Inventories of Multi-type Landslides for Deep Leaning- and Remote Sensing-Based Mapping: A Case of the 2024 Noto Peninsula Events, Japan

This dataset contains two multi-class landslide inventories developed for the 2024 Noto Peninsula earthquake and rainfall events in Ishikawa Prefecture, Japan.

The datasets integrate optical (SPOT-6, PlanetScope) imagery with slope for deep learning–based multi-class landslide mapping and analysis.

Each dataset covers the entire study area and can be spatially subset for regional applications.

Dataset Contents

File: Ishikawa_20240518 (Post-earthquake, PlanetScope)This dataset corresponds to the co-seismic landslides following the Jan. 2024 event. It contains six bands:

Red band – PlanetScope red reflectance

Green band – PlanetScope green reflectance

Blue band – PlanetScope blue reflectance

NDVI band – Normalized Difference Vegetation Index derived from PlanetScope imagery

Slope band – Topographic slope derived from post-earthquake DEM (MLIT)

Mask band – Manually labeled multi-class landslide inventory used for model training and validation

File: Ishikawa_20241024 (Post-Rainfall, SPOT-6)This dataset corresponds to the post-rainfall imagery acquired after the September 2024 rainfall event. It contains seven bands:

Red band – SPOT-6 red reflectance

Green band – SPOT-6 green reflectance

Blue band – SPOT-6 blue reflectance

Near-infrared (NIR) band – SPOT-6 near-infrared reflectance

NDVI band – Normalized Difference Vegetation Index derived from red and NIR bands

DSM band – Digital Surface Model from dual-angle SPOT-6 imagery

Mask band – Manually labeled multi-class landslide inventory used for model training and validation