Bayesian Updating of Landcover Deviation (BULC-D)
BULC-D is based on the logic of the Bayesian Updating of Land Cover (BULC) algorithm from Cardille & Fortin, 2016. It uses Bayesian updating to estimate whether every pixel in a study area is stable or changing from year to the next. Using an earlier year to build a harmonic expectation, BULC-D weighs the deviance of every image in the later year from the expected value at that same point in the year. Values that are more unexpected (for example, an unusually low NBR) are stronger evidence of change.
Scripts in Google Earth Engine (GEE)
Access the full repository: https://code.earthengine.google.com/?accept_repo=users/alemlakes/r-2909-BULC-Releases
Link to presentation slides: https://docs.google.com/presentation/d/1sIAz_xrgomnVB-C9oxbNnHvRGAzlP-qpfRN6JUtpmqc/edit?usp=sharing
Link to GEE script for Belize: https://code.earthengine.google.com/b0ed8807b36787b5aaca2a5db690a2e2
Link to view cloud-free data by country: https://africa-uah.users.earthengine.app/view/spatialtemporalcloudfree
BULC-D interface in GEE: https://code.earthengine.google.com/ab64b87e06ce223d1587dc95623cbc69
Contacts
Jeffrey Cardille
jeffrey.cardille@mcgill.ca
Africa Flores-Anderson
africa.flores@nasa.gov
Flavie Pelletier
flavie.pelletier@mail.mcgill.ca