Name: Modeling the impact of global change on regional agricultural land use through an activity-based non-linear programming approach. Authors: Martin Henseler (Spain), Alexander Wirsig (Germany), Sylvia Herrmann (Germany), Tatjana Krimly (Germany), Stephan Dabbert (Germany). Publication: Agricultural System. Year: 2009. Keywords: Global change, Regional optimization model, Global change scenarios, Agricultural production, Nonlinear programming.

1. 04.07.2012
END 603E- Operations Res. & Mathem. Mod.
Article Presentation
507102103 Özcan ÇAVUŞOĞLU
504112306 Nevin DÖNMEZ 1

2. Article Information
• Name: Modeling the impact of global change on regional
agricultural land use through an activity-based non-linear
programming approach.
04.07.2012
• Authors: Martin Henseler (Spain), Alexander Wirsig (Germany),
Sylvia Herrmann (Germany), Tatjana Krimly (Germany), Stephan
Dabbert (Germany).
• Publication: Agricultural System.
• Year: 2009.
• Keywords: Global change, Regional optimization model, Global
change scenarios, Agricultural production, Nonlinear programming. 2

3. Agenda
• Abstract
• Introduction
04.07.2012
• Literature Review
• Methodology
• Implementation
• Results and Findings
• Future Research, Discussion and
Implementations
3
• Conclusion

4. Abstract
• The impact of climate change will vary strongly across regions depending on
• pre-existing climatic,
• agronomic,
04.07.2012
• and political conditions.
• Most of the present modeling approaches, which aim to analyze the impact of global change on
agriculture, deliver aggregated results both with regard to content and spatial resolution.
• To deliver results with a higher spatial resolution and to produce a more detailed picture of
agricultural production, the county-based agro-economic model known as ACRE-Danube was
developed.
4

5. Abstract
• The German and Austrian part of the Upper Danube basin, a study area with great diversity in
agricultural landscapes and climatic conditions, was chosen for study.
04.07.2012
• For the analysis, two scenarios of climatic and socio-economic change were derived.
• The first and more economically and globally oriented scenario, termed ‘‘Full Liberalization,”
included significant temperature increases.
• The second and more environmentally and regionally oriented ‘‘Full Protection” scenario
included a moderate temperature increase.
• Both scenarios produce different results regarding agricultural income and land use.
5

6. Introduction
• The influence of climate change on agriculture thus represents a new challenge to quantitative
model-based policy analysis.
• With regard to agriculture, ‘‘location and context-specific modeling” (Buysse et al., 2007) is now
04.07.2012
more important than ever, as the ability to adapt to climate change will be strongly linked to
location and region- and farm-specific behavior.
• The aim of this study is to give a more spatially detailed analysis on the impact of global change
scenarios on agricultural land use in order to better inform future policy decisions. For this
reason, in this study developed the county-based ACRE (Agro-eConomic pRoduction model at
rEgional-level) model.
• The investigation area of ACRE-Danube, which is presented in this study, represents the German
and Austrian part of the Upper Danube basin, the research area of the GLOWA-Danube project
(Global Change in the Hydrological Cycle).
6

7. LiteratureApproaches
• Existing Modeling
Review
04.07.2012
7

8. LiteratureApproaches
• Existing Modeling
Review
04.07.2012
8

9. Methodology
The ACRE (Agro-eConomic pRoduction model at rEgional ) Model
• For the Upper Danube basin uses a bottom-up approach to simulate
regional agricultural land use.
04.07.2012
• This model offers a high level of detail, both in terms of spatial
resolution and the number of agricultural production activities
included.
• The main objective of ACRE is to produce a regional analysis of the
impact of global change and agricultural policy measures (e.g.,
quotas, subsidies) on agricultural land use.
• In ACRE, 24 food and non-food crops with different production
intensities per crop as well as 15 production processes for livestock
9
are considered at the county (NUTS3) level.

10. Methodology
The ACRE (Agro-eConomic pRoduction model at rEgional ) Model
1. Process Analytical Approach
04.07.2012
• ACRE is a comparative static optimization model that maximizes total gross
margin at the regional level by calculating the optimal combination of different
production activities for each county. The prices for crops and animal products
influence the total gross margin.
• Production factors within each county are aggregated to create a ‘single farm’
(regional farm approach). The shortest simulation period is one year.
• The model analyzes the most important processes and interactions in
agricultural production. On arable land, cash crops or fodder crops for livestock
production may be produced. The animals produce manure, which is used as
fertilizer in crop production. Mineral fertilizer and feed concentrates are
purchased.
• As the Upper Danube basin has a good climatic water balance and is not
expected to experience severe water problems within the next decade, irrigation 10
is not integrated.

