Optimization of spray application on bed-grown vegetables. On-going developments within the OPTIMA project

Abstract

The development of a smart sprayer for carrot is one of the issues of the OPTIMA IPM project (www.optima-h2020.eu). The general design of this sprayer is combining several smart components with optimized application technologies adapted to crops grown in beds like carrot. A Decision Support System (DSS) uses local weather forecasts (resolution of about 1 km²) combined with an epidemiological model for Alternaria in order to predict potential infection periods. Alternaria disease is identified by using an Early Detection System (EDS) where disease spots are identified in images from a multispectral camera and dynamic detection is possible with deep learning. Besides the development and implementation of these smart components, the sprayer itself is also optimized. A first series of experiments was aimed at selecting the most appropriate nozzles and nozzle arrangements. Clusters of four nozzles are defined to spray on a single crop bed. Within a cluster, the distance between nozzles equals the boom height in order to focus sprays on crop beds in a way that distances/heights are adapted to crop development stage. Assuming the broadcast application is deemed to provide an acceptable biological efficacy, a potential dosage reduction of up to 60% can be reached when nozzle distance/height is 0.4 m corresponding to early development stage (Zwertvaegher et al., 2020).
Additional functionalities of the OPTIMA sprayer for carrots are studied in the project. A variable air support mounted on the boom sprayer is specifically designed using an electrical driven fan able to adapt the airflow in order to increase spray deposition and penetration inside the canopy, but also to reduce spray drift. Preliminary investigation with a 2 m wide prototype boom including four nozzles was tested showing contrasted results depending on the spray configuration at early and late development stages of carrots. Field tests will be conducted in 2021 to cross compare the reference situation using FF110 04 nozzles – 12 km h-1 – 158 L ha-1 without any smart devices with the smart sprayer including the optimized set of low drift nozzles and the air support. Deposition and spray drift will be tested according to ISO 25522 and ISO 22866 protocols respectively. PWM nozzle control is implemented on the sprayer in order to allow variable rate applications (VRA) according to a prescription map based on the EDS and DSS. This VRA functionality will be operated in the fields with either synthetic fungicides or BioPPP selected for their efficacy against Alternaria in laboratory conditions. Results of these field investigations will be presented.

Full text

Aspects of Applied Biology 147, 2022
International Advances in Pesticide Application
133
Optimization of spray application on bed-grown vegetables.
On-going developments within the OPTIMA project
By J P DOUZALS1, S FOUNTAS2, L ATHANASAKOS2, N MYLONAS2, A LAMARE1,
I ZWERTVAEGHER3, D NUYTTENS3, P BALSARI4, P MARUCCO4, M GRELLA4
and A CAFFINI5
1INRAE UMR ITAP, Université de Montpellier, France
2AUA, Greece,
3ILVO, Belgium
4DISAFA, Torino University, Italy
5CAFFINA Spa, Italy
Corresponding Author Email: jean-paul.douzals@inrae.fr
Summary
The development of a smart sprayer for carrot is one of the issues of the OPTIMA IPM
project (www.optima-h2020.eu). The general design of this sprayer is combining several
smart components with optimized application technologies adapted to crops grown in beds
like carrot. A Decision Support System (DSS) uses local weather forecasts (resolution of
about 1 km²) combined with an epidemiological model for Alternaria in order to predict
potential infection periods. Alternaria disease is identied by using an Early Detection System
(EDS) where disease spots are identied in images from a multispectral camera and dynamic
detection is possible with deep learning. Besides the development and implementation of
these smart components, the sprayer itself is also optimized. A rst series of experiments
was aimed at selecting the most appropriate nozzles and nozzle arrangements. Clusters of
four nozzles are dened to spray on a single crop bed. Within a cluster, the distance between
nozzles equals the boom height in order to focus sprays on crop beds in a way that distances/
heights are adapted to crop development stage. Assuming the broadcast application is deemed
to provide an acceptable biological ecacy, a potential dosage reduction of up to 60% can
be reached when nozzle distance/height is 0.4 m corresponding to early development stage
(Zwertvaegher et al., 2020).
Additional functionalities of the OPTIMA sprayer for carrots are studied in the project. A
variable air support mounted on the boom sprayer is specically designed using an electrical
driven fan able to adapt the airow in order to increase spray deposition and penetration
inside the canopy, but also to reduce spray drift. Preliminary investigation with a 2 m wide
prototype boom including four nozzles was tested showing contrasted results depending
on the spray conguration at early and late development stages of carrots. Field tests will
be conducted in 2021 to cross compare the reference situation using FF110 04 nozzles
– 12 km h-1 – 158 L ha-1 without any smart devices with the smart sprayer including the
optimized set of low drift nozzles and the air support. Deposition and spray drift will be
tested according to ISO 25522 and ISO 22866 protocols respectively. PWM nozzle control
is implemented on the sprayer in order to allow variable rate applications (VRA) according
to a prescription map based on the EDS and DSS. This VRA functionality will be operated
in the elds with either synthetic fungicides or BioPPP selected for their ecacy against
Alternaria in laboratory conditions. Results of these eld investigations will be presented.
Keywords: Bed-grown carrots, spray optimization, IPM

134
Introduction
Carrot production in Europe represents about 4.7 million tons and about 103 thousand hectares
(Cook, 2020) mostly found in Poland, Germany, France and UK. Carrots represent a typical food
component for humans and especially baby food. Several diseases are aecting carrot crop among
them Alternaria Leaf Blight (A. dauci) a fungus that causes epidemic damages on leaves leading
to a reduction of the crop yield and crop quality. Several fungicide applications can signicantly
reduce the risks (Callens et al., 2005) but raise the issue of Plant Protection Products (PPP) residues
in carrots (Horska et al., 2021). Alternative solutions may also rely on less sensitive cultivars
(Boedo et al., 2008) or the use of bio PPP (Koch et al., 2010). Conventional boom sprayers are
generally used to spray on bed-grown vegetables through a broadcast application. However carrots
are seeded considering three to four rows per bed of about 1.80 m, each row including three or
two sub-rows respectively, the crop occupies a variable surface on beds according to the growth
stage. Broadcast applications involve then substantial losses of product, especially during the early
stages of the crop growth. Two ways are considered to mitigate the impact of PPP, the rst one
focuses on the improvement of the spray application, i.e. to maximize PPP deposits on the target
crop and the other one aims at limiting losses to the environment through ground losses and spray
drift. Giles & Slaughter (1997) developed a precision band spraying system for small plants using
vision sensors and the automatic adjustment of the yaw angle of a at fan nozzle. Modifying the
spray footprint direction involved a better adjustment of the spray to the target width. In this case,
a reduction in spray application rates of 66–80% was operated considering an increase in the target
deposition eciency of 2.5–3.7 times followed by a reduction of non-target deposition on the
ground of 72–90% compared to conventional spraying. In addition, Giles & Slaughter (1997) also
suggested a signicant decrease in airborne spray drift with such precision spraying techniques.
Other research by Holterman et al. (2018) was based on the adaptation of the spray pattern to
dierent bed widths and the ability to apply dierent dose rates depending on the crop canopy
height. However, they did not study the possibility of adjusting nozzle spacing and boom height to
the canopy width depending on crop growth stage. The eect on canopy deposition, ground losses
and spray drift potential, possibly further reducing PPP use and environmental contamination were
the goal of this study
Specic cropping system and necessity to adapt the spray conguration
Another way to optimise spray applications on bed grown crops relies on the adaptation of the
nozzle spacing and nozzle height to the bed width. This strategy theoretically involves target
sprays on the bed and to minimise losses between the beds. Laboratory tests were conducted by
Zwertvaegher et al. (2020) leading to the denition of optimal spray congurations. It is based
on the improvement of spray deposit on crop with subsequent economic benets and as well as
biological ecacy.
These are among the goals of the H2020-project OPTIMA (OPTimised Integrated pest MAnagement
for precise detection and control of plant diseases in perennial crops and open-eld vegetables,
www.optima-h2020.eu). Indeed, a smart sprayer is specically developed for bed-grown carrots.
Materials and Methods
Necessity to develop a fully integrated IPM system from the risk evaluation to the application
The development of a smart sprayer for carrots is one of the issue of the OPTIMA IPM project
(www.optima-h2020.eu). The general design of this sprayer is combining several smart components
with optimized application technologies adapted to crops grown in beds like carrot. A Decision
Support System (DSS) uses local weather forecasts (resolution of about 1 km²) combined with
an epidemiological model for Alternaria in order to predict potential infection periods. Alternaria

135
disease is identied by using an Early Detection System (EDS) where disease spots are identied
in images from a multispectral camera and dynamic detection is possible with deep learning.
These preliminary experiments aims at dening ve spray congurations with dierent practical
objectives:
• Broadcast application using FF spray at constant nozzle distance and nozzle height with
an application rate of 160 L ha-1 (FF)
• Bed adapted conguration with wide angle FF sprays with reduced application rate (80 L
ha-1) (FFred)
• Bed adapted conguration with wide angle drift reducing air inclusion nozzles with an
application rate of 160 L ha-1 (AIw)
• Bed adapted conguration with narrow angle drift reducing air inclusion nozzles with an
application rate of 160 L ha-1 (AIn)
• Bed adapted conguration with drift reducing nozzles combining O-Centre and wide
angle air inclusion nozzles with an application rate of 160 L ha-1 (AIC)
In this case, a bed adapted conguration means that both nozzle distance and nozzle height are
adjusted to focus the spraying on a variable target within the bed according to the crop stage. All
congurations were tested under a pressure of 3 bar, a travel speed of 12 km h-1 and with three
levels of air support (No – Low or High air speed) as possibly delivered by the fan.
Phase 2 : Characterization of spray ecacy
Each spray conguration was tested in semi-eld conditions for validation at INRAE Montpellier
using a prototype boom of 2 m width provided by Cani Spa, Italy. The spray mix composed
of dye tracer with water was distribute to nozzles using air pressurized container with a pressure
controller. The air support was provided by a fan controlled with a two-phase inverter.
Fig 1. Deposition experiments on articial collectors – source INRAE.
Carrots were grown in bins and experiments were achieved either at early development stage
(BBCH 19) or at later stage closer to maturity (BBCH 49). Two dierent varieties (Soprano®
and Maestro®, Vilmorin, France) were compared considering their dierences in terms of leaves
architecture (spread leaves or upright).
Deposition on carrot leaves and ground losses
Spray deposits on the crop leaves or ground losses were evaluated for the dierent spray congurations
using a dye tracer (Brillant SuloFlavine BSF 1g L-1) in solution with water. Plastic collectors (of
7.5cm × 2.5 cm were distributed on carrot foliage and under the foliage for deposition and ground

136
Fig. 2. Carrots grown in bins and articial collectors used for dye collection and quantication – Source
INRAE.
losses evaluation respectively. Six collectors were placed on the foliage per bin (considering six
bins in total) plus six collectors placed under the foliage. After each spray application, exposed
collectors were let to dry and stored before analysis. The quantity of dye tracer on the collector
was determined by spectrouorometry after washing the collector with a known volume of water.
For this purpose, a calibration curve was dened using successive dilutions of spray mix. Results
were expressed as relative deposition in reference to the application rate.
Potential spray drift
Potential spray drift of the dierent spray congurations was evaluated using ISO 22401 (2017)
methodology (Fig. 6). A test bench consisting of 20 dishes spaced 0.5 m were covered while the
boom passed over the test bench. The spray mix consisted of BSF dye tracer at a concentration
of 1 g L-1. After the boom reached a distance of 2 m beyond the test bench, the dishes were
automatically uncovered and droplet remaining in the atmosphere were collected. The advantages
of this methodology were the absence of both a target vegetation and a natural wind. A global
indicator of Potential Spray Drift was calculated with the addition of all deposits normalized by
the application rate.
Fig. 3. Methodology for Potential Spray Drift measurement according to ISO 22401 (2017).
Field experiments
Field experiments were operated in Cestas, South West France, estate Planete Vegetal. Spray
deposition and spray drift tests were conducted in a eld with the comparison of the reference
broadcast application (XR11004/XR11004/XR11004) with the Optima spray conguration
(AIUB8504/AI11004/AIUB8504).
Results and Discussion
In an objective of simplicity, main results were expressed relatively to the reference spray
conguration.
Compared to the reference, droplet size results followed some eects linked to nozzle angle, nozzle
size or nozzle type (Table 1). For the same nozzle size (full application rate), a narrower angle




137
Table 1. Median volume diameter and median volume velocity of a set of nozzles
Spray conguration - 3 bar Volume Median
Diameter (µm)
Volume Median
Velocity (m s-1)
Broadcast - FF Sprays wide angle - 160 L ha-1 Ref Ref
Bed adapted - Narrow angle - FF - 160 L ha-1 +13% +50%
Bed adapted - Narrow angle - FF - 80 L ha-1 -13% -12%
Bed adapted - Drift reducing Narrow/wide angle - 160 L
ha-1 +53% -9%
induced an increase in the volume median diameter. Conversely, a reduction of the nozzle size
(reduced application rate) induced a reduction of the volume median diameter (Nuyttens et al., 2009).
Another eect was visible when comparing the reference FF nozzle with air inclusion nozzles. In
this last case, the volume median diameter increased by 53% compared to the reference.
Droplet velocity was much higher for the reduced application rate nozzle compared to other
nozzles and the reference nozzle. Sensitivity to the air support was also much visible in this case
(results not shown).
Table 2. Lateral Spray Distribution indicators considering a bed width of 1.60 m
Spray conguration CoV (%) % O target losses
Broadcast - FF Sprays wide angle - 160 L ha-1 5.3% 33%
Bed adapted - Narrow angle - FF - 160 L ha-1 8% 12%
Bed adapted - Narrow angle - FF - 80 L ha-1 7% 12%
Bed adapted - Drift reducing Narrow/wide angle - 160 L ha-1 5.9% 9%
The situation of a bed width of 1.60 m corresponded to an intermediate situation between the
extreme values of 1.4 m and 2.2 m. As shown in Table 2, the CoV for all spray congurations was
maintained in a range of 5% to 8% that was complying with the requirement of ISO 5682-2 with
a maximum acceptable CoV under the boom of 9%. Beyond the acceptable evenness of the liquid
distribution over the bed, the main benet of bed adapted spray congurations was the signicant
reduction of the losses on both sides of the bed that can be up to three times less than the reference.
When comparing the results on a wider range of bed widths, it was observed that the smaller the
bed width, the higher the lag between the reference spray conguration and bed adapted ones.
Conversely, increasing the bed width logically led to more homogeneous distribution performance
(lower CoV) for the dierent spray congurations including the reference due to the increase in
the boom height.
Table 3. Relative deposition (late crop stage)
Spray conguration Relative
deposition (%)
Ground
Losses (%)
Broadcast - FF Sprays wide angle - 160 L ha-1 Ref Ref
Bed adapted - Narrow angle - FF - 160 L ha-1 +25% +50%
Bed adapted - Narrow angle - FF - 80 L ha-1 +50% +50%
Bed adapted - Drift reducing Narrow/wide angle - 160 L ha-1 +62.5% +50%
The deposition on the crop leaves shown in Table 3 increased for all bed adapted spray
congurations compared to the reference. Surprisingly, the reduced application rate case gave higher
relative deposition compared to the full application rate situation. Bed adapted spray conguration

138
with air inclusion nozzles gave the highest relative deposition. Ground losses under the leaves
were the lowest for the reference conguration. All spray congurations that were susceptible to
the deposition on the crop also increased the ground losses probably due to the vertical porosity
of the foliage.
Although some specic eect was observed, the global inuence of the air support was not
statistically signicant for any of the bed adapted spray congurations. It is to consider the size of
the prototype boom and the relatively high travel speed.
Table 4. Potential spray drift
Spray conguration % drift reduction
Broadcast - FF Sprays wide angle - 160 L ha-1 0
Bed adapted - Narrow angle - FF - 160 L ha-1 24%
Bed adapted - Narrow angle - FF - 80 L ha-1 -97%
Bed adapted - Drift reducing Narrow/wide angle - 160 L ha-1 70%
As shown in Table 4, bed adapted spray congurations generally reduced spray drift compared to
the reference except for the reduced application case involving a higher drift due to smaller droplets
(cf. Table 1). The more visible eect on drift reduction was logically found for spray congurations
using air inclusion nozzles. The inuence of the air support was probably dierent depending on
the spray conguration while no clear eect of the air support was demonstrated.
Elements for the denition of the specications
The rst objective of the project was to dene optimal spray congurations in order to maximize
the deposition of the spray liquid on the bed following the development of the crop during the
season. The results showed that several spray congurations were found to cope with this objective
assuming a proper adjustment of the nozzle distance and nozzle height. The second objective was
to reduce the environmental impact in terms of spray drift. In this case, droplet size was probably
the most important criteria leading to the exclusive choice of air inclusion nozzles.
However, these experiments did not demonstrate a clear and constant eect of the air support. A
possible explanation lies on the small size of the prototype boom (1/10th of the boom width on the
scale 1 sprayer) where with potentially high boundary eects of the air currents due to the travel
speed compared to the air speed.
Additional specications were considered in order to better adjust the application rate to a variable
disease infestation. For this purpose, a PWM control of individual nozzles was implemented on
the system and was tested with the dierent spray congurations inducing no signicant change
of the droplet size nor of the lateral spray distribution.
Conclusions
This project aims at developing a global solution against Alternaria combining IPM functionalities
with a smart sprayer optimized for bed-grown carrots. The main proposals issued from a grower
focus group were converted into technical specications through laboratory measurements of
nozzle owrate, droplet size and velocity and lateral spray distribution and semi-eld experiments
for deposition and potential spray drift evaluation. The results demonstrated the benet of bed
adapted spray congurations where the nozzle spacing, and nozzle height are adjusted according
to the crop development as expressed in terms of bed width. In this context, the use of air inclusion
nozzles greatly helps in the potential reduction of spray drift. Field tests conrmed the benet of
an OPTIMA spray conguration based on the association of o-centre and air inclusion nozzles
compared to a broadcast application.

139
Acknowledgement
This project has received funding from the European Union’s Horizon 2020 research and innovation
program under grant agreement No 773718 (OPTIMA-project).
References
Boedo C, Le Clerc V, Briard M, Simoneau P, Chevalier M, Georgeault S, Poupard P. 2008.
Impact of carrot resistance on development of the Alternaria leaf blight pathogen (Alternaria dauci).
European Journal of Plant Pathology 121:55–66. DOI: 10.1007/s10658-007-9241-6.
Callens D, Sarrazyn R, Evens W. 2005. Signum, a new fungicide for control of leaf diseases in
outdoor vegetables. Communications in Agricultural and Applied Biological Science 70(3):199–207.
Giles D K, Slaughter D C. 1997. Precision band spraying with machine vision guidance and
adjustable yaw nozzles. Transactions of the ASAE 40(1):29–36. DOI: 10.13031/2013.21240.
Holterman H J, van de Zande J C, van Velde P. 2018. Optimizing sprayer boom design for bed-
grown crops. Aspects of Applied Biology 137, International Advances in Pesticide Application,
pp. 123–130.
Horská T, Kocourek F, Stará J, Holý K, Mráz P, Krátký F, Kocourek V, Hajšlová J. 2020.
Evaluation of pesticide residue dynamics in lettuce, onion, leek, carrot and parsley. Foods 9(5):680.
DOI: 10.3390/foods9050680.
Koch E, Schmitt A, Stephan D, Kromphardt C, Jahn M, Krauthausen H, Forsberg G, Werner,
S, Amein T, Wright S A I, Tinivella F, Gullino M, Robert S J, van der Wolf J, Groot S P C.
2010. Evaluation of non-chemical seed treatment methods for the control of Alternaria dauci and
A. radicina on carrot seeds European Journal of Plant Pathology 127:99–112. DOI: 10.1007/
s10658-009-9575-3.
Zwertvaegher I, Fountas S, Mylonas N, Douzals J P, Balsari P, Marucco P, Cani A, Nuyttens
D. 2020. OPTIMA project: Development of a smart sprayer for bed-grown carrots. Aspects of
Applied Biology 144, International Advances in Pesticide Application, pp. 85–92.

140