Abstract: With the implementation of major projects such as the comprehensive management of rocky desertification and ecological restoration, the karst region of southwest China has become one of the regions with the fastest growth of vegetation cover and aboveground biomass in the world, and is a first potential region in China to achieve the goal of carbon neutrality. However, due to the complex geological background, the estimation of aboveground vegetations biomass in this region is uncertain. Therefore, accurate estimation of aboveground biomass of vegetations in Southwest China and monitoring and evaluation of carbon sequestration under ecological restoration, are urgently needed to provide scientific guidance for optimising ecological management projects, achieving carbon neutrality and forest management.In this data set, MODIS and SMOS satellite observations are used to estimate aboveground carbon density changes in southern China from 2002 to 2017. A static benchmark map of carbon density for 2015 was used to train a machine learning algorithm applied to annual MODIS imagery to estimate carbon density changes. SMOS low-frequency passive microwave data at 25 km × 25 km resolution was used to independently assess biomass changes and soil moisture trends from 2010–2017. The carbon density dataset covers southern China at 500 m × 500 m spatial resolution from 2002–2017. It was validated against independent tree cover data at 30 m and 5.6 km resolutions, with correlation coefficient of 0.9. The dataset quantifies the carbon sequestration impact of different forest management strategies like afforestation, harvesting, and natural regrowth, and provides data on biomass carbon density of multiple ecosystems, vegetation types, and vegetation structures at different temporal and spatial scales in the southern region of China, which can support ecological and environmental management and research at multiple scales. It demonstrates southern China's forests offset 33% of regional fossil fuel emissions, with implications for climate change mitigation. Meanwhile, the results of this dataset can also be applied to ecological environmental protection, climate change, ecosystem service assessment, carbon emission accounting, land use planning and other thematic research areas, which can provide important decision-making support and analyses for government departments, scientific research institutions and environmental protection organisations.
Keywords: Carbon density; eight south-western provinces of China; multi-source remote sensing; machine learning