Using smart phones to improve agronomic decision making in

Using smart phones to improve agronomic decision making in potato crops
Dr. Robert Allen
Talking Data South West Conference
20th June 2013
It’s been a collaborative effort…
Agronomy – The science of soil management and crop production
http://faostat.fao.org
“Understanding how potatoes grow
determines how to grow potatoes”
(Allen & Scott, 2001)
Potato Crop – Process of yield formation
Incident radiation
1. Groundcover
Absorbed radiation
2. Radiation use efficiency (RUE)
Total DM yield
3. Partitioning of DM
Haulm DM yield
Tuber DM yield
4. Tuber DM concentration
Tuber FW yield
5. Number of tubers and size distribution
Marketable yield
Water balance in crop – soil system
Weather
meteorological data
Incident radiation
max-min temperature
max-min humidity
wind run
Reference ET0
Crop
Groundcover
Root length
Soil Water Balance
Water Inputs
Rainfall
Irrigation
(Capillary Rise)
(Run-on)
Potential ETp
Soil moisture deficit
Water Outputs
Drainage
(Run-off)
Actual water usage
by crop ETact
Allison, Cambridge University Farm, pers. comm.
Potato crop groundcover
The percentage of ground covered by green leaf
22 % GC
47 % GC
99 % GC
Plant growth, water demand and yield
Practical considerations
Time
• Data needs to be collected weekly
• Decisions made in near real time
Geography
• Large scale operations (10,000’s acres)
• Fields distributed over large areas
Man power
• Operations are lean, system must be quick and easy to use (both data
collection and interpretation)
• Data often collected by field assistants and not agronomists
• Decision makers are often remote from the field and each other
RDO Farms example
Time
• Data needs to be collected weekly
• Decisions made in near real time
Geography
• Large scale operations (10,000’s acres)
• Fields spatially distributed over large areas
Man power
• Operations are lean, system must be quick and easy to use (both data
collection and interpretation)
• Data often collected by field assistants and not agronomists
• Decision makers are often remote from the field and each other
Measuring potato groundcover – manually
Measuring potato groundcover – remote sensing
Measuring potato groundcover – smart phone
Measuring potato groundcover – smart phone
Comparison of manual and iPhone groundcover measurements
100
Estimate of ground cover (%)
Manual Grid
iPhone
80
60
40
20
0
1 May
31 May
30 Jun
30 Jul
29 Aug
28 Sep
Comparison of iPhone and grid groundcovers
Ground cover by IPhone (%)
100
80
60
40
1 : 1 Relationship
20
Y = -0.0037x2 + 1.347x; R2 = 95.8
0
0
20
40
60
Ground cover by grid (%)
80
100
CanopyCheck Design – key functionality
The App
1. Fast and easy to use
2. Capture reliable and
accurate images
3. Collect field and crop
information
4. Record location and date
The Website
1. Real time analysis of
images
2. Easy interpretation of
results
3. Data security & access
4. Linked to NIAB
CanopyCheck – app functionality
CanopyCheck - Website
Processor
Grower Group 1
Grower Group 2
Grower Group 3
Grower Group 4
Grower Group 5
Grower Group 6
Grower Group 7
Grower Group 8
CanopyCheck - Website
CanopyCheck – Design Challenges
• Data security and access
• Squashing different user requirements into single data model
• Architectural convenience versus usability
•
Fully relational data model
•
Easier to develop
•
Restricts usability in the field
• Reliability of data capture and transfer
• Working in areas of limited of no network coverage
• Engineering robust data transfer from app to database
In conclusion
• Groundcover is crucial to good agronomic management of potato crops
• Collecting reliable data isn’t easy:
• Large scale operations with many locations
• Cost of data collection (human and financial)
• Distribution of data is important
• Decision makers often (very) remote from the field
• Real time results required
• Smart phone & digital technology offer a good hybrid solution
• Users will be in the field anyway – exploit them
• Rapid results provided by automatic upload & analysis
• All users referencing the same data point