Deciduous Cover and Canopy in Boreal North America, 1992-2015

NASA's ABoVE project produced 30-meter resolution maps of deciduous tree fraction and total canopy cover across the North American boreal domain for five nominal years. The data, derived from Landsat imagery using Random Forest models trained on in-situ samples from Alaska and Canada, includes per-pixel uncertainty estimates. These geospatial time series are designed to track forest composition changes over decades.

Use Cases

• Tracking shifts in forest composition from deciduous to evergreen species based on multi-seasonal NDVI composites.
• Modeling carbon storage and albedo changes using high-resolution tree canopy cover maps.
• Validating and calibrating regional or global vegetation models with in-situ trained, per-pixel uncertainty data.
• Assessing the impact of disturbances like fire or insect outbreaks on forest structure across the boreal domain.

Strengths

• Provides a long-term time series covering 23 years (1992-2015) at a high 30-meter spatial resolution.
• Includes per-pixel uncertainty estimates derived from the standard deviation across all Random Forest decision trees.
• Models were trained and validated using a 70/30 split of in-situ samples collected across Alaska and Canada.

Limitations

• Key metadata such as row count, file size, and specific license details are missing or conflicting across platforms.
• The dataset's temporal coverage is constructed from multi-year composites, which may smooth over annual variability.
• Deciduous fraction values are only valid for pixels where tree canopy cover exceeds 25%, limiting analysis in sparse areas.

Provenance

Source

National Aeronautics and Space Administration (NASA), ABoVE project.

Collection Method

Random Forest regression models applied to NDVI-based median-value composites from Landsat 5, 7, and 8 surface reflectance data.

Time Range

1992 to 2015

Freshness

2026-04-10 03:59:07.062252

Geography

North American boreal domain (Alaska and Canada).