Using hyper-spectral derivative indices to inverse Colletotrichumgloeosporioides disease indices

Remote sensing technology has made it possible to monitor vegetation under a range of environmental stress conditions. Several research results have illustrated the superior indicator functions of plant spectral reflectivity derivative over original data. To further investigate the application of remote sensing technology in monitoring Colletotrichum gloeosporioides, this paper used spectral reflectivity of oil camellia canopy measured by hand-held outdoor spectral radiometer (ASD, made in USA) in C. gloeosporioides disease index (DI) field survey. The first derivative of hyper-spectral data integrated with a moving average filter was pretreated. Through relevant analysis, the first derivatives highly related with DI were selected. Then using single variable linear and nonlinear regression methods, partial samples were chosen to build an inversion model. Accuracy test was subsequently accomplished using other tests. The results showed that reflection peaks and valleys of the first derivative of oil camelliae canopies in the visible-light region vanished gradually along with decreasing red-edge slope. A high correlation was noted between DI and the first derivative data in the regions of 480~513 nm, 526~569 nm, 583~607 nm and 669~727 nm. Using SDr′ as independent variable, the logarithmic model of inversed DI was the most accurate. The correlation coefficient R and RMSE between the predictive and observed values were 0.869 and 0.067, respectively, and also with much higher prediction accuracy. This study showed the feasibility of using the first derivative of hyper-spectral data to inverse C. gloeosporioides DI. This approach could be used to assess the health of oil camellia.