Alsat-2: calibrate the image before trusting the map

The Alsat-2 program comprises ALSAT-2A, launched in July 2010, and ALSAT-2B, launched in September 2016. Airbus describes a progression from joint work on the first satellite in France to integration of the second by ASAL engineers in Algeria. The two spacecraft supported a coordinated imaging program, while their construction history also recorded a change in the responsibilities carried by the Algerian team.

ASAL’s account of Alsat-2B describes orbital adjustment and in-orbit tests before image products entered wider use. A camera’s ability to sample fine detail is not the same as knowing exactly where that detail belongs on a map. Nor does a bright pixel automatically correspond to a known surface brightness. Those questions require calibration — open definition: establishing how the instrument and processing system relate measured values to the quantities users want.

Radiometric — open definition calibration concerns the instrument’s response to light, including offsets and differences between detector elements. Geometric — open definition calibration concerns the placement of observations, including alignment between spectral bands and reference locations on Earth. A map can be visually sharp yet shifted, and a well-positioned image can still have brightness errors. Correcting one problem does not prove the other is solved.

In an invented alignment exercise, a recognizable road crossing appears four pixels east of its surveyed reference. If each pixel represents 2.5 metres in this simplified map, the offset is ten metres. The arithmetic converts a measured displacement into ground distance; it does not diagnose whether the cause is pointing, timing, terrain treatment or processing. The team should test additional reference features before choosing a correction.

Detail does not remove map error

Suppose an analyst adjusts the image until that one crossing lines up. Checking only the same crossing would mainly confirm that the adjustment did what it was designed to do. A stronger validation — open definition uses other suitable reference points, distributed across the scene, and reports the remaining errors. It also checks whether the reference data themselves are accurate and appropriate for the observation date.

ASAL’s historical commissioning description includes both brightness-related and geometric work. Its figures describe a particular assessment stage, not a guarantee for every later image or every terrain type. The Alsat-2 lesson is to treat a finished-looking map as a measurement product with a documented chain of corrections. Users should be able to ask how it was calibrated, how its accuracy was tested and whether those tests address the decision they need to make.