1. Agricultural modelling and assessments in a changing climate
Olivier Crespo
Climate System Analysis Group
University of Cape Town

2. Keep in mind that crop models are
Partial : simplified representation of a system
Biased : a specific perspective on the system
Mostly mechanistic (describe the processes)
Mostly dynamic (across time)
Mostly deterministic (no randomness)

3. Inputs and Outputs of a model
Uncontrollable variables
Environment definition
Weather
Biophysical conditions
Outcome
Decision rules
Crop model
Crop response
Resources consumed
Controllable variables
Calendar applied
Decision thresholds
Limitations

4. A crop model
Air
Plant
Biophysical
model
Model the decision making process of crop actions : sowing, irrigation, fertilisation, harvest …
Decision model
Soil

5. The biophysical part of the model
A biophysical model describes the chemical and biological subsystems of the crop model.
It usually includes :
a soil model : water fluxes within soil layers, from soil to plant roots
an air model : wind, transpiration, evapotranspiration
a plant model : the plant growth according both to soil and air interactions

6. The decisional part of the model
A decisional model describes the decision making process.
It usually consists in :
a sequence/loop
of decision rules
if condition then action
where
condition: “variable (operator) threshold”
action: application details

7. Example of decision rule
Sowing decision
condition: Within D1 weeks surrounding my usual planting date, if D2 mm of rain falls within a week and D3 mm of rain falls in the 2 following weeks,
then action: plant with D4 density, D5 deep, etc..
You have control
the rule structure and the rule variables Dx

8. Biophysical conditions
Inputs and outputs
Weather
Biophysical conditions
Decision rules
Crop response
Resources consumed
Decision thresholds
Calendar applied
Limitations

9. Environmental conditions: Controllable variables:
More about the inputs
Environmental conditions:
soil composition, water limitations
Controllable variables:
biophysical (crop, cultivar), decision (rules, condition threshold), action (application details)
Uncontrollable variables:
mostly the weather affecting the crop (temperatures, rainfall, solar radiation) but also soil inconsistency in the field, pest/disease spatialisation, ground level and natural pools

10. Crop Consumption Calendar biomass, yield quantity, quality, N residue
More about the outputs
Crop
biomass, yield quantity, quality, N residue
Consumption
what sowing density, what amount of irrigation water, of fertiliser
Calendar
when was the crop sown, what was the irrigation schedule, fertilisation

11. Crop models Pros and Cons to keep in mind
Advantages :
Predictions based on physiological principles
valid for different conditions
Complementary to field experiments
number of conditions, possible corrections
More predictive indicators
Weaknesses :
Complex (to understand and to use)
Based on current understanding (limited)

12. Useful for operational decisions
At a few days time scale, it impact the execution of a decision:
Calculate non measured quantities
e.g. soil water
Predict decision efficiency
e.g. washed fertiliser
Test alternative applications
e.g. irrigation amount

13. Useful for tactical decisions
At a few months time scale, it impact the procedure decisions:
Adapt the calendar
e.g. regarding weather forecasts
Predict the outcome
e.g. yield quantity and quality
Test alternative decisions
e.g. alternative crop, irrigation schedule

14. Useful for strategic decisions
At a few years time scale, it impacts policy decisions:
Predict the outcome over years
e.g. crop suitability in a region
Rotation management
e.g. soil composition over the years
Regulation change assessments
e.g. water demand, pesticide use

15. The strategic time scale is particularly relevant for CC
Crop impact assessment
e.g. permanent yield reduction
Resources availability
e.g. water competition
Adaptation alternatives
e.g. alternative crops, relocation
Vulnerability
Copping potential

16. A model can be simulated
which makes its prediction ability
a useful tool for :
Exploitation:
Improving current systems
Optimising the outcomes
Exploration:
Assessing innovative systems
Assessing uncontrollable variable impacts

18. Example of soil data Multiple layers describing Structure and texture
Fine, coarse, sand, silt, clay
Soil water description
Lower limit, drained upper limit
Soil carbon, nitrogen ratio
Soil pH

19. Example of plant data
e.g. APSIM big advantage is to provide lots of plant modules, so that you probably will find what you need
Plant Available Water Capacity (PAWC)
Water uptake limits

20. Example of air data Mainly your weather data set.
Depending on the time step of the model
Hourly, daily, monthly
The usual suspects
Min and max temperatures, rainfall, solar radiation/ETo
Sometimes
Wind, relative humidity, ...

21. Example of rules block Loop Sequential
More open to innovating managements
While (1 month around my usual planting date)
Test irrigation rule
Test fertilisation rule
Sequential
Control of operation number
20 < DAS < 30 : test irrigation rule
30 < DAS < 40 : test fertilisation rule
...

22. Back to the simplified crop model
Weather
Biophysical conditions
Decision rules
Crop model
Crop response
Resources consumed
Decision thresholds
Calendar applied
Limitations

23. Useful tool for agriculture and CC
historical
future
Weather
Biophysical conditions
Agricultural system
Crop response
Resources status
Actions
Adapting to expectations
Coping with impacts
current
innovative

24. Some crop models Water-balance models Plant-based models
AgroMetShell
Aquacrop
Plant-based models
CERES
STICS
Soil-based models (ease modularity)
APSIM
DSSAT (?)

25. In any case simulated / potential Best possible in the model
achievable
Best possible in the field,
Field constraints, limitations ...
actual
But there is also weeds, pests, diseases ...
Example : Yield

26. Time step : Actions Variable state Interactions
hourly, twice a day, daily, weekly, monthly ..
Biophysical
model
Actions
Variable state
Decision model