Quantification and mapping of the carbon sequestration potential of soils via a quantile regression forest model

Abstract

Understanding the soil carbon sequestration potential is vital for decision-making related to crop and soil management and for prioritizing the area for carbon sequestration and climate change mitigation. In the present study, we mapped the soil carbon sequestration potential (CSP) along with its uncertainty over two depth ranges (0–30 cm and 0–100 cm) in parts of Western Ghats, Kerala, India, using datasets from 150 soil profiles. The difference between the soil organic carbon (SOC) saturation potential of the finer soil fractions (clay and silt) and the current soil organic carbon content of finer particles is considered as the CSP. The actual SOC stock and CSP in the study area were mapped using the quantile regression forest (QRF) algorithm. The model yielded better predictions of the SOC stock (R² = 0.52–0.55) than did the CSP model (R² = 0.22–0.36). The predicted SOC stock and CSP for 100 cm depth in the study area ranged from 5.2 kg m⁻² to 26.18 kg m⁻² and from 4.89 kg m⁻² to 28.69 kg m⁻², respectively. In total, approximately 256 Tg and 1089 Tg CO₂ equivalents could theoretically be stored in the top 30 cm and 100 cm of the study area, respectively. A relatively greater CSP was detected for soils of annual crops (18.6 kg m⁻²) than for soils of plantations (10.5 kg m⁻²) and forests (9.5 kg m⁻²) in 100 cm soil depth. The high sequestration potential of soils of annual crops could be met with agronomic practices such as mulching and conservation tillage. Improved agricultural management practices could have a significant impact on CO₂ mitigation through the construction of SOC stocks, which has also added the advantage of high agricultural productivity.