Padrão espectro-temporal da cultura do crambe (Crambe abyssinica Hochst) e estimativa de rendimento de grãos por sensores terrestres

Abstract

Crambe is an annual crop with high oil content in the seeds. Cultivated in winter, it presents a short cycle and low implantation cost. The extracted oil is used in the production of industrial lubricants, biofuels and electrical insulating fluids. In Brazil, several researches have been carried out mainly choosing experiments that evaluate grain yield in different regions. However, data related to the spectral-temporal behavior of the crambe during the developmental phases are non-existent. The knowledge of these data enables the monitoring of the crop by remote sensing techniques, which subsidize official organisms in the estimation of grain crops of the country. In this context, the present study aims to (i) characterize the spectral-temporal pattern of crambe obtained by terrestrial sensors in the city of Cascavel, state of Paraná; and (ii) to evaluate the relationship between spectral bands and vegetation indices obtained in different stages phenological measurements by terrestrial spectral sensors, with the grain yield of the crambe crop. The experiment was carried out in Cascavel, PR in the winter crops of 2014 and 2015. During these two years hyperspectral information was collected with the FieldSpec® 4 Hi-Res (FS4) terrestrial sensor through the phenological phases of crambe development, generating spectral-temporal and temporal profile of vegetation indexes by normalized difference (NDVI) and adjusted to soil (SAVI). The daily averages of the temporal profiles were submitted to the exploratory analysis of the data. In 2015, spectral information was collected with active and passive ground sensors (FS4 and Greenseeker® 505 Handheld) during all phenological phases of the plant. Separated in spectral bands of red, near and medium infrared and generated indices NDVI, SAVI, and the Normalized Difference Moisture Index (NDMI). At the maturation stage, plants were collected to determine grain yield. Linear regression models between spectral data and crop yield were elaborated and evaluated. During the seedling and vegetative growth phases the spectral band of red showed higher reflectances, due to the soil present in the sensed area. In the vegetative growth and early flowering phases, the highest reflectance in the near infrared (NIR) occurred, related to the green biomass of the canopy. Phases of granulation and maturation were characterized by the reduction in reflectance in the IVP and medium infrared (SWIR), due to leaf senescence and loss of cellular water content. The temporal profiles of NDVI and SAVI demonstrate linear growth up to the vegetative peak of the crambe, characterized between the end of the phenological phase of the vegetative growth and the beginning of the flowering. The model generated by NDMI in 66 days after sowing (DAS), at the beginning of flowering, showed a negative correlation with grain yield (R2adj: 0,49; RMSE: 134,80 kg ha-1). The spectral range of red (648 to 672 nm) in 66 DAS presented positive linear correlation with grain yield (R2adj: 0.36; RMSE: 151.38 kg ha-1). The flowers present in the canopy of the plant, in the flowering phase, contributed to the increase of reflectance in the spectral range of red and infrared. The data of terrestrial sensors enabled the spectral-temporal characterization of the crambe crop, generating subsidy for the delimitation of the cultivated areas with the culture by orbital sensors. Excessive precipitation and wind during the flowering, granulation, and maturation phases during the 2015 harvest contributed to variability in grain yield, reducing the explanatory capacity of production with the models.