Doctor of Philosophy / Department of Agronomy / Ignacio Ciampitti / The adoption of Remote Sensing (RS) in agriculture have been mainly utilized to inference about biological processes in a scalable manner over space and time. In this context, this work first explores two non-traditional approaches for rapid derivation of plant performance under field conditions. Both approaches focus on plant metrics extraction exploiting high spatial resolution from Unmanned Aerial Systems (UAS). Second, we investigate the spatial-temporal dynamics of corn (Zea mays L.) phenology and yield in the corn belt region utilizing high temporal resolution from satellite. To evaluate the impact of the adoption of RS for deriving plant/crop performance the following objectives were established: i) investigate the implementation of digital aerial photogrammetry to derive plant metrics (plant height and biomass) in corn; ii) implement and test a methodology for detecting and counting corn plants via very high spatial resolution imagery in the context of precision agriculture; iii) derive key phenological metrics of corn via high temporal resolution satellite imagery and identify links between the derived metrics and yield trends over the last 14 years for corn within the corn belt region. For the first objective, main findings indicate that digital aerial photogrammetry can be utilized to derive plant height and assist in plant biomass estimation. Results also suggest that plant biomass predictability significantly increases when integrating the aerial plant height estimate and ground stem diameter. For the second objective, the workflow implemented demostrates adequate performance to detect and count corn plants in the image. Its robustness highly dependends on the spatial resolution of the image, limitations and future research paths are further discussed. Lastly, for the third objective, outcomes evidenced that for a long-term perspective (14 years), an extended reproductive stage significantly correlates with high yield for corn. When considering a shorter-term period (last 4 years) mainly characterized by optimal growth conditions, early season green-up rate and late season senescence rate positively describe yield trend in the region. The significance of the variables changed according to the time-span considered. It is noticed that when optimal growth conditions are met, modern-hybrids can capitalize by increasing yield, due to primarily a faster (green-up) rate before flowering and on senescence rate better describes yield under these conditions. The entire research project investigates opportunities and needs for integrating remote sensing into the agronomic-based inference process.

Digital image processing is one of the potential technologies used in precision agriculture to gather information, such as seed emergence, plant health, and phenology from the digital images. Despite its potential, the rate of adoption is slow due to limited accessibility, unsuitability to specific issues, unaffordability, and high technical knowledge requirement from the clientele. Therefore, the development of open source image processing applications that are task-specific, easy-to-use, requiring fewer inputs, and rich with features will be beneficial to the users/farmers for adoption. The Fiji software, an open source free image processing ImageJ platform, was used in this application development study. A collection of four different agricultural field applications were selected to address the existing issues and develop image processing tools by applying novel approaches and simple mathematical principles. First, an automated application, using a digital image and “pixel-march” method, performed multiple radial measurements of sunflower floral components. At least 32 measurements for ray florets and eight for the disc were required statistically for accurate dimensions. Second, the color calibration of digital images addressed the light intensity variations of images using standard calibration chart and derived color calibration matrix from selected color patches. Calibration using just three-color patches: red, green, and blue was sufficient to obtain images of uniform intensity. Third, plant stand count and their spatial distribution from UAS images were determined with an accuracy of ≈96 %, through pixel-profile identification method and plant cluster segmentation. Fourth, the soybean phenological stages from the PhenoCam time-lapse imagery were analyzed and they matched with the manual visual observation. The green leaf index produced the minimum variations from its smoothed curve. The time of image capture and PhenoCam distances had significant effects on the vegetation indices analyzed. A simplified approach using kymograph was developed, which was quick and efficient for phenological observations. Based on the study, these tools can be equally applied to other scenarios, or new user-coded, user-friendly, image processing tools can be developed to address specific requirements. In conclusion, these successful results demonstrated the suitability and possibility of task-specific, open source, digital image processing tools development for agricultural field applications. / United States. Agricultural Research Service / National Institute of Food and Agriculture (U.S.)

