Prediction of total nitrogen content of loblolly and slash pine biomass using near infrared spectroscopy

Leticia Sonon, Uttam Saha, Jake Mowrer, Yuangen Yang, and David Kissel

Agricultural and Environmental Services Laboratories, The University of Georgia Cooperative Extension 
2300-2400 College Station Road, Athens, GA 30602, USA

Abstract

With the recent increase in energy demand and a consequent increased use of biofuels, new technologies and techniques are needed to increase production of biomass suitable for the biofuel industry. Loblolly pine (Pinus taeda) and slash pine (Pinus elliotti) biomass have great potential to be used in biofuel industry because of their encouraging gross caloric values. Total nitrogen content in the plant at various stages of growth is a useful and effective tool for developing appropriate nitrogen management strategies to maximize biomass production and nitrogen use efficiency. Therefore, non-destructive spectroscopic sensing techniques have the potential to be a useful and accurate method to rapidly determine the total nitrogen content. In this study, we used Near-infrared Reflectance Spectroscopy (NIRS) to determine the total nitrogen content of woody biomass of 8-17 year old pine  trees of Georgia. The 104 samples included in this study were chips of bark, branch, wood, needle as well as whole trees. The samples were oven-dried (at 60 ^oC), coarse-ground, and passed through a 2 mm sieve. The processed samples were scanned on a Foss NIRSystem model 6500 scanning monochromator in the reflectance mode. The scan covered the wavelength range from 400 to 2500 nm at 2 nm intervals to give a total of 1050 data points per sample. The reference total nitrogen contents were determined on an Elementer Combustion Nitrogen Analyzer.  The calibration equation was developed using modified partial least-squares regression with internal cross validation using 79 randomly chosen samples. The reference total nitrogen content values of these calibration samples ranged from 0.02 to 0.22%, with standard deviation of 0.04. The equation developed had low standard error of calibration (SEC = 0.0090%) and cross-validation (SECV = 0.0139%) with high coefficient of determination in both calibration (R² = 0.8479) and cross validation (1-VR = 0.8479). Prediction of an independent validation set of 25 samples showed significant correlation between the NIRS predicted values and the reference total nitrogen content based on the standard error of prediction (SEP = 0.007%), coefficient of determination in prediction (r² = 0.9620), and the ratio of standard deviation for prediction (RSP = the ratio of standard deviation of reference data to SEP(c) = 1.94).  The results suggest that NIRS could be used to rapidly determine total nitrogen contents of woody biomass of loblolly and slash pines.