Abstract:
Agarwood is a highly valued fragrant resin produced inside a few tree species belonging to the family
Thymalaeaceae as a self-defense response to plant stress. The amount of resin developed inside the tree cannot be
estimated by outside inspection. Consequently, harvesting trees before they reach their potential yield is a severe
drawback to the agarwood industry. Therefore, developing effective techniques for detecting agarwood resin
status inside the tree species has become a critically important task for the agarwood industry to increase
productivity. The present study evaluates the factors affecting Near-Infrared Spectroscopy (NIRS) models when
predicting agarwood formation inside A. crassna trunks using NIR spectroscopy. The research used 110 wood
specimens obtained from well-grown agarwood trees in a commercial plantation in Nawimana GS Division,
Matara District, Sri Lanka. NIR meter FQA-NIR Gun (588-1100nm) with a custom-made probe was used to
acquire NIR reflectance spectra without outside light interference. SIMCA models were built to identify the agar
resin-developed wood log areas from the normal wood areas in the tree trunk. SIMCA prediction models were
built to investigate three influencing factors, namely present or absent outside tree bark, surface roughness and
wood thickness agarwood prediction. Better prediction results were obtained from the bark-removed samples (at
the accuracy rates of 97%) to the bark present (85%), smooth wood surfaces (98%) to the rough surface (90%)
and 2mm thickness (98%) to the other thickness. The most effective wavelength for the separation of agarwood
present and absent samples was located at 978 nm of NIR. The study has demonstrated the potential possibility of
using NIR spectroscopy to identify the agarwood formation in A. crassna in non-destructive and rapid mode.
