Supervised Classification and Change Detection¶
Lab Materials¶
Download this:
Lab Content¶
Objective¶
To detect and quantify the change in Land Use/Cover in the Lake Lanier Watershed, GA area between 1991 and 2005
Procedure¶
In Lab#7, you Subsetted an 1991 Landsat image of the Lake Lanier Watershed Area to match a 2005 subsetted Landsat image, also of the Lake Lanier Watershed
Do a supervised classification of this subset file with the same land cover categories as in Lab#6.
You will have to develop a new set of signatures for the land cover types.
Port the classified 1991 land cover file to ArcMap and do a map composition like you did for the 2005 land cover of the Lake Liner Watershed
Find the area corresponding to each land cover type for 1999, for the Gainesville area.
Write a formal report on the change in land cover in the Lake Lanier Watershed area from 1991 to 2005. Present the areas of the land cover classes in a table as follows
Land Cover Type 1991 (Acres) 2005 (Acres) % Change Urban Forest Water Pasture/Ag
Now you will do a spatial change analysis by using the ‘Change Detection’ feature in Imagine software: 1) Open up Imagine 2010 and click on the Raster tab and click on Zonal Change then Image Difference
- Make sure that you input the 1999 image as the before image and the 2005 as the after image.
- Select Layer 4 for the Layer for each of the images, which is the NIR band.
- Name the Image Difference Layer as 91to05diff.img and name the Highlight Change as 91-05Changes.img
- Leave everything else as default and click OK.
The change detection is done by the difference in the digital number for each pixel.
Do a map composition of both the Image Difference and the Highlight Changes file.
Attach both the 91 and 05 classified maps in this report. Write a formal lab report that includes an analysis of the land cover changes in this watershed and the trends that are apparent in the change of land cover. What reasons would you attribute to the regions of the greatest change in land cover?
% Change for a particular Land Cover Class =
100 x (2005