Frequent occurrence of drought disaster will seriously affect the growth and development of winter wheat (Triticum aestivum). We set different water stress treatments (80%, 60%, 45%, 35%, 30% of field water capacity) to simulate the severity of drought disaster. We measured free proline content (Pro) of winter wheat, and investigated the responses of Pro to canopy spectral reflectance under water stress. Three methods, i.e., correlation analysis and stepwise multiple linear regression (CA+SMLR), partial least squares and stepwise multiple linear regression (PLS+SMLR), and successive projections algorithm (SPA) were used to extract the hyperspectral cha-racteristic region and characteristic band of proline. Furthermore, partial least square regression (PLSR) and multiple linear regression (MLR) methods were used to establish the predicted models. The results showed that Pro content of winter wheat was higher under water stress, and that the spectral reflectance of canopy changed regularly in different bands, indicating that Pro content of winter wheat was sensitive to water stress. The content of Pro was highly correlated with the red edge of canopy spectral reflectance, with the 754, 756 and 761 nm bands being sensitive to Pro change. The PLSR model performed good, followed by the MLR model, both showing good predictive ability and high model accuracy. In general, it was feasible to monitor Pro content of winter wheat by hyperspectral technique.
Read full abstract