Reflectance Spectroscopy of Biochemical Components as Indicators of Tea (Camellia Sinensis) Quality

meng bian,andrew k skidmore,martin schlerf,fei teng,yanfang liu,tiejun wang

DOI: https://doi.org/10.14358/PERS.76.12.1385

2010-01-01

Abstract:The potential of reflectance spectroscopy to estimate the concentration of biochemical compounds related to tea (Camellia sinensis (L.)) quality (total tea polyphenols and free amino acids) is demonstrated. Partial least squares regression (PLSR) was performed to establish the relationship between reflectance and biochemicals for leaf powders as well as fresh leaves. Highest accuracy was found for tea powders with a cross-validated r(2) of 0.97 for tea polyphenols and 0.99 for free amino acids, and the root mean square error of cross validations (RMSECVS) are 8.36 mg g(-1) and 1.01 mg g(-1) for the two chemicals. The accuracy achieved at leaf level was slightly lower, with results yielding cross-validated r(2) of 0.91 and 0.93 with RMSECVS of 13.74 mg g(-1) and 2.32 mg(-1) for tea polyphenols and free amino acids, respectively. Important wavelengths for prediction of the two biochemicals from powder and leaf spectra were identified using the PLSR bcoefficients as indicators. Wavelengths of 1,131 nm, 1,654 nm, 1,666 nm, 1,738 nm and 1,752 nm were identified as bonds related to absorption by total tea polyphenols, while 1,492 nm represented the absorption feature office amino acids. The results obtained using fresh leaves indicate that hyperspectral remote sensing may be useful for routine monitoring of tea chemistry at landscape scale.