Skip to content

Glossary

Key terms used in the course, explained for planners.

Band
One layer of a satellite image, recording one colour of light (for example red, or near-infrared). A colour photo has three bands. Sentinel-2 has thirteen.
Bounding box (bbox)
The rectangle that encloses an area, written as four numbers: west, south, east, north.
Centroid
The centre point of a shape.
Circuity
How much longer a route along the streets is than the straight-line distance. A value of 1.0 means the streets are perfectly direct.
CRS (Coordinate Reference System)
The system used to give locations coordinates. Latitude and longitude (EPSG:4326) are good for storing data. A projected CRS such as UTM zone 43N (EPSG:32643) is used for measuring distances and areas in metres around Bengaluru.
DEM (Digital Elevation Model)
A raster where each pixel holds the ground height above sea level.
Footprint
The outline of a building on the ground, stored as a polygon.
GeoDataFrame
A table (like a spreadsheet) where each row also has a shape: a point, line or polygon. It comes from the GeoPandas package.
Ground coverage
The share of a plot or area covered by building footprints, in percent.
Intersection density
The number of street intersections per square kilometre. Higher values usually mean smaller blocks and a more walkable area.
Isochrone
The area you can reach from a point within a set time, for example a 15-minute walk.
Land cover
What physically covers the ground: trees, water, buildings, bare soil. It differs from land use, which is what people use the land for.
LST (Land Surface Temperature)
How hot the ground surface is, measured by a satellite's thermal sensor. On a sunny day it is often much hotter than the air temperature.
NDBI (Normalised Difference Built-up Index)
A satellite index that highlights built-up areas. It is calculated from short-wave infrared and near-infrared bands.
NDVI (Normalised Difference Vegetation Index)
A satellite index that measures how green and healthy vegetation is. It ranges from -1 to 1, and values above about 0.4 usually mean dense vegetation.
Network graph
A street network stored as nodes (intersections) and edges (street segments between them).
OpenStreetMap (OSM)
A free, editable map of the world built by volunteers, like a Wikipedia for maps.
Pixel / resolution
The smallest square in a raster. "10 m resolution" means each pixel covers 10 m by 10 m on the ground.
Random Forest
A machine learning method that learns from labelled examples and then predicts labels for new data. It works by combining many simple decision trees.
Raster
Data stored as a grid of pixels, like a photo. Satellite images and elevation models are rasters.
Remote sensing
Collecting information about the Earth from a distance, usually with satellites or drones.
Slope
How steep the ground is, in degrees.
STAC (SpatioTemporal Asset Catalog)
A standard way to search catalogues of satellite images by place, date and cloud cover.
Suitability analysis
Combining several map layers, each given a weight, to score how suitable every location is for a use.
Vector
Data stored as points, lines and polygons, for example bus stops, roads and building footprints.
Zonal statistics
Summarising raster values inside each zone, for example the average NDVI in each ward.