Spatially Explicit Large Area Biomass Estimation: Three Approaches Using Forest Inventory and Remotely Sensed Imagery in a GIS

• DOI: 10.3390/s8010529
• Publication Date: 2008-12-01T14:42:49Z
• Authors: Michael A. Wulder; Joanne C. White; Richard A. Fournier; Joan E. Luther; Steen Magnussen
• Keywords: Forest Inventory

Forest inventory data often provide the required base to enable largearea mapping of biomass over a range scales. However, spatially explicit estimates ofabove-ground (AGB) large areas may be limited by spatial extent theforest relative area interest (i.e., inventories not exhaustive), orby omission attributes for estimation. These andattributional gaps in forest result an underestimation areaAGB. The continuous nature and synoptic coverage remotely sensed have led totheir increased application AGB estimation areas, although use thesedata remains challenging complex environments. In this paper, we present anapproach generating integrating AGBestimates from multiple sources; 1. using lookup table conversion factors appliedto non-spatially exhaustive dataset (R2 = 0.64; RMSE 16.95 t/ha), 2.applying unique combinations land cover vegetation densityoutputs derived 0.52; 19.97 3. hybridmapping augmenting with where there are or attributional data. Over our714,852 ha study central Saskatchewan, Canada, estimate generated fromthe was approximately 40 Mega tonnes (Mt); however, inventoryestimate represents only 51% total area. theremotely outputs that overlap those made based approachdiffer 2 %; however total, is 30 % greater (58 Mt)than when entire isaccounted for. Finally, hybrid approach, whereby inputswere used fill inventory, wasestimated at 62 Mt. example presented, integration facilitates comprehensiveand