11. Methodology
The ACRE (Agro-eConomic pRoduction model at rEgional ) Model
2. Calibration Method and Model
04.07.2012
• ACRE is based on the calibration method of Positive Mathematical Programming
(PMP).
• A PMP model optimizes agricultural production by maximizing the objective
value of a non-linear total gross margin function (Howitt, 1995).
• In comparison to Linear Programming (LP) models, PMP models have the
following advantages:
• they are calibrated by the reference situation and avoid overspecialization;
• they react continuously to parameter variations and allow a flexible result
calculation;
• they tend to require fewer data.
• These features make PMP models particularly suitable for modeling regional
agricultural production. 11

12. Methodology
The ACRE (Agro-eConomic pRoduction model at rEgional ) Model
04.07.2012
2. Calibration Method and Model
• Generally, a PMP model is built in two steps:
• an LP model representing the observed statistical situation calculates
dual values, which are then used to calibrate the non-linear functions
of the PMP model.
• The system of non-linear functions has its optimum at the point
where the marginal gross margins are equal. Graphically, this is
where the non-linear functions intersect. Thus, the optimum value,
or the maximum objective value, is determined by the non-linear
function parameters (e.g., the slopes of the non-linear functions).
• In other words, the LP model produces shadow prices that are used
to calculate non-linear function parameters. Dual values ensure the 12
replication of production patterns as simulated by the LP model.

13. Methodology
2. Calibration Method and Model
a.Creating LP Model
Objective Function(Total Gross Marginal (TGM))
04.07.2012
13

14. Methodology
2.Calibration Method and Model
1.Creating LP Model
Subject to
04.07.2012
14

15. Methodology
2. Calibration Method and Model
a.Creating LP Model
Subject to
04.07.2012
15

16. Methodology
2.Calibration Method and Model
b.Transforming from LP to NLP
04.07.2012
16

17. Methodology
2. Calibration Method and Model
c.Creating NLP Model
Objective Function:
04.07.2012
Parameters
17

18. Methodology
Model Data and Study Area • ACRE-Danube was calibrated using statistical
production data at the county level for 1995,
which was the base year for the GLOWA-Danube
project.
• Agricultural data (e.g., yields, crop acreages, and
livestock) at the county level are available and
sufficiently precise.
04.07.2012
• The production processes in ACRE are
formulated according to the publications of the
German Association for Technology and
Structures in Agriculture (KTBL, 1995, 1997,
1999).
• These data collections represent an accurate
standardized database for agricultural
production in Germany.
• Soil differences are integrated into the ACRE
model using the Agricultural Comparability
Index(LVZ).
• The Upper Danube basin, which is the research
area of the GLOWA-Danube project,
• covers an area of 77,000 km2
• Approximately 55% of the study area is used for agriculture.
• extend across five countries (Fig. 1).
• The largest portion of land lies in the south 18
• Overall, the study area includes 74 counties (NUTS3-level), of Germany, with an area of 56,000 km2,
with 58 belonging to Germany and 16 to Austria. The followed by that of Austria, with ~20,000
counties belong 12 administrative units based on the km2
NUTS2-level classification.

19. Methodology
Development of Global Change Scenarios
04.07.2012
• The storylines A1 and B2 of the Intergovernmental Panel on
Climate Change (IPCC) Special Report on Emission Scenarios
(SRES) constitute the scenario framework of this study (Nakic
´enovic ´et al., 2000).
19

20. Methodology
Development of Global Change Scenarios
• The scenarios were labeled according to their level of market liberalization and their protection of agricultural
production through public expenditures.
• ‘‘Full Liberalization” is characterized by high technological advances and low public expenditures;
• ‘‘Full Protection” is characterized by low technological advances and high public expenditures
• Table 2 presents the selected percentage yield changes for cereals and grassland as available for the NUTS2
regions of the study area.
04.07.2012
• Table 2 presents the scenario assumptions for
• climate change (i.e., percentage crop yield changes)
• socio-economic change (i.e., crop yield changes due to technological progress, subsidies, and prices).
20

21. Methodology
Development of Global
Change Scenarios
• Table 3, shows that in the
04.07.2012
Full Liberalization scenario,
the subsidies were
cancelled, while in the Full
Protection scenario they
were increased.
• Market prices tended to
decrease in the Full
Liberalization scenario and
to increase in the Full
Protection scenario.
21

22. Methodology
• Reference Scenario:
04.07.2012
• The CAP (Common Agricultural Policy)
• reform 2003 begun in 2005 and projected to end in 2013 is used as a
reference scenario.
• This policy is assumed to remain constant until 2020, without
changes in crop yields.
• Changes in subsidies were calculated based on the single farm
payments (SFP) under the CAP scenario.
22

