Skip to main content

Dataset Processing Level

Data level processed selects the image layer used to generate the dataset. Every assigned image must provide that level. A processing level is metadata describing what the values are intended to represent; it does not by itself certify units, calibration, atmospheric correction, or provenance.

The dataset editor currently supports:

  • Radiance: values intended to represent spectral radiance; verify their units and calibration provenance
  • Reflectance: reflectance values as imported or produced; depending on provenance, these may be relative white-reference reflectance or atmospherically corrected surface reflectance
  • TOA Reflectance: values intended to represent top-of-atmosphere reflectance
  • Absorbance: values intended to represent absorbance derived from a compatible source layer

Raw and Dark layers can exist on an image but are not dataset processing-level choices.

Choose the level that matches the scientific objective and the layers actually available across all assigned images. When assigned training images share one processing level, Clarity may auto-select it; review the selection before finalizing. Changing the level can mark already assigned images as incompatible. Matching level names across images do not establish matching units or correction methods.

See Process Hyperspectral Data to create needed image layers before assigning them to a dataset.