Embedded Files
Problem
Black Water National Wildlife Refuge seeks information regarding land classification data in order to develop a management plan for the refuge. This project involves land classification from an existing aerial photo.
Analysis Procedures
The Black Water National Wildlife Refuge provided an aerial photo which was used for image classification in ArcMap. Following appropriate procedures for supervised classification, I selected training polygons from the existing aerial photo and then ran the classification using the Interactive Supervised Classification tool. The image was updated with additional polygons and the classification rerun. Maps were symbolized for visual display.
I added the aerial photo to ArcMap and changed the color to false color composite. I began the classification process by selecting training sample polygons for each of the six land cover classes. Once I completed selecting the training polygons, I used the Interactive Supervised Classification tool in the Classification toolbar to classify the image. I saved the classification as a GRID file. I then added a field to the attribute table of the classified image and calculated the area for each land cover class. A map showing the training polygons and the image classification were symbolized for visual display. I added additional training polygons to a water spot in the western area of the image and polygons for additional developed/impervious areas in the eastern area of the image. I then used the Interactive Supervised Classification tool to reclassify the image. I saved the classification as a GRID file. I then added a field to the attribute table of the classified image and calculated the area for each land cover class. A map showing the additional training polygons and the reclassification were symbolized for visual display.
Workflow diagram (Click to enlarge)
Results
Map displaying results of first Interactive Supervised Classification (Click to enlarge)
Map displaying results of second Interactive Supervised Classification (Click to enlarge)
Application & Reflection
Image classification allows one to create data from a simple aerial photo. In this sense, image classification can be used for small projects, but also as a starting point for larger projects. I would anticipate this being a useful skill when looking at changes in land cover over time. Looking at land cover change over time can inform questions regarding deforestation, wetland loss/growth, as well as spatial questions regarding productive land. For example, if I was interested in looking at deforestation of a particular area over time, I could use image classification to classify the land cover of the study area at different points in time. Satellite imaging could be obtained online from NASA’s Landsat imaging database for various time periods. These would then be classified for land cover using either unsupervised or supervised classification (depending on familiarity with the area) and then compared across time. Additionally, one could use area calculations to determine the extent of deforestation. Researchers interested in the effects of deforestation, particularly with regard to climate change, may find this a useful first step.
Page updated
Report abuse