Below is an image of Germantown, Maryland that features recoded LULC information.  I used tools and skills taught to us throughout the class, but primarily given to us in module 5 of GIS4035.  We generated seed classes using provided coordinates and region growing tools in ERDAS. We then used the maximum likelihood algorithm to extrapolate our seed classes to the entire image.  We then recoded our classifications into the combined classes you see below.  I actually combined Urban and Roads classes, although it was not part of the directive, to make the image more visually accurate.  I did this because there was some spectral confusion between those two classes, likely due to my error in generating seed classes.

GIS4035 Lab 4

The following map shows three images depicting three separate geographic features using the spatial analysis tools available to us in ERDAS Imagine, and using the skills taught to us in lab 4 and throughout the semester in GIS4035.

Here is a brief description of each of the images and how I found each of these features:

Water Body: The feature creating this data spike in layer 4 is a water body, and judging from its geographic location it appears to be the Puget Sound.  I found this feature by opening the metadata for the given image and examining the histogram for layer_4.  I then looked for obviously dark areas in the true color image and used the inquire tool to find the specific DN information for the suspected areas in layer_4.  I dragged the cross-hairs over multiple locations within the large water body and found that the DN values all fell within 12 and 18 as described in the lab directions.  I also dragged the cross-hairs to dark areas outside this water body, but couldn’t find any areas outside that fell within the range (aside from other small water bodies).  I changed the band combinations to False Color Infrared because water stands out so well in this type of band usage.  The water body should be easily discernible in my map.

Snow-Capped Peaks: The feature matching these data spikes were snow-capped mountains.  I found this feature by using the same method as with the first feature (opening the metadata for the given image and examining the histogram for layers 1-6, and looked instead for obviously bright areas in the true color image).  When using the inquire tool I paid attention to the differences between layers 5 and 6, and layers 1-4 to make sure they were corresponding with their respective spikes when in this obviously bright area in the true color image.  I then set green band to layer_3, the red band to layer_5, and the blue band to layer_6. I selected these because I wanted one layer to be in the bright range and the other two to be in the dark range of these data spikes.  This layer combination turned out to be great for accentuating the desired feature as bright green against a deep red/purple background. 

Contaminated Runoff: I happened to notice that some areas of water stood out as a bright green when I changed the color bands to red/layer_5, green/Layer_3, and blue/layer_6.  This seemed fishy, so I went ahead and checked out those areas first.  My suspicions were confirmed when I moved the Inquire cursor over those areas.  Layers 1-3 became much brighter (mid-80’s to mid-90’s), layer 4 became somewhat brighter (mid-20’s), and layers 5 and 6 remained in the 10-12 range.  The area I selected appears to be caused from drainage on the south end of the Sound.  I chose to represent this area using the following layers: red/layer_2, green/layer_3, blue/layer_6.  I played around with different band combinations a bit, but this combination made the area of interest a bright yellow/green against a dark blue background.  Having the red being set at layer_2 also made the surrounding water darker and allowed the yellow/green to contrast better with the non-contaminated water.

This map represents a subset image of Thematic Mapper data provided in a lab exercise that allowed us to familiarize ourselves with ERDAS Imagine.  We selected this area and used ERDAS Imagine to calculate the area of each vegetation type in hectares.  We then transferred that image and it's associated data over to ArcGIS Pro and created this layout.  I listed the vegetation types in the legend from low to high in terms of total hectares.  I rounded up to the nearest hectare to keep it looking neat.