top of page
Unsupervised Classification Over Time
The above maps, show two layouts of land cover classification in the immediate area of Lake Tahoe, California. The image rasters were created using the Unsupervised Classification method, and the accompanying charts were auto-generated by the program based on the similarity of pixels. The overall spatial cover of land by each category, was estimated by the ArcMap, and the categories were later assigned by me after examining the overlap with the original aerial photograph. This model allows us to see the change in land cover and land use in the Lake Tahoe region over the period of 11 years. The "unclassifiable" category accounts for land that could not be generalized in this particular model.
Unsupervised Classification
This is another example of using the Unsupervised Classification technique to generalize land cover of a general area. The map to the left shows a raster aerial of the general area of the Pensacola campus of the University of West Florida.
​
The area is divided into three parts that can be established with confidence, as well as two which constitute for a margin of error. The "shadows" category does not account for land cover, which could not be sorted due to the discoloration in the photograph. The "mixed" category is different, in the way that it could not be classified with confidence.
This image was created using two programs: ArcMap and ERDAS Imagine.
Supervised Classification
The map to the right is an example of the Supervised Classification method. The raster produced is based off of an aerial photograph of Germantown, Maryland.
​
The main distinction between the Supervised and Unsupervised methods for land classification, is whether land cover categories are established automatically, or by human intervention. In this Supervised Classification example, I chose several points based on visible features on the aerial photograph which would account for a representative sample used to generalize the map based on the categories listed below. Afterwards, the area of each category was calculated in ArcMap.
This image was created using mostly ERDAS Imagine, with the final layout and aesthetics done in ArcMap.
Thermal and Multispectral Analysis
The example to the left shows a heat bump map, created in the ArcMap program.
​
It was generated based on a previously generated ETM composite image, which was selected mainly due to its clear-to-see distinctions in land cover which were essential for the purpose of result verification.
​
The color spectrum reflects the amount of radiance exhibited in specific areas of the aerial, with the most heat-emitting placed coming up as red. This allows us to clearly track human activity across the regional geography.
​
Near-Infrared Multispectral Analysis
The example on the right demonstrates a very simple multispectral analysis using the false color (near infrared) color spectrum)
​
False color is highly useful in making distinctions between living and dead vegetation by highlighting the level of photosynthesis visible. It makes it quite easy to distinguish water, as well as non-photosynthetic regions across the general area clearly.
​
The image was made and processed using ArcMap.
​
bottom of page