An Uncertainty-Aware Explainable AI Framework for Landslide Susceptibility Mapping Using Dempster-Shafer Evidence Fusion

The five GeoTIFF output maps (~30-50 MB each) : | File | Description | Values | |------|-------------|--------| | `PR_belief_landslide.tif` | DST belief in landslide occurrence, Bel(L) = m({L}) | 0 to 1 | | `PR_plausibility_landslide.tif` | DST plausibility, Pl(L) = m({L}) + m({L,N}) | 0 to 1 | | `PR_ignorance.tif` | Residual ignorance mass after fusion, m({L,N}) | 0 to 1 | | `PR_conflict_K.tif` | Maximum inter-modality conflict coefficient | 0 to 1 | | `PR_susceptibility_classes.tif` | 4-class susceptibility (Jenks breaks) | 1-4 |