I am working on a project where I need to identify the area of a hilltop that is relatively flat within which stone structures could be built before the slope becomes too great to build without having to modify/terrace the hill.
Specifically, I am doing a project with known pre-historic settlements, all of which are located on hilltops or spurs. I am trying to get an estimate of settlement size, assuming the inhabitants only built structures following the natural topography. This isn't always the case - there is evidence of intensive terracing in some areas, but this is relatively uncommon and only exists in the largest settlements. In all others (following physical visits to 10% of the total number of settlements I am working with), the limits of the settlement follow the topography - once it starts to become too steep to build houses, a wall was built around the settlement.
I have been trying to think through how I could use GIS to identify the settlement extent for settlements that have never been formally studied following what I have seen. I have a relatively accurate DEM with 1 arc-second horizontal resolution. I am using ArcGIS 10.5.
I am thinking Topographic Position Index will be my best bet, but have had issues that I think are connected with neighborhood size and have never quite figured this out, despite trying to use various TPI tools.
Or perhaps TPI is more complex than what I need? Perhaps I could just generate slope lines and identify the line where the slope reaches a certain point and identify the point at which the slope goes over a certain value (e.g. 10%)?