In the lush landscapes of Southern Assam, India, a groundbreaking study is reshaping our understanding of deep soil organic carbon (SOC) stocks and their implications for carbon accounting and land management. Led by Jintu Kumar Bania from the Department of Ecology and Environmental Science at Assam University, this research leverages advanced machine-learning models to predict the spatial distribution of deep SOC stocks across various land use systems.
The study, published in the journal ‘Scientific Reports’ (translated to English as ‘Scientific Reports’), highlights the critical role of deep SOC in the global carbon cycle, a component often overlooked in previous research. “Deep soil organic carbon is a vital but understudied aspect of the carbon cycle,” Bania explains. “Our work aims to fill this knowledge gap by providing high-resolution spatial analysis of SOC pools up to 5 meters deep.”
Using a combination of geospatial technology and machine-learning models, the research team evaluated the SOC stocks of five different land use systems: forested areas, open forest, tea agroforestry, other plantations, and agricultural lands. A total of 86 pits were excavated, and 6,450 samples were collected for analysis. Sentinel-2 imagery, obtained during the post-monsoon season for optimal visibility, provided crucial data for the study.
The findings reveal that the Random Forest (RF) model outperformed the Support Vector Machine (SVM) model, with a higher R² value of 0.92 and a lower RMSE of 40.59 Mg C ha⁻¹. “The Random Forest model proved to be more accurate in predicting the spatial distribution of deep SOC stocks,” Bania notes. “This model identified elevation as a key factor influencing SOC distribution across the region.”
The study’s predictions show significant variations in SOC stocks across different land use systems. Agricultural areas had the lowest SOC stock at 147.78 Mg C ha⁻¹, while forested areas had the highest at 483.43 Mg C ha⁻¹. The total SOC stored in the study area was estimated at 170.53 Teragrams (Tg), with forested areas storing the most carbon (76.79 Tg) and open forests the least (17.45 Tg).
These findings have profound implications for the energy sector and carbon budgeting. Accurate predictions of deep SOC stocks can inform sustainable land use practices, climate change mitigation strategies, and soil conservation efforts. “Understanding the spatial distribution of deep SOC stocks is crucial for developing effective carbon accounting and land management strategies,” Bania emphasizes. “This research provides a critical basis for informed decision-making in these areas.”
The study also highlights the importance of temporal land use transitions, which directly impact regional carbon accounting and land management strategies. By offering high-resolution spatial analysis of SOC pools, this research paves the way for more precise and effective carbon budgeting and land use planning.
As the world grapples with the challenges of climate change and sustainable development, studies like this one are essential for guiding policy and practice. The insights gained from this research can help shape future developments in the field, ensuring that land use practices are both environmentally sustainable and economically viable. “Our hope is that this research will contribute to a more sustainable future for Northeast India and beyond,” Bania concludes.