Abstract:

In order to monitor Pestalotiopsis funerea desm fast, accurately and extensively and promote application of hyper-spectral remote sensing in monitoring forest diseases and insect pests, through high spectral data in different severity of P. funerea desm, relevant analysis was performed among canopy spectrum, first derivative data and the disease severity data, the inversion models of the P. funerea desm severity were built by single variable linear regression and multivariate stepwise regression techniques. The results showed that with the increase of severity level, the canopy reflectivity in visible region enhanced gradually, but the reflectivity in near infrared region weakened gently, and the greatest change in the region of 680-780 occurred. There was a significant linear relationship between the disease severity and the red edge feature parameters, and the multivariate stepwise regression model established by using the red edge feature parameters as independent variables was better than the single-variable linear model inversing severity level of the P. funerea desm. The fitting R2, forecasting R2 and RMSE was 0.815, 0.778 and 0.053 respectively, which showed that the red edge feature parameters had excellent indication function on severity level.