SENSORIAMENTO REMOTO COMO FERRAMENTA PARA AUXILIAR NA PRESCRIÇÃO DE FERTILIZANTE NITROGENADO EM CEREAIS

Abstract

The development of models to convert spectral reflectance measurement to Nitrogen (N) rate in cerals, as well as to identify factors that may interfere with these measurements, may increase the efficiency of this technique. This study aimed to: (i) study factors that can influence the reflectance measure and decrease the accuracy of N rate indication, and (ii) propose a model for the prescription of N requirements for the maize crop, from spectral reflectance measure data. For that, It was evaluated: (i) the influence of the positioning height of the sensor, the hours of operation and the stage of development of the wheat crop in spectral reflectance; (ii) the influence of the stage of development, the maize plant population and the soil cover in spectral reflectance; (iii) it was also proposed a model for prescription N rates in topdressing for maize combining spectral reflectance measure with Agronomic Efficiency of Nitrogen in Topdress (EANT). Spectral Reflectance data was obtained by commercial ground sensor GreenSeeker and converted into NDVI (Normalized Difference Vegetation Index) and INSEY (In-Season Estimated Yield) indices. In the wheat crop, it was observed that: (i) there were differences in NDVI values in relation to the heights in which the GreenSeeker sensor was positioned. Between the evaluated space, from 10 cm to 120 cm of height of the culture canopy, the average values stabilized from 30 cm up to 120 cm in height; (ii) before the N topdress application, NDVI presented a significant difference between the stages of development of the crop. After N topdress application the NDVI presented significant difference only at a more advanced stage of development of the crop; (iii) the averages of spectral reflectance measurement were different according to the reading times, with distinct photosynthetically active radiation, in the following order: night> morning> afternoon. In the maize crop, it was observed that: (i) NDVI values showed a significant difference in all levels of N applied at the sowing for all stages of development of the crop until the topdress N application; (ii) the NDVI was not influenced by soil cover, and increased linearly (p <0.01) with the increase of the maize population in the V5 stage. It was identified a quadratic relation between maize productivity gain with N topdress and EANT. This relationship was used to propose the model of prescription of N in coverage from spectral reflectance measurement. The results of the proposed model showed productivity close to the maximum values indicated by the regression equation, and economic result close to that obtained with the economic dose, respecting the productivity limitation found by the field situation. It was concluded that remote sensing presents potential to improve the efficiency of application of nitrogen fertilizers in cereals. Factors that interfere with spectral reflectance measurement should be controlled; otherwise they may affect the amount of N indicated by the models. The EANT is an important indicator and its use can increase the effectiveness of the models.