Data Concepts — Satellite Imagery Glossary

Sensor

GSD — Ground Sample Distance: The size of one pixel on the ground. 30 cm GSD means each pixel covers a 30 cm square.
Off-nadir angle: How far the satellite tilted from straight-down when capturing. Higher angles give more collection opportunities but degrade quality.
Spectral bands: Different sensors capture different parts of the electromagnetic spectrum. More bands means more analytical options.
Bit depth — 8-bit vs 16-bit: Determines how many distinct values the sensor can record per band. 16-bit preserves dynamic range needed for quantitative work.
Optical vs SAR (synthetic aperture radar): Optical sensors capture reflected sunlight. SAR sensors emit radar pulses and measure the echo — they see through clouds and at night.
Sun-synchronous orbit and LEO: Why Earth observation satellites all live in similar orbits at similar altitudes — and why that drives revisit times.
Revisit rate — theoretical vs real: How often a given point on Earth can be imaged. The number on a spec sheet is the theoretical maximum; real revisit at acceptable quality is usually half that.
Swath width: The width of the strip of ground the satellite images in a single pass. Determines coverage per acquisition and how quickly a constellation can cover wide areas.
Stereo and tri-stereo capture: When the satellite captures the same area from two or three angles in a single pass — required for DSM and DTM generation.

Cloud cover: Cloud percentage matters per-AOI, not per-scene. Thin cirrus is the silent killer.
Sun elevation angle: How high the sun was at capture time. Affects shadows, illumination consistency, and time-series comparability.
Sensor noise, speckle, and artefacts: Real-world imagery has imperfections. How they are handled determines whether downstream analytics produce reliable results.

Orthorectification: Geometric correction that projects imagery onto a digital elevation model so every pixel sits at its true ground position.
Surface reflectance (and atmospheric correction): Converts raw sensor counts into the actual fraction of light reflected by the ground — required for any quantitative or multi-date work.
Pansharpening: Fuses high-resolution panchromatic with lower-resolution multispectral to produce sharper colour imagery.
Mosaicking: Stitches many scenes into a seamless, cloud-removed composite over a large AOI.
Processing levels — L0, L1, L2, L3: Industry shorthand for how much processing has been applied. Definitions vary by provider but the rough hierarchy is consistent.

DSM — Digital Surface Model: A 3D model of everything visible from above — ground, trees, buildings, infrastructure, vehicles. Built from stereo or tri-stereo captures.
DTM — Digital Terrain Model: Bare-earth 3D model with trees, buildings and other surface features removed or interpolated through.

Ground stations and downlink: Satellites can only send data to Earth when overhead a ground station. Ground station geography drives delivery latency.

Tasking tiers — regular, priority, speculative: Three commercial tiers determine how aggressively the satellite operator chases your acquisition window.
Catalog (archive) imagery: Imagery that already exists in the archive — fastest, cheapest path when you can flex on quality parameters.
Feasibility check: Operations matches a tasking request against constellation orbit predictions to produce a list of collection opportunities.
AOI — Area of Interest: The geographic region you want imaged. Shape and size both affect feasibility and pricing.

CRS — Coordinate Reference Systems: How geographic coordinates are projected onto a map. UTM-WGS84 is the most common for satellite imagery deliveries.
Positional accuracy (CE90): How closely a pixel's reported geographic position matches its true ground position. Reported as CE90 — the radius within which 90% of points fall.