Deploying an unmanned aerial vehicle (UAV) equipped with a multispectral sensor for detecting between differing levels of barley (H vulgare L.) fungal diseases in Northern Ontario

Abstract

To date, no studies exist exploring the multispectral detection of fungal diseases in barley (H Vulgare L.) with UAV imagery. The purpose of this work was to determine if the spectral response (VI) of barley fungicide treatment levels (Control, Stratego, Stratego + Prosaro) could be distinguished between two farmers’ fields and four growth stages (Feekes 8 to Feekes 11.4), when evaluating UAV multispectral imagery (Blue- 475 nm +/- 20 nm, Green- 560 nm +/- 20 nm, Red- 668 nm +/- 10 nm, Red Edge- 717 nm +/- 10 nm, Near Infrared-840 nm +/- 40 nm) of 6.7 cm/pixel spatial resolution. Radiometrically and geometrically corrected orthomosaics were generated and intrusive features such as weeds and crop damage were classified and extracted before the analyses of canopy level barley. Each photo was registered with RTK accuracy to ensure the analysis of identical pixelated areas between dates. A randomized complete block design was performed for 5 separate VIs: NDVI, RE-NDVI, RDVI, RE-RDVI, and TGI. 3-way interactions (Field x Growth Stage x Treatment) were found to be non-significant for NDVI (p=0.415), RE-NDVI (p=0.383) and TGI (p=0.780), with RDVI (p=0.003) and RE-RDVI (p=0.005) being significant. Despite some differences, a consistent trend in the spectral separability of fungal intensity by treatment type was observed regardless of field, from Feekes 10.51 onwards. With ground truthing, the mapping of fungicide intensity is possible with potential towards savings and environmental benefits from reduced fungicide use.