After many years of successful operation in military domains, Unmanned Aerial Systems (UASs) are generating significant interest amongst civilian operators in sectors such as law enforcement, search and rescue, aerial photography and mapping. To maximise the benefits brought by UASs to sectors such as these, a high level of autonomy is desirable to reduce the need for highly skilled operators. Highly autonomous UASs require a high level of situation awareness in order to make appropriate decisions. This is of particular importance to civilian UASs where transparency and equivalence of operation to current manned aircraft is a requirement, particularly in the terminal area immediately surrounding an airfield. This thesis presents an artificial situation awareness system for an autonomous UAS capable of comprehending both the current continuous and discrete states of traffic vehicles. This estimate forms the basis of the projection element of situation awareness, predicting the future states of traffic. Projection is subject to a large degree of uncertainty in both continuous state variables and in the execution of intent information by the pilot. Both of these sources of uncertainty are captured to fully quantify the future positions of traffic. Based upon the projection of future traffic positions a self separation system is designed which allows an UAS to quantify its separation to traffic vehicles up to some future time and manoeuvre appropriately to minimise the potential for conflict. A high fidelity simulation environment has been developed to test the performance of the artificial situation awareness and self separation system. The system has demonstrated good performance under all situations, with an equivalent level of safety to that of a human pilot.

As part of a collaborative project at a stream segment reach on Proctor Creek in Atlanta, GA, UAV-based SfM photogrammetry was tested as a method for collecting fluvial topographic data relative to traditional USGS total station surveying methods. According to the USGS method, 11 transects were surveyed, and imagery was collected via a UAV to create a SfM DEM. The resulting DEM was incomplete but showed promise for the SfM method. Two additional stream segments were chosen in the Atlanta area, the first along SFPC in DHCL and the second along NFPC near Buford Hwy. For each site 11 transects were surveyed along with submerged GCPs so that the SfM DEMs could be compared to the surveyed data. The BW and BD values were collected from the TS transects and the DEM transects and compared according to the percent difference between the two. For SFPC, the average percent difference values for BW and BD were, respectively, 15.9 and 26.0 with standard deviations of 15.7 and 18.0. For NFPC, the BW and BD average percent difference values were 6.8 and 7.5 with standard deviations of 3.9 and 5.9. The GCPs were also compared for each site using linear regressions. There was no strong correlation for SFPC (R2 = 0.31 and p-value > 0.05), but there was a strong relationship indicated for NFPC (R2 = 0.78 and p-value < 0.05). While the results of this study are variable, the results do indicate promise for future work on this emerging method.

Doctor of Philosophy / Department of Geography / Douglas G. Goodin / Jesse A. Poland / Kevin Price / Wheat breeders are in a race for genetic gain to secure the future nutritional needs of a growing population. Multiple barriers exist in the acceleration of crop improvement. Emerging technologies are reducing these obstacles. Advances in genotyping technologies have significantly decreased the cost of characterizing the genetic make-up of candidate breeding lines. However, this is just part of the equation. Field-based phenotyping informs a breeder’s decision as to which lines move forward in the breeding cycle. This has long been the most expensive and time-consuming, though most critical, aspect of breeding. The grand challenge remains in connecting genetic variants to observed phenotypes followed by predicting phenotypes based on the genetic composition of lines or cultivars. In this context, the current study was undertaken to investigate the utility of UAS in assessment field trials in wheat breeding programs. The major objective was to integrate remotely sensed data with geospatial analysis for high throughput phenotyping of large wheat breeding nurseries. The initial step was to develop and validate a semi-automated high-throughput phenotyping pipeline using a low-cost UAS and NIR camera, image processing, and radiometric calibration to build orthomosaic imagery and 3D models. The relationship between plot-level data (vegetation indices and height) extracted from UAS imagery and manual measurements were examined and found to have a high correlation. Data derived from UAS imagery performed as well as manual measurements while exponentially increasing the amount of data available. The high-resolution, high-temporal HTP data extracted from this pipeline offered the opportunity to develop a within season grain yield prediction model. Due to the variety in genotypes and environmental conditions, breeding trials are inherently spatial in nature and vary non-randomly across the field. This makes geographically weighted regression models a good choice as a geospatial prediction model. Finally, with the addition of georeferenced and spatial data integral in HTP and imagery, we were able to reduce the environmental effect from the data and increase the accuracy of UAS plot-level data. The models developed through this research, when combined with genotyping technologies, increase the volume, accuracy, and reliability of phenotypic data to better inform breeder selections. This increased accuracy with evaluating and predicting grain yield will help breeders to rapidly identify and advance the most promising candidate wheat varieties.

