Image Classification in land cover for Black Water National Wildlife Refuge

Problem

The Black Water National Wildlife Refuge managers would like to develop a plan for the refuge and would like land cover data. The task at hand is to perform a supervised classification using one aerial photograph from the Refuge. The aerial photograph is a 4-band (true-color and color-infrared) 1-foot resolution image from August 2010.

Analysis Procedure

Projection for all data is NAD83, State Plane, Maryland FIPS 1900 meters. The aerial photograph as mentioned above is important in helping to develop a training sample as it provides the land use class categories. The original aerial photo is changed to RGB imagery where Red: Band_4, Green: Band_3 and Blue: Band_2. The Supervised Classification is when a training sample is created and used in the Classification. The first training sample is created with a minimum of two polygons for each class. After classification, the resulting imagery is compared against the original photo to note the areas incorrectly classified. A second training sample is generated and the imagery again classified to correct those discrepancies. An area field is calculated for all classes after each classification to compare the difference in those incorrectly classified areas. ArcGIS Pro is used here.

Workflow Diagram

Results

Application and Reflection

Image classification is very useful for classifying land use categories.

Problem Description For the Brownfield project, land use raster is useful when layered over the Interpolation surface to glean information on the effect of Brownfields’ contamination on the type of land.

Data Needed The Landsat imagery for the Charlotte Mecklenburg area can be downloaded from the USGS Earth Explorer home page.

Analysis Procedures The Landsat raster image is used in the creation of the training sample which is used in the supervised classification. Multiple training samples would need to be generated till the Landsat image is correctly classified.