USGS provides a dataset from over 40 hours of submarine video collected in April 2004 to ground-truth geophysical data in Glacier Bay, Alaska. The video examines geologic characteristics like rock, boulders, and bioturbated mud, and investigates relationships between substrate and benthic communities. A supervised decision-tree classification was used to predict the distribution of four principal sea-floor geomorphic types.
Use Cases
- Train models to classify sea-floor geomorphic types (e.g., rock, boulders, rippled sand) from video observations and acoustic reflectance data.
- Analyze the correlation between substrate types (e.g., bioturbated mud, mussel beds) and the distribution of benthic communities.
- Validate and improve bathymetric and acoustic reflectance maps using georeferenced video ground-truth data.
- Build predictive habitat distribution maps by applying hierarchical decision-tree classification to geophysical features.
Strengths
- Over 40 hours of georeferenced submarine video provides direct observational ground-truth.
- Covers water depths from 15 to 370 meters across lower and central Glacier Bay.
- Analysis distinguishes four principal sea-floor geomorphic types from video data.
Limitations
- Sample size and temporal coverage are limited to a single collection period in April 2004.
- Geographic scope is restricted to Glacier Bay, Southeast Alaska, limiting generalizability.
Provenance
- Source
- CEOS_EXTRA, with summary provided by the USGS.
- Collection Method
- Georeferenced submarine digital video collection and supervised hierarchical decision-tree statistical classification of geophysical data.
- Time Range
- April 2004.
- Freshness
- null
- Geography
- Glacier Bay, Southeast Alaska.