1. The Dynamic Aerial Survey
Algorithm
Precision Fertiliser Applications
Falzon GA, Lamb DW & Schneider DA
2. Overview
The Need for Top-dress Fertilisers.
The Advent of Active Optical Sensors.
Real-Time Aerial Crop Survey & its
Challenges.
The Solution: DAS.
Future Developments.
3. Top-Dress Fertilisers
Irrigated durum wheat: 2 applications of
N/season
322 kg N/ha, $A309.84/ha (July 2011).
300 ha cereal feld requires $A92,952
N/application and $A185,904 total.
Extra on-costs! Aircraft cost $A7000-
$A10000/9ight.
We need greater effciency!!!
5. Real-Time Aerial Survey
Fletcher FU24954 aircraft
Garmin GPS18 5Hz
On-The-Go Sensing: 80 kts = 144 km/h
Can we create an accurate prescription map after each
pass?
6. The DAS Algorithm: Pt I
Paddock Polygons
cumulative passes
Ragged Arrays
geo-referenced samples: different dimensions
8. The DAS Algorithm: Pt III
Support Vector Machines
Nk
f ( z)=∑i=1 [β −β K ( z i , z)+β0 ]
1
i
2
i
Kernels
Polynomial Radial Sigmoid
T n
(γ z⋅z +κ) −∣z−z T∣ T
tanh (γ z⋅z +κ)
exp( 2
)
2σ
10. The DAS Algorithm: Pt V
Divide f ( z) into prescription zones
M P =( M 1, P ,… , M N , P )
Pre-set Levels e.g. NDVI = {0.0,0.2,0.4,0.6,0.8}
Divide into N levels e.g. n = 10
K-means clustering, e.g. cluster into three groups (low, med, high).
12. Future Developments
Incorporate Prior Information
Prior survey results
Additional information such as soil surveys
and yield maps
Expert knowledge
13. Future Developments
Joint Air-Ground Operations
Extension of Prior Survey Models.
Ground Validation, Small Scale Areas, Combined
Air/Ground Fertiliser Teams.
Field Calibration & Validations
Wind shear, height, droplet size (I. Yule), servo latency.
Compare drop zone to management map.
14. Thank you!
M. Trotter (UNE PARG)
SUPERAIR
CRC for Spatial Information (CRCSI)
Sugar Research & Development Corporation (SRDC)
Nick Barton (Twynam Agriculture)
Andrew Smart (Precision Cropping Technologies)
Nick Gillingham (Sundown Pastoral Company)
Dr. F. Honey (SpecTerra Services)