Home       Chase County Overview       Makesig       Maxlike       Mindist       Piped       Cluster       Isocluster       Results   References

  


Results


After completing maxlike, mindist, piped, cluster, and isocluster classification techniques on Chase County, Kansas. I discovered that the maxlike module gave the most accurate results. I believe the maxlike image I created is very near the existing Chase County landcover. The mindist image was fairly accurate except that it misrepresented rock/pavement and bare soil in many areas. The piped supervised classification technique took little time to accomplish, but much of the image was left unclassified so the results were inconclusive. I found the unsupervised techniques were easy to perform but the results were inaccurate, especially at distinguishing water bodies from other features.

One thing that was important in doing the supervised classification was knowing the study area well. When creating the signature files I had to ‘ground truth’ the sites that I digitized to be sure they were classified correctly. Also, I discovered that classifying an image involves many processes with many variables. Thus, the end results may differ substantially by creating new signature files, using different composites for seeding, or adjusting other options.
















  
Home       Chase County Overview       Makesig       Maxlike       Mindist       Piped       Cluster       Isocluster       Results   References