Module 4: Ground Truthing

Week four was another building week.  The objectives of this lab was to collect sample points, and check the accuracy of the classifications using Google Street View.

I decided to select 30 points at random.  I was concerned it would be difficult to be consistent with the systematic approach.  Randomly selecting points, I felt it would be very unbiased.  Once the points were selected, I selected one point and matched it up to Google Maps.  Once I located the point on Google, I zoomed in as much as possible.  Street view allowed me to see the front of the buildings if the point was on a building.  From seeing where the point actually fell, I could determine if my original classification was correct.

Most of my inaccuracy occurred when dots fell in the polygons that were classified as residential.  A lot of the dots ended up falling on trees or buildings that look like houses.  After using street view, it was clear they were commercial properties or industrial.  My accuracy rating was 67%.

Ground Truthing of LULC Classification