G.Hudson Mastering GIS

  The focus of this week's class and lab was unsupervised and supervised classification. Once we covered all of the theory during lectures we were tasked with working through a guided unsupervised and supervised classification. We then had to perform our very own supervised classification on Germantown, Maryland. This involved creating areas of interest (AOI's) which act as training sites, performing the classification and then recoding the classification to limit the number of classes. The map above is my end result.  This lab was pretty enjoyable and it was nice to work through the classification process. My downfall, however, is sometimes overthinking steps and processes. While it is important to be careful and attentive, it's also important to trust the process. 

 As the title of this week's blog suggests, week 4 of Remote Sensing and Photo Interpretation was focused on various types of spatial enhancement, multispectral data and band indices. Spatial enhancement is a key preprocessing step that usually takes place before using or conducting detailed analysis of imagery. Some of the spatial enhancement processes explored include low pass filter, high pass filter and focal statistics. In addition to spatial enhancement, geospatial professionals also usually view the properties of the multispectral data to help understand what the imagery is depicting. To garner more information about the imagery we explored the histogram and manipulated it to bring attention to features that may have otherwise not been seen, and we learned about the inquire cursor that displays information on individual pixels in an image. Lastly, we learned how to create a Normalized Differential Vegetation Index (NDVI) and the various band combinations used to highlight ce

  Map of classified subset of Olympic National Park The third week of Photo Interpretation and Remote Sensing introduced students to detailed information about Electomagnetic Radiation (EMR), the various types of remote sensors and the basics of Digital Image Processing. As such, the two part lab allowed students to put the newly acquired information to use. We were introduced to the ERDAS imagine software, where we learned the interface and some of the various functions that will be required for future use. The result of exploring the software is the map pictured above. We were tasked with creating a subset of a classified image, along with area calculations, which would be used in ArcGIS Pro to produce the map. 

The second week of Photo Interpretation and Remote Sensing was focused on Land Use/Land Cover Classification and Accuracy Assessments. Students were tasked with using the USGS Land Use/Land Cover classification system to classify a provided image of Pascagoula, Mississippi at level I and II. Once the classification was complete we then had to determine the accuracy by 'ground truthing'. As this is an online course we used Google Maps/Earth to verify random points and then calculated the accuracy.  I ended up with a fairly low accuracy. This was mostly due to errors in my residential class. I had a few buildings that were on the cusp of a residential and commercial area but ended up in the wrong class. I also had an instance of a church being located in the middle of a residential area. My highest accuracy was in the water class, 100%.