D01
Scenario development
PD Dr. Wolfgang Britz
University of Bonn | +49 228 73-2912 |
Summary
Project D01 aims to provide ex-ante land use and land cover (LULC) scenarios reflecting different socio-economic developments. This scenarios with an annual resolution up to 2050 provide results at continental scale with sub-national detail for the Euro-Cordex domain. To consider interactions at regional and global scale due to market feedbacks, a recursive-dynamic global Computable General Equilibrium (CGE) model is applied for the scenario quantification. This model covers all sectors of the economy with detail relevant for land use and considers important drivers of structural change, such as income and demography dependent changes in consumption including diets. The storylines are developed based on the Shared Socio-Economic Pathways (SSPs) from the Intergovernmental Panel on Climate Change (IPCC).
Graphical summary
Contribution to the CRC
The LULC information provided by project D01 will be down scaled to grid level applying the methodology developed in project B04. The joint aim of both projects is to inform the coupled Earth-System-Model.
Approach
Project D01 uses and further develops CGEBox (https://www.ilr1.uni-bonn.de/en/research/research-groups/economic-modeling-of-agricultural-systems/cgebox), a modular platform for CGE modeling. In addition to updating the database, the land use drivers covered in the model are expanded. Within the framework of the five SSPs, scenario narratives with a focus on LULC are developed and quantified. The LULC results are subject to a comprehensive sensitivity analysis and a socio-economic and ecological welfare analysis.
Main results in 2022
Major drivers of LULC change related to the five SSPs are identified based on literature research. The following graphic shows the selection which is implemented in CGEBox and reflects the available model mechanism and the possibility to arrive at quantified assumptions for each driver. The arrows indicate the strength of an assumption in the respective SSP.
Note: LMH: Different assumptions for low, middle and high-income countries. Low-income countries use the assumptions of SSP3, middle- income countries of SSP2 and high-income countries of SSP1.
The literature research and the selection of the drivers cannot be published as a separate paper and are therefore comprised in the publication of the land use results related later. The work on the database development foreseen in 2023 started earlier, see Britz, W. (2022).
Main results in 2023
Database development
Databases for global CGE analyses start from data from the global trade analysis project (GTAP) center, here from the latest version GTAP v11 Power (https://www.gtap.agecon.purdue.edu/databases/v11/) which provides a snapshot of the global economy for 2017. For the work in D01, the GTAP regions outside the Euro-Cordex are grouped into continental aggregates, while their full resolution is retained within the Cordex, resulting in a database with 47 regions. In addition, the agro-food detail of GTAP is extended by two split mechanisms (Britz 2022). Beyond the work in the cited paper, crops are split into irrigated and rainfed variants. To improve the economic and land use detail for Europe, a database for the so-called NUTS2[1] sub-national regions is developed by combining various economic statistics into a consistent data set of estimated value added by sector and sub-regions. This includes the use of CORINE data to estimate land cover at the level of NUTS2. The process of the database development is graphically summarized in the upper right part of the figure in the next section.
As the GTAP database requires a paid-for license, the data can be shared with license holders, only. However, the underlying data for the splits are publicly available from FAOSTAT (https://svn1.agp.uni-bonn.de/svn/cgebox/data/FAO_2017_forSplits.gdx ). Further, CGEBox including the split utility and the NUTS2 data (https://svn1.agp.uni-bonn.de/svn/cgebox/gams/GTAPNuts2/gdp_NUTS2_V11.gdx) are open-source and can be downloaded from: https://www.ilr1.uni-bonn.de/en/research/research-groups/economic-modeling-of-agricultural-systems/cgebox.
Methodological improvements
Methodologically, CGEBox and its tool for creating recursive-dynamic future baselines (G-RDEM) were further developed in various respects:
- Update to the latest version of the SSP-specific projections for gross domestic product (GDP) and demographics (https://svn1.agp.uni-bonn.de/svn/cgebox/data/ssp_v3_1_CGEBox.gdx).
- Introduction of a more detailed nesting structure of the land use allocation (https://svn1.agp.uni-bonn.de/svn/cgebox/gams/GtapAEZ/gtapAez_model.gms).
- Improvement of the targeting of the FAO-based cropland (https://svn1.agp.uni-bonn.de/svn/cgebox/gams/GTAPRdem/fao_baseline.gms).
- Introduction of a GDP-dependent change in the calorie content of various products.
- Algorithm improvement to solve the model in large-scale recursive analysis.
All code improvements are publicly available on the SVN repository of CGEBox (it can be accessed in read-only mode with the userid cgebox and the password cgebox). These improvements are used by the community of CGEBox users which includes, for instance, a group at the World Bank in Washington. They are used in work involving the PI of D01, for instance, for the macro-economic analysis as part of the Country Climate and Development Report for Romania (https://www.worldbank.org/en/country/romania/publication/country-climate-and-development-report-for-romania) which shows the successful transfer of project outcomes into policy relevant analysis.
Main results in 2024
The identified LULC drivers are mapped into quantified assumptions (https://svn1.agp.uni-bonn.de/svn/cgebox/gams/scen/user_scenarios/ssp.gms). Using the database and improved model, global scenario runs are conducted. The following figures provide an overview of this process and a snapshot of the results.
Figure 1: Overview of modelling process set-up
Figure 2: Change in Agricultural Area in 2050 based on several SSPs
The approach and results will be presented in a paper which is close to submission. The land use projections are planned to be made afterwards available as a data in brief publication. The sensitivity analysis as foreseen in the work plan is conceptualized. The results are compared for all SSPs for runs with all drivers at place as the default against a set-up where one of the drivers is not considered each time, resulting in 19 sensitivity scenarios for each SSP. Technical details on how the model is solved can be in found in Britz W. (2025).
On-going work in 2025
The work on down-scaling the Euro-Cordex results to grid-level using the methodology and software developed in B04 has started by designing the interface in close cooperation with B04. The improved recursive-dynamic module of CGEBox is presented in a one-week interactive block course to PhD students from different European countries in March 2025.
Footnote
[1] NUTS: Nomenclature des Unités territoriales statistiques. For Germany, the 36 NUTS2 regions refer to the so-called “Regierungsbezirke” respectively “Laender” where these are both aNUTS1 and NUTS2 region.
References
Britz, W. (2021). Comparing penalty functions in balancing and dis-aggregating Social Accounting Matrices. Journal of Global Economic Analysis, 6(1), 34-81. https://doi.org/10.21642/JGEA.060102AF
Britz, W. (2022). Disaggregating agro-food sectors in the GTAP database. Journal of Global Economic Analysis, 7(1), 44-75. https://doi.org/10.21642/JGEA.070102AF
Britz W. (2025). How to solve large recursive-dynamic Computable General Equilibrium models, Discussion Paper 2025:01, Institute for Food and Resource Economics, University of Bonn.
