Identifying grain-size dependent errors on global forest area estimates and carbon studies

Abstract

Satellite-derived coarse-resolution data are typically used for conducting global analyses. But the forest areas estimated from coarse-resolution maps (e.g., 1 km) inevitably differ from a corresponding fine-resolution map (such as a 30-m map) that would be closer to ground truth. A better understanding of changes in grain size on area estimation will improve our ability to quantify bias and uncertainty, and provide more accurate estimates of forest area and associated carbon stocks and fluxes. We simulated that global forest area estimated from a 1-km land-cover map (the most practical and finest resolution currently used for global applications) was 947,573 km² less than that of its corresponding 30-m map (excluding Antarctic and Greenland).

Citation

Zheng, Daolan; Heath, Linda S.; Ducey, Mark J. 2008. Identifying grain-size dependent errors on global forest area estimates and carbon studies. Geophysical Research Letters. 35: L21403. doi:10.1029/2008GL035746