1
High-Resolution UAS Imagery in Agricultural Research
Concepts, Issues, and Research Directions
Michael P. Bishop, Muthukumar V. Bagavathiannan, Dale A. Cope, Da Huo, Seth C. Murray, Jeffrey A. Olsenholler, William L. Rooney, J. Alex Thomasson, John Valasek, Brennan W. Young, Anthony M. Filippi, Dirk B. Hays, Lonesome Malambo, Sorin C. Popescu, Nithya Rajan, Vijay P. Singh, Bill McCutchen, Bob Avant, and Misty Vidrine
Contents
1.1Introduction
1.2Background
1.2.1Phenotyping
1.2.2Precision Agriculture
1.2.3Geospatial Technologies
1.3Study Area
1.4Image Acquisition
1.4.1Flight Operations and Sensors
1.4.2Data Acquisition Flight Strategies
1.4.2.1Prevailing Wind
1.4.2.2Waypoint Scheduling
1.4.2.3Waypoint Auto-Triggering
1.5Geometric Correction: UAS Photogrammetry
1.6Crop Assessment and Mapping
1.7Weed Assessment
1.8Conclusions
Acknowledgments
References
1.1Introduction
Crop yield (production per acre) has increased up to eightfold over more than a century of concerted scientific research into agricultural systems and genetic improvement of crops (Brummer et al. 2011; Hall and Richards 2013). The global population is expected to increase to over 9 billion people by 2050, and increases in the standard of living will require more food, fiber, and fuel (Godfray et al. 2010; FAO 2017). There will also be new constraints from climate change, decreased availability of fertilizer and irrigation water inputs, and external pressure from consumers, companies, and governments to produce agricultural products in a more sustainable way to address food quality and security issues (Godfray et al. 2010; Tilman et al. 2011; Challinor et al. 2014; FAO 2017). This will require new developments and utilization of agricultural and information technologies.
Given the advent of unmanned aerial systems (UASs) and high-resolution multispectral imagery, managers and agricultural specialists have new tools and information for optimizing management decisions and enabling precision agriculture solutions. Enabling technologies include global positioning systems (GPSs), high-resolution multispectral and hyperspectral sensors, geographic information system (GIS) technology, and field sensor and network monitoring capabilities. In addition, these information technologies also support improved information production and decision support capabilities using knowledge representation, artificial intelligence, and visualization techniques. Collectively, such enabling geospatial technologies permit mapping and monitoring of crops (Hunt et al. 2010; Torres-Sánchez et al. 2014, 2015), weed assessment and management (Armstrong et al. 2007; Gray et al. 2008; López-Granados 2010; Eddy et al. 2013; Peña et al. 2013; Torres-Sánchez et al. 2013), plant stress detection (Hunt et al. 2010; Zarco-Tejada et al. 2012), and many other precision agricultural applications throughout every aspect of preplanting through postharvest (Herwitz et al. 2004; Blackburn 2006; Castaldi et al. 2017). Nevertheless, although the potential for geospatial technologies to significantly impact agricultural research and management practices is high, numerous concepts and issues must be addressed to generate diagnostic and actionable information that can be relied upon for scientific inquiry, management, and optimization problem solving. There are numerous issues associated with UAS data acquisition strategies, image preprocessing, geospatial data management, information production, information synthesis, and the development and evaluation of agricultural decision support systems that are specific to agriculture.
Perhaps the greatest agriculture challenge is understanding the complex interactions between plants and their agricultural environment. Understanding these interactions may enable optimal management of plant growth by controlling input factors to maximize crop yield, sustainability, safety, and nutrition. Using objective information on these and other measurable factors, and applying site-specific management, is the basis of precision agriculture. Another major challenge is to breed and select the best genetics in a cultivar to meet crop production goals. Automated and semiautomated methods of plant phenotypes are termed high-throughput phenotyping (HTP), which seeks to use sensors deployed on various types of platforms to conduct measurements rapidly so that larger populations and/or more replicates can be measured and assessed. Consequently, developing and evaluating UAS technology and high-resolution imaging systems can significantly assist in precision agriculture and high-throughput phenotyping activities to help managers make better decisions and assist plant breeders and geneticists to screen more varieties more quickly and accurately, and with less cost. Even more exciting is the possibility of allowing plant breeders to observe traits or phenotypes that have never before been possible, such as the utility of UAS technology to temporally screen large fields and estimate plant growth curve trajectories.
Although UAS technology and the collection of high-resolution imagery can help address primary agricultural challenges, numerous concepts and researc...
