Peatland Land Cover Map of Michigan's Upper Peninsula

NASA's dataset provides a 12.5-meter resolution land cover map specifically for peatland ecosystems in Michigan's Upper Peninsula. It was created using a random forest classifier trained on field data and aerial photography, applied to fused optical and L-band SAR imagery from Landsat-5 TM and ALOS PALSAR satellites. The map covers the period from 2007 to 2011 and achieved a 92% overall accuracy according to field-sampled assessments.

Use Cases

• Classifying peatland vegetation types using the pixel-based land cover classes.
• Validating regional carbon stock models for wetland ecosystems.
• Monitoring changes in wetland extent and health over time using the 2007-2011 temporal baseline.
• Training machine learning models for land cover classification with the field-validated random forest methodology.

Strengths

• High overall map accuracy of 92% based on field-sampled validation sites.
• Uses a multi-sensor fusion approach combining optical (Landsat-5 TM) and L-band SAR (ALOS PALSAR) data for improved classification.
• Derived from training data that includes polygons delineated from both field surveys and aerial photography.

Limitations

• Specific land cover class names and column definitions are not provided in the source metadata.
• There is a conflict in the 'last updated' date between platforms, with Data.gov listing 2026 and NASA Earthdata listing 2011.

Provenance

Source

National Aeronautics and Space Administration

Collection Method

Random forest classifier applied to fused Landsat-5 TM and ALOS PALSAR imagery, trained on field and aerial photo polygons.

Time Range

2007 to 2011

Freshness

2026-03-13 00:02:20.820987

Geography

Upper peninsula of Michigan, USA