Open AccessIgnoring variation in wood density drives substantial bias in biomass estimates across spatial scales
Author(s) -
Jørgen Sand Sæbø,
Jacob B. Socolar,
Euclides Sánchez,
Paul Woodcock,
Christopher G. Bousfield,
Claudia A. Medina Uribe,
David P. Edwards,
Torbjørn Haugaasen
Publication year - 2022
Publication title -
environmental research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.37
H-Index - 124
ISSN - 1748-9326
DOI - 10.1088/1748-9326/ac62ae
Subject(s) - environmental science , biomass (ecology) , spatial variability , spatial distribution , lidar , extrapolation , sampling (signal processing) , spatial ecology , range (aeronautics) , atmospheric sciences , spatial heterogeneity , physical geography , soil science , remote sensing , ecology , statistics , geography , mathematics , geology , biology , materials science , filter (signal processing) , computer science , composite material , computer vision
Rapid development of remote sensing and Light Detection and Ranging (LiDAR) technology has refined estimates of tree architecture and extrapolation of biomass across large spatial scales. Yet, current biomass maps show significant discrepancies and mismatch to independent ground data. A potential obstacle to accurate biomass estimation is the loss of information on wood density, which can vary at local and regional scales, in the extrapolation process. Here we investigate if variation in wood specific gravity (WSG) substantially impacts the distribution of above-ground biomass (AGB) across a range of scales from local plots to large regions. We collected wood cores and measured tree volume in 341 forest sites across large altitudinal and climatic gradients in Colombia. At all spatial scales, variation in WSG was substantial compared to variation in volume. Imputing study-wide average values of WSG induced regional biases in AGB estimates of almost 30%, consequently undervaluing the difference between forest areas of low and high average wood density. Further, neither stem size nor climate usefully predicted WSG when accounting for spatial dependencies among our sampling plots. These results suggest that remote sensing- and LiDAR-based projections to biomass estimates can be considerably improved by explicitly accounting for spatial variation in WSG, necessitating further research on the spatial distribution of WSG and potential environmental predictors to advance efficient and accurate large-scale mapping of biomass.