--- title: "Earth observation & GeoAI, explained | Nexus Observatory" description: "Interactive explainer: earth observation & geoai in 3D, chapter by chapter: altitude sets the inclination, the period and the revisit…" url: https://observatory.global/earth-observation/explainer/ publisher: Nexus Observatory Inc. updated: 2026-10-28 --- **3.6° N** *· 705 km · swath 185 km* **Inclination 98.21°** *· retrograde, sun-synchronous* **Satellite** **Footprint** **Mapping drone** **L0** *· Raw instrument counts* **L1** *· Calibrated, geolocated* **L2** *· Analysis-ready: reflectance or backscatter* **L3** *· Composite of many acquisitions* **Catalog** **Collection** **Item · one scene** **Assets · COG bands, metadata** **COG overviews** *· coarse to full resolution* **Ground station** *· tracking antenna* **Bow Valley, Alberta** 01 02 03 04 05 06 07 Intro. Earth observation & GeoAI · Explainer # From orbit to a tile you can query. Seven steps of an Earth-observation chain, each worked live: the orbit that sets the revisit, the geometry that sets the resolution, the pass that sets the data volume, the processing levels, a change-detection pass, a drone campaign and the catalogue that serves it. Every number below is computed live from your inputs. **01** Orbit and revisit **02** Acquisition geometry **03** Downlink **04** Processing levels **05** GeoAI change detection **06** Drone campaign **07** Catalogue and tiles **01** · Orbit and revisit ## Altitude sets the inclination, the period and the revisit. A sun-synchronous orbit precesses with the Sun, about 0.9856° a day, so every pass crosses a latitude at the same local time. The Earth's oblateness supplies the precession, so altitude fixes the inclination: cos i = −2Ω̇a7/2 / (3J₂Re²√μ). Kepler gives the period, T = 2π√(a³/μ). With the swath and the orbits a day, that sets how long the equator takes to cover at nadir. **Sun-synchronous inclination**: 98.21° **Orbital period**: 98.9 min **Orbits per day**: 14.56 **Track spacing at the equator**: 2,759 km **Equator fully covered every**: 14.9 days **02** · Acquisition geometry ## Radar measures range. Optics measure angle. A synthetic aperture radar looks sideways and resolves the ground by the timing of its echoes, so its ground resolution improves as the incidence angle grows. An optical imager resolves by angle, so its pixels stretch as it points away from nadir. Radar slant-range resolution is c/2B and its ground-range resolution divides that by sin θ; the optical across-track pixel grows as 1/cos²θ. Sensor Chirp bandwidth **Slant-range resolution**: 1.50 m **Ground-range resolution**: 2.61 m **Slant range to the scene**: 861 km **03** · Downlink ## The elevation mask decides how much data comes down. A ground station can only work a satellite above its elevation mask: terrain, buildings and the atmosphere set it. The pass is the arc of the orbit above that cone; its length times the link rate is the data volume. The overhead pass is the longest: the Earth central angle λ = arccos(Re cos ε / (Re + h)) − ε sets it, and every pass off the ground track is shorter. Downlink rate **Longest pass (overhead)**: 11.7 min **Slant range at the mask**: 2,574 km **Data per pass**: 52.5 GB **04** · Processing levels ## From raw counts to analysis-ready data. Level 0 is the instrument's raw stream. Level 1 is calibrated and located. Level 2 is a physical quantity per pixel, surface reflectance or backscatter, and is analysis-ready when it meets the CEOS ARD requirements. Level 3 composites many acquisitions into one. A scene's volume is its pixels times bands times bit depth. Scene width Steps: 1 · 3 · 10 · 20 · 30 · 60 m Bands × bit depth **Pixels per band**: 100.0 M **Raw volume per scene**: 1.95 GB **As a COG, 2:1 lossless**: 975.0 MB **05** · GeoAI change detection ## A threshold trades missed changes for false alarms. On the Bow Valley terrain, slope failures are seeded on ground steeper than 32°, and a detector scores every analysis cell (369 × 590 m). Raise the threshold and false alarms fall while misses rise. Red cells are detections; outlined cells are the seeded failures. **Seeded slope failures**: 66 **Detections**: 78 **Precision · recall**: 68 % · 80 % **F1 score**: 0.74 **06** · Drone campaign ## Ground sample distance is a flight plan decision. For a mapping camera, the ground sample distance is sensor width times altitude over focal length times image width. Altitude and overlap then set the line spacing, the photo count and the time in the air. Camera class **Ground sample distance**: 2.74 cm **Line spacing**: 37.5 m **Photos per km²**: 1,067 **Flight time per km² at 10 m/s**: 44.4 min **07** · Catalogue and tiles ## STAC finds the scene. COG serves only the pixels you ask for. A SpatioTemporal Asset Catalog describes every scene as an item with assets. A Cloud-Optimized GeoTIFF stores overviews and internal tiles, so a client reads a few kilobytes over HTTP range requests instead of the whole file. At zoom z a Web Mercator pixel spans 2πR / (256 · 2z) at the equator. **Map pixel at the equator**: 38.2 m **Overview read for a 10 m image**: Overview 1 (20 m) **Tiles for a 1024 × 768 view**: 20 [Nexus Observatory](https://observatory.global/)**Earth observation & GeoAI** . The ground, measured and modelled.[Talk to us](https://observatory.global/contact/) Earth observation & GeoAI - [Ground segment](https://observatory.global/earth-observation/ground-segment/) - [Drones and imagery](https://observatory.global/earth-observation/drones-and-imagery/) - [Site twins](https://observatory.global/earth-observation/site-twins/) - [GeoAI](https://observatory.global/earth-observation/geoai/) Technologies - [Platforms and orbits](https://observatory.global/technologies/eo/platforms-and-orbits/) - [Optical, spectral and thermal sensors](https://observatory.global/technologies/eo/optical-spectral-and-thermal-sensors/) - [Reference systems and geodesy](https://observatory.global/technologies/gis/reference-systems-and-geodesy/) - [Vector, raster and array formats](https://observatory.global/technologies/gis/vector-raster-and-array-formats/) Industries - [Cities](https://observatory.global/industries/cities/) - [Civil resilience](https://observatory.global/industries/resilience/) - [Humanitarian](https://observatory.global/industries/humanitarian/) - [Climate](https://observatory.global/industries/climate/) - [Water](https://observatory.global/industries/water/) - [Nature](https://observatory.global/industries/nature/) - [Oceans](https://observatory.global/industries/oceans/) Service platforms - [Observatory Protocol](https://observatory.global/protocol/) - [Protocol & Grid](https://observatory.global/grid-services/) - [Software](https://observatory.global/software/) - [Hardware & power](https://observatory.global/hardware/) - [Sensors & instruments](https://observatory.global/sensors/) - [Connectivity](https://observatory.global/connectivity/) - [Edge compute & AI](https://observatory.global/edge-ai/) © 2026 Nexus Observatory Inc. [Privacy](https://observatory.global/legal/privacy/) [Terms](https://observatory.global/legal/terms/) [Accessibility](https://observatory.global/legal/accessibility/) [Cookie settings](https://observatory.global/privacy-choices/) [Data credits](https://observatory.global/credits/)