Measuring Canopy Structure and Condition Using Multi-Spectral UAS Imagery in a Horticultural Environment
Tree condition, pruning and orchard management practices within intensive horticultural tree crop systems can be determined via measurements of tree structure. Multi-spectral imagery acquired from an unmanned aerial system (UAS) has been demonstrated as an accurate and efficient platform for measuring various tree structural attributes, but research in complex horticultural environments has been limited. This research established a methodology for accurately estimating tree crown height, extent, plant projective cover (PPC) and condition of avocado tree crops, from a UAS platform. Individual tree crowns were delineated using object-based image analysis. In comparison to field measured canopy heights, an image-derived canopy height model provided a coefficient of determination (R2) of 0.65 and relative root mean squared error of 6%. Tree crown length perpendicular to the hedgerow was accurately mapped. PPC was measured using spectral and textural image information and produced an R2 value of 0.62 against field data. A random forest classifier was applied to assign tree condition into four categories in accordance with industry standards, producing out-of-bag accuracies >96%. Our results demonstrate the potential of UAS-based mapping for the provision of information to support the horticulture industry and facilitate orchard-based assessment and management.
A Comparison of Analytical Approaches for the Spectral Discrimination and Characterisation of Mite Infestations on Banana Plants
This research investigates the capability of field-based spectroscopy (350–2500 nm) for discriminating banana plants (Cavendish subgroup Williams) infested with spider mites from those unaffected. Spider mites are considered a major threat to agricultural production, as they occur on over 1000 plant species, including banana plant varieties. Plants were grown under a controlled glasshouse environment to remove any influence other than the imposed treatment (presence or absence of spider mites). The spectroradiometer measurements were undertaken with a leaf clip over three infestation events. From the resultant spectral data, various classification models were evaluated including partial least squares discriminant analysis (PLSDA), K-nearest neighbour, support vector machines and back propagation neural network. Wavelengths found to have a significant response to the presence of spider mites were extracted using competitive adaptive reweighted sampling (CARS), sub-window permutation analysis (SPA) and random frog (RF) and benchmarked using the classification models. CARS and SPA provided high detection success (86% prediction accuracy), with the wavelengths found to be significant corresponding with the red edge and near-infrared portions of the spectrum. As there is limited access to operational commercial hyperspectral imaging and additional complexity, a multispectral camera (Sequoia) was assessed for detecting spider mite impacts on banana plants. Simulated multispectral bands were able to provide a high level of detection accuracy (prediction accuracy of 82%) based on a PLSDA model, with the near-infrared band being most important, followed by the red edge, green and red bands. Multispectral vegetation indices were trialled using a simple threshold-based classification method using the green normalised difference vegetation index (GNDVI), which achieved 82% accuracy. This investigation determined that remote sensing approaches can provide an accurate method of detecting mite infestations, with multispectral sensors having the potential to provide a more commercially accessible means of detecting outbreaks.
Assessing Radiometric Correction Approaches for Multi-Spectral UAS Imagery for Horticultural Applications
Multi-spectral imagery captured from unmanned aerial systems (UAS) is becoming increasingly popular for the improved monitoring and managing of various horticultural crops. However, for UAS-based data to be used as an industry standard for assessing tree structure and condition as well as production parameters, it is imperative that the appropriate data collection and pre-processing protocols are established to enable multi-temporal comparison. There are several UAS-based radiometric correction methods commonly used for precision agricultural purposes. However, their relative accuracies have not been assessed for data acquired in complex horticultural environments. This study assessed the variations in estimated surface reflectance values of different radiometric corrections applied to multi-spectral UAS imagery acquired in both avocado and banana orchards. We found that inaccurate calibration panel measurements, inaccurate signal-to-reflectance conversion, and high variation in geometry between illumination, surface, and sensor viewing produced significant radiometric variations in at-surface reflectance estimates. Potential solutions to address these limitations included appropriate panel deployment, site-specific sensor calibration, and appropriate bidirectional reflectance distribution function (BRDF) correction. Future UAS-based horticultural crop monitoring can benefit from the proposed solutions to radiometric corrections to ensure they are using comparable image-based maps of multi-temporal biophysical properties.
Optimising drone flight planning for measuring horticultural tree crop structure
In recent times, multi-spectral drone imagery has proved to be a useful tool for measuring tree crop canopy structure. In this context, establishing the most appropriate flight planning variable settings is an essential consideration due to their controls on the quality of the imagery and derived maps of tree and crop biophysical properties. During flight planning, variables including flight altitude, image overlap, flying direction, flying speed and solar elevation, require careful consideration in order to produce the most suitable drone imagery. Previous studies have assessed the influence of individual variables on image quality, but the interaction of multiple variables has yet to be examined. This study assesses the influence of several flight variables on measures of data quality in each processing step, i.e. photo alignment, point cloud densification, 3D model building, and ortho-mosaicking. The analysis produced a drone flight planning and image processing workflow that delivers accurate measurements of tree crops, including the tie point quality, densified point cloud density, and the measurement accuracy of height and plant projective cover derived from individual trees within a commercial avocado orchard. Results showed that flying along the hedgerow, at high solar elevation and with low image pitch angles improved the data quality. Optimal flying speed needs to be set to achieve the required forward overlap. The impacts of each image acquisition variable are discussed in detail and protocols for flight planning optimisation for three scenarios with different drone settings are suggested. Establishing protocols that deliver optimal image acquisitions for the collection of drone data over horticultural tree crops, will create greater confidence in the accuracy of subsequent algorithms and resultant maps of biophysical properties.
Suitability of Airborne and Terrestrial Laser Scanning for Mapping Tree Crop Structural Metrics for Improved Orchard Management
Airborne Laser Scanning (ALS) and Terrestrial Laser Scanning (TLS) systems are useful tools for deriving horticultural tree structure estimates. However, there are limited studies to guide growers and agronomists on different applications of the two technologies for horticultural tree crops, despite the importance of measuring tree structure for pruning practices, yield forecasting, tree condition assessment, irrigation and fertilization optimization. Here, we evaluated ALS data against near coincident TLS data in avocado, macadamia and mango orchards to demonstrate and assess their accuracies and potential application for mapping crown area, fractional cover, maximum crown height, and crown volume. ALS and TLS measurements were similar for crown area, fractional cover and maximum crown height (coefficient of determination (R2 ) ≥ 0.94, relative root mean square error (rRMSE) ≤ 4.47%). Due to the limited ability of ALS data to measure lower branches and within crown structure, crown volume estimates from ALS and TLS data were less correlated (R 2 = 0.81, rRMSE = 42.66%) with the ALS data found to consistently underestimate crown volume. To illustrate the effects of different spatial resolution, capacity and coverage of ALS and TLS data, we also calculated leaf area, leaf area density and vertical leaf area profile from the TLS data, while canopy height, tree row dimensions and tree counts) at the orchard level were calculated from ALS data. Our results showed that ALS data have the ability to accurately measure horticultural crown structural parameters, which mainly rely on top of crown information, and measurements of hedgerow width, length and tree counts at the orchard scale is also achievable. While the use of TLS data to map crown structure can only cover a limited number of trees, the assessment of all crown strata is achievable, allowing measurements of crown volume, leaf area density and vertical leaf area profile to be derived for individual trees. This study provides information for growers and horticultural industries on the capacities and achievable mapping accuracies of standard ALS data for calculating crown structural attributes of horticultural tree crops.
Assessing radiometric corrections for UAS multi-spectral imagery in horticultural environments
UAS-based multi-spectral imagery is becoming ubiquitous for monitoring and managing various horticultural crops. To accurately measure and monitor their structure and condition and estimate yields, appropriately corrected data must be used to drive the necessary algorithms. There are several popular radiometric correction methods commonly used for UAS-based data correction. However, their relative and absolute accuracies are not known. This study used three flight datasets, including along- and across-tree-row flight patterns in an avocado orchard. Four correction methods were applied to produce at-surface reflectance image mosaics for each flight pattern as well as the grid pattern and the results were compared to assess the reflectance consistency. Results show that no method provided consistently correct at-surface reflectance for the same features. A BRDF correction workflow was being developed to address these limitations. Preliminary application of the BRDF correction shows that it significantly improves the brightness consistency of features across different images.
Detection of Banana Plants Using Multi-Temporal Multispectral UAV Imagery
Unoccupied aerial vehicles (UAVs) have become increasingly commonplace in aiding planning and management decisions in agricultural and horticultural crop production. The ability of UAV-based sensing technologies to provide high spatial (<1 m) and temporal (on-demand) resolution data facilitates monitoring of individual plants over time and can provide essential information about health, yield, and growth in a timely and quantifiable manner. Such applications would be beneficial for cropped banana plants due to their distinctive growth characteristics. Limited studies have employed UAV data for mapping banana crops and to our knowledge only one other investigation features multi-temporal detection of banana crowns. The purpose of this study was to determine the suitability of multiple-date UAV-captured multi-spectral data for the automated detection of individual plants using convolutional neural network (CNN), template matching (TM), and local maximum filter (LMF) methods in a geographic object-based image analysis (GEOBIA) software framework coupled with basic classification refinement. The results indicate that CNN returns the highest plant detection accuracies, with the developed rule set and model providing greater transferability between dates (F-score ranging between 0.93 and 0.85) than TM (0.86-0.74) and LMF (0.86-0.73) approaches. The findings provide a foundation for UAV-based individual banana plant counting and crop monitoring, which may be used for precision agricultural applications to monitor health, estimate yield, and to inform on fertilizer, pesticide, and other input requirements for optimized farm management.
