Supervised Image Classification
Composite Bands
- Option 1: Windows > Image Analysis:
- Select all bands
- Composite
- Option 2: Open and apply Composite Band tool
- Map RGB with bands on generated layer (landsat 8)
- 4 3 2 for natural color
- 5 6 2 for health vegetation
- 7 6 4 for urban
- Map what color indicates what with a reference (e.g., base map)
Extract Study Area
- Enable label feature of the map
- Select Dhaka
- Right click on map layer > Data > Export Data
Classification
- Perform KML conversion using "Layer to KML" tool
- Drag and open the KML file on Google Earth Pro (Places)
- Customize > Toolbars > Image Classification
- Open Training Sample Manager: Select Composite Band layer
- Draw Polygon (or circle)
- Merge multiple training samples into one (e.g., Building)
- Repeat for building, water body, vegetation, etc.
- Create a signature file [Ocassionally problem arises if not saved on drive C:]
- Image Classification > Interactive: Automatic
- Image Classification > Maximum Likelyhood:
- Input "Composite"
- Input the produced signature file
- Clear training data to remove polygons
Extract Classification
- Open "Extract by Mask" tool
- Input 1: Interactive Classification output
- Input 2: Dhaka
Area Calculation
- Raster to Polygon on the Dhaka Interactive Classification layer
- Use field = Class_name
- Dissolve
- Open attribute table of dissolved layer
- If area not calculated, produce area column
Assignment
- Of assigned district
- Three years (e.g., 2015, 2020, 2025)
- Create layout
- Graph: Show area comparison across the years
- 14th week, tuesday