Intact forest carbon dynamics and indicative carbon economy in Southeast Asia: A remote sensing and random forest approach

Intact tropical forests store high amounts of carbon biomass, but this long-stored carbon has not been widely quantified at the regional scale. This study integrates multi-source remote sensing data and Random Forest (RF) to model above- and below-ground biomass carbon (AGBC and BGBC) in intact forests across Southeast Asia from 2000 to 2020. Intact forests were operationally defined as continuously forested areas that remained free from deforestation, cropland, and oil-palm expansion—criteria tailored to the dominant anthropogenic drivers of forest loss in Southeast Asia. Over the study period, intact forest extent declined by approximately 59.35%, reflecting major disturbance pressures across the region. The RF model demonstrated strong predictive performance against the reference data and the ESA CCI biomass dataset, with results of R AGBC 2 = 0.899 and R BGBC 2 = 0.874, then R AGBC 2 = 0.645 and R BGBC 2 = 0.632, respectively. Total forest biomass carbon declined from 32,786.73 ± 6.15 MgC (2000) to 19,913.93 ± 4.23 MgC (2020). In addition, to illustrate the magnitude of these losses, an indicative non-market valuation was applied using two carbon price scenarios. The results show a decrease from USD 776 billion to USD 471 billion based on regional prices and from USD 9.02 trillion to USD 5.48 trillion based on global recommendations. These findings provide quantitative evidence for urgent prioritisation of forest conservation in Southeast Asia, and the spatially explicit carbon stock estimates produced here offer a science-based foundation for strengthening REDD+ commitments, NDCs, and equitable carbon economy frameworks supporting indigenous and forest-dependent communities across Southeast Asia.