Using Multispectral Aerial Platforms to Manage Soybean Cyst Nematode Disease

The soybean cyst nematode (SCN) is one of the most damaging soybean pathogens in the United States. However, it is difficult to be detected and monitored in production fields. This study develops an early detection methodology to provide timely precise information to the controlling and management of SCN disease development. Field trails were established at SCN infested farms in Southern Illinois, and microplots were set up at the university agronomy research center. Ground truth data including soil SCN eggs and yields were collected. Hyperspectral data were collected with an ASD hand-held spectroradiometer from microplots with varying SCN infections. Multispectral image data at the field level were collected with a DJI Matrice 210 drone mounted with an Altum multispectral camera on a weekly basis. A total of 16 common UAV-based Vegetation Indices (VI) were selected to examine their statistical relationships with SCN infestation and yields. Enhanced Vegetation Index (EVI) and Difference Vegetation Index (DVI) were found to have the highest correlations with both SCN egg counts and crop yields. Based on the hyperspectral reflectance analysis, we found that red and near infrared bands were most sensitive to SCN infections. Based on the microplot data analyses, an unsupervised machine learning and a supervised deep learning approach were applied to identify SCN-infested areas in the field level. The proposed research framework is expected to benefit the monitoring and management of other soybean pathogens.