The application of near infrared (NIR) spectroscopy for the quantitative assessment of soil organic matter fraction in forests

In this study we investigated if near infrared (NIR) spectroscopy can be effectively used to predict content of various fraction (LF - light fraction and MAF - mineral associated fractions) of soil organic matter, also in the context of their spatial distribution. Additionally, we used NIR spec- troscopy to evaluate basic properties of forest soils. We analyzed 256 soil samples from the topsoil of plots in central Poland. Using laboratory techniques, we divided the soil samples into two soil carbon fractions: the light fraction (LF) and mineral-associated fraction (MAF). A calibration model was developed using the spectra from 171 soil samples and their corresponding measured values. The regression model was validated using 85 independent soil samples. Using this model, we estimated the following forest soil properties: carbon concentration in light fraction (CLF), carbon concentration in mineral-associated fraction (CMAF), ratio of the CLF to the total carbon content of the soil sample (CLF/C), ratio of the CMAF to the total carbon content of the soil sam- ple (CMAF/C), the total concentration of carbon (Ct) and nitrogen (Nt), C:N ratio (CN), pH, the concentration of exchangeable base cations (BC) and cation exchange capacity (CEC). The best calibration results were obtained for CLF, CLF/C and CMAF/C. The largest adjusted coefficients of determination for validation were obtained for Nt, CN, BC and CEC. Model developed for the CLF was characterized by inaccurate value prediction. The paper shows also the relationship between the optimum number of soil sample spectra and the absolute and relative measure- ment error. Comparison of the measured and predicted values show that NIR spectroscopy has potential for determining soil parameters. The statistical assessment and spatial distribution analysis of the modelled CMAF demonstrated relatively good agreement with measured values. However, the model's assessment of the CLF was less accurate. We conclude that NIR spectroscopy is most applicable for use in soil science to determine the parameters: Ct, Nt, C/N, pH, CEC, CMAF and CMAF/C.

Automated summary

This study investigated the use of NIR spectroscopy to predict the content of different fractions of soil organic matter and evaluate forest soil properties. The results showed the potential of NIR spectroscopy in determining soil parameters, although it had less accuracy in assessing the light fraction.