Assessment of sudden death syndrome in soybean through multispectral broadband remote sensing aboard small unmanned aerial systems

Abstract

Sudden Death Syndrome (SDS) has spread from the US to other countries which is causing yield loss of 10–15% and 70% in extreme cases from infected plants. Currently, SDS impacts are scored by visual assessment of infection severity and percent of crop diseased. The quality of manually collected row based coarse data collected over several hours can be impacted by assessment errors and changes in diurnal environmental conditions. Small unmanned aerial systems (sUAS) offer an alternative method to provide a more accurate and reliable measurement of crop disease. A platform designed to collect high throughput aerial imagery data to quantify SDS is proposed. A comparative evaluation of ground-based and aerial remote sensing methods for scoring of SDS is proposed to evaluate efficacy. The purpose of this research was to (1) compare accuracy and benefits of ground-based and aerial remote sensing methods for scoring of SDS, (2) determine if pigment index (PI) can be used for the assessment and quantification of SDS, and (3) assess if PI can be utilized for determination of maturity. A seven-acre field was selected as test plots to collect reflectance using both a ground-based spectrometer and sUAS aerial imagery using broadband modified color infrared sensor over a two-year period. Aerial imagery was collected once in 2016 (FD1) and twice in 2017 (FD2 and FD3) using a sUAS late in the growing season and at maturity each year. PI values were compared to manual collected ground-based data. Results from check plots indicated that the PI derived using aerial imagery and the ground-based spectrometer data explained 80 and 78% of the variation in SDS scores, respectively. When analyzing only high instances of SDS (SDS score >25) in the check plots aerial data and ground-based data explained 84 and 71% of the variation in SDS scores, respectively. Correlations between SDS scores and different indices analyzed showed that only PI and Blue Normalized Difference Vegetation Index (BNDVI) were significantly correlated with SDS score, with PI showing significantly greater correlations (−0.79 (FD2) and −0.72 (FD3)) to SDS on field scale than BNDVI (−0.36 (FD2) and −0.35 (FD3)). The PI derived from aerial imagery data showed strong correlations with SDS score, SDS severity and plant maturity, indicating that PI can be used in field studies to quantify critical growth indicators in soybean plots.