23. Results and Findings
• In each of the two Global Change scenarios, changes in
agricultural income and land were analyzed in comparison to
04.07.2012
the baseline (CAP) scenario.
23

24. Results and Findings
• While the developments in the Full Protection scenario are small, the Full Liberalization
scenario yields extreme regional changes in agricultural income.
04.07.2012
24

25. Results and Findings
• While the developments in the Full Protection scenario are small, the Full
Liberalization scenario yields an increase in cereal production.
04.07.2012
25

26. Results and Findings
• While the developments in the Full Protection scenario are small, the Full
Liberalization scenario yields extensive grassland farming.
04.07.2012
26

27. Future Research, Discussion and
Implementations
1. Implications of scenario assumptions for results
• The results from the ACRE calculations correspond with recent findings for
04.07.2012
Middle European regions published by several other authors.
• Given these results, terminating subsidies and boosting yields through
technological development may lead to strong reductions in grassland farming.
• The result would be intensification in favorable regions and de-intensification in
marginal regions.
• In contrast, a rise in public spending may ensure the maintenance of the current
landscape, in particular the preservation of cultural landscapes and agricultural
income levels, despite small increases in productivity.
• If recent developments in agricultural output price levels (OECD-FAO, 2008)
persist in the long-term, this would be a counteractive force.
27

28. Future Research, Discussion and
Implementations
2. The use of positive Mathematical Programming Methods
• The use of Positive Mathematical Programming (PMP) allows
04.07.2012
projections based on past observations of the cost function and
reflects real farmer behavior.
• However, PMP models more suitable to modeling medium-term
future scenarios than to making long-term projections, because of
the difficulty in calibrating.
• The most valuable use of PMP is to model modifications of the
existing (calibrated) situation; in such situations, it can give a reliable
projection of the consequences of change.
• The inclusion of new measures or activities is only possible if they
are in line with the existing calibration . 28

29. Future Research, Discussion and
Implementations
3. Implications of the use of a quadratic cost function on the
results
04.07.2012
• The formulation in ACRE produce a quadratic gross marginal
function.
• The quadratic cost function is assumed due to simplicity , the lack
of data and the lack of strong arguments for other types of
functions.
• To examine the impact of using different cost functions, a less
complex model could be developed for future work.
29

30. Future Research, Discussion and
Implementations
4. Additional benefits of ACRE to current modeling of agricultural land
use changes
04.07.2012
• The highly detailed picture of agricultural activities that ACRE provides
may be used as the basis for diversification strategies in rural areas.
• The ACRE model is flexible and allows additional indicators, which might
address the demands of different groups of society, e.g., protection of
cultural landscapes, ecological services and socio-economic benefits, to
be integrated.
• ACRE covers a large part of southern Germany. Therefore, it allows for
coverage of a substantial area of highly diversified regions in Europe in
terms of landscape and farm structure. This makes it particularly
suitable for modeling different scenarios which examine the effects of
funds for societal functions of agriculture (e.g., preservation of extensive
grassland, sustainable resource management). 30

31. Future Research, Discussion and
Implementations
5. Limitations of ACRE and implications for simulation results
04.07.2012
• Limitations of the current model include the exclusion of
trade, the use of exogenous prices, and the lack of common
alternative land use options.
• The lack of common alternative land use options, such as
energy crops or reforestation, hampers the results of land use
changes, in particular for abandoned agricultural land.
• The lack of irrigation in the current model is unlikely to affect
the results significantly 31

32. Future Research, Discussion and
Implementations
6. Further development and potential of ACRE in regional land
use modeling
04.07.2012
• To improve the description of farmer reactions, the coupling
of ACRE with a multi-agent system (MAS) may be useful. The
first steps have been taken in the GLOWA-Danube project to
develop a farming decision-making framework that works on a
1 km2 grid and interacts with other (ecological) models.
• Furthermore, upcoming activities such as energy crops should
be considered in the model
32

33. Conclusion
• Detailed regional information on the consequences of global change
is essential to regional decision-makers (e.g., agri-environmental
programs, less favored areas, and creation of protected areas).
04.07.2012
• ACRE takes economic decision-making into account by using a
normative optimization approach.
• The model describes the actors of regional agricultural land use with
a great amount of detail.
• Its particular strengths consist of a high regional resolution on which
regional decision-makers can be addressed, an increase in reliability
through the exact reproduction of the reference situation (crop and
livestock production activities) and the prevention of jumpy model
33
behavior.

34. ?
?
?
?
04.07.2012
34