Improvements in assessments of disease severity in conventional scouting using UAV-assisted multispectral imaging in watermelon
Melanie Kalischuk: North Florida Research and Education Center, University of Florida
<div>Agriculture-based multispectral imaging is popular but documentation of operational field-based benefits are limited. Multispectral imaging data were collected and analyzed from two commercial watermelon fields located in north Florida. The fields were rated for disease incidence and severity at random locations (conventional scouting) first followed by assessments at locations that were identified by differences in stress index and Normalized Difference Vegetation Index (NDVI) through multispectral imagery on an Unmanned Aerial Vehicle (UAV) platform (UAV-assisted scouting). Diseases identified in the watermelon fields included gummy stem blight, anthracnose, fusarium wilt, Phytophthora fruit rot, Alternaria leaf spot, and Cucurbit leaf crumple disease. A test for marginal homogeneity indicated that the severity ratings were different between conventional and UAV-assisted scouting. Higher severity ratings of 3 and 4 (on a scale of 0-5 from no disease to complete loss of the canopy) were more common after the scouts used the stress images in determining the locations for sampling. Experienced scouts tended to rate higher severities than lesser-experienced scouts. The UAV-assisted scouting locations had significantly greater green, red and red edge NDVI and lower stress index values than the scouted areas located using multispectral imagery. While progress has been made to identify some diseases using multispectral imaging, conventional scouting involving human evaluation remains necessary to validate other types of diseases. Multispectral imagery improved the output of watermelon field scouting due to the increased ability to identify hot spots more rapidly than conventional scouting practices.</div>
|
