dssatslidesphil-110930024540-phpapp01.pdf

Crop modeling

1. Crop modelling with the DSSAT
September 2011

2. • Use for manipulations and experiments that are impractical,
too expensive, too lengthy or impossible (in real-world social
and economic systems)
• Address dynamic complexity (“emergent properties”) of
systems in a way that reductionist science may not be able to
do
• Identify “best management” strategies (through optimization)
• Study the long-term effects of options (predictions,
projections)
Why model?

3. • Allow the researcher to control environmental and
experimental conditions
• Allow hypothetical and exploratory situations to be
investigated
• Allow insight to be gained into the relative importance of
different system elements
• Assemble and synthesise what is known about particular
processes
Why model? -2-
Nicholson (2008)

4. What can models produce?
Inputs OutputsModel
“Predictions”
• Point prediction: temperature in Kathmandu tomorrow
• Behaviour: trends, patterns in space and time
• Differences: system response with/without an intervention
“Understanding”
• Best bet: optimised performance of the system (N application rate)
• Trade-offs: household income and range condition
• Syntheses: what do we know about these processes, and which are still
black boxes?

5. Floodwater
Reduced
soil layers
Oxidised soil zone
A complicated system …

6. … but it can be modelled to a useful extent
INPUTS
Genotype information
Soil information
Weather information
Management information
…
OUTPUTS
Biomass, yield
Water use
Nitrogen use
Carbon balance
…
CROP MODEL
Based on mechanisms
of plant growth and
development (some
may be represented
empirically)
Use in
some way
Things that apply to
the biophysical
world in general
Things that apply
to one particular
situation (e.g. a
field plot)

7. Simulated and observed biomass accretion (kg DM/ha) for cowpea cultivar
TVU 3046 grown in Griffin, Georgia, in 1998
Hoogenboom et al., 2000
canopy
stem
leaf

8. Observed grain yield (t / ha)
Simulated grain yield (t / ha)
Comparison of observed
and simulated grain
yield for 5 wheat
models
(a) AFRC-WHEAT2
(b) CERES-Wheat
(c) Sirius
(d) SUCROS2
(e) SWHEAT
The solid lines represent the 1:1
relationship
Jamieson et al., 1998

9. defining factors: CO
2
radiation
temperature
crop characteristics
- physiology, phenology
- canopy architecture
limiting factors: water
nutrients
reducing factors: weeds
pests
diseases
pollutants
potential
actual
attainable
1
2
3
Production situation
Production level (t/ha)
Yield-increasing measures
Yield-protecting measures

10. defining factors: CO
2
radiation
temperature
crop characteristics
- physiology, phenology
- canopy architecture
limiting factors: water
nutrients
reducing factors: weeds
pests
diseases
pollutants
potential
actual
attainable
1
2
3
Production situation
Production level (t/ha)
Yield-increasing measures
Yield-protecting measures
“Realism” increases:
but so does complexity

11. Runoff
Evaporation
Deep drainage
Rainfall,
Irrigation
Transpiration
Capillary rise
Bypass flow
Plant water uptake
Crop modelling is 50 years old: some of it is “mature science”
Crop model water balance in a layered soil (from late 1970s): Ritchie’s tipping bucket

12. DSSAT
Decision Support System for
Agrotechnology Transfer

13. DSSAT v2.1 in 1989  DSSAT v4.5 2010
About 2000 users in over 90 countries

14. Components of DSSAT
DSSAT User Interface
Crop Models
MODELS
Weather
Pests
Genetics
Soil
Experiments
DATABASES
Economics
Graphics
Experiments
Soil
Weather
Pests
Genetics
SUPPORT SOFTWARE
Economics
Validation /
Sensitivity
Analysis
Crop Rotation /
Sequence
Analysis
Seasonal Strategy
Analysis
Spatial Analysis /
GIS Linkage
APPLICATIONS

15. DSSAT v4.5
• Windows-based
• Incorporates DSSAT CSM (+ Legacy Models)
• Field scale
• Data management tools
• XBuild: Input crop management information in standard format
• SBuild: Create and edit soil profiles
• GBuild: Display graphs of simulated and observed data, compute
statistics
• ATCreate: Create and edit observations from experiments, formatted
correctly
• WeatherMan: Assist users in cleaning, formating, generating weather
data
• ICSim – Introductory tool to demonstrate potential yield concepts

16. Several different analytical capabilities
• Sensitivity Analysis: vary soil, weather, management or variety
characteristics for insight
• Seasonal Analysis: multiple-year simulations to evaluate uncertainty
in biophysical and economic responses
• Rotation/Sequence Analysis: long-term simulations to analyze
changes in productivity and soil conditions associated with cropping
systems
• Spatial Analysis: define spatially variable soil, weather, management
characteristics across a field or region for analysis
DSSAT v4.5

17. Main window in DSSAT v4.5

18. Selection of maize
experiment, all treatments
selected for simulation.
Circle shows button for
running the model and
for graphing results.

