GIS5027 Module 5: Unsupervised & Supervised Classification

    Lab 5 in Aerial Photo Interpretation and Remote Sensing focused on unsupervised and supervised classification. 

    We used a provided a high resolution satellite image to determine unsupervised classification processes by first creating a cluster layer and comparing the cluster layer to the original high resolution image. We then reclassified the cluster layer by recoding within the Recode Table from 50 to 5 classes. Some pixels were shown to be a part of two classes relatively equally when we used the Swipe, Flicker, Blend, and Highlight tools. and we added a fifth class to account for Mixed (M) features that were very split between two classes.

Unsupervised Classification

    To utilize supervise classification, we used a new provided image to determine a new spectral signature with Signature Editor. Using both the “Drawing Polygon” method and “Creating Signature from Seed” method, we were able to look at the Spectral Euclidean Distance (used to determine which pixels will be selected based on the distance chosen from the “seed” pixel) and Neighborhood (how the AOI will form around of “seed” pixel) to best determine a spectral signature, and then we recoded similarly as we did for unsupervised classification to merge together similar classes.

Supervised Classification

     Ultimately, I really don’t think this map is accurate (there is a number of bright spots in the Distance Image map, which indicate several areas are probably misclassified) – unfortunately during this lab assignment, several unexpected life and family events took place that greatly impacted the ability to do this assignment to the best of my ability. There is only so much time. I find myself having trouble with the histogram section of this assignment, and hope to further study that because I think that is where I went wrong. 





Comments

Post a Comment

Popular posts from this blog

GIS5007 Orientation: About Me

  Hello! My name is Abby Starling, and I’m currently in the GIS Graduate Certificate program. I started this program to move into a career shift, as I currently work for a county government doing more admin related items full-time. While my job is fine, it isn't what I envisioned - it is time for a change! Through this program, I hope to gain more understanding in GIS and form a solid foundation for a career in GIS. I took several GIS classes in my undergraduate Geology program at Valdosta State University, but am certainly no pro. In my free time, I always end up playing copious amounts of Stardew Valley.    Abby Starling's Story Map Links to an external site.
Read more

GIS 5007 Module 6: Isarithmic Mapping

Image
     Module 6 of Computer Cartography focused on isarithmic mapping and using PRISM (Parameter-elevation Regressions on Independent Slopes Model) derived data to showcase the annual precipitation of Washington state, USA over a ~30 year period (1981-2010). Originally created by the PRISM Group at Oregon State University, the PRISM system has proved to be a leading system in producing quality precipitation maps. A climate-elevation regression function is calculated in PRISM with consideration of climate variables that may influence the data, such as proximity to coast, atmospheric layer, topography of region, and more. As precipitation tends to increase with elevation, elevation is an included variable in PRISM using a digital elevation model (DEM) and each DEM grid cell has its own climate-elevation regression function calculated within. Annual Precipitation of Washington, USA from 1981-2010      The above isarithmic map depicts the annual precipitation of...
Read more

GIS5935 Module 2.1: Surfaces - TINs and DEMs

Image
                In this lab, we explored two types of elevation data models - Digital Elevation Model (DEM) and Triangulated Irregular Network (TIN). The DEM is raster-based and information is stored in grid format, while TIN is vector-based and uses irregularly spaced points that were calculated according to the connections of sets of triangles for contouring. Because of this, the DEM has smoother curves in its contour lines, while TIN has pointed/jagged contour lines. The DEM contour lines in this instance also have a few more lines due to the data parameters.                 The TIN contour lines are likely more accurate, as TIN can differentiate between more complex terrain than DEM contour typically can (depending on resolution). TIN uses irregularly placed elevation points that vary based on data density, but DEM is a grid with elevation values stored within the grid cel...
Read more