Content uploaded by Anice Cheraiet
Author content
All content in this area was uploaded by Anice Cheraiet on Feb 19, 2021
Content may be subject to copyright.
A preview of the PDF is not available
Experimental and statistical modeling of relations between morphological characteristics of grapevine and spraying deposits : application to precision agriculture.
Abstract and Figures
The social demand for a reduction in the use of plant protection products (PPP) is a driver to reconsider the vine protection process. Today agro-chemical doses are expressed as a fixed quantity per unit area of ground in the field. Dose rates are independent of the conditions of application, and this includes the quantity of foliage to be protected. The agro-environmental performance of pesticides application in viticulture exhibits great variability according to the sprayer type and the growth stage, and training of vegetation.
The aim of this thesis was to develop models for predicting the quantities and distributions of deposits in grapevines according to the canopy structure and sprayer characteristics. The canopy structure, according to its dimensions and density, was characterised on a set of vine blocks using a proximal on-the-go LiDAR sensor (Light Detection And Ranging). Data about PPP distributions in the different strata of the canopy was acquired on the same set of plots.
LiDAR sensors have been shown to be of interest in previous works related to viticulture, but their use is still scarce in commercial perennial crops, primarily due to the lack of robust protocols for processing and interpreting the data. One of the results of the thesis is an automated method for filtering and classifying point clouds from a LiDAR sensor. This makes it easier to estimate vine canopy dimensions (height, thickness) and its apparent density. Importantly, the method requires scanning on only one side (half) of the canopy. The developed method was compared to a non-automated reference method and to classical manual canopy measurements with good agreement. The proposed method was shown to be able to characterise canopy dimensions from LiDAR data in an automated and robust manner throughout the growing season.
A second set of results of the thesis deals with the study at the vine scale of the statistical distribution of leaf deposits intercepted onto leaves and artificial targets within the vine canopy. It is the first such high-resolution study of the variability of within canopy spray depositions. The observed distribution of deposits was shown to be highly variable. It follows therefore, that in order to manage the parts of the canopy that are under treated, and more susceptible to pathogens, using the mean of the deposits, even at plant scale, is insufficient information. Precision spraying should be based on the statistical distribution of deposits, not the mean.
The thesis proposed a novel multivariate statistical modelling approach to predict the distribution of deposition taking into account the sprayer application technology and the structure of the vegetation evaluated by a LiDAR sensor. The new model was compared with existing univariate models and offered better accuracy and robustness in predicting deposition distributions, especially under evolving and varying canopy sizes. The proposed novel modelling approach is general enough to be applied to other spraying technologies.
Finally, a chapter of the thesis is devoted to assess the interest of developing variable rate precision spraying technologies, in terms of PPP savings and their ability to control the phytosanitary risk. The canopy of a vineyard estate was assessed periodically over an entire growing season with a LiDAR sensor. The multivariate statistical models constructed were inverted with the high-resolution canopy information at different dates in order to produce site-specific PPP prescription maps. These prescription maps were interpreted at different decision scales, according to different technological scenarios.
Figures - uploaded by Anice Cheraiet
Author content
All figure content in this area was uploaded by Anice Cheraiet
Content may be subject to copyright.
ResearchGate has not been able to resolve any citations for this publication.
Reducing pesticide use is an important concern for many including the European Commission. One way to achieve this goal is to adjust the amount of pesticides in relation to canopy geometry and foliage. This objective currently poses an important challenge in vineyards with uniform vegetation but it is an added difficulty when the canopy shows spatial variability within the field. Is it possible to set a constant volume rate adjusted to this variability? Or is it more convenient to adjusting different volume rates based on a prescription map? Assuming a plant cell density (PCD) vegetation index from multispectral images to be optimal for detecting variations in vigour, two methods to adjust volume rates in spatially variable vineyards were proposed and tested: i) adjustment of a constant volume rate uniformly applied using a conventional sprayer, and ii) adjustment of two volume rates adapted to two vigour classes according to a prescription map. In both methods, PCD was previously correlated to the leaf area index (LAI), then taking the 70th percentile of LAI to determine adjusted volume rates through DOSA3D decision support system (www.dosa3d.cat/en). Leaf deposition with tracer was analysed to compare the proposed methods with the standard volume rate commonly used in the area. Statistical analysis showed no significant differences between treatments. Since pesticide savings can be achieved using the two methods, specifically 25.6% in adjusted uniform and 25.3% in adjusted map-based treatments, adjusted volume rate strategies can be recommended in vineyards with spatial variability.
A geostatistical methodology is presented to optimise the dosage of plant protection products (PPP) in vineyards with spatial variability. Sprayers are commonly used in viticulture to apply a constant volume rate per unit ground area (l ha−1). This can be a problem in vineyard plots with remarkable spatial variability in vine vigour, being necessary guiding winegrowers through a decision-making tool to determine an appropriate uniform volume rate. The leaf area index (LAI), measured by a terrestrial LiDAR scanner at high spatial resolution along crop rows, can be used to determine the optimum volume application rate. The proposed method is based on obtaining different probability maps of LAI by applying an indicator kriging to the original LAI data. As a result, this method allows winegrowers to i) map and locate areas within the plot that, within a given confidence level (70% or 90%), exceed or do not exceed different values (percentiles) of the original LAI, and ii) set the LAI and the corresponding volume rate seeking, for example, to balance the probability (risk) of areas with lower and higher doses than required. In more conservative protection strategies, the method also allows farmers to set the values of LAI and volume rate that greatly minimise the probability of vulnerable areas being underdosed.
[Ce travail est consultable en ligne : https://pastel.archives-ouvertes.fr/tel-02457148 ]. Résumé : L’utilisation de pesticides permet de réduire les pertes de récolte mais génère des impacts environnementaux négatifs. Il est important de fournir des informations précises sur les risques épidémiques concernant les bioagresseurs afin de raisonner l’utilisation des pesticides, en particulier dans le cas du mildiou de la vigne, responsable en moyenne de 43% des traitements utilisés dans le Bordelais. Cette thèse évalue l’intérêt de la date d’apparition des symptômes de mildiou de la vigne pour raisonner l’usage des fongicides dans la lutte contre cette maladie.En nous basant sur des observations régionales et de l’expertise locale, nous montrons que dans le Bordelais, les premiers traitements sont réalisés en moyenne quatre semaines avant l’apparition des premiers symptômes. Nous montrons que reporter la date du premier traitement anti-mildiou à la date d’apparition de la maladie permet d’économiser en moyenne 56% des traitements, par rapport aux pratiques actuelles de cette région. Nos résultats montrent que combiner cette stratégie avec le port d’équipements de protection réduit l’exposition des opérateurs de plus de 70%.En utilisant des méthodes de machine learning, nous montrons que la précocité et la gravité des épidémies de mildiou sont fortement liées. Les prévisions de nos modèles peuvent être utilisées pour déclencher les traitements contre la maladie dans les cas de risques élevés, entraînant une réduction de plus de 50% des traitements anti-mildiou par rapport aux pratiques actuelles.Nos résultats et les méthodes employées sont discutés et mis en perspective avec d’autres moyens de réduction de l’usage des pesticides en viticulture.
Canopy characteristics are crucial for accurately and safely determining the pesticide quantity and volume of water used for spray applications in vineyards. The inevitably high degree of intraplot variability makes it difficult to develop a global solution for the optimal volume application rate. Here, the design procedure of, and the results obtained from, a variable rate application (VRA) sprayer are presented. Prescription maps were generated after detailed canopy characterization, using a multispectral camera embedded on an unmanned aerial vehicle, throughout the entire growing season in Torrelavit (Barcelona) in four vineyard plots of Chardonnay (2.35 ha), Merlot (2.97 ha), and Cabernet Sauvignonn (4.67 ha). The maps were obtained by merging multispectral images with information provided by DOSAVIÑA®, a decision support system, to determine the optimal volume rate. They were then uploaded to the VRA prototype, obtaining actual variable application maps after the application processes were complete. The prototype had an adequate spray distribution quality, with coverage values in the range of 20–40% and exhibited similar results in terms of biological efficacy on powdery mildew compared to conventional (and constant) application volumes. The VRA results demonstrated an accurate and reasonable pesticide distribution, with potential for reduced disease damage even in cases with reduced amounts of plant protection products and water.
One strategy to reduce usage of crop protection agrochemicals in viticulture is to adjust the spray rate according to local vegetative characteristics. Acquisitions of mobile 2D LiDAR data and deposition rate measurements with a face-to-face sprayer were made on plant blocks in 5 fields and at 4 dates in a Mediterranean vineyard. Two indicators were calculated: the leaf wall area by points and the tree area index. Univariate deposition prediction models using either indicator were estimated from data acquired in 2 fields, and evaluated with regards to data from 3 other fields. In this experiment with a 3m spatial resolution, the model based on LWApts was found more a little more accurate.
The 3D characterisation of individual vine canopies with a LiDAR sensor requires point cloud classification. A Bayesian point cloud classification algorithm (BPCC) is proposed that combines an automatic filtering method (AFM) and a classification method based on clustering to process LiDAR data. Data were collected on several grape varieties with two different modes of training. To evaluate the quality of the BPCC algorithm and its influence on the estimation of canopy parameters (height and width), it was compared to an expert manual method and to an established semi-automatic research method requiring interactive pre-treatment (PROTOLIDAR). The results showed that the AFM filtering was similar to the expert manual method and retained on average 9% more points than the PROTO-LIDAR method over the whole growing season. Estimates of vegetation height and width that were obtained from classification of the AFM-filtered LiDAR data were strongly correlated with estimates made by the PROTOLIDAR method (R 2 ¼ 0.94 and 0.89, respectively). The classification algorithm was most effective if its parameters were permitted to be variable through the season. Optimal values for classification parameters were established for both height and width at different phenological stages. On the whole, the results demonstrated that although the BPCC algorithm operates at a higher level of automation than PROTOLIDAR, the estimates of canopy dimensions in the vineyards were equivalent. BPCC enables the possibility to adjust the spray rate according to local vegetative characteristics in an automated way.
Keywords: Point clouds partition; Automatic filtration; Canopy dimensions; Variable rate; Crop protection
Citrus and olive orchards respectively represent 5% and 29% of the total area of 3D crops grown in the EU (Eurostat, 2018). All of these orchards are located in the Southern Zone for the registration of pesticides (Reg. EC 1107/2009). Every year, the number of
treatments ranges from 4-10 for citrus fruits and 2-6 for olives. For the majority of the PPP currently applied to these crops, the label-recommended dose is expressed as a concentration (%), and only occasionally in relation to the ground surface (kg or L/ha). The
majority of new citrus orchards are managed very intensively, with adult trees forming a near row-continuous canopy which can have a crosswise mid-width of to, or over, 3.0 m. The sprayers work at a constant flow rate. In contrast, olives are traditionally grown as
isolated trees, with canopies up to 4.0 m high and 6.0 m wide (Fig. 1). In this case, the liquid flow between trees tends to be, either manually or automatically, turned off. In compliance with the SUD Directive (2009/128/EC), when applying chemical treatments to 3D crops, experts (Cruz et al. 2017; Miranda et al. 2016; Planas et al., 2015) and authorities recommend adjusting the dose applied in line with crop dimensions, to ensure that a minimal (safe) but sufficient (efficient) quantity of chemicals is sprayed.
The development of reliable fruit detection and localization systems provides an opportunity to improve the crop value and management by limiting fruit spoilage and optimised harvesting practices. Most proposed systems for fruit detection are based on RGB cameras and thus are affected by intrinsic constraints, such as variable lighting conditions. This work presents a new technique that uses a mobile terrestrial laser scanner (MTLS) to detect and localise Fuji apples. An experimental test focused on Fuji apple trees (Malus domestica Borkh. cv. Fuji) was carried out. A 3D point cloud of the scene was generated using an MTLS composed of a Velodyne VLP-16 LiDAR sensor synchronised with an RTK-GNSS satellite navigation receiver. A reflectance analysis of tree elements was performed, obtaining mean apparent reflectance values of 28.9%, 29.1%, and 44.3% for leaves, branches and trunks, and apples, respectively. These results suggest that the apparent reflectance parameter (at 905 nm wavelength) can be useful to detect apples. For that purpose, a four-step fruit detection algorithm was developed. By applying this algorithm, a localization success of 87.5%, an identification success of 82.4%, and an F1-score of 0.858 were obtained in relation to the total amount of fruits. These detection rates are similar to those obtained by RGB-based systems, but with the additional advantages of providing direct 3D fruit location information, which is not affected by sunlight variations. From the experimental results, it can be concluded that LiDAR-based technology and, particularly, its reflectance information, has potential for remote apple detection and 3D location.
DOSAVIÑA is a new tool (website and app for smartphones) developed for calculating the optimal volume rates and pesticide doses to apply during spray application processes in vineyards. DOSAVIÑA also calculates and recommends the optimal working parameters for working pressure, forward speed, and number and types of nozzles. DOSAVIÑA was developed by the Unit of Agricultural Machinery at the Universitat Politècnica de Catalunya, and is available for iOS and Android devices. It is also available on the DOSAVIÑA website (https://dosavina.upc.edu). The developed tool can be used also for the calibration of spray applications on fruit trees (as well as on citrus orchards, olive trees, almond trees, and many other vertical crops) once the volume rate has been established. The system, which is based on a modified version of the leaf wall area (LWA) method, calculates the optimal volume rate for vineyards by considering the effects of leaf density, canopy width, and sprayer type. System testing took biological efficacy into consideration and measured the main factors used for characterizing spray processes, coverage, and distribution over the entire canopy. Results showed that water and pesticide use could be reduced by more than 20% while still meeting economic, environmental, and food quality requirements. The design of the tool is aligned with European requirements concerning pesticide use, as established in the European Directive for a Sustainable Use of Pesticides.