Improving in-season nitrogen recommendations for maize using an active sensor.

摘要

An active crop canopy reflectance sensor could be used to increase N-use efficiency in maize (Zea mays L.), if temporal and spatial variability in soil N availability and plant demand are adequately accounted for with an in-season N application. Our objective was to evaluate the success of using an active canopy sensor for developing maize N recommendations. This study was conducted in 21 farmers' fields from 2007 to 2009, representing the maize production regions of east central and southeastern Pennsylvania, USA. Four blocks at each site included seven sidedress N rates (0-280 kg N ha-1) and one at-planting N rate of 280 kg N ha-1. Canopy reflectance in the 590 nm and 880 nm wavelengths, soil samples, chlorophyll meter (SPAD) measurements and above-ground biomass were collected at the 6th-7th-leaf growth stage (V6-V7). Relative amber normalized difference vegetative index (ANDVI_relative) and relative SPAD (SPAD_relative) were determined based on the relative measurements from the zero sidedress treatment to the 280 kg N ha-1 at-planting treatment. Observations from the current study were compared to relationships between economic optimum N rate (EONR) and ANDVI_relative, presidedress NO₃ test (PSNT), or SPAD_relative that were developed from a previous study. These comparisons were based on an absolute mean difference (AMD) between observed EONR and the previously determined predicted relationships. The AMD for the relationship between EONR and ANDVI_relative in the current study was 46 kg N ha-1. Neither the PSNT (AMD=66 kg N ha-1) nor the SPAD_relative (AMD=72 kg N ha-1) provided as good an indicator of EONR. When using all the observations from the two studies for the relationships between EONR and the various measurements, ANDVI_relative (R2=0.65) provided a better estimate of EONR than PSNT (R2=0.49) or SPAD_relative (not significant). Crop reflectance captured similar information as the PSNT and SPAD_relative, as reflected in strong relationships (R2relative) provided as good or better an indicator of EONR than PSNT or SPAD_relative, and provides an opportunity to easily adjust in-season N applications spatially.