Quantifying Spatial Variability of Peat Soil Carbon and Nitrogen using Infrared Spectroscopy, Statistical and Geo-Statistical Models

Chemical and biological analyses of soil carbon (C) and nitrogen (N) are time-consuming or require fresh material in cases of intensive in situ sampling. Infrared spectroscopy is one of the rapid and non-destructive methods that can be applied to a large number of soil samples. In this study, Attenuated Total Reflectance (ATR) Fourier transform infrared (FTIR) spectroscopy in the range of 600 to 4000 cm(-1) was assessed using partial least squares (PLS) regression model to predict total C and N of peat soils. ATR/FTIR-based PLS models had r(2) values of about 0.8 for fitted functions and 0.7 for leave-one-out cross-validations. Using an independent dataset to compare soil C and N values estimated by ATR/FTIR-based PLS models versus those measured using CHN elemental analyzer led to r(2) value of 0.97 for both soil C and N. The combined use of ATR/FTIR-based PLS models and inverse distance weighting (IDW) interpolation appears to be a promising method to estimate total C and N of peat soils for rapid data acquisition across spatially extensive areas.