Abstract:

Both reflection spectroscopy and transmission spectroscopy data were recorded by HR4000 high-resolution portable spectrometer and leaf folder with built-in tungsten halogen light source. Using a variety of pretreatment methods on the spectral data pre-processing, mathematical models between reflection spectroscopy, transmission spectroscopy and chlorophyll content of wheat leaves were established using partial least squares (PLS) method, respectively. The effect of different pretreatment methods on predicted performance was analyzed. The results showed that the best detection model to predict chlorophyll content in wheat leaves by reflection spectroscopy with 0.950 and 0.926 as the correlation coefficient (r) of calibration and predicted set, 0.1287 and 0.1892 as the standard deviation (SD) of calibration and predicted set, 5.67% and 8.26% as the relative standard deviation (RSD) of calibration and predicted set, respectively. These parameters of the best detection model to predict chlorophyll content in wheat leaves with transmission spectroscopy were 0.846 and 0.880 as the correlation coefficient (r) of calibration and predicted set, 0.2201 and 0.2714 as the standard deviation (SD) of calibration and predicted set, 9.69% and 11.84% as the relative standard deviation (RSD) of calibration and predicted set, respectively. The results indicated that the application of reflection spectroscopy and transmission spectroscopy to quantitative analysis the chlorophyll content of wheat leaf was feasible. Through comparison, the model based on reflection spectroscopy was shown to have better predictability than the model based on transmission spectroscopy. At the same time, it provided an effective way for the quick non-destructive determination of chlorophyll content in leaf.