Consequences of climate change on South American biomes

GoViLa Modelling Workshop
23.09.2014 Darmstadt
Technical Session
Model coupling within the GoViLa project
Rüdiger Schaldach1, David Laborde2, Florian Wimmer1
1Center
for Environmental Systems Research (CESR), Universität Kassel
Food Policy Research Institute (IFPRI), Washington D.C.
2International
Overview
• Research questions and objectives
• Methodology
− Modelling framework
− MIRAGE-BIOF model
− LANDSHIFT model
− Model coupling
• Scenario analysis
• Summary
Research questions and objectives
• How can countries produce or import the raw materials for biofuel
production without triggering adverse land use changes, leading to a
release of CO2 that would worsen the footprint of biofuels in terms of
climate change.
• How can alternative governance scenarios lead to better or worse
outcomes, and how can policy makers in the EU act to improve the
environment in which the biofuel target will take place?
• Assess (direct and indirect) land use change in the most critical
regions, namely Brazil, Indonesia and Ukraine.
• Provide information that help to identify the room for maneuver
through several scenarios for the mitigation of LUC effects assuming
an increasing demand for biofuels in the EU
Methodology
•
Model-based assessment of land-use change globally and within
the focus countries under the GoViLa governance scenarios.
• Combination of a global economic model (MIRAGE-BIOF) with a
spatially explicit land-use model (LANDSHIFT)
− Linkage of global trade and markets with regional land-use
decisions and spatial details of the biophysical environment .
− Spatial information of land suitability and land-use constraints
provide a more detailed picture of land availability.
− Incorporation of spatially explicit crop yield data into
economic analysis.
− The generated land-use maps will allow more detailed
assessment of CO2 emissions from LUC.
Socio-economy module
Scenarios
MIRAGE-BIOF
Population
Agricultural
production
and trade
State variables
Modelling framework
Land-use change
module
LANDSHIFT
Land-use
activities
Settlement
Biophysical module
Biomass
productivity
Crop yields
Hydrology
Grazing
State variables
Grassland
NPP
Climate scenarios
GAEZ
Crop
cultivation
+
Irrigation
Water
availability
Water stress
Time series
of maps and
statistics
(Schaldach und Koch, 2009)
MIRAGE-BIOF
•
•
•
•
The MIRAGE model has started to be developed in 2001 in CEPII,
Paris. Focusing on EU Integration and Trade Policy analysis of the
beginning
Now used by several institutions around the World, numerous
versions ( trade policy focused, FDI, Services, Climate Change etc.)
Biofuels assessment started in 2008
On land use:
•
•
•
•
First study for the DG Trade in 2009 (limited to ethanol)
Second study for DG Trade in 2010 (part of the public consultation)
2011-2012 study for the EC: Impact Assessment and draft legislation
But other applications: mandates of other countries, comparison of
“traditional” ag policies and biofuels etc., food prices and price
stability consequences
MIRAGE-BIOF: Special features

MIRAGE model
– Multi country, Multi sectoral, and global
– Recursive dynamic set-up

