INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Marco de Estrategias de Desarrollo Bajo en Emisiones: Resultados finales de escenariode referencia.LEDS: Low Emissions Development Strategies
Dr. Alex De Pinto - Senior Research Fellow
Proyecto de USAID
LEDS: Estrategias de desarrollo económico con bajas emisiones
Herramientas de evaluación de LEDS en el sector de uso de la tierra
Colombia, Vietnam, Bangladesh, Zambia
Colaboración entre IFPRI, CIAT, y sociosnacionales
Marco de Estrategias de DesarrolloBajo en Emisiones
La idea es presentar a los paises un portafolio de tecnologias que ayuden al desarrollo economico, con diferentesemisiones causadas.
Simulación a largo plazo del uso de tierra,
emisiones y secuestro de carbono, debidas a la implementación de politicasque afectan uso del suelo.
Incluye efecto de economía y mercado global
Capaz del modelo en resumen
Some Lessons Learned
To a very complex problem follows a complex analysis, therefore:
Modeling framework and tools should be adaptable
Modeling framework and tools should be flexible so that new information can be incorporated in the analysis as it becomes available
Modeling framework and tools should be fully transparent, so that trust in the results can be built
Parameter estimates for
determinants of land use change
Change in carbon stock and GHG emissions
Policy scenario:Ex. land use allocation targets, infrastructure, adoption of low-emission agronomic practices
Land use change
Future commodity prices and
rate of growth of crop areas
IMPACT modelMacroeconomic scenario: Ex. GDP and population growth
Model of Land Use Choices
Model of Land Use Choices
Ancillary data:Ex. Soil type, climate, road network, slope, population, local ag. statistics
Satellite data
General Circulation Model
Climate scenario:Ex. Precipitation and temperature
Change in carbon stock and GHG
emissions. Economic trade-offs
Land use change
Baseline
Policy Simulation
Crop Model
Crop Model
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
The Theory Behind our Land Use Modeling Approach
Land use:
rice
Land use:
vegetables
Farm price of rice
Farm price of
vegetables
The von
Thünen ModelA Featureless Plain
A Parcel of Land…
Is operated by individual or group to maximize benefits deriving from using the land (utility)
Has an exogenous set of geophysical and socioeconomic characteristics that influence production choices and productivity
Has a set of possible production choices (relates inputs to possible outputs) that determine land use
Benefits at One Location from several possible uses
𝑙– location of a parcel h – possible use at location 𝑙 𝑿𝒍 is vector of factors that affect the stream of
benefits deriving from farming
X= Elevation, slope, soil pH, precipitation, temperature, crop suitability, crop price
f – technology function that relates inputs to each possible output h
𝐵𝑙,ℎ = 𝑓 𝑋𝑙,ℎ
Farmer at One Location Makes a Choice
Farmer chooses the use that returns the highest benefit B
𝐵𝑙,𝑀𝑎𝑖𝑧𝑒 = 𝑓 𝑋𝑙,𝑀𝑎𝑖𝑧𝑒
𝐵𝑙,𝑅𝑖𝑐𝑒 = 𝑓 𝑋𝑙,𝑅𝑖𝑐𝑒
𝐵𝑙,𝐶𝑎𝑠𝑠𝑎𝑣𝑎 = 𝑓 𝑋𝑙,𝐶𝑜𝑓𝑓𝑒𝑒.
.
.
𝐵𝑙,𝐹𝑜𝑟𝑒𝑠𝑡 = 𝑓 𝑋𝑙,𝐹𝑜𝑟𝑒𝑠𝑡
Choose Land Use to Maximize Net Present Value
Net Present
Value (NPV)
Value of
Output
Cost of
Inputs
Discount
effect
One optimization for each possible land use h at location l
at time T
Choose land use (categorical variable) with highest benefit
minusequals
𝐵𝑙ℎ,𝑇 =
𝑡=0
∞
𝑃𝑙ℎ𝑇_𝑡𝑄𝑙ℎ𝑇+𝑡 − 𝐶𝑙ℎ𝑇+𝑡𝑋𝑙ℎ𝑇+𝑡 𝑒𝑖,𝑡𝑑𝑡
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
The Data
Satellite Images as a source of data
An incredible amount of data
Land-use Choices Are Observed
Data from satellite images are processed into land-use maps.
We observe the choices made by whomever has control over the land.
Many Sources of Data
But also statistical data on hectares
allocated to rice, maize, cassava, etc, at
the province or municipio level.
Method of estimation: discrete choice models, e.g. multinomial logit, nested logit, etc.
For each land use we estimate the probability for that use to be chosen
We statistically evaluate the effect of each explanatory variable
.....3
210
lmaizelmaize
lmaizemaizelmaizemaizejlmaize
price
soilslopeB
Prob. Forest
Prob. Agriculture
Prob. Pasture
Model Specification:Two-level Nested Logit
Cocoa
Coffee
Palm
Plantain
Other Perennials
Pasture ForestPerennial
Crops
Annual
Crops
Forest Other
Uses
Cassava
Maize
Potato
Rice
Sugarcane
Other Annuals
Land-use
choice
Crop
choice
Model Specification:Two-level Nested Logit
Prob. Forest
Prob. Agriculture
Prob. Pasture
Prob. MaizeProb. Rice
Prob. Sugarcane
Land-use
choice
Crop
choice
The estimated probabilities are used to
allocate total area changes predicted by
the IMPACT model
Colombia’s Unit of Analysis: Municipio
Variables dependientes de uso de la tierra
Uso de tierra
año Nivel Fuente
Area cultivo
2008 Municipio Ministerio de
Agricultura y
Desarrollo Rural 2008
Area pasto 2007 Municipio IDEAM, IGAC, IAvH, Invemar, I. Sinchi e IIAP
Area bosque
2007 Municipio IDEAM, IGAC, IAvH, Invemar, I. Sinchi e IIAP
Explanatory Variables for Colombia
Explanatory variables
Lower level:
Choice variable: crop shares within
provinces (2008 and 2030)
Crop area, crop suitability, commodity producer price,
elevation, terrain slope, soil pH, annual precipitation,
annual mean temperature
Upper level:
Choice variable: land cover (2008 and
2030)
Land cover, elevation, terrain slope, soil pH, annual
precipitation, annual mean temperature, population
density, travel time to major cities, conserved areas,
indigenous reserves
Variables independientes, capa de usodel suelo (2008)
Variables año Nivel Fuente
Suitabilidad para cultivo 2009 10kmsquare
IIASA/FAO
Altitud 2012 1km square World HarmonizedSoil DB
Gradiente 2012 1km square World HarmonizedSoil DB
PH de suelo 2012 10km square
ISRIC-WISE
Precipitación anual 1950-2000 1km square WorldClim
Temparetura promedioanual
1950-2000 1km square WorldClim
Population density 2000 CIESIN
Tiempo de viaje a ciudades grandes
-2000 1km square JRC-IES-LRM
Precio de carne 2007 nacionál FAO
Valores inclusivos paracultivo
Derived from lower model
Variables independentes, capa de usodel suelo (2008)
Variables año Nivel Fuente
Valores inclusivos paracultivo
Derived from lower model
Inertia variable
Parques nacionáles 2012 250m square
RUNAP / SINAP
Areas conservaciónesregionales
2012 250m square
RUNAP / SINAP
Reservas Naturales de la Sociedad Civil
2012 250m square
RUNAP / SINAP
Reservas forestales 2011 250m square
RUNAP / SINAP
Reserva Indigena 2012 250m square
SIGOT
Areas Afrodescendentes
2012 250m square
SIGOT
Assumptions, Problems, and Shortcomings Stationary state and dynamic processes
• Processes of land use change are inherently dynamic
Property rights• Competitive bidding process breaks down when property rights are
poorly defined or inexistent.
Spatial effects and interdependent behavior• Possible interactions among several decision makers and eco-
biological processes can transcend parcel boundaries
Nonlinearity in the objective function • Problem specifications force a linear relationship between dependent
and independent variables.
Profit/Utility-maximizing operator• Self-sufficiency or risk minimization might be the goal in land use
decisions
Assumptions, Problems, and Shortcomings
Data on carbon stock and GHG emissions that are specific for Colombia
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Thank you
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
RESULTS
Enfoque tecnico
Modelo de uso del suelo espacialmenteexplicito
IMPACT: modelo global de equilibrio partial para el sector de agropecuario
DNDC: Modelo cultivo que simula cosecha, emisiones y cambio de almacenamiento de carbono
Combina y reconcilia los modelos
Parameter estimates for
determinants of land use change
Change in carbon stock and GHG emissions
Policy scenario:Ex. land use allocation targets, infrastructure, adoption of low-emission agronomic practices
Land use change
Future commodity prices and
rate of growth of crop areas
IMPACT modelMacroeconomic scenario: Ex. GDP and population growth
Model of Land Use Choices
Model of Land Use Choices
Ancillary data:Ex. Soil type, climate, road network, slope, population, local ag. statistics
Satellite data
General Circulation Model
Climate scenario:Ex. Precipitation and temperature
Change in carbon stock and GHG
emissions. Economic trade-offs
Land use change
Baseline
Policy Simulation
Crop Model
Crop Model
Land Use Model: Assessment of Predicted Land Use Choices
Table 3c. Summary Statistics of Municipal-level Predicted Percent Errors
Crop Mean Min Q1 Median Q3 Max
Perennial crop (N=927)
Cacao 2% -78% 0% 2% 7% 52%
Coffee 3% -74% -11% 1% 12% 90%
Palm 9% -80% 0% 1% 12% 88%
Plantain -4% -94% -16% 5% 15% 47%
Other crops -10% -99% -12% 1% 5% 47%
Annual crop (N=1080)
Cassava -4% -91% -9% 0% 3% 27%
Maize -4% -85% -23% 0% 13% 68%
Potato 2% -90% 0% 0% 2% 74%
Rice 7% -78% 1% 5% 14% 94%
Sugarcane 4% -96% -2% 2% