Grand Mesa National Park Unguided Landcover Classification

William Hecker IV | February 9, 2026 | GPHY 426 - Remote Sensing | Montana State University

Preface

I have decided to hand write my own website rather than use Story Maps because:

  1. I am fed up with the Story Maps to the point of pettiness, and
  2. I did not want to come it after hours to finish this lab with ARCgis, as I use QGIS, and ESRI does not allow uploading other formats of feature tiles to ArcGIS Online.

I hope this does not create too much difficulty for your grading, as all my answers to the questions and maps should all still be here. - Bill Hecker IV

Question 1

  1. What data are you using mission/bands (ie. What does each band represent).
    2.
    Band 1 | Coastal/Aerosol
    Band 2 | Blue
    Band 3 | Green
    Band 4 | Red
    Band 5 | NIR
    Band 6 | SWIR 1
    Band 7 | SWIR 2
    Band 10 | Thermal IR 1
    Band 11 | Thermal IR 2
  2. Why are we not using bands 8 and 9 for classifications?
    3.
    Band 8 is panchromatic and uses a different resolution and that is the only reason I see for not using it(it also overlaps the visible spectrum). Band 9 is Cirrus used for detecting clouds which is not our focus.

Map 1 | RGB Image

Map 1: An RGB map of Grand Mesa National Park.
There are areas that are plains of rock and there are also area that have a very vibrant greenery. There are a couple of small lakes and reservoirs also in this area.

Question 2

  1. Where is your ROI located?
    2.
    Grand Mesa National Park
  2. When was the image acquired? Why did you choose this area? What classes do you expect to find/delineate from your unsupervised analysis?
    3.
    Image Acquired: September 25, 2025
    Reason: It was a part of Colorado that having looked over on google maps has very distinct and high contrast landforms.
    Expectations: I expect to see:
    an exposed rock class,
    a deciduous forest class,
    a boreal forest class,
    a shallow water class,
    a deep water class,
    an exposed reservoir class,
    and a cliff class.

Map 2 | Classified Image

Map 2: A map of the unsupervised classification of land cover in Grand Mesa National Park.
Notable failures of unsupervised classification is the combined classification of water and evergreen forest. It distinguished the barren land from grasslands and scrublands quite well, as well as deciduous forest from evergreen forest (though it created another class for the transition between the two).

Question 3

  • How well do these classes match up with your predicted classes? Give an example of the model meeting your expectations and an example of it missing expectations.

    • It classified 6 classes(7 if including the null region, It was unable to distinguish the two water classes, and in fact classified it the same as (1)boreal forest. There were 2 rock(2,3) classes. A light vegetation(4) class, a deciduous forest(5) class, and one class that is difficult to ID, but I would call a sparsely forested class(6), this also appears to be most prominent in the Thermal IR bands.