A Quick and Dirty GIS Archaeological Probability Model for Broadband Installation

I am in the process of doing an archaeological survey for some fiber optic installation; almost all of the installation will occur in existing road right-of-way (think road ditches).  The projects are part of the USDA Rural Broadband initiative.  Since Federal money is involved, the environmental review process includes a cultural resource survey.  For some surveys, the appropriate level of effort has been determined to be: a)watch out for known sites, and b)take a quick look at high probability areas.  [Readers should note that their has been vigorous debate about what level of survey should be done; I’m describing what has been deemed appropriate for the projects I am working on today.   This may or may not be the best or most acceptable approach.]

The technical issues of such surveys are substantial, as installation can include hundreds of miles of cable (I’ve done about 7000 miles of such survey in the past 5 years).  The most cost effective way to proceed is, of course, using GIS.  As I sit here watching data process, I’ll spell out what I am doing.

1)Get the cable route layout from the client.  Usually the route is traced onto USGS Topo sheets.  If I get the topo sheets in paper format, I scan them; more often than not I get digital versions from my client.  I then georegister the topos and digitize the routes.  I usually do this using Delorme XMap 7 editor.

2)Identify known sites within 1 mile of the corridor.  This is done by a standard record search at the State Historic Preservation Office files.  Right now I am working on projects in Minnesota.  Minnesota does not have an archaeological site GIS, which makes this step time consuming and cumbersome.  In Minnesota, this step usually takes about a week and a half.  In states with a site GIS (e.g. North Dakota, South Dakota, Iowa) this step usually takes less than a day.

3)Induct site boundaries of known sites in the GIS.   For states with an archaeological GIS, this is easy, and usually involves asking the State GIS manager to send the data (takes a few hours).  In Minnesota, this is much more cumbersome, and takes a week or so. I pull and scan the site files (because MN still uses paper records).  I then have to go through the site files, find the maps, and manually enter the site boundaries in the GIS.  I use Delorme XMap 7 for this. 

4)Identify sites within 1000ft of the corridor.  Now that the route and the sites are in the GIS, things get a bit easier.  I create a 1000ft buffer around the cable corridor and run a query that identifies sites that intersect that buffer.  These sites are then flagged for a site visit and individual consideration.

5)Iimagedentify areas of high probability based on proximity to water.  For these projects I am using some shortcuts rather than a detailed archaeological probability model.  For most of Minnesota, 80-90% of the known archaeological sites are within 200 meters of a perennial water source.  I download the relevant 1:24k lake and streams files from the MN DNR (at the DataDeli).  I bring those files into QGIS (an open source GIS program).  I create a 200m buffer around the lakes, and do the same for the streams.  Some of the streams are actually drainage ditches, but I just leave them in as it is faster to check those spots in the field than mess with the GIS coverage.  This generates 2 buffers which I merge into one polygon coverage.  This coverage defines my high probability areas.

6)Add some Historic Map data.  I already have GIS coverages of the trails and villages in Minnesota at the time the General Land Office (GLO) Survey was conducted (1890s).  In Minnesota we are fortunate to have the work of J. WIlliam Trygg, who created overview maps based on the GLO maps and a few other sources.  These Trygg maps made my work pretty easy.  I add a 500m buffer around these trails and villages and add it into my already defined high probability areas (using XMap 7).

7)Add some caution.  I create a buffer from known archaeological sites, defining anything within 500m of a known site as high probability. Using XMap 7, I add this to my already defined high probability area.

9)Identify cable in high probability areas.  I bring the cable route as digitized in XMap 7 into QGIS. I use the high probability area coverage as created by the steps above to clip out the areas of the route with high probability to contain archaeological sites.  The sections of the route clipped are exported and put back into XMAP. These are the high probability areas.

10)Field check!  I go out and do a visual inspection of the route within 1000ft of known sites and the route with high probability areas.  I document areas were the route avoids the sites.  I also document areas where the high probability concerns are moot (e.g. the road has been ditched to a depth greater than 3 ft).  Any areas of concern remaining are dealt with in a case-by-case basis.


(And if you are wondering, yes I am giving some info away to the competition. If that improves the quality of fieldwork being done, I’m glad for it.  There is also the reality that doing the above seems pretty simple, but it takes a reasonably expert GIS user and some reasonably serious hardware and software to do this.  I inevitably encounter some projection issues, and some of the dataset sizes are large.  Buffers, merging and clipping are processor/memory intensive.) 

2 thoughts on “A Quick and Dirty GIS Archaeological Probability Model for Broadband Installation”

  1. As site predictive models is one of my areas of study I really loved the post. Not sure if you were using this as an example and so you might know this but Minnesota has a statewide site predictive model- http://www.dot.state.mn.us/mnmodel/

    From my understanding it works pretty well and anyone can access it (at least it use to be that way). That might save you some time in the future