Mapping and documentation of land features and landcover types is a very important process that many government municipalities do in the United States. However, in some places they face certain challenges such as Intra-Class Variation as well as scale variation. There are of course more but I’m going to be looking at a bigger obstacle, clouds. In this project I will be attempting an image classification of Maui, HI. My main target months were July and June however because of cloud coverage I was only able to find two images. One in 2001 and one in 2021 I will perform an unsupervised as well as supervised classification. My main take away is to see how difficult and to see if an image classification is accurate on imagery with a specific date. There is a workaround you can do but it requires at least three years of data and are not available day to day. When getting my images and noticing all the clouds I decided to look up some weather data as well as some tourism data on Maui. I used the weather data to get some averages to create a graphic and di the same with the tourism data. After I got the images downloaded, I performed an unsupervised classification on both images and then a supervised classification on both images. I added one class more to one image because of the number of shadows in the image. After this I went and did an accuracy assessment using ground truthing points. For this I had aerial imagery from 2001 as well as base map imagery for 2021.
Additional Maps
 |
 |
 |