A model of spray drift for the 21st Century - Defra

A model of spray drift for the 21 st Century
A model of
spray drift for
the 21 st Century
Clare Butler Ellis
• Why do we need a spray drift model
• How the Silsoe spray drift model has been developed
–The BREAM project
• Potential model uses
–Regulatory exposure assessment
–Drift mitigation

Spray drift matters
• Drifting pesticide is a problem for biodiversity, public health, water,
sensitive neighbouring crops
• By law, pesticide should be confined to the area to be treated
• Regulatory process recognises that zero drift is not feasible and
therefore attempts to quantify and restrict exposure to spray drift
Spray drift highly dependent on
• Environmental conditions
• Application equipment
• Application practice

Ignoring the factors that influence drift
in the risk assessment process:
• Is often over-protective if a genuine worst case is used
• Is sometimes under-protective if not
• Risks exposure assessment becoming out of date if practice
changes
• Does not allow or encourage risk reduction through
appropriate mitigation
Silsoe Spray Drift Model
• Specifically to address the issue of exposure of
bystanders/residents to spray drift from boom sprayers
• BREAM project – 3 ½ years developing and validating a model of
exposure of people to spray and vapour
• BREAM model uses the Silsoe spray drift model, and based around
specific scenarios.

Development of Silsoe spray
drift model
• Original Silsoe spray drift model developed in 1980s and 1990s
–Droplet tracking model
–Ballistic near nozzle; random walk further downwind
• Now updated
–Includes multiple nozzles and a moving sprayer
–User-friendly interface for inputs
• Specific ‘bystander’ scenarios
–12 – 16 kph, 24 m boom
–Bystanders 2 – 10 m from sprayer
Model validation
• New data obtained
• Ensure ‘high drift’ conditions are included
• Commonly-used nozzle designs
• Airborne and ground deposit
• Bystander contamination

Ground deposit, ml/m2
BREAM field measurements
2.5
2
1.5
1
0.5
0
Ground deposit
0 5 10
Distance downwind, m
15 20
Standard flat fan
nozzle
Airborne spray, ml/m2
12
10
8
6
4
2
0
Airborne spray 2 m downwind
Comparison with
experimental data
0 0.5 1
Height above ground, m
1.5 2

Ground deposit, ml/m2
Ground deposit, ml/m2
3.5
3
2.5
2
1.5
1
0.5
0
XR nozzle
1.4
1.2
1
0.8
0.6
0.4
0.2
0
Ground deposit
0 5 10
Distance downwind, m
15 20
Air induction
nozzle
Airborne spray, ml/m2
25
20
15
10
5
Ground deposit
Airborne spray 2 m downwind
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Height above ground, m
0 5 10
Distance downwind, m
15 20
Airbrorne spray, ml/m2
6
5
4
3
2
1
0
Airborne spray at 2 m downwind
0 0.5 1
Height above ground, m
1.5 2

Comparison between predicted and
measured ground deposit – one field trial
Predicted ground deposit,
ml/m2
0.01 0.1 1 10
10
1
0.1
0.01
Measured ground deposit, ml/m2
Comparison between predicted and
measured mean airborne spray up to
2.0 m height, measured 2.0 m downwind
Predicted mean airborne
spray, ml/m2
16
14
12
10
8
6
4
2
0
0 2 4 6 8 10 12 14 16
Measured mean airborne spray, ml/m2

Relationship between airborne spray
and bystander contamination
Bystander contamination, ml
10.00
1.00
0.01 0.10 1.00 10.00 100.00
Modelling variability
0.10
0.01
0.00
Airborne spray integrated up to bystander height, ml/m
• Uncertainty over relationship between airborne spray and bystander
contamination
• Wind turbulence/ variation
– Short timescale, built into model
– Medium timescale (few seconds)
– Long timescale (minutes or more)
• Variation in sprayer operation
– Boom height
– Forward speed – pressure – spray characteristics
• Emulator being created to allow inputs to be selected from a distribution
– Include boom height instability and medium timescale wind speed
variation
– Predicted bystander contamination has a distribution

Example of use for drift onto
surface water
• Currently data from Rautmann et al (2001) are used in regulatory
process
• Not necessarily representative of application practice across
Europe
–0.5 m boom
–10 m boom width
–6 km/h forward speed
• UK conditions now very different
Rautmann et al (2001) ground deposit
data
Ground depost, % applied dose
10.000
1.000
0.100
0.010
0 5 10 15 20 25 30 35 40 45 50
Distance downwind, m
Rautmann mean values - error bars show 95th%ile

Model simulation: “04” nozzle, 6 km/h speed,
0.5 m boom height, 2.0 m/s wind speed
Ground depost, % applied dose
Ground depost, % applied dose
10.000
1.000
0.100
0.010
0.001
0 5 10 15 20 25 30 35 40 45 50
Distance downwind, m
Rautmann mean values - error bars show 95th%ile Model simulation - typical Rautmann conditions
Typical UK scenario: “03” nozzle, 12 km/h
speed, 0.7 m boom height, 2 m/s wind speed
10.000
1.000
0.100
0.010
0.001
0 5 10 15 20 25 30 35 40 45 50
Distance downwind, m
Rautmann mean values - error bars show 95th%ile Model simulation - typical Rautmann conditions
Model simulation - typical UK scenario

Higher drift conditions: “XR 03” nozzle, 16 km/h
speed, 0.9 m boom height, 3 m/s wind speed
Ground depost, % applied dose
10.000
1.000
0.100
0.010
0.001
0 5 10 15 20 25 30 35 40 45 50
Distance downwind, m
Rautmann mean values - error bars show 95th%ile Model simulation - typical Rautmann conditions
Model simulation - typical UK scenario Model simulation - higher drift conditions
Potential uses for a spray drift model:
• Model can be used to generate more appropriate data sets for regulation
– Same model across EU
– Different scenarios for member states
• Model can be used to explore mitigation approaches
– Distance (buffer zones)
– Nozzle type
– Application practice (boom height, forward speed)
• Empirical data relating spray drift to contamination of species will allow
model to be used for all assessments of exposure to spray drift.

Conclusions
• A model is essential for estimating exposures from spray drift
• The Silsoe spray drift model has been validated for UK conditions and can
provide more realistic data than is currently used
• The model can already be used for estimating exposures to bystanders
and to surface water
• Empirical data can be combined with model predictions to give estimates
of all other non-target species’ exposure
• Emulators can be developed to allow variability and uncertainty to be taken
into account giving distributions of outputs
Thanks to the team at Silsoe and
Marc Kennedy, Helen Owen and Richard Glass at
Fera.
Work on the BREAM project was funded by
CRD/Defra