Subglacial Topography Training Images: 166 Elevation Maps at 500m Resolution

Emma MacKie of Stanford University provides a database of 166 topographic training images for subglacial terrain. The data is a 40,000 x 166 matrix, where each column reshapes to a 200x200 pixel image representing a 100x100 km area with a 500-meter resolution. Elevation data originates from the ArcticDEM and IBCSO bathymetric compilations, as cited in Yin et al. (2022).

Use Cases

• Training non-stationary multiple-point geostatistical models based on the provided topographic training images.
• Developing computer vision models for terrain feature recognition based on high-resolution elevation data.
• Benchmarking geospatial interpolation algorithms for Antarctic subglacial topography.
• Creating synthetic subglacial landscapes for geological simulation studies.

Strengths

• Contains 166 distinct topographic training images.
• Each image covers a 100x100 km area at a consistent 500-meter resolution.
• Data is sourced from authoritative elevation compilations: ArcticDEM and IBCSO.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

ArcticDEM (Porter et al., 2018) and IBCSO (Arndt et al., 2013)

Collection Method

Elevation data compiled from radar and bathymetric sources.

Geography

Likely focuses on West Antarctic and circum-Antarctic regions based on source citations.