Application of Near-Infrared (NIR) Spectroscopy for Predicting the Density of Rattan (Calamus spp.) from Transverse and Bark Surfaces

Rattan is a vital organic biomaterial alongside wood and bamboo and is used in furniture, handicrafts, and construction. The industry necessitates a rapid, noninvasive assessment of rattan quality for optimal efficiency. This study aims to identify an optimal prediction model for estimating rattan density using near-infrared (NIR) spectroscopy. The rattan laboratory data included measurements of green and air-dry densities. A prediction model was constructed using NIR spectra obtained from the transverse and bark surfaces. Multivariate data analysis of cross-validation partial least squares regression was applied with many pretreatment choices for the spectra. This study yielded the most accurate prediction model for rattan density, with a coefficient of determination for cross-validation (R2CV) of 0.51. This was achieved by analyzing the air-dry density and spectra acquired on the bark with a pretreatment using the first derivative with 25 smoothing points. Distinct events were observed in the original NIR spectra of both surfaces. The bark spectra exhibited prominent peaks at 1,728 and 1,762 nm, followed by additional peaks at 2,308 and 2,348 nm, which were absent in the spectra of the transverse surface. The best density prediction model for rattan derived from the bark has considerable industrial applicability.