Remote Sensing applications for Banana Crops - Dataset
Bananas are considered vital for economic development and food security in many countries. Whilst the application of remote sensing for the improved management of some abiotic and biotic constraints and for production forecasting has been investigated, there still remains a substantial knowledge gap. This study addressed this gap by developing and testing the following remote sensing applications i) to establish a methodology for the accurate detection and delineation of individual banana crowns from unoccupied aerial vehicle (UAV) imagery to support the monitoring of individual plants within mixed age, asynchronous commercial banana plantations; ii) to derive a new time series approach for differentiating and quantifying key phenology growth stages, plant morphology and physiology in commercial banana plantations; iii) to assess the accuracies of hyperspectral and multispectral remote sensing for measuring the presence and severity of pest mite infestations on banana plants. Datasets include UAV Parrot Sequoia Multispectral Camera captured over 22 flight campaigns (i and ii) and Hyperspectral datasets used for mite monitoring (iii).
Mapping the condition of macadamia tree crops using multi-spectral UAV and WorldView-3 imagery
Australia is one of the world's largest producers of macadamia nuts. As macadamia trees can take up to 15 years to mature and produce maximum yield, it is important to optimize tree condition. Field based assessment of macadamia tree condition is time-consuming and often inconsistent. Using remotely sensed imagery may allow for faster, more extensive, and more consistent assessment of macadamia tree condition. To identify individual macadamia tree crowns, high spatial resolution imagery is required. Hence, the objective of this work was to develop and test an approach to map the condition of individual macadamia tree crowns using both multi-spectral Unmanned Aerial Vehicle (UAV) and WorldView-3 imagery for different macadamia varieties and three different sites located near Bundaberg, Australia. A random forest classifier, based on all available spectral bands and selected vegetation indices was used to predict five condition categories, ranging from excellent (category 1) to poor (category 5). Various combinations of the developed models were tested between the three sites and over time. The results showed that the multi-spectral WorldView-3 imagery produced the lowest out of bag (OOB) classification errors in most cases. However, for both the UAV and the WorldView-3 imagery, more than 98.5% of predicted macadamia condition categories were either correctly mapped or offset by a single category out of the five condition categories (excellent, good, moderate, fair and poor) for trees of the same variety and at one point in time. Multi-temporally, the WorldView-3 imagery performed better than the UAV data for predicting the condition of the same macadamia tree variety. Applying a model from one site to another site with the same macadamia tree variety produced OOB classification between 31.20 and 42.74%, but with >98.63% of trees predicted within a single condition category. Importantly, models trained based on one type of macadamia tree variety could not be successfully applied to a site with another variety. The developed classification models may be used as a decision and management support tool for the macadamia industry to inform management practices and improve on-demand irrigation, fertilization, and pest inspection at the individual tree level.
Remote Sensing Applications for Banana Crops
Bananas are the fourth most important staple food source globally and are considered vital for economic development and food security in many countries. Current management of commercial banana crops is largely based on in-field visual appraisal and manual record keeping, with targeted agronomic activities guided by the manual tagging of individual plants in the field. Such activities can be labour-intensive, subjective and lacking rigour as they often rely on the experience of the individual undertaking the assessment. Remote sensing technologies play a fundamental, enabling role in precision agriculture and are becoming increasingly commonplace. Applications such as the monitoring of crop phenology to guide management activities, determining harvest readiness, pest and disease detection and yield forecasting using remote sensing have been adopted by other industries, but for banana these tasks are still currently undertaken manually. Little to no adoption of remote sensing applications exist in the banana industry and research into new applications is minimal. The low level of adoption is largely due to the unique phenology, morphology, propagation, and growing properties of banana plants that limit the use of whole-field remote sensing applications common in other crops. To address these knowledge gaps, this thesis developed methods and investigated the accuracies of ground and unoccupied aerial vehicle (UAV) based sensors for measuring several key needs of the Australian banana industry.
Robust spatio-temporal detection and delineation methods were developed and assessed for their ability to accurately represent individual banana plant crowns from UAV multispectral imagery. Furthering this concept of individual plant monitoring, a time series based on a 15-month UAV flight campaign was used to create and compare spectral and morphological data of individual plants over time, from initial establishment to harvest. Verification against infield measurements determined that UAV-based multi-temporal crop monitoring models of individual banana plants can be used for the determination of key phenological growth stages of banana plants (including establishment, flower emergence and harvest) and offer pre-harvest yield forecasts. Finally, the accuracies of both hyperspectral and multispectral data for measuring mite infestations on banana plants were investigated, with both sensors providing promising results. Overall, this research's findings and developed methods contribute important information that enhances crop knowledge and understanding. The methods presented have the potential to add novel precision agriculture applications to the banana industry that compensate for the unique growth and propagation of banana crops. These outcomes have the potential to improve and promote economic advancement and food security in the banana industry.