01A view of the whole Earth#
Earth observation is the collection and interpretation of measurements about the planet’s land, oceans, ice, and atmosphere. Satellites provide broad, repeat views of places that are difficult to reach, while aircraft and instruments on the ground add closer or more local measurements. Together, these observations help people study weather, ecosystems, hazards, and long-term environmental change.[1]
A satellite record can show change across a large region, but a measurement is not automatically an explanation. A bright pixel might reflect a surface material, cloud, viewing angle, or processing choice. Earth observation works best when measurements are interpreted with their context and, where possible, compared with other evidence on the ground.[1][2]
Earth observation spans remote measurements from satellites and aircraft as well as in-situ observations used to characterize or validate conditions at specific locations. A satellite instrument samples a footprint through a viewing geometry and acquisition time; the resulting observation represents neither every point in that footprint nor every moment between visits. Combining observations across platforms can improve coverage, but differences in calibration, scale, and timing must be accounted for.[1][3]
02Different instruments answer different questions#
Some instruments record sunlight reflected by Earth or heat naturally emitted from its surface and atmosphere. Others send out a signal and measure its return. Optical sensors can distinguish reflected energy across visible and infrared wavelengths; radar sends microwave pulses and can collect observations at night and through many clouds. Altimeters time a returning pulse to estimate height, while atmospheric instruments examine how gases absorb or emit particular wavelengths.[1]
No single instrument sees every property of the planet. A radar image, a thermal measurement, and a colour image encode different physical interactions, so their values cannot be read as if they were the same kind of photograph. The instrument and its orbit are designed around the question being asked, as with other kinds of satellite missions.[1]
Passive sensors measure naturally available reflected or emitted radiation; active sensors transmit energy and measure the returned signal. For optical instruments, band placement and spectral response determine which portions of the electromagnetic spectrum are sampled. Radar backscatter depends on the wavelength, polarization, viewing geometry, and surface properties, so it is not a direct measurement of a single universal quantity such as roughness. Interferometric radar can compare phase between compatible acquisitions to estimate displacement along the radar line of sight, subject to coherence and geometry limits.[1]
Orbit and instrument design jointly set observation opportunities. Spatial resolution describes the size of the smallest distinguishable detail in a product; spectral resolution describes the separability of wavelength information; temporal resolution concerns observation frequency; and radiometric resolution concerns the ability to distinguish signal levels. These properties trade against one another: a narrower swath can limit area covered per pass, while revisit also depends on orbit, pointing, and constellation design. Basic orbital geometry is therefore part of interpreting coverage and revisit.[1]
03From measured signal to usable data#
What users often call a satellite image is a processed data product. Processing can correct instrument and viewing effects, place measurements on a map, and combine spectral bands into a display or a quantity such as surface reflectance. The choices matter: a false-colour image assigns measured bands to display colours, and a surface-reflectance product estimates light at the surface after accounting for atmospheric effects.[1][2]
For example, USGS Landsat surface-reflectance products are designed to improve comparisons between images by accounting for atmospheric effects such as aerosol scattering and thin clouds. Their stored digital numbers need the product’s scaling factors before they represent meaningful reflectance values. A map is thus the end of a measurement and processing chain, not an unmediated view of the ground.[2]
A product-processing chain may transform instrument counts into calibrated radiance, apply geometric and atmospheric corrections, then retrieve a geophysical variable. Each step uses assumptions and auxiliary information, so product level and algorithm version are essential metadata. For Landsat Collection 2, USGS describes surface reflectance as a Level-2 product generated from suitable Level-1 inputs and notes that scaling factors are required to convert stored digital numbers to reflectance.[2]
Calibration establishes how instrument indications relate to standards or known targets; validation assesses product quality by independent evidence. NASA’s NISAR calibration and validation plan describes pre-launch instrument and algorithm work and post-launch checks against measurement requirements, including field measurements and validation sites. These steps support stated accuracy and stability, but they do not remove all uncertainty or make every retrieval valid for every surface or condition. Such engineering trade-offs are part of mission systems engineering.[3]
