Geolocation accuracy is how closely the ground position assigned to an image or pixel agrees with its true position on the Earth.12

USGS characterisation work assesses it by comparing ground coordinates measured in orthorectified images with well-surveyed ground control points used as checkpoints. Results are reported as absolute error, for example a circular error at a stated percentile, with confidence bounds that reflect the number of checkpoints in the scene.2

Accuracy can change with the reference data used. For Landsat Collection-2, USGS describes changes to the ground control points and elevation datasets used as the geometric reference, and reports that the geometric accuracy of Tier-1 terrain-corrected products improved. That assessment is described as preliminary.3

Geometric corrections are also among the criteria in CEOS Analysis Ready Data specifications, assessed per product family.1 Calibration and georeferencing parameters, including platform ephemeris, are carried with data at early processing levels.4 The measured values in the sensor report relate to one sensor and test site, and the sources do not give a general figure.2

Footnotes

  1. CEOS, “CEOS Analysis Ready Data”. https://ceos.org/ard/ ↩ ↩2

  2. USGS, “System characterization report on Vision-1”, Open-File Report 2021-1030, chapter Q. https://pubs.usgs.gov/publication/ofr20211030Q/full ↩ ↩2 ↩3

  3. USGS, “Preliminary assessment of the geometric improvements to the Landsat Collection-2 archive”, SPIE 2021. https://pubs.usgs.gov/publication/70222614 ↩

  4. NASA Earthdata, “Data Processing Levels”. https://www.earthdata.nasa.gov/learn/earth-observation-data-basics/data-processing-levels ↩