Modified model and data components
–
–
–
–
–
–
–
Improvement in demand system (food and energy)
Improved sector disaggregation
New modeling of ethanol sectors
Co-products of ethanols and vegetable oils
Modeling of fertilizers
Modeling of livestocks (extensification/intensification)
Land market and land extensions at the AEZ level
MIRAGE-BIOF: New developments
•
New data
Higher level of crop disaggregation
Higher level of regional disaggregation
Double cropping
Carbon markets (all sectors, including LULUCF)
•
Explicit FQD and RED modelling
•
•
•
•
Products
Sector
Description
Sector
Description
Sector
Description
Rice
Rice
Permcrops
Permanents crops
EthanolB
Ethanol - Sugar Beet
Wheat
Wheat
Fodder
Fodder crops
EthanolM
Ethanol - Maize
Maize
Maize
SoybnOil
Soy Oil
EthanolW
Ethanol - Wheat
PalmFruit
Palm Fruit
SunOil
Sunflower Oil
Biodiesel
Biodiesel
Rapeseed
Rapeseed
OthFood
Other Food sectors
Manuf
Other Manufacturing
activities
Soybeans
Soybeans
MeatDairy
Meat and Dairy
WoodPape
products
r
Wood and Paper
Sunflower
Sunflower
Sugar
Sugar
Fuel
OthOilSds
Other oilseeds
Forestry
Forestry
PetrNoFuel Petroleum products,
Fuel
except fuel
Vegetable
Vegetable
Fishing
Fishing
Fertiliz
Fertilizers
OthCrop
Other crops
Coal
Coal
ElecGas
Electricity and Gas
Sugar_cb
Sugar beet or cane
Oil
Oil
Constructi
Construction
on
Cattle
Cattle
Gas
Gas
PrivServ
Private services
OthAnim
Other animals (inc.
OthMin
Other minerals
RoadTrans
Road Transportation
Ethanol
Ethanol - Main sector AirSeaTran
hogs and poultry)
PalmOil
Palm Oil
Air & Sea
transportation
RpSdOil
Rapeseed Oil
PubServ
Public services
Illustration Biodiesel sectors
Feedstock
Crops
Veg.Oil
sector
(+meals)
Sunflower
seed
Sunflower
oil
Soybean
Soybean
oil
Rapeseed
Rapeseed
oil
Palm fruit
& Kernel
Palm oil
Biofuel
Biodiesel
Agricultural Production (1 sector)
Land Markets – at the AEZ Level
Wheat
Corn
Oilseeds
CET
Sugar
crops
Substitutable
crops
Vegetables
and fruits
Other
crops
CET
Livestock1
LivestockN
CET
Cropland
Pasture
CET
Managed
forest
Agricultural
land
CET
Unmanaged land
Natural forest - Grasslands
Land extension
Managed land
Technical issue: Land Extension
Crop
Land
price
Cropland
Total land available
for agriculture
Land
Land Extension Allocation: Old Method
Forest
Other
Primary
Argentina
Savannah &
Grassland
0.0%
24.7%
23.3%
Brazil
16.3%
11.2%
48.5%
CAMCarib
30.4%
10.7%
42.9%
Canada
7.8%
42.5%
16.1%
China
2.2%
27.3%
26.0%
CIS
5.6%
33.3%
26.7%
EU27
0.4%
23.5%
30.9%
IndoMalay
51.7%
7.0%
31.0%
LAC
10.8%
14.3%
33.8%
Oceania
0.0%
32.6%
22.5%
RoOECD
0.0%
18.8%
45.8%
RoW
3.7%
36.9%
16.7%
SEasia
20.4%
21.5%
33.8%
SouthAfrica
5.1%
28.4%
22.2%
SouthAsia
0.0%
32.4%
23.9%
SSA
13.0%
16.7%
41.7%
USA
2.5%
21.1%
23.7%
Methodology
•
•
•
Amount of land extension:
“isoelastic” land supply based
on cropland price
Evolution of the elasticity
Where the land is taken:
• Ad Hoc coefficients:
Winrock
• Limitations
• Done at the AEZ level
• RAS procedure to consider
land availability constraint
at the AEZ level
Pag
Yield dynamics in MIRAGE-Biof
•
•
•
An exogenous factor that accounts for technical change (defined in
the baseline(s) and scenario(s));
Economic drivers
• Factors of production (capital, labor) used by unit of land;
• Fertilizer use (amount of fertilizer by ha);
Intrinsic quality of the land by crop  Landshift
LANDSHIFT
Land Simulation to Harmonize and Integrate Freshwater availability
and the Terrestrial environment
•
Spatially explicit approach
•
Multiple spatial scales
•
Integration of socio-economic and environmental aspects
•
Land-use change on the global scale
•
Land-use intensity and competition between activities
•
Spatial resolution of 5 arc minutes (9 km x 9 km at the Equator)
Spatial simulation with LANDSHIFT
MIRAGE-BIOF
Socio-economic drivers
(Population, agricultural
production, governance)
Macro level
(Countries / regions)
t+1
t
GAEZ
Land-use change
Micro level
(5‘ Raster = 9 x 9 km)
LANDSHIFT
Potential crop yields
Environmental data
Crop cultivation activity
Driving factors for quantitative
land-use change:
Driving factors for location of
land-use change:
- Crop production (t)
- Topography
- Yield increases (t)
- Road infrastructure
- Conservation area
Suitability map (t)
Land allocation
„Multi-Objective Land Allocation“ Heuristics
Spatial distribution of crop types
Land-use map (t)
[Fischer et al. 2002]
Suitability assessment
Crop yields (t)
(AEZ)
Feedback to suitability
assessment (t+1)
Suitability assessment
Multi-criteria Analysis (MCA)
suit k = ∑ wi f i ( pi ,k )× ∏ g j (c j ,k )
i =1
j =1

