Archaeological Site Potential in the Ottawa Region

An Exercise



Once the data was in raster format, the DISTANCE module was run on the rivers, streams, lakes and sites layers (see results).

Next, each variable was assessed for its potential to contain an archaeological site. When possible, I QUERIED the variable against the site location and viewed the resulting HISTOGRAM. Limitations of the site location data meant that I could only do this with some degree of confidence for the distance to rivers, slope, surface geology and hydrology layers. Even here however, I made some adjustments. (View the histograms for rivers, slope, surface geology and hydrology.)

From the histogram results, I assigned a buffer of 540 metres for the river layer (this is the point when the number of site declines rapidly on the histogram). While the slope histogram indicated a buffer of 1 degree, I determined this to be too restrictive. Many predictive models define slopes up to 15 degrees as favourable. I chose 5 degrees as the cutoff for favourable slope - although I suspect this still very conservative. I used the frequencies of site location on the various surface geology and hydrology types to reclass the layers in terms of 'favourable' location on a 0 to 255 scale (see results).

Where I was not confident of the histogram results, I turned to other archaoelogical predictive models. The histograms of sites against distance to lake and streams called for larger buffers than those selected for rivers (and in the case of the lakes histogram indicated some sort of bimodal distribution). This is likely an artifact of the selection of located sites - i.e. near rivers in the Ottawa area. The discovery that sites were farther away from lakes and streams than rivers is not surprising in this circumstance and it would make no sense to use the histogram to assign the buffer sizes in these cases.

Simple predictive models place buffers of various sizes around the cultural and environmental factors selected. While the size and number of the buffers and the cut-off points for high, medium and low buffers are generally determined by an examination of known sites, it is the case that larger buffers are created around larger features - i.e. large lakes get larger buffers than smaller lakers, rivers get larger buffers than streams. The rational behind this is as follows: the larger feature is more likely to contain a larger amount of any one resource - be it fish, plants or transportation opportunities and therefore be able to support a larger number of people and generally be more appealing as an area to contain a site. Following this logic it would again be misleading to create larger buffers for streams and large lakes than those created for rivers.

I chose a buffer of 250 metres around the large lake layer and a buffer of 200 metres around the streams layer. These buffers were arrived at by examining a predictive model for the boreal forest of Ontario (Dalla Bona and Larcombe 1996*). Finally, I chose a south facing aspect as the most favourable aspect for site location - again this was selected by examining the predictive model for the boreal forest in northern Ontario.

In terms of cultural factors used to predict site location, a buffer of 250 metres was selected for the location of known sites. In my opinion, this buffer distance was adequate for the resource.

Once the various buffer sizes were chosen, the FUZZY module was used (sigmoidal, monotonicaly decreasing, byte output data) for rivers, streams, lakes and slope. This model was chosen to indicate that suitability may not decrease with distance in a constant fashion. The FUZZY module (J-shaped, monotonically decreasing, byte output data) was run for the sites layer. This module was chosen based on the assumption that as distance from the site increases, the likelihood of finding a site decreases rapidly, but only reaches zero at infinity. The FUZZY module (sigmoidal, symmetric, byte output data) was run for the aspect layer. This module was chose as aspect favourablity increase as you go from 0 to 180 degrees, and then decreases as you move from 180 degrees to 360 degrees (see results).

Back           Onward

*Dalla Bona, L. and L. Larcombe
1996           Modeling Prehistoric Land Use in Northern Ontario in New Methods, Old Problems: Geographic Information Systems in Modern Archaeological Research, H.D.G. Maschner ed. Center for Archaeological Investigations, Occasional Paper No. 23, Carbondale Illinois