19. DSSAT4.5 graphics screens

20. Assessing Risk and Ways to Reduce it
•Crop simulation models integrate the interaction of
weather, soil, management and genetic factors
•Use the crop simulation models to run “what if”
scenarios
•Develop alternate management practices that will
benefit the farmer
•Risk factors: weather and price uncertainty, two of the
major sources

21. Context
•Next season’s weather is uncertain
•Variability in historical weather data can be assumed to describe
uncertainty in next season’s weather
•“Experiment” is run by specifying a possible management system
over a number of prior years of weather data
•Thus, a distribution of yields (& other outputs) is produced,
converting uncertainty in weather into uncertainty in yield—for the
specific management
•Other management “treatments” can be simulated in the
experiment

22. Using DSSAT to Analyze Uncertainty
lSimulate n years of the management being analyzed,
using historical years of weather data and soil properties
for the site
lEach year starts with the same initial soil conditions
lEach yield value is assumed to have an equal probability
of happening in the future (assuming future weather
statistical properties are the same)
lCreate cumulative probability distribution
lCompute statistical properties (mean, variance, etc.)

23. Annual Yield Variability
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
12345678910111213141516171819202122232425
Years
Yield, t/ha

24. Developing Cumulative Probability Distributions
from Simulated Results
Year Yield (t/ha) prob
1 1.83 0.04
2 2.78 0.04
3 1.9 0.04
4 2.3 0.04
5 4.7 0.04
6 2.4 0.04
7 1.3 0.04
8 4.1 0.04
9 3.5 0.04
10 0.3 0.04
11 2.6 0.04
12 2.05 0.04
13 3.04 0.04
14 3.28 0.04
15 1.69 0.04
16 0.9 0.04
17 1.1 0.04
18 3.24 0.04
19 3.95 0.04
20 4.2 0.04
21 4 0.04
22 2.75 0.04
23 1.75 0.04
24 3.67 0.04
25 3 0.04
Cumulative
Ranked YieldProbability
0.3 0.04
0.9 0.08
1.1 0.12
1.3 0.16
1.69 0.2
1.75 0.24
1.83 0.28
1.9 0.32
2.05 0.36
2.3 0.4
2.4 0.44
2.6 0.48
2.75 0.52
2.78 0.56
3 0.6
3.04 0.64
3.24 0.68
3.28 0.72
3.5 0.76
3.67 0.8
3.95 0.84
4 0.88
4.1 0.92
4.2 0.96
4.7 1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Yield, t/ha
Cumulative Probability
Raw
Yield Data
Ranked
Yield Data
Cumulative
Probability Distribution
Mean=2.65 t/ha
Var=1.31 (t/ha)
2

25. CPFs of monetary return ($/ha) for three treatments

26. DAS CO32- Irrigated conditions
Planting date
Feb-01Feb-15Mar-01Mar-15Apr-01Apr-15
Yield (kg ha
-
1
)
0
2000
4000
6000
8000
DAS CO32- Rainfed conditions
Yield (kg ha
-
1
)
0
2000
4000
6000
8000
Planting date evaluation
Simulated yields for
different planting
dates under rainfed
(top) and irrigated
(bottom) conditions

27. a) AG9010
Forecast date
Mar-01 Apr-01 May-01 Jun-01 Jul-01 Aug-01 Sep-01
Yield (kg ha
-1
)
0
1000
2000
3000
4000
5000
6000
7000
Simulated yield
Observed yield
b) DKB 333B
Forecast date
Mar-01 Apr-01 May-01 Jun-01 Jul-01 Aug-01 Sep-01
Yield (kg ha
-1
)
1000
3000
5000
7000
0
2000
4000
6000
Simulated yield
Observed yield
c) DAS CO32
Forecast date
Mar-01 Apr-01 May-01 Jun-01 Jul-01 Aug-01 Sep-01
Yield (kg ha
-1
)
0
1000
2000
3000
4000
5000
6000
7000
Simulated yield
Observed yield
d) Exceler
Forecast date
Mar-01 Apr-01 May-01 Jun-01 Jul-01 Aug-01 Sep-01
Yield (kg ha
-1
)
0
1000
2000
3000
4000
5000
6000
7000
Simulated yield
Observed yield
Yield forecasting
Average forecasted
yield and standard
deviation for 2002
as a function of the
forecast date and
observed yield (kg/
ha) for four maize
hybrids

28. DSSAT and other crop modelling systems
Used in many different ways around the world:
Crop management
Fertilizer management
Irrigation management
Pest management
Tillage management
Variety evaluation
Precision agriculture
Sustainability studies
Climate change studies
Yield forecasting
Education
…

29. International climate change study: implications
•Crop yields in mid- and high-latitude regions are less adversely
affected than yields in low-latitude regions
•Will simple farm-level adaptations in the temperate regions be
able to offset the detrimental effects of climate change?
•For the tropics, appropriate adaptations need to be developed
and tested further at the household level; the role of genetic
resources and information provision?
•Regional impact analyses: discussion tomorrow

30. • DSSAT training course sponsored by the University of
Florida and ICRISAT, Hyderabad, 5-9 December 2011
(open for applicants)
• Possible: DSSAT training course at CRIDA during the
week of 13-17 February 2012
DSSAT v4.5 training

31. Prediction of milk production from cows consuming
tropical diets
4035302520151050
40
35
30
25
20
15
10
5
0
observed milk production (kg/d)
p
r
e
d
i
c
t
e
d
m
i
l
k
p
r
o
d
u
c
t
i
o
n
(
k
g
/
d
)
grain supplements
pastures
stovers
grass/legume
Herrero (1997)

33. “All models are wrong, but some are useful”
- GEP Box
“… the practical question is, how wrong
do they have to be to not be useful.”

34. A simple interface for running complex crop models: