Orr, Ryan, McBeath, Anna V., Dieleman, Wouter I.J., Bird, Michael I., and Nelson, Paul N. (2017) Estimating organic carbon content of soil in Papua New Guinea using infrared spectroscopy. Soil Research, 55 (8). pp. 735-742.

PDF (Published Version) - Published Version Restricted to Repository staff only

 

5

3

Abstract

Quantification of soil organic carbon (SOC) content is important for sustainable agricultural management and accurate carbon accounting. Infrared (IR) absorbance can be used to estimate SOC content, but the relationship differs between regions due to matrix effects. We developed an IR-based model specific for SOC in Papua New Guinean soils. A total of 437 samples from 0.0–0.3 m depth were analysed for SOC using Dumas combustion. IR absorption spectra were collected from the same samples, and a predictive regression model was developed using the 6000–1030 cm–1 spectral range. Using a validation set, predicted SOC values resulting from the IR-based model compared well with values from Dumas combustion (R2 = 0.905; ratio of performance-to-deviation = 5.64). Constraining wavelengths to positively correlated regions of the spectra was also explored and showed improved model performance (R2 = 0.932). Overall, IR analysis provides a robust method for estimating SOC content for a range of Papua New Guinean soils.Quantification of soil organic carbon (SOC) content is important for sustainable agricultural management and accurate carbon accounting. Infrared (IR) absorbance can be used to estimate SOC content, but the relationship differs between regions due to matrix effects. We developed an IR-based model specific for SOC in Papua New Guinean soils. A total of 437 samples from 0.0–0.3 m depth were analysed for SOC using Dumas combustion. IR absorption spectra were collected from the same samples, and a predictive regression model was developed using the 6000–1030 cm–1 spectral range. Using a validation set, predicted SOC values resulting from the IR-based model compared well with values from Dumas combustion (R2 = 0.905; ratio of performance-to-deviation = 5.64). Constraining wavelengths to positively correlated regions of the spectra was also explored and showed improved model performance (R2 = 0.932). Overall, IR analysis provides a robust method for estimating SOC content for a range of Papua New Guinean soils.

Actions (Repository Staff Only)

Item Control Page