STADOS: Radiometric Harmonization for Multi-Temporal NDVI
39% and 31% reduction in temporal variance for water and rooftop anchors
ProblemRaw NDVI drifts across satellite acquisition dates because of atmospheric path radiance that changes with each pass, making it hard to trust whether a crop's vegetation trend is real or just an artefact of the atmosphere that day.
SolutionA five-stage framework that identifies stable reference objects in the scene, weights them by reliability, and solves for a per-date correction factor that removes the atmospheric drift while leaving NDVI's mathematical definition untouched.
Impact39% and 31% less temporal noise on the two most reliable reference classes, with a documented, honestly quantified trade-off for the anchor that doesn't fully cooperate with a single shared correction.
Process simulation
Naziru explains this step
Full conversation walkthrough
A complete, in-depth explanation of this project, turn by turn. Runs about 5 minutes if you let it play.
Ask a question about this work
Send Naziru a question directly, it's delivered straight to his inbox.
Commonly asked questions
Private, for the author only.
This section is password-protected.
Incorrect password.