Le foreste tropicali rappresentano circa la metà del carbonio (al di sopra del livello del suolo, aboveground carbon) presente nella vegetazione globale; esse sono oggi sottoposte a notevoli pressioni per effetto dei cambiamenti di uso del suolo, come deforestazione e degradazione. Resta tuttavia un’incertezza nella determinazione delle emissioni di carbonio delle foreste tropicali dovute a queste attività umane. Quantificare l’incertezza nella stima della biomassa e del carbonio immagazzinato (carbon stock) appare pertanto fondamentale per comprendere il ciclo globale del carbonio e sostenere il programma REDD (Reducing Emissions from Deforestation and Forest Degradation) delle Nazioni Unite.
In uno studio pubblicato di recente sulla rivista Remote Sensing of Environment un gruppo di ricercatori (fra cui G. Vaglio Laurin and R. Valentini della Divisione IAFES del CMCC) ha proposto un nuovo metodo per quantificare gli errori nelle misurazioni della biomassa delle foreste con sensori LIDAR.

L’abstract dell’articolo:
Quantifying the uncertainty of the aboveground biomass (AGB) and carbon (C) stock is crucial for understanding the global C cycle and implementing the United Nations Program on Reducing Emissions from Deforestation and Forest Degradation (UN-REDD). The uncertainty analysis of remotely sensed AGB is tricky because, if validation plots or cross-validation is used for error assessment, the AGB of validation plots does not necessarily represent the actual measurements but estimates of the true AGB. Leveraging a recently published pan-tropical destructively measured tree AGB database, this study proposed a new method of characterizing the uncertainty of the remotely sensed AGB. The method propagates errors from tree- to landscape-level by considering errors in the whole workflow of the AGB mapping process, including allometric model development, tree measurements, tree-level AGB prediction, plot-level AGB estimation, plot-level remote sensing based biomass model development, remote sensing feature extraction, and pixel-level AGB prediction. Applying such a method to the tree AGB mapped using airborne lidar over tropical forests in Ghana, we found that the AGB prediction error is over 20% at 1 ha spatial resolution, larger than the results reported in previous studies for other tropical forests. The discrepancy between our studies and others reflects not only our focus on African tropical forests but also the methodological differences in our uncertainty analysis, especially in the aspect of comprehensively addressing more sources of uncertainty. This study also highlights the importance of considering the plot-level AGB estimate uncertainty when field plots are used to calibrate remote sensing based biomass models.

Leggi la versione integrale dell’articolo:
Chen Q., Vaglio Laurin G., Valentini R.

Uncertainty of remotely sensed aboveground biomass over an African tropical forest: Propagating errors from trees to plots to pixels
2015, Remote Sensing of Environment, in press, DOI: 10.1016/j.rse.2015.01.009