GIS5027 Module 5: Unsupervised & Supervised Classification

-
This week's module covered all things classification from unsupervised to supervised. In ERDAS, we got to see how the two differed and how to preform each. It was really cool to see how the program would classify pixel areas for you based on what you essentially feed into it. Since unsupervised classification has users classify after the program categorized pixel values, it felt much easier and quicker; however, accuracy may not be as high when compared to supervised. Supervised classification took me much longer with all the steps we had to go through to identify to the program the classes/features themselves, although it was the most rewarding trying to wrap my head around everything.

After learning how to preform both, the map deliverable this time around involved us taking a raster image from Germantown, Maryland and classifying its land use/land cover based on supervised classification. By the time we got to this image, I figured out that ERDAS had some nice preset color themes, one of which was Land Use -- this came in handy when picking colors for this particular land use focused analysis! Overall, when comparing my results to the distance image created during the process of recoding (in order to merge features together), I found that the patterns for the most part followed true to my supervised classification. There are definitely areas where I wonder if the roads could be better defined, but due to how skinny they were in comparison, I can only imagine how difficult it is to accurately be caught. Especially when you look at large, expansive areas like the forests -- those areas are quite clear to interpret.

Ultimately, I'm pretty content with the map I created this time around. I do hope I can get a better grasp of fine tuning the supervised classification when it comes to my final project, where I hope to do something similar!
Screenshot of the Module 5 Supervised Classification map deliverable of Germantown, Maryland. Land Use/Land Cover has been divided into 8 classes with their respective acreages in acres.

Comments

Post a Comment

Popular posts from this blog

GIS5050 Module 4: Vector Analysis

-
Image
We're hitting the halfway mark of the Introduction to GIS course, and this week's lab is on vector analysis. We focused on running spatial queries in the first half of the lab and ran buffers in the second. I learned a lot about other ways of running buffers that I haven't yet to perform in my workplace, such as compound/multi-ring buffer. I can see how this would be useful in cases where certain distance thresholds must be met, like distance from road that is safe to build sidewalks or enough cushion for a safe bike lane, etc. The map that was created this week centered around the buffer tools, where we had to take De Soto National Forest feature classes and find out possible site locations for future amenities, such as campsites, pavilions, and the sort. We did this by running various buffers on both the water and road feature classes and finding where the buffers overlapped, while also taking into consideration conservation areas, which are non-construction zones. Screen...
Read more

GIS5050 Module 6: Georeferencing, Editing, & 3D Mapping

-
Image
It's the last module before the final project assignment -- it's both exciting and a bit nerve-racking since the deadlines are coming up so fast! Module 6 covered georeferencing, editing, and 3D mapping. Since 3D mapping was not a requirement, I didn't do it; however, if I have time later in the week, I'd like to make an addendum to this post and add my 2nd map deliverable if possible. As someone who hasn't made a 3D map, it intrigues me and would love the opportunity to finally get one underneath my belt. Since the lab assignment focused largely on georeferencing and editing vector data, we worked a lot with satellite aerial imagery of the University of West Florida campus. We had to take 2 JPEGs of the north and south sides of the UWF campus and georectify them using different transformations -- the north aerial I performed in 1st order polynomial while the south aerial I did in 2nd order polynomial. Georeferencing was one of the first things I learned at my workp...
Read more

GIS5050 Final Project: The Bobwhite-Manatee Transmission Line

-
Image
The time has come for the final project to be completed and the class done! It's a bit wild for me to think that much time has passed where my first GIS grad course is over. Throughout the entire final project, I had a lot of fun actually creating these maps -- maybe too much since I ended up losing sleep some days because I was so in a flow state, so into just constantly trying and tweaking ideas of how the basemap should be or how the insets should look, etc. I think the biggest takeaway I have from my first grad course in a long time is that I'm way too much of a perfectionist. I even chuckled at myself how frequently I wanted to redo the recordings of my final presentation just because I didn't like how I said something, stuttered, or so on. At this point, no amount of minor editing will change my grade that drastically -- it's not worth it to mull over to no end. Ultimately, I'm still very proud of the map deliverables I created and are maps I would be content ...
Read more