Can leaf area index and plant height measurement improve sensor-based nitrogen recommendations and yield prediction for corn?
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Nitrogen (N) management remains a significant challenge for corn growers due to the unpredictability and influence of weather conditions, soil properties, and soil biological activity on N transformations in the soil. Innovative technology is needed to assist farmers in making accurate in-season N recommendations to improve N use efficiency (NUE) and reduce the environmental impacts of N losses. Sensor-based aerial imagery can be collected using unmanned aerial vehicles (UAVs) to assist with N management decisions and help improve NUE. However, there are limitations associated with vegetative indices from aerial imagery in guiding N decisions because the indices can reach a “saturation point” once the corn canopy closes. We hypothesized that adding leaf area index (LAI) data and plant height measurements could improve our understanding of how plant biomass is related to the vegetative indices for predicting corn N response by adding in a third dimension to the analysis. Corn N rate trials were established in Delaware, Maryland, and Pennsylvania (0, 30, 60, 90, 120, and 150% of university-based N rates; DE and MD did not have a 0 N rate) and four replicates in a randomized complete block design. In-season UAV-multispectral imagery, LAI, and plant height measurements were obtained at the V6 and R2 corn growth stages. Plant height was also derived from UAV imagery using structure from motion (UAV-SFM) using Pix4D photogrammetry software. Drone-derived vegetative indices, UAV-SFM, and LAI were used to predict the sidedress N rates and grain yields, which were compared to yield data at harvest. ☐ The PA and DE sites exhibited a yield response to the N rate; however, the MD site was non-responsive. At the PA location, the N rate at the yield plateau occurred at 166 kg ha-1, with an in-season sensor-based N recommendation of 159 kg ha-1 and an economic optimum N rate (EONR) of 180 kg ha-1. At the DE site, the N rate at the yield plateau was 315 kg ha-1, with rates of 81 and 219 kg ha-1 for the in-season recommendation and EONR, respectively. We saw a significant LAI response to N rate only at the PA site (P-value = .0027), whereas there was a significant plant height response to N rate at the DE site (P-value = .0029). The in-field plant height measurements and the UAV-SFM did not correlate 1:1 across all three sites and relationships were inconsistent across sites. As such, UAV-SFM derived plant height was deemed as an unsuitable proxy for the in-field plant height measurements across sites. The normalized difference red edge index (NDRE) was the best predictor of corn grain yield at the R2 stage (r2 = .951, P-value < .0001); the addition of LAI and plant height to the predictive model marginally improved upon the yield prediction of NDRE (r2 = .9621 and .9708, respectively). Therefore, NDRE alone is still the recommended vegetative index for predicting corn yield. Further studies are needed to explore the contribution of biomass estimates on in-season N recommendations and yield prediction.
Description
Keywords
Nitrogen, Weather conditions, Soil properties, Unmanned aerial vehicles, Leaf area index