关键词: 生物质燃烧, 碳排放, 气候变化, 随机森林, 东南亚

Abstract: Southeast Asia harbors one of the world’s richest stocks of aboveground biomass. However, the recurrence of biomass burning in this region not only degrades habitat quality but also accelerates global climate change. Analyzing the spatiotemporal patterns of carbon emissions from biomass burning in Southeast Asia can help improve regional habitat quality and enhance resilience to global climate change. In this study, we integrated observational data on aboveground biomass carbon density in Southeast Asia with multisource remote sensing information and applied a random forest algorithm to model and reconstruct annual gridded aboveground biomass data for 2001-2020. By combining vegetation-specific combustion efficiencies and emission factors, we calculated the spatiotemporal distribution of biomass burning-induced carbon emissions in the region for each year over this 20year period. The results show that biomass burning in Southeast Asia is the most frequent in the winter and spring, and that the total burned area has been slowly declining over time. Over the past two decades, biomass burning carbon emission peaked at 0.15 Pg C in 2004 and reached a lowest value of 0.04 Pg C in 2001, which exhibited a slow downward trend. During the study period, the mean carbon emission from biomass burning was 801.55 g C·m^-2, showing a slightly increasing trend amid interannual fluctuations. Biomass burning activity also showed marked spatial heterogeneity, being concentrated mainly in the central and northern Indochinese Peninsula, the south-central region of Kalimantan (Borneo), and the southeastern Indochinese Peninsula. In the first two regions, most fires were forest fires, whereas in the latter region they were predominantly agricultural waste burning caused by traditional farming practices. These findings provide a scientific basis for developing fire management strategies and carbon emission reduction measures in Southeast Asia.

Key words: biomass burning, carbon emission, climate change, random forest algorithm, Southeast Asia