Mapping Above-Ground Carbon Stocks at the Landscape Scale to Support a Carbon Compensation Mechanism: The Chocó Andino Case Study

Abstract

(1) Background: Tropical Mountain forests (TMF) constitute a threatened major carbon sink due to deforestation. Carbon compensation projects could significantly aid in preserving these ecosystems. Consequently, we need a better understanding of the above-ground carbon (AGC) spatial distribution in TMFs to provide project developers with accurate estimations of their mitigation potential; (2) Methods: integrating field measurements and remote sensing data into a random forest (RF) modelling framework, we present the first high-resolution estimates of AGC density (Mg C ha−1) over the western Ecuadorian Andes to inform an ongoing carbon compensation mechanism; (3) Results: In 2021, the total landscape carbon storage was 13.65 Tg in 194,795 ha. We found a broad regional partitioning of AGC density mediated primarily by elevation. We report RF-estimated AGC density errors of 15% (RMSE = 23.8 Mg C ha−1) on any 10 m pixel along 3000 m of elevation gradient covering a wide range of ecological conditions; (4) Conclusions: Our approach showed that AGC high-resolution maps displaying carbon stocks on a per-pixel level with high accuracy (85%) could be obtained with a minimum of 14 ground-truth plots enriched with AGC density data from published regional studies. Likewise, our maps increased precision and reduced uncertainty concerning current methodologies used by international standards in the Voluntary Carbon Market.

Keywords

• remote sensing
• permanent plots
• Andes
• Sentinel
• artificial intelligence
• mountain forests