[Study on the influence of temperature on near infrared spectra]

Near-infrared (NIR) spectroscopy technique is one of qualitative and quantitative analysis techniques that were developed quickly in recent years. However, the characteristic of NIR spectral region determines its low absorbance, low sign-to-noise and peak overlapping, which results in the phenomenon that th NIR spectrum is affected by sample source, sample varieties, sample state, location and sample temperature and uncertainty of spectrum. In this paper, the influence of temperature on NIR spectra was studied and calibration and validation models of chlorophyll at different temperatures were built. The results show that temperature does have effect on the model precision. High precision could be achieved at 10 and 20 degrees C and the former was the best though using more principal components. Calibration and validation precisions of models were poor when it was 25 degrees C. At the same time, spectra obtained at different temperatures were classified using discriminate analysis. It was found that the spectra were misclassified exception ones obtained at 20 degrees C which meant that there were not clear differences between the spectra exception the ones at 20 degrees C. The primary research work about the condition and application of detecting leaves was studied which helps to provide theory foundation for proposing temperature modification model in the future.