Derivation and evaluation of coarse resolution LAI estimates over Canada

Leaf area index (LAI) is a fundamental land surface parameter. Coarse (1 km) resolution sensors currently offer sufficient repeat frequency to provide large area estimates of LAI within the growing season. We describe an approach to generate Canada-wide LAI estimates using SPOT-Vegetation imagery. Relationships between surface LAI and optical vegetation indices are developed using a Canada-wide surface LAI data set and Landsat TM and ETM+ imagery. LAI is estimated using indices derived from infrared reflectances so as to minimize atmospheric correction uncertainties. Additional indices based on visible reflectances are used to screen for environmental influences on the infrared surface reflectances. Supplemental 1 km resolution land use maps are used to optimise LAI estimates. Comparison to a sampling of Landsat based LAI fields suggests total errors in LAI of 30% (1σ). Absolute validation is performed by cross-validation using the surface data set. The derived LAI estimates are ∼33% lower than similar MODIS LAI estimates over the BOREAS region.