Folie 1 - Teagasc

Download Report

Transcript Folie 1 - Teagasc

UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Modelling climate change impacts on
short term agricultural management
decisions
Jun.-Prof. Dr. Joachim Aurbacher
MBA Phillip S. Parker
M.Sc. Germán Calberto Sanchez
Prof. Dr. Stephan Dabbert
[email protected]
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Overall Project Targets
Interdisciplinary project to
 improve the understanding of processes and structures
of climate change on the regional scale
 combine basic research with measurements and
advanced modelling
 improve the modelling ensemble of meteorology, crop
and soil science and economics
 close the feedback cycle among atmosphere, soil,
vegetation and land use
 funded by DFG (German research council) under
FOR 1695 (see: klimawandel.uni-hohenheim.de)
2
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Targets of the economic subproject
Improve the understanding and the modelling of farmer
reactions to climate change
 Will there be changes in crop shares and profitability due
to climate change?
 Will traditional cropping schemes and processes change
with climate change?
 How quick will farmers adapt to a changing climate given
certain behavioural characteristics?
 How do farmers consider risk related to climate change?
3
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Model characteristics
Model “FarmActor” ready to work with unique characteristics
 Daily time steps
 Spatially explicit on field level
 Coupled to Expert-N crop growth model
 Management in interaction with soil, weather and plant
growth
 Feedbacks between crop growth and farm management
 Learning from previously observed weather and from
observed crop yields
4
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Modelling Procedure
Main modules
of present model
“FarmActor”
Weather
Weather
Year
Learning
(after
harvest)
Initialisation
(1st year)
daily time steps
Crop
Management
Phase
Spring
daily time steps
Expert-N
Weather
Weather
Crop
Management
Phase
Autumn
Expert-N
Distribution
Annual of Crops on
Planning
Fields
(August)
5
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Interaction between FarmActor and
Expert-N
1 Look up weather for
the present model day
2 Look up plant status
for the present (= end
of previous) day
3 Calculate applicability
of actions
4 Execute actions
5 Look up weather
(Expert-N)
6 Simulate plant growth
with respect to
executed action
Weather
1
5
2
FarmActor
3
4
Expert-N
6
6
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Is it the right time for
sowing winter wheat?
Requirements:
· soil preparation has to
be performed
Is the field prepared?
Yes
Requirements:
According to experience
from previous years:
·from today to the end of
the year there is an
expected amount of
cumulative temperature
No
No
Yes
No
Do not
sow
today
Requirements:
· each crop must be
planted once
Has the crop already been
sown?
Yes
No
These include:
 time periods
 stage of crop growth
 air and soil temperature
(immediate and several
days)
 precipitation
 soil moisture
 completion of other actions
Does the actor want to
grow winter wheat?
Yes
All actions are controlled by
specific criteria (“triggers”)
Requirements:
· a certain crop is next
in the field’s rotation
Requirements:
· no rain
· soil not too wet or dry
· soil not too cold
Is the current weather
amenable?
No
Yes
Decision Tree to
model the daily
Management Actions
Decision tree: Sowing of Winter Wheat
Seed today
7
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Adaptation to observed results („Learning“)
Model farms observe the outcome of production and adapt
their expectations for the subsequent years:
 Yield expectations are adapted to observed yields
 Fertilization is adapted to yield expectations
 Sowing periods are adapted to observed weather: For
spring crops a soil temperature threshold is relevant; for
winter crops the remaining growing degree days
 The timing of the other activities is less dependent on
time periods and thus adapted endogenously
 Several learning methods and rates are possible
8
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Model Calibration and Data
The model has been calibrated to a project area in the
Swabian Jura (southern Germany)
Triggers have been set up according to agronomic
literature and further calibrated to best fit the observed
sowing and harvesting dates in the region
A compromise was chosen between reproducing means
and achieving correlation
Climate data was taken from the German Weather Service
(DWD) and from climate scenario models (WETTREG)
9
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Results: Seeding and
harvesting dates of Silage Maize
modelled weather (WETTREG)
day of year
observed weather (DWD)
10
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Yield of Silage Maize
11
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Seeding and Harvest of Winter Wheat
observed weather (DWD)
modelled weather (WETTREG)
12
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Yield of Winter Wheat
observed weather (DWD)
modelled weather (WETTREG)
13
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Correspondence between Simulation and
Observation
Silage Maize - Years 1980-2010
average seeding day
modelled
observed
122.1
121.6
correlation coefficient of seeding days
average harvesting day
0.47
259.1
correlation coeff. of harvesting days
average yield
0.45
457 dt/ha
correlation coefficient of yields
Winter Wheat - Years 1980-2010
average seeding day
correlation coefficient of seeding days
average harvesting day
correlation coeff. of harvesting days
average yield
correlation coefficient of yields
265.0
475 dt/ha
0.36
modelled
observed
261.6
270.4
0.11 (last 10 years: 0.44)
228.5
224.9
0.57 (last 10 years: 0.64)
84.9 dt/ha
66.9 dt/ha
-0.08 (last 10 years: 0.42)
14
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Conclusions
 The model can reproduce sowing and harvesting dates
 Accuracy for summer crops is better than for winter
crops
 The model reacts to climate variations and can thus
provide insight into future trends
 There is still room for improvement: systematic
relaxation of criteria, inclusion of risk consideration,
parameterisation of changing genetics and technology…
15
UNIVERSITÄT HOHENHEIM
Inst. für Landwirtschaftliche Betriebslehre
Prof. Dr. Dabbert
Thank you for your attention!
16