twelve first level administrative regions and even has a nice map.

How to resolve the conflict?  At first, surfing only made me more confused.

A credible source, the FAO, provided a 2004 map of first level administrative boundaries in Georgia that agrees with our version.  The source for the FAO map is the United Nations Office for Coordination of Humanitarian Affairs (UNOCHA), also credible.

Unfortunately an equally credible 2006 map from the UN High Commission on Refugee showed twelve administrative units.

Then this page from the 2008 quarterly census of Georgia listed Tbilisi, the Adjari Autonomous Region, and nine provinces: eleven!  What is  missing from the Georgian census is the Autonomous Republic of Abkhazia or, as it is now referred to by Russia, the Republic of Abkhazia.

What was going on?  It turned out that the five-province version of Georgia corresponds with Georgia’s administrative boundaries when it was a province of the former Soviet Union.  See this excerpt from a 1989 map of the former Soviet Union.

We did a little digging at our end, and discovered that, for Georgia’s boundaries, we had relied on the the 2001 Oak Ridge National Laboratories administrative boundaries data set, which used the Soviet borders.

Once we realized that we had an out-of-date set of boundaries for Georgia, we went looking for a better set.  We found them at the Global Administrative Units database at the University of California Berkeley. As always, there was a wrinkle … it turns out that while Berkeley lets users download individual countries, that was version 0.8 of their dataset, and when you download their whole global dataset, that’s version 0.9, which is better.  Once we overcame that issue, we incorporated the maps into TerraViva, with the following result.

We have now incorporated the new set of boundaries in our Sub-National Political Boundaries data set. Since this data set is a part of the TerraViva! core product, Global Data Viewer users will receive the new data as part of the next automatic software update (i.e, whenever you launch GDV and you are connected to the Internet).

It sometimes strikes me as amazing that fifty-plus years after the invention of Geographic Information Systems (GIS), representing a cumulative global investment of tens if not hundreds of billions of dollars, and the creation of surely millions of unique global maps, there is no single,  official, authoritative, centrally distributed digital set of political boundaries for the world.  Wouldn’t all global mapping be a lot easier if there was a single source for administrative boundaries, like an atomic clock?

Unfortunately, the world is a lot messier than that. The fundamental reason why there is no single source of boundary data is that there is no world government: sovereignty is divided out among the same nations that debate their boundaries with each other.

To be sure, there are a few quasi-standards, like the Country Boundaries that come standard with ARC/GIS ,  VMAP0, and the UN’s incomplete Second Administrative Boundaries Level (SALB) data set, but none of them is truly definitive, and they are only as good as the effort put into updating them.

As the story above indicates, we keep a close eye on our political boundaries data, and we update it as often as we see a need. We regularly compare our list of 1st level entities for each country against other sources, such as the CIA World Factbook. When we see that a change has occurred, we then try to locate a database that contains the new border definitions. This definition may explicitly define the correct 1st level unit, or it may contain smaller administrative units that can be combined to obtain the needed 1st level entities. For example, Denmark reorganized their 1st level units from 14 counties to 5 regions in January 2007. The borders for the new regions couldn’t be obtained by a combination of the existing counties, but they could be created by the correct combinations of the 2nd level municipalities.

Once we are satisfied that we have good borders for the new 1st level units, we then fit these borders to our coastline mask, which we’ve derived from the SRTM Waterbodies dataset at 1 arc-second of resolution. ISciences also takes pride in the high quality of our coastline boundary data.  Fitting the borders to our water mask involves not only substituting our coastline for any existing in the new border definitions, but also making sure that borders along estuaries and between islands correctly fit our coastline. After the borders are fitted to our coastline, we then merge them into our global political boundaries definition, meshing the borders of this corrected country with its neighbors.