<div> <div> <div> <p>This paper introduced a near real-time acoustic unmanned aerial vehicle detection system with multiple listening nodes using machine learning models. An audio dataset was collected in person by recording the sound of an unmanned aerial vehicle flying around as well as the sound of background noises. After the data collection phase, support vector machines and convolutional neural networks were built with two features, Mel-frequency cepstral coefficients and short-time Fourier transform. Considering the near real-time environment, the features were calculated after cutting the audio stream into chunks of two, one or half seconds. There are four combinations of features and models as well as three versions per combination based on the chunk size, returning twelve models in total. To train support vector machines, the exhaustive search method was used to find the best parameter while convolutional neural networks were built by selecting the parameters manually. Four node configurations were devised to find the best way to place six listening nodes. Twelve models were run for each configuration, generating color maps to show the paths of the unmanned aerial vehicle flying along the nodes. The model of short-time Fourier transform and support vector machines showed the path most clearly with the least false negatives with 2-second chunk size. Among the four configurations, the configuration for experiment 3 showed the best results in terms of the distance of detection results on the color maps. Web-based monitoring dashboards were provided to enable users to monitor detection results. </p> </div> </div> </div>

In the coming 20 years, unmanned aerial data collection will be of great importance to many sectors of civilian life. Of these systems, Personal Remote Sensing (PRS) Small Unmanned Aerial Systems (sUASs), which are designed for scientic data collection, will need special attention due to their low cost and high value for farming, scientic, and search-andrescue uses, among countless others. Cyber-Physical Systems (CPSs: large-scale, pervasive automated systems that tightly couple sensing and actuation through technology and the environment) can use sUASs as sensors and actuators, leading to even greater possibilities for benet from sUASs. However, this nascent robotic technology presents as many problems as possibilities due to the challenges surrounding the abilities of these systems to perform safely and eectively for personal, academic, and business use. For these systems, whose missions are dened by the data they are sent to collect, safe and reliable mission quality is of highest importance. Much like the dawning of civil manned aviation, civilian sUAS ights demand privacy, accountability, and other ethical factors for societal integration, while safety of the civilian National Airspace (NAS) is always of utmost importance. While the growing popularity of this technology will drive a great effort to integrate sUASs into the NAS, the only long-term solution to this integration problem is one of proper architecture. In this research, a set of architectural requirements for this integration is presented: the Architecture for Ethical Aerial Information Sensing or AERIS. AERIS provides a cohesive set of requirements for any architecture or set of architectures designed for safe, ethical, accurate aerial data collection. In addition to an overview and showcase of possibilities for sUAS-enabled CPSs, specific examples of AERIS-compatible sUAS architectures using various aerospace design methods are shown. Technical contributions include specic improvements to sUAS payload architecture and control software, inertial navigation and complementary lters, and online energy and health state estimation for lithium-polymer batteries in sUAS missions. Several existing sUASs are proled for their ability to comply with AERIS, and the possibilities of AERIS data-driven missions overall is addressed.

Une gestion raisonnée et multifonctionnelle des forêts n'est possible qu'avec une description à jour de l'état de la ressource naturelle.Les inventaires forestiers traditionnels, réalisés sur le terrain, sont couteux et ne couvrent qu'un échantillonnage de la surface boisée.L'essor des drones civils pour la cartographie a initié une révolution dans le domaine de la télédétection environnementale.La polyvalence et la diversité des systèmes drones sont une aubaine pour la foresterie de précision.Ceux-ci sont utilisé pour la réalisation de cartographie très fine des habitats naturels avec une résolution temporelle et spatiale sans précédent.Nous explorons les possibilités d’utilisation de mini-drones pour la caractérisation quantitative et qualitative de la ressource forestière.Nous nous intéressons en particulier à l’estimation de la hauteur des arbres et à la caractérisation de la composition spécifique au sein de peuplements forestiers.La hauteur de la canopée est une variable dendrométrique de première importance : elle est un bon indicateur du stade de développement des peuplements et intervient notamment dans les estimations de biomasse ou de niveau de productivité.La composition spécifique est une information essentielle en regard des principales fonctions que remplit la forêt (conservation, production, récréation, etc).Nous avons comparé l'estimation de la hauteur des peuplements à partir de mesures LiDAR et celles obtenues par photogrammétrie.Bien que permettant une mesure de hauteur individuelle avec une incertitude de l'ordre de 1.04 m (RMSE) en feuillus, la photogrammétrie par drone sur des zones forestières est systématiquement moins précise que les mesures par LiDAR (RMSE de 0.83 m).Néanmoins, la grande flexibilité que confère les petits drones permet d'acquérir, au moment propice du stade de végétation, et l'information de relief de la canopée, et l'information spectrale.La période de fin de feuillaison, au début du mois de juin, s'est avéré le moment le plus propice à une discrimination automatique de cinq groupes d'essences feuillues (le chênes pédonculé, les bouleaux, l'érable sycomore, le frêne commun et les peupliers).Une erreur globale de classification des houppiers de 16% est obtenue avec des acquisitions monotemporelles, alors que l'utilisation d'imageries acquises à différentes dates permet encore d'améliorer cette classification.Les contraintes de la législation régissant l'utilisation des aéronefs sans pilote à bord restreignent le champs d'action des drones civils.Ainsi, les opérations avec un drone sont limitées sous un seuil d'altitude et à une distance maximale du télépilote, ce qui ne permet pas une utilisation optimale de cette technologie pour la couverture de grands domaines forestiers (plusieurs milliers d'hectares).C'est pourquoi nous pensons que les drones resterons un outils d'analyse de petites surfaces (dizaines voire centaines d'hectares), plus utiles à des fins de recherches scientifiques qu'à une utilisation en gestion forestière / The recent development of operational small unmanned aerial systems (UASs) opens the door for their extensive use in forest mapping, as both the spatial and temporal resolution of UAS imagery better suit local-scale investigation than traditional remote sensing tools.Along with this rising use of drones, dense three-dimensional reconstruction through the combined use of photogrammetry and textit{Structure from Motion} techniques enables now the fine modelization of the canopy surface relief from a set of overlapping images.Forest management is evolving and has to cope with numerous news demands.A sustainable managemnent practice requires beforehand up-to-date and comprehenvise forest inventory.Traditionnal forest ressources inventories are carried out on the field.They are expensive and focus only on an sample of the forest.Information is delivered at the stand level, and specific measurements for individual tree is missing.The use of mapping drones can potentially changes the story by describing forest ecosystems on a tree-level.This thesis aims at investigating the use of unmanned aerial systems for the characterization of temperate forests (in Wallonia, Belgium).Modelization of the vegetation heigth also is investigated by the combinaison of photogrammetric canopy surface measurements with digital terrain elevation acquired by LiDAR.Eventually, the study of a time series of 20 drone fligths through the growing season enables to determine when is the optimal period for automatic classification of deciduous species.Photogrammetric measurements of individual deciduous tree heigth are always less accurate than high density LiDAR measurements (RMSE of 1.04 m versus 0.83 m for the latter).Nevertheless, the versatility of drones is far higher than LiDAR data, with the possibility of flying at the appropriate time and delivering both spectral and 3D information with a very high resolution.Spetral information is relevant among other for tree species identification.The optimal phenology state for the discrimination of deciduous species was demonstrated to be the end of leaf flush.The intra-species phenology is indeed well synchronized during this time windows ranging from late spring to early summer.A global classification error of 16% is reached by using single date UAS imagery, and multitemporal UAS acquisitions still improve the process of species discrimination.Altough precision forestry can largely benefits from UAS technology, legislation constraints limit the operationnal use of drones.Thus, UAS flights are most of the time restricted under a specific altitude and within a certain distance from the remote pilot.These constraints are sub-optimal for the mapping of forest, which requires beyond line of sigth fligth at relatively high altitude.We thus believe that the drone technology will be more developped for scientific investigations at a local scale (dozens or hundreds of hectares) than for forest inventory of large forest estate (thousands of hectares)

The purpose of the study was to compile a map of Furuvik theme park using UAS-photogrammetry and evaluate the products quality by performing a map control. The map control is carried out with guidelines from new and old HMK-documents and how such an evaluation is carried out in our neighbouring countries. At the same time, a time study was carried out on the project's workflow as a request from the University of Gävle (HiG) for a future Master's degree program in Land Surveying. The flight was carried out with a multicopter from Altigator. Prior to the flights, flight signals were placed and as well as, known points (stompunkter), were measured with SWEPOS network-RTK (real-time kinematic). The flight resulted in 1036 images, which in PhotoScan were joined together by block adjustment and generated an orthophotomosaic and a digital elevation model were generated. In ArcMap, from the orthomosaic, a map was produced, which was then controlled using measured control points. The results in the plan points show that the difference between objects in the produced map and their known coordinates varies radially between 0.0014 m and 0.029 m. The mean deviation is 0.009 m with the standard uncertainty (Sp) 0.014 m and the root mean square (RMS) 0,014 m. All requirements in HMK-Geodatakvalitet (Geodata Quality), HMK-Flygfotografering (Aerial Photography), HMK-Kartografi (Cartography), and similar documents from the Norwegian and Finnish national land survey were fulfilled. The requirements of the newer HMK documents on geodata quality and aerial photography are reasonable while HMK cartography needs updating as the requirements are too low, 0.07 m To control the height model, 18 control profiles were measured in according to the Swedish technical specification SIS-TS 21144: 2016. RMS in height for the entire area was 0.032 m. The duration of the study's implementation was documented to produce a time study that resulted in 374 hours of work during nine weeks. / Syftet med studien var att med hjälp av UAS-fotogrammetri framställa en karta över Furuviks nöjespark och utvärdera produktens kvalitet i form av en kartkontroll. Kartkontrollen genomfördes med riktlinjer från nya och gamla HMK-dokument samt hur en sådan utvärdering utförs i våra grannländer. Samtidigt utfördes en tidsstudie över projektets arbetsgång som ett önskemål från Högskolan i Gävle (HiG) för ett framtida civilingenjörsprogram inom lantmäteriteknik. Flygningen genomfördes med en multikopter från Altigator. Inför flygningarna placerades flygsignaler ut som liksom stompunkter mättes in med SWEPOS nätverks-RTK (real time kinematic). Flygningen resulterade i 1036 bilder som fogades samman i PhotoScan genom blockutjämning och genererade en ortotfotomosaik samt en markmodell. I ArcMap framställdes, ur ortofotomosaiken, en karta som sedan kontrollerades med hjälp av inmätta markpunkter i form av stickprov. Resultatet i plan av stickproven visar att skillnaden mellan objekt i den producerade kartan och motsvarande objekt inmätta i området varierar radiellt mellan 0,0014 m och 0,029 m. Medelavvikelsen radiellt är 0,014 m med standardosäkerheten (Sp) 0,014 m. Samtliga krav i HMK-Geodatakvalitet, HMK-Flygfotografering, HMK-Kartografi samt norska och finska styrdokument uppfylldes. Kraven i de nyare HMK-dokumenten om geodatakvalitet och flygfotografering har följt den tekniska utvecklingen medans HMK-Kartografi behöver uppdateras då kraven är för låga, 0,07 m. För att kontrollera markmodellen mättes 18 kontrollprofiler in i enlighet med den tekniska specifikationen SIS-TS 21144:2016. Standardosäkerheten i höjd för hela området resulterade i 0,032 m. Tidsåtgången för studiens genomförande dokumenterades för att framställa en tidsstudie som resulterade i 374 arbetstimmar under nio veckor.

Unmanned Aerial Systems (UASs) offer a new era of local-scale environmental monitoring where access to invaluable aerial data no longer comes at a substantial cost. This provides the opportunity to vastly expand the ability to detect natural hazards impacts, observe environmental conditions, quantify restoration efforts, track species propagation, monitor land surface changes, cross-validate existing platforms, and identify hazardous situations. While UASs have the potential to accelerate understanding of natural processes, much of the research using UASs has applied current remote sensing image processing techniques without questioning the validity of these in UAS applications. With new scientific tools comes a need to affirm that previous techniques are still valid for the new systems. To this end, the objective of the current study is to provide an assessment regarding the use of current remote sensing image processing techniques in UAS applications. The research reported herein finds that atmospheric effects have a statistically significant impact on low altitude UAS imagery. Correcting for these external factors affecting the imagery was successful using an empirical line calibration (ELC) image correction technique and required little modification for use in a complex UAS application. Finally, it was found that classification performance of UAS imagery was reliant on training sample size more than classification technique, and that training sample size requirements are larger than previous remote sensing studies suggest.