Regression modeling nitrogen fertilization requirement for maize crop by combining spectral reflectance and agronomic efficiency

Abstract

This study aimed to develop a novel regression model for prescription of required nitrogen (N) for maize by combining spectral reflectance data from GreenSeeker sensor and agronomic efficiency. For this end, two field trials at two sites were carried out with maize (Zea mays L.) on Oxisols under no-till systems in Southern Brazil. For both experiments, a randomized complete block design, with three replications in a split-plot arrangement, was used. The main plots consisted of four urea-N rates at sowing (0, 20, 40, and 60 kg N ha−1) and subplots included four urea-N rates in top dressing (0, 80, 160, and 240 kg N ha−1). Based on the results, a novel model for prescription of N rate in top dressing for maize was defined. The increases in maize yields with the N rates estimated by the model (16% to 39%) were similar to that obtained with the N rates used for maximum economic yield (15% to 42%). The estimated N rate by the model provided an economic return 36.5% and 7.0% higher than the N rate for maximum yield, and only 6.6% and 2.0% lower than the N rate for maximum economic yield, at sites 1 and 2, respectively. Thus, the economic return obtained using the model was closer to that reached with the N rate for maximum economic yield than that for maximum yield. Therefore, the developed model, combining spectral reflectance and agronomic efficiency, exhibited great potential to improve the maize N fertilization efficiency.