For this week's lab, focusing on remote sensing, we worked with an aerial photograph of the area around Cahokia. Cahokia is a large Mississippian site in what is now southern Illinois. Constructed and occupied from about the 800s to about 1300, the city originally covered about 6 square miles and contained some 120 earthen mounds. Part of this area, including about 80 mounds, are now preserved as a National Historic Landmark and state historic site. It is the largest pre-Columbian archaeological site north of Mexico.
For the lab assignment, we obtained an aerial photo from the USGS database and performed two different land cover classifications in ArcMap, one unsupervised (classified entirely by the program's algorithms) and one supervised (classified with the guidance of a set of points with user-assigned values). The resulting maps are below. The unsupervised classification resulted in only three easily defined land cover classes, while I was able to create five using the supervised method and produce a more detailed image, but both contain many errors resulting from the algorithm being unable to tell the difference between similar pixels representing different land cover types (e.g. water and dark-colored trees, or pavement vs. a barren agricultural field). However, it might be possible to refine the supervised classification by adding additional control points.
No comments:
Post a Comment