The application of unmanned aerial vehicle to precision agriculture: Chlorophyll, nitrogen, and evapotranspiration estimation

Precision agriculture (PA) is an integration of a set of technologies aiming to improve productivity and profitability while sustaining the quality of the surrounding environment. It is a process that vastly relies on high-resolution information to enable greater precision in the management of inputs to production. This dissertation explored the usage of multispectral high resolution aerial imagery acquired by an unmanned aerial systems (UAS) platform to serve precision agriculture application. The UAS acquired imagery in the visual, near infrared and thermal infrared spectra with a resolution of less than a meter (15--60 cm). This research focused on developing two models to estimate cm-scale chlorophyll content and leaf nitrogen. To achieve the estimations a well-established machine learning algorithm (relevance vector machine) was used. The two models were trained on a dataset of in situ collected leaf chlorophyll and leaf nitrogen measurements, and the machine learning algorithm intelligently selected the most appropriate bands and indices for building regressions with the highest prediction accuracy. In addition, this research explored the usage of the high resolution imagery to estimate crop evapotranspiration (ET) at 15 cm resolution. A comparison was also made between the high resolution ET and Landsat derived ET over two different crop cover (field crops and vineyards) to assess the advantages of UAS based high resolution ET. This research aimed to bridge the information embedded in the high resolution imagery with ground crop parameters to provide site specific information to assist farmers adopting precision agriculture. The framework of this dissertation consisted of three components that provide tools to support precision agriculture operational decisions. In general, the results for each of the methods developed were satisfactory, relevant, and encouraging.