We continued with our attempt to discover Mayan pyramids in the Guatemalan jungle using Landsat images. This involved using NDVI (Normalized Difference Vegetation Index) and also mixing some of the Landsat bands to try to find a combination that allowed us to identify a pattern we could use for searching for additional pyramids. Once we identified a pattern (not so easy, more below) we would train ArcGIS to identify this pattern with pyramids and analyze the entire 4-5-1 band composite for us.
This was a difficult lab for a couple reasons. First, ArcGIS just seems to have some inherent performance bugs when dealing with the raster operations for these large files. I have a fairly high-performance computer and it really came to a crawl in a few instances (restart/reboot sorts of instances). Second, the known Mayan pyramids that we were looking at were never more than mere smudges on the (fairly) low resolution Landsat images. I could see the Mirador pyramids in the ERSE imagery basemap but really couldn't see anything definitive in the Landsat images regardless of the processing we put it through.
Overall, I liked the idea of the lab and I think it could be used to good effect if we had higher resolution images to manipulate and train.
No comments:
Post a Comment