This week's maps were produced by running spatial analysis tools to run, first an unsupervised raster classification on an ortho-rectified aerial photo and then a supervised raster classification (wherein a point file was created that identified various points in the raster as having a particular classification. The goal of this would be to create high contrast classes of land cover in aid of using remote sensing as an archaeological prospection tool.
The results indicate that while this type of analysis may be helpful in the right circumstance and with the right kind of data, it is not necessarily the most useful approach. Land classes need to be differentiable enough to prevent mixed classes that cross different classes, causing the resulting classification scheme to obscure more than reveal.
Unsupervised Raster Classification
Supervised Raster Classification
No comments:
Post a Comment