These raster files represent land cover classifications in Trinidad, Bolivia at two different time periods: March 7, 2007 and June 15, 2015.
In order to better understand the changing landscape of Trinidad, imagery covering the entire city was acquired at two different time periods (2007 and 2015). These high resolution (50cm) scenes were then transformed into land cover maps using a methodology developed by Graesser et al (2012). Originally created to accurately detect shanties in major cities throughout the world, this method has been proven effective in a diverse set of cities (Kandahar, Kabul, Caracas, and La Paz). Furthermore, it has been shown to be effective at capturing land cover change in 5 primary cities in Africa by Antos et al 2016. Since its creation, it has been adopted by the US Census Bureau, US Department of Energy’s Oakridge Laboratory and The George Washington University.
This semi-automated classification approach, examines the texture and structural composition of various neighborhoods, and then groups land with similar patterns into a single class. For the city of Trinidad, the images were divided into 9 distinct classes: Regular residential, Sparse residential, Flooded residential (only detected in 2005), commercial/industrial, roads, bare soil/dry grass, sand, vegetation, and water.
Raster is coded by number, defined below:
1. Sparse Residential
2. Regular Residential,
3. Commercial/Industrial
4. Bare Soil/Dry Grass
5. Vegetation
6. Water
7. Sand
8. Road
9. Flooded residential (only detected in 2007)