 

m
n
Suitability
factors
Constraints
Factor weights
∑w
i
i
=1
Evaluation
functions
Constraints
f i ( pi ) ∈ [0,1]
g j (c j ) ∈ [0,1]
Evaluation
factors
Crop yields
Terrain slope
…
Constraining
factors
LU-transitions
Conservation
areas
…
Model coupling
Common data (2012)
production, area,
available land
1
Model initialization
MIRAGE-BIOF / LANDSHIFT
MIRAGE-BIOF
update
assumptions on
biophysical yield
change
2
MIRAGE-BIOF
area (A), production (P),
yield change (econ. input)
4
3
Land use map
Carbon storage
LANDSHIFT
area (A*), production (P*),
yield change (biophysical)
No
Test:
A==A*
P==P*
Yes
stop
• Initialization of both models with common land-use data set (1)
• Simulation of scenarios
–
Agricultural production from MIRAGE-BIOF on regional level (2)
–
Calculation of land-use change on raster level (3)
–
Iteration until model results converge (4)
Model initialization
1. Regions and agricultural products in MIRAGE-BIOF
Modelled crops
• Ukraine
fruits and nuts
palm tree
• Brazil 8 sub-regions
• Indonesia, 6 sub-regions
olive tree
other permanent crops
rice
• Other countries and regions
(100+)
corn
• 24 products
− 19 crop types
− Rangeland
− Forest area
− Settlement
− Primary forest & savannah
cotton-soybean
• (Sub-)national statistical data
vegetables
wheat
corn-soybean
other cereals
soybeans
sunflower
rapeseed
other oilseeds
sugar beet
sugar cane
fiber
fodder
Model initialization
2. Global remote sensing data: MODIS – land cover
• 300x300m aggregated to 5 arc-minutes cells
• Map only shows land-cover, not land-use
− no spatial distribution of crops
− no grazing areas
Model initialization
3. Merging of remote sensing data and
census data
−
Spatial distribution of 19 crop types
−
Rangeland and stocking density
MODIS
(GAEZ)
−
Result is a land-use map
Base year land-use map
Harmonized initial conditions for simulation with MIRAGE-BIOF/LANDSHIFT
–
Spatial distribution of crop types and rangeland
–
Agricultural production (and implicitly mean crop yields)
–
Potentially available area for cropland and rangeland
Data exchange during a simulation
Design of the scenario analysis
GoViLa scenarios
− International climate policy
− Regional governance
− European biofuel policy
Translation of
scenario assumptions
MIRAGE-BIOF
LANDSHIFT
− Change of crop production
− Technological change /crop yield increases
− Livestock numbers
− Direct and indirect land-use change
− Maps for Brazil, Indonesia, Ukraine
GIS Analysis and Evaluation
− CO2-Emissions: IPCC Tier 1 approach
− Effectiveness of governance
− Guidelines, room to maneuver