Doctor of Philosophy (PhD)
School of Plant, Environment, and Soil Sciences
Nitrogen (N) management is being conducted at flat rate in Louisiana due to practicality and convenience, but the price of N fertilizer and high breakeven costs are forcing producers to find ways to reduce costs and optimize N application. In this scenario, precision agriculture technologies, specifically the use of optical sensors on board of unmanned aerial systems (UAS) to variable rate N application on farm is showing a promising approach to save inputs and reduce environmental impacts. However, the general goal of this research was to develop and evaluate in-season N management approaches for N recommendation in corn (Zea mays L.) fields using plant canopy sensors and UAS. The specific objectives were to: 1) investigate the differences in spectral reflectance bands and vegetation indices for sensing the N status of corn, through different hours of the day, under different weather conditions and sun irradiation angulation; and 2) evaluate an in-season N fertilizer recommendation algorithm based on an approach that reflects local conditions and needs for N fertilization using active crop canopy sensor and unmanned aircraft systems coupled with multispectral camera, and to validate and compare the algorithm proposed with other approaches. The experiments were conducted in three fields at the LSU Doyle Chambers Central Research Station located at Ben Hur Road, Baton Rouge, LA, 30.365°N, -91.166°W, with continuous corn during the growing seasons from 2018 to 2021. To investigate time of the day effects on active and passive sensor systems the experiment was conducted at the same location in corn during a day with percentages of cloudy coverage conditions varying from 80 to 100 %, with very few moments of cloud dispersion resulting in 100% of clear sky at the target area. The conclusion in this experiment addressing objective 1 is that the data obtained from passive sensors (commercial UAS camera and spectroradiometer), contrarily to the active crop canopy sensor, presented prominent significant variations in measurements at different times of the day, especially observed when ambient conditions changed solar radiation. This indicates higher sensitivity to changes during the day for the wavebands and vegetation indices derived using these sensors. For objective 2, the main conclusions are: (i) a practical and easy to implement algorithm approach was proposed and validated considering local conditions and implemented in-season, (ii) the use of the Chlorophyl Red Edge Vegetation Index (CIRE) obtained from the crop canopy reflectance with the approaches developed from local data to manage N status, can address spatial variability presented in fields through the different responses obtained for N fertilization across the sites analyzed, and (iii) the virtual approaches using both active and passive sensors, indicated relatively better performances based on yield and partial factor productivity (PFP) responses. Due to the easy implementation this finding suggests that this approach has great potential to be applied for N recommendations regardless of the type of sensor used to collect data.
Martins, Murilo de Santana, "Integration of Remote Sensing Approaches for In-Season Nitrogen Management" (2022). LSU Doctoral Dissertations. 5939.
Shiratsuchi, Luciano S.