Landsat Burned Area Product Validation Across the Conterminous U.S., 1984-2018

19% omission and 41% commission error rates characterize this Landsat-derived burned area dataset validated with high-resolution commercial satellite imagery. The United States Geological Survey algorithm identifies burned areas from 1984 through 2018 using machine learning and image segmentation on Landsat analysis ready data. This product provides a unique data source for monitoring spatial and temporal fire patterns across the contiguous United States.

Use Cases

• Documenting spatial and temporal patterns of fire based on Landsat imagery time-series
• Quantifying drivers of wildfire based on burned area maps across the conterminous U.S.
• Assessing fire impacts on human and natural systems based on validated burned area data
• Comparing fire activity with other burned area datasets based on the provided validation metrics

Strengths

• Validation with high-resolution commercial satellite imagery resulted in omission and commission error rates averaging 19% and 41%, respectively
• Covers a long temporal extent from 1984 through 2018
• Algorithm uses machine learning, thresholding, and image segmentation on Landsat analysis ready data
• Products will be routinely produced as new Landsat data are collected

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count, file formats, and sample data are unknown, which may limit suitability assessment

Provenance

Source

United States Geological Survey

Collection Method

Algorithm identifies burned areas in Landsat analysis ready data time-series using machine learning, thresholding, and image segmentation

Time Range

1984 through 2018

Geography

Conterminous United States