Abstract
Nature-based climate solutions (NbCS) generate carbon credits through forest protection to reduce carbon emissions, or afforestation/reforestation to capture atmospheric carbon dioxide. NbCS can potentially unlock ~2 billion tons of carbon credits per year across the global tropics, which businesses can purchase to offset their carbon footprint and meet climate goals. Remote sensing technology provides a unique opportunity to transparently estimate aboveground carbon in tropical forests at multiple spatiotemporal scales and increase investor’ confidence in NbCS. However, there is little consensus on the remote sensing data and models used to accurately estimate aboveground forest carbon. To fill this knowledge gap, we conducted a quantitative review of remote sensing datatypes (optical, radar, lidar) and models (regression, machine-learning) that best predicted aboveground carbon in tropical forests. We compared the coefficient of determination (R2) results from 95 studies (501 field sites) across the tropics. We found that combining optical and lidar datatypes and machine-learning models best predicted aboveground forest carbon. However, only 36 studies (151 field sites) conducted validation of their products. Forest plot sizes used for field calibration and assessment also did not affect R2 values. Our findings provide insights for transparent and robust assessments of carbon projects for effective climate change mitigation.
Jose Don T De Alban
Global Head of Remote Sensing and Geospatial Analysis
land change | remote sensing | conservation | political ecology | sustainability | Southeast Asia