SO503, Spring 2011

Satellite Math and Image Analysis Lab

Insure that you are running version 2009.3.8 of MICRODEM.  Update it if necessary.  The help file has been expanded to cover the  topics we will be using, so insure that is current as well.

We will use the 6 band Landsat Enhanced Thematic Mapper (ETM) image of  Normandy in France.  It will be available on the data download option of MICRODEM.

The satellite image, like the DEM we used earlier, are raster data sets.  We also have the roads, a vector data set.

For the requirments below, insure that you have a map at the full resolution of the data for the region around Omaha beach, and that your images have a scale bar. 

Do both a smoothing and a sharpening filter.

• Discuss how they are different.

Do a principal components analysis of the imagery, and answer the following questions:

• How many of the images contain useful information?  How can you justify your assessment? Include some screen captures.

• Which bands are most different from the others?  How can you see this in the principal components and the statistics presented by the program?

Do an unsupervised classification of the scene.

• Discuss what it shows.

Do a supervised classification of the scene, using five categories (your choice, but think about what appears to be there).  Google Earth can help you see what Normandy looks like today.  Answer the following questions:

• What is the effect of changing the std dev allowed in the classification?  Are there tradeoffs, or is there an automatic value you should select?

• How much do you trust your classification?

• Turn in a map showing the results of the classification in the region around the D-Day beaches, a histogram showing the locations of the training sets on the two bands you think best discriminate the categories, and a table showing the results of the classification.