Abstract:Small unoccupied aircraft systems (sUAS) are becoming popular for mapping applications in agriculture, and photogrammetry software is available for developing orthorectified imagery and three-dimensional surface models. Ground control points (GCPs), which are objects or locations with known geographic coordinates, may be required for accurate image georeferencing. However, few studies have compared global position equipment among sUAS or investigated the effects of GCP number or arrangement on georeferencing a… Show more
“…Ground control points (GCPs) were collected using two paired Emlid Reach RS2+ (Emlid Tech Korlátolt, Budapest, Hungary) devices, where one device was placed on a fixed, permanent base location and the other was used as a rover to collect the GPS data for each GCP. Because prior research has demonstrated that four accurate and precise GCPs near the corners of each flight mission are sufficient to georeference photogrammetry projects, that is the number that was used in this study ( Pugh et al., 2021 ).…”
Peanut is a critical food crop worldwide, and the development of high-throughput phenotyping techniques is essential for enhancing the crop’s genetic gain rate. Given the obvious challenges of directly estimating peanut yields through remote sensing, an approach that utilizes above-ground phenotypes to estimate underground yield is necessary. To that end, this study leveraged unmanned aerial vehicles (UAVs) for high-throughput phenotyping of surface traits in peanut. Using a diverse set of peanut germplasm planted in 2021 and 2022, UAV flight missions were repeatedly conducted to capture image data that were used to construct high-resolution multitemporal sigmoidal growth curves based on apparent characteristics, such as canopy cover and canopy height. Latent phenotypes extracted from these growth curves and their first derivatives informed the development of advanced machine learning models, specifically random forest and eXtreme Gradient Boosting (XGBoost), to estimate yield in the peanut plots. The random forest model exhibited exceptional predictive accuracy (R2 = 0.93), while XGBoost was also reasonably effective (R2 = 0.88). When using confusion matrices to evaluate the classification abilities of each model, the two models proved valuable in a breeding pipeline, particularly for filtering out underperforming genotypes. In addition, the random forest model excelled in identifying top-performing material while minimizing Type I and Type II errors. Overall, these findings underscore the potential of machine learning models, especially random forests and XGBoost, in predicting peanut yield and improving the efficiency of peanut breeding programs.
“…Ground control points (GCPs) were collected using two paired Emlid Reach RS2+ (Emlid Tech Korlátolt, Budapest, Hungary) devices, where one device was placed on a fixed, permanent base location and the other was used as a rover to collect the GPS data for each GCP. Because prior research has demonstrated that four accurate and precise GCPs near the corners of each flight mission are sufficient to georeference photogrammetry projects, that is the number that was used in this study ( Pugh et al., 2021 ).…”
Peanut is a critical food crop worldwide, and the development of high-throughput phenotyping techniques is essential for enhancing the crop’s genetic gain rate. Given the obvious challenges of directly estimating peanut yields through remote sensing, an approach that utilizes above-ground phenotypes to estimate underground yield is necessary. To that end, this study leveraged unmanned aerial vehicles (UAVs) for high-throughput phenotyping of surface traits in peanut. Using a diverse set of peanut germplasm planted in 2021 and 2022, UAV flight missions were repeatedly conducted to capture image data that were used to construct high-resolution multitemporal sigmoidal growth curves based on apparent characteristics, such as canopy cover and canopy height. Latent phenotypes extracted from these growth curves and their first derivatives informed the development of advanced machine learning models, specifically random forest and eXtreme Gradient Boosting (XGBoost), to estimate yield in the peanut plots. The random forest model exhibited exceptional predictive accuracy (R2 = 0.93), while XGBoost was also reasonably effective (R2 = 0.88). When using confusion matrices to evaluate the classification abilities of each model, the two models proved valuable in a breeding pipeline, particularly for filtering out underperforming genotypes. In addition, the random forest model excelled in identifying top-performing material while minimizing Type I and Type II errors. Overall, these findings underscore the potential of machine learning models, especially random forests and XGBoost, in predicting peanut yield and improving the efficiency of peanut breeding programs.
“…Despite direct georeferencing being possible with the onboard-RTK, the use of some GCPs is recommended for full accuracy. In the case of small square mapping extents, at least four GCPs are recommended [63]. For larger extents, at least one additional GCP is recommended in the central area [64,65].…”
Peatland restoration aims to achieve pristine water pathway conditions to recover dispersed wetness, water quality, biodiversity and carbon sequestration. Restoration monitoring needs new methods for understanding the spatial effects of restoration in peatlands. We introduce an approach using high-resolution data produced with an unmanned aircraft system (UAS) and supported by the available light detection and ranging (LiDAR) data to reveal the hydrological impacts of elevation changes in peatlands due to restoration. The impacts were assessed by analyzing flow accumulation and the SAGA Wetness Index (SWI). UAS campaigns were implemented at two boreal minerotrophic peatland sites in degraded and restored states. Simultaneously, the control campaigns mapped pristine sites to reveal the method sensitivity of external factors. The results revealed that the data accuracy is sufficient for describing the primary elevation changes caused by excavation. The cell-wise root mean square error in elevation was on average 48 mm when two pristine UAS campaigns were compared with each other, and 98 mm when each UAS campaign was compared with the LiDAR data. Furthermore, spatial patterns of more subtle peat swelling and subsidence were found. The restorations were assessed as successful, as dispersing the flows increased the mean wetness by 2.9–6.9%, while the absolute changes at the pristine sites were 0.4–2.4%. The wetness also became more evenly distributed as the standard deviation decreased by 13–15% (a 3.1–3.6% change for pristine). The total length of the main flow routes increased by 25–37% (a 3.1–8.1% change for pristine), representing the increased dispersion and convolution of flow. The validity of the method was supported by the field-determined soil water content (SWC), which showed a statistically significant correlation (R2 = 0.26–0.42) for the restoration sites but not for the control sites, possibly due to their upslope catchment areas being too small. Despite the uncertainties related to the heterogenic soil properties and complex groundwater interactions, we conclude the method to have potential for estimating changed flow paths and wetness following peatland restoration.
“…More specifically, GCPs are often laid out around the perimeter as most distortion occurs in these areas during model construction (Gabrlik, 2015;Sanz-Ablanedo et al, 2018). It is important to note, that having more GCPs that are evenly dispersed increases horizontal and vertical georeferenced accuracy of the spatial data products (DTM/DSM models, orthomosaic images, or point clouds) (Pugh et al, 2021). For example, in the works of Forlani et al, (2019), implementing just one GCP decreased error from 10 cm to 3 cm.…”
Section: Direct and Indirect Georeferencing In Sfm Photogrammetrymentioning
Remotely piloted aircraft systems are increasingly used as a remote sensing platform for peatland researchers to monitor changes in vegetation height/composition/structure. In this work, an RPAS was flown to collect aerial images over Alfred Bog, a domed peatland complex located in Eastern Ontario. The images were processed with a photogrammetry technique referred to as Structurefrom-Motion (SfM) which can be used to create 3D point clouds of a x by y transect. The point cloud results were used to assess the utility of extracting ground terrain and vegetation height as compared to a transect field survey. This field survey was completed with a Trimble Catalyst RTK GNSS to record ground elevation and maximum vegetation height of the canopy top. The results from this research suggest that terrain information could not be extracted at all from the generated point clouds. Although, a digital surface model can be generated to model the canopy top and crown area.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.