Spatial Enhancement, Multispectral Data, and Band Indices

This week, we learned about image preprocessing and spatial enhancements to aid in interpreting data. We completed several exercises to learn about the various aspects of this and then completed a final exercise to put it all together and create 3 maps as deliverables for the assignment. 

The first few execises focused on obtaining and prepping the data. First, we learned how to download data. In this, we used Glovis to download Level 1 products, which have been terrain corrected, orthorectified and georeferenced, and radiometrically calibrated. Level 2 generally also would include atmospheric corrections. The next step included formatting the files, importing the imagery as a tiff and saving them as image files in ERDAS Imagine.

Then we moved on to the pre-processing. We applied basic filters in Imagine including low passes for broader patterns, and high passes and sharpening for finer details. We also looked at how to use the Focal Statistics tool in ArcGIS for similar filtering. Next we explored histograms in both Imagine and ArcGIS to understand how to find spikes and manipulate breakpoints in the data. Then we explored how to change the bands to explore different visualizations of the imagery as well as how to create band indices to enhance the appearances of certain features. Finally, we looked at the Metadata and used the Inquiry Cursor to examine specific pixels.

For final exercise, we put all of these skills to use to examine three features within an image from Northwest Washington state. We had to examine the historgrams to look at areas with spikes, then explore the imagery to figure out where these pixels were in the data. We looked at individual bands in grey scale as well as explored the various band combinations to see which helped visualize the features best. Finally, I used the Inquiry Cursor to make sure I had the correct pixels as noted in the histogram based on there brightness. After identifying the areas, I used the Inquiry Box and Subset tools to create a new file that contained only the feature data. I then brought these into ArcGIS to create maps for each.

Feature 1 was identified as a spike of dark pixels in Layer 4 of the data. It turned out to be the rivers. I visualized this using False Natural Color as this emphasized the contrast between the river and the landscape.

Feature 2 was identified as bright pixels in layers 1-4 and dark pixels in layers 5-6. This feature turned out to be the snow on the mountains. I used the False Color IR to visualize this as this gave the best contrast between the snow and the surrounding ground cover.

Feature 3 included areas of water that looked much lighter in layers 1-3 but was not visible in layers 4-6. I used True Color to visualize this as it only used the layers that have the lighter pixels visible. This had the best combination of RGB to show the area.