Friday, April 26, 2013

The really Big picture

from Jason Ur's website at Harvard University

Bjoern Menze and Jason Ur have combined archaeological data, satellite imagery and spatial analysis in a way that really brings Big Data to settlement archaeology. The scope of their project is really staggering: 8,000 years of settlement history, about 14,000 sites, covering about 23,000 square kilometers at a 15 meter resolution. I am not going to claim any expertise on the big questions that relate to Mesopotamian archaeology, although I certainly am interested in them. What I like in particular about this article is that it puts the pressure on the rest of us to put these data sources (which are getting more and more accessible, and less and less expensive) together with the "traditional" archaeological data we already have. All I can come up with for an excuse as to why I don't have this kind of Big Data for the Llanos de Mojos is that there are more trees and more vegetation, but I am beginning to wonder if that excuse really works anymore.

photo of flooded forest near the confluence of the IruyaƱez and Omi Rivers (John Walker)


  1. Not an excuse! Your sites and landscape features have spatial and spectral characteristics with highly seasonal variability- the methods should be transferable. The biggest problem for archaeologists is to find a collaborator like Bjoern, who is both interested in archaeology and has the requisite computer science/vision background.

  2. Oh hurray! A comment! Thanks for stopping by. I do think you are right, because the rate of our forest islands having evidence of occupation is something like 80%, and they are very easy to spot. And of course the fields are fairly easy to spot, too. I do think you have more of a backlog of data to sort through, but then I could at least imagine that as a problem as well as a blessing.