This weeks lecture/lab focused on the USGS Land Use Land Cover Classification System and ground truthing. It was a very interesting, but time consuming project. I could have easily spent 3x as long on this lab as I did, maybe more, but the lab said to spend close to 4 hours which I had already gone over. Every time I looked at the picture more features stood out and I would use different context clues, parking lot size, surrounding buildings, etc. to try to figure out what I was looking at. I would have loved to have gone more into level 3 classification as there were some specific features I was able to pick out but that would have been beyond the scope of the project. Going into this program I had absolutely 0 exposure to GIS and it's been stunning to me how quickly the time goes and how much work goes into each project/map. It's made me stop and think when I'm just perusing the internet and stumble upon a map clearly made with GIS.
After we created our classification data on an image of Pascagoula, MS, we created random sample points and use ground truthing via Google Maps to determine the accuracy of our maps. I wasn't surprised that out of 30 random points, my 2 errors were in the residential and commercial fields. These were the hardest to distinguish for me. There are obvious features that would be classified as commercial- business with large parking lots, and even schools and churches, but then there are those buildings that are smaller or used to be homes that were rezoned, or buildings surrounded by homes on all sides although they themselves are commercial that are confusing. Water, trees, and even industrial areas are a lot easier to identify. I feel like I got a good grasp on this classification system with this hands-on experience.
No comments:
Post a Comment