Sugarcane mosaic virus detection in maize using UAS multispectral imagery

Authors: BEVERS, NOAH - The Ohio State University; Ohlson, Erik; KUSHAL, K - The Ohio State University; Jones, Mark; KHANAL, SAMI - The Ohio State University

Submitted to: Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/29/2024
Publication Date: 9/5/2024
Citation: Bevers, N., Ohlson, E.W., Kushal, K.C., Jones, M.W., Khanal, S. 2024. Sugarcane mosaic virus detection in maize using UAS multispectral imagery. Remote Sensing. 16(17) Article 3296. https://doi.org/10.3390/rs16173296.
DOI: https://doi.org/10.3390/rs16173296

Interpretive Summary: Maize dwarf mosaic (MDM) disease is one of the most important virus disease of maize worldwide. Detection of disease and monitoring its spread are critical for developing and deploying disease management strategies. However, traditional disease scouting is labor intensive, time consuming, and expensive. In this study, we explored whether unmanned aerial vehicles (UAVs) could be used to detect virus presence using multispectral imagery. A machine learning approach known as support vector machine (SVM), was able to predict the presence or absence of virus infection with greater than 70% accuracy. Our discovery indicates that UAVs can facilitate the rapid and widespread detection of virus spread and emergence, facilitating better disease management and leading to improved yields for growers.

Technical Abstract: One of the most important and widespread maize virus diseases is maize dwarf mosaic (MDM), which can be induced by sugarcane mosaic virus (SCMV). This study explores a machine learning analysis of five-band multispectral imagery collected via unmanned aerial system (UAS) during 2021 and 2022 for SCMV disease detection in maize fields. The three primary objectives are to: (i) determine the spectral bands and vegetation indices that are most important/correlated with SCMV infection in maize, (ii) compare spectral signatures of mock-inoculated and SCMV-inoculated plants, and (iii) compare model prediction performance for early and late in-fection of SCMV. On average, SCMV-inoculated plants had higher reflectance values for blue, green, red, and red-edge bands and lower reflectance for near-infrared as compared to mock-inoculated samples. Machine learning models explored included ridge regression, support vector machine (SVM), random forest, and XGBoost. Across both years, XGBoost regression per-formed best for predicting disease incidence percentage (R2 =0.29, RMSE = 29.26) and SVM clas-sification performed best for binary prediction of SCMV-inoculated vs mock-inoculated samples (72.9% accuracy). Generally, model performances appeared to increase as the season progressed into August or September. According to SHAP analysis of the top performing models, simplified canopy chlorophyll content index (SCCCI) and saturation index (SI) were the vegetation indices that consistently had the strongest impacts on model behavior for SCMV disease regression and classification prediction. The findings of this study demonstrate the potential for the development of UAS image-based tools for farmers, aiming to facilitate precise identification and mapping of SCMV infection in maize.