Earth observation satellite
Also known as: EO satellite, remote sensing satellite, imaging satellite
An Earth observation satellite is a spacecraft that collects images or measurements of the Earth's surface, oceans, and atmosphere from orbit.
These satellites power weather forecasts, wildfire maps, crop monitoring, sea-ice charts, and the satellite view in mapping apps. While commercial constellations now image most of the planet every day, turning that flood of pixels into reliable answers is still the hard part.
Earth observation satellite, explained at three levels
At home
Think of a scanner reading a page line by line as the paper moves. Many Earth observation satellites work the same way: as they orbit, a sensor sweeps strips of ground below, building long images one line at a time. Some see visible colours like a camera, others see heat or infrared, and some use radar. The satellite view in your map app is stitched together from strips like these.
At work
An Earth observation satellite measures the planet from orbit and sends that data to the ground as images or readings. Businesses and governments use it to watch crops, forests, floods, wildfires, ice, construction, and shipping across huge areas, repeatedly, without sending people out. Free sources such as Landsat and Sentinel cover broad needs, while commercial firms sell sharper or more frequent imagery.
At home
Satellites split light into bands. A regular camera records red, green, and blue; multispectral sensors add near-infrared, where healthy plants reflect strongly, plus shortwave and thermal infrared, which reveal moisture and fire heat. Optical sensors depend on sunlight and clear skies. Radar satellites, such as Canada's RADARSAT Constellation, send their own microwaves and see through clouds at night. Weather satellites in geostationary orbit watch one hemisphere continuously, while low-orbit imagers see more detail but pass less often.
At work
Three numbers define a sensor. Spatial resolution is the ground size of one pixel, from about 30 centimetres for top commercial satellites to 10 metres for Sentinel-2 and 30 metres for Landsat. Revisit time is how often the same spot is imaged, from daily to every couple of weeks. Spectral bands are the wavelengths captured. Data comes in processing levels, from raw sensor counts to corrected surface reflectance, delivered through platforms such as Copernicus Data Space, Google Earth Engine, or vendor APIs.
At home
There is a trade-off built into orbit. A sensor that sees fine detail covers a narrow strip, so it revisits each place less often, while wide-swath sensors see everything with coarser pixels. Commercial firms work around it by flying many satellites. While top commercial images can show individual cars, in my opinion most privacy fears outrun reality, because a satellite captures a moment, not a continuous feed of your backyard.
At work
Turning pixels into answers is a pipeline: radiometric calibration, atmospheric correction, orthorectification to remove terrain distortion, cloud masking, then analysis, increasingly with machine learning for change detection and land-cover classification. Failure modes include persistent cloud, differences between sensors, and models trained in one region misreading another. Licences often limit resale, and Canada's Remote Sensing Space Systems Act regulates Canadian operators. I'd validate any product against ground truth before using it for claims or compliance.
Real-world examples
At home
Marcus checked a wildfire map during a summer evacuation alert in British Columbia. The fire perimeter was drawn from satellite heat detections updated several times a day.
At work
Fatima works for a Prairie crop insurer. Her team uses Earth observation imagery to confirm hail and drought damage across fields before sending adjusters, speeding up claims.
How it compares
| feature | Earth observation satellite | Aerial or drone imaging |
|---|---|---|
| Coverage | Huge areas, repeatedly | Small areas on demand |
| Resolution | Tens of centimetres to tens of metres | Centimetres |
| Weather dependence | Optical blocked by clouds; radar is not | Needs flyable weather |
| Cost per area | Low at scale | Higher |
Satellites win on scale and repetition. Drones and aircraft win on detail.
For analysts
Satellite data can replace a lot of site visits, but I'd test it against ground truth first. Clouds, resolution limits, and processing errors can turn a confident map into a wrong one.
History & origin
TIROS-1, launched by the United States in 1960, sent back the first television images of Earth's weather from orbit. Landsat 1 followed in 1972, beginning the longest continuous record of Earth's land surface, and its archive was made free to the public in 2008. Canada launched RADARSAT-1 in 1995, then RADARSAT-2 in 2007 and the three-satellite RADARSAT Constellation Mission in 2019, used heavily for sea-ice and Arctic monitoring. Europe's Copernicus program began launching Sentinel satellites in 2014 with free, open data. Commercial firms such as Planet and Maxar pushed resolution and revisit rates much further. In my view, the decision to make Landsat and Sentinel data free did more for environmental science than any single satellite.