The definition of processing levels for optical data depends on the considered sensor. Most common satellite optical imagery are available in three distinct processing levels, from level 0 to level 2. The most used processing levels are level 1 and level 2, depending on the user and the application.
In Level 0, the raw data are processed in a way that they are ready to be archived. Processing operations generally includes telemetry analysis, error detections and granule concatenation. Furthermore, relevant parameters such as acquisition date and geographical reference are annotated in the form of metadata, this information being necessary for processing higher levels. Additionally, a quicklook of the image is generated. No correction is performed at this level.
Level 1 is often divided in several sublevels. Generally, both radiometric correction and geometric refinement are performed at this level. The radiometric processing includes several radiometric corrections such as dark signal correction or spectral band binning. The radiometric correction allows the determination of physical variables (e.g. reflectance) from the digital numbers. The geometric processing includes tiles association and resampling grid computation, in order to link for each image band its native image geometry to the target geometry. The result of this processing steps is usually a geocoded, Top of Atmosphere product.
Level 2 data usually consist of atmospherically corrected Level 1 data, i.e. Bottom-of-Atmosphere data. These surface reflectance products may be accompanied by additional outputs, such as scene classification, water vapor or surface temperature maps.
For specific applications and sensors, Level 3 application ready data are available. These are derivated products such as burned area, dynamic surface water content and snow cover maps.
Depending on the considered sensor and level, the name of the sublevels can differ: Sentinel 2 defines Level-1B as radiometrically corrected data. Level 1C are radiometrically and geometrically corrected data, i.e Top-Of-Atmosphere (TOA) orthoimage products. Landsat sensors distinguish between Terrain precision correction (L1TP), systematic Terrain Correction (L1GT) and Geometric systematic Correction (L1GS) depending on the quality of the reference data for geometric correction. These are usually separated into Tier 1 and Tier 2 datasets.