Predicting Soil Organic Carbon from Loss-On-Ignition Across Four Regions of Nebraska

Abstract

ABSTRACT Soil organic carbon (SOC) is a dynamic property that is a key indicator of soil management effectiveness within efforts toward climate change mitigation. This study was conducted to create locally-informed equations for predicting SOC from loss-on-ignition (LOI) analysis. Soil samples were collected from four counties across Nebraska. Total carbon, inorganic carbon, and LOI were analyzed in the lab, and SOC was calculated as the difference between total and inorganic carbon. Locally-trained equations relating SOC and LOI were evaluated. The accuracy of predictions based on the equations was assessed by calculating the root mean square error (RMSE). Linear relationships were observed between SOC and LOI, and were found to be weaker in western Nebraska (R2 = 0.59), compared to central and eastern parts of the state (R2 = 0.93). Predictions were more accurate using the locally-trained predictions (RMSE = 0.05–0.13) compared to conventional estimates (assuming organic matter is 58% carbon) (RMSE = 0.30–0.53), and were further improved by developing separate equations for surface (0–25 cm) and subsurface (25–180 cm) soils (RMSE = 0.02 to 0.07). Locally-trained, depth-specific predictive equations can be used to improve the accuracy of SOC analysis using the routine LOI analysis.