Multispectral radiometric characterization of coffee rust epidemic in different irrigation management systems

Abstract

Coffee rust (Hemileia vastatrix Berk. & Br.) is one of the most prominent diseases in coffee (Coffea arabica L.) and causes serious damage to the crop. The pathogen incubation period may be long for about 30 days and 10% incidence of rust may result in 3 times more disease few days after the signals of rust appear in the leaves, even in the absence of new infections. The objective of this study was to evaluate the applicability of coffee crop monitoring under different irrigation systems by orbital radiometry, exploring the spectral signature and spectral, spatial and temporal pattern of rust incidence in the coffee field. The study was carried out in four areas of coffee plantations in Carmo do Rio Claro, Minas Geris, Brazil, between August 2012 and December 2014, under self-propelled, drip, center pivot irrigation systems and rainfed system. Fifteen Landsat-7/ETM + and Landsat-8/OLI-TIRS images were used, trying to establish a better sequence of images between in situ data of rust incidence in the coffee leaves and coffee leaf growth evaluated by sample meshes in the field along the time. Space-time disease incidence distribution maps, Pearson correlation and reflectance spectral signatures were used to evaluate coffee rust progress in the different irrigation fields. The highest coffee rust incidence occurred in August and corresponded to the values of lower NIR reflectance for all evaluated areas, independently of the irrigation management system. In the visible, SWIR-1 and SWIR-2 spectral regions, there were higher reflectance values in the rainfed area when compared to irrigated areas in rainy periods. There was a greater spectral and temporal variation of rust in the center pivot irrigation system when compared to the other irrigation management systems, presenting high values ​​of average incidence of rust above 30% in the periods close to the harvest period, from June to August 2013 and 2014. The high incidence of rust associated with coffee fruit harvest probably led to a reduction in plant leaf growth in center pivot and rainfed fields. There was a negative correlation between near infrared and rust in the self-propelled and center pivot management systems. High coffee rust incidence values mainly in center pivot and rainfed coffee fields determined reduction in the average reflectance of NIR and green and increase reflectance in red, SWIR-1 and SWIR-2 when compared to periods with lower rust in the coffee fields.