For the case study, we will provide a few datasets. Moreover, participants are not restricted to use only the official datasets provided. They may incorporate any freely accessible and open geospatial or satellite data, such as Landsat or Sentinel-1/2, that they deem appropriate. There are no limitations on the methodology used for data pre-processing. Nevertheless, participants must disclose the data sources they utilised and their pre-processing methodology if they employ more data than the official dataset provided, and they must share these datasets on the discussion forum for others to access.

The datasets provided for the case study include:

1. a Digital Terrain Model (DTM) in a raster format at a 5 m/pixel scale (source: Lombardy Region),

2. training dataset (Train.gpkg) which is a combination of:

2.1. a landslide inventory comprising the boundaries of translational, rotational shallow landslides, and debris flows at a 1:10,000 scale in vector format. In the Train.gpkg has an assigned value of 1 in the “Target” field. (modified from source: ISPRA – Regione Lombardia. Inventario dei Fenomeni Franosi in Italia - IFFI).

2.2. No Landslide Zone - a vector layer depicting areas with low probability of shallow landslide occurrence at a 1:10,000 scale in vector format. The NLZ is computed using slope angle and lithology for the AOI. (source: GISGeolab@Politecnico di Milano). In the Train.gpkg has an assigned value of 0 in the “Target” field.

3. a road network at a 1:10,000 scale in vector format (source: Lombardy Region),

4. a river network at a 1:10,000 scale in vector format (source: Lombardy Region),

5. geological fault zones map at a 1:10,000 scale in vector format (source: Lombardy Region),

6. a land use/land cover map at a 1:10,000 scale in vector format (source: Lombardy Region),

7. Meteorological data:

7.1. Interpolated yearly averaged hour precipitation for the year of 2020 (source: ARPA Lombardia),

7.2. 90th percentile of the hour precipitation for the year of 2020 (source: ARPA Lombardia).

License: CC-BY

Checklist for data submission

A landslide susceptibility map at a 5 m spatial resolution should be computed for the case study.

An executable and technical document describing the algorithm and procedure is also required.

Participants must also specify the criteria for defining areas with a low probability of landslide occurrence (zero-case scenario).

The code utilised for data processing must be submitted, and it should be limited to Python and GEE JavaScript. The submitted code's potential for data processing will be evaluated based on the technical document, from the raw data to the resulting maps. The submission will not be accepted if the methodology is deemed unrepeatable.