[Quantitative determination of parameters of substrate using near-infrared spectroscopy technique]

Soilless culture has many virtues, such as space saving, time saving, etc.. It has become one of the technologies which developed fastest in agricultural products. The selection of substrate is one of the keys to determining the success of soilless culture. Therefore, it is important to rapidly determine the parameters of substrate. In the present paper, moisture, electronic conductivity and pH values of substrate were tested by near-infrared (NIR) spectroscopy. The spectra were preprocessed by baseline correction and derivative. Partial least squares (PLS) regression model was built using different wave bands. It was found that baseline drift was improved after correction. NIR spectroscopy can be used to determine EC value of substrate. The correlation coefficient r, root-mean-square error of the cross validation (RMSECV), relative percent difference (RPD) and bias of the optimum PLS was 0.923 6, 634 micros x cm(-1), 3.11 and 19.8 micros x cm(-1), respectively, when the best wave band was 4 246.7 - 7 502.2 cm(-1) and the best factor was 7. NIR spectroscopy technique can also be used to predict moisture of substrate although the accuracy of model should be improved. However, it can not be used to predict pH value of substrate.