Relationship between Sentinel-2 orbital data and in situ monitoring of coffee rust

Abstract

Coffee rust (Hemileia vastatrix Berkeley & Broome) is the main coffee disease in Brazil. Coffee rust control is calendar-based and performed by applying chemicals in order to avoid the outburst of disease epidemics. The aim of this work was to identify coffee rust using multispectral orbital sensing through analyses utilising vegetation indices and coffee rust incidence, defoliation and yield data obtained in situ. Field samples were georeferenced using a high-accuracy global navigation satellite system receiver in two plots, in a 42-year-old, leaf rust susceptible coffee cultivar. Conventional chemical control of rust was performed in only one of the plots to serve as control. Coffee rust incidence in areas with and without chemical control was assessed over 5 months, from December 2018 to April 2019, a period with optimal environmental conditions for disease occurrence. Following analyses of different vegetation indices and data sampled in the coffee crop, Pearson’s correlations between the variables were verified. Reported correlations occurred mainly among coffee rust incidence levels in February 2019 and vegetation indices calculated using Sentinel-2 images from August 2018, September 2018 and February 2019 [IRECI r = 0.566; IRECI r = 0.493; NDMI r = − 0.518; NDVI(RE1) r = − 0.562; CI(RE1) r = − 0.573; MSR(RE1) r = − 0.569], in areas without coffee rust control. Indices based on relationships between bands in the Red-Edge and Near Infra-Red regions were more sensitive to spectral changes in vegetation due to variation in coffee rust occurrence over time.