Towards detection and understanding of paraglacial landslides in Southern Alaska using surface velocity data and deep learning. Supplementary files

Inventory

landslides_southernAK.gpkg

Landslide inventory derived through manual mapping

includes attributes:

landslide_ID: ID

movement_confidence: confidence in active movement

delineation_confidence: confidence of delineation

ident_aerial: identifiability in multispectral satellite imagery

ident_DEM: identifiability in DEM

ident_GSSICB: identifiability in radar coherence

ident_ts: identifiability in multispectral satellite imagery time series

Model

mrcnn_landslide_detection_notebook.ipynb

Jupyter notebook for model training and inference

model_weights.h5

Best performing weights from model training, can be used together with mrcnn_landslide notebook

dataset_landslide_detection.zip

Contains normalized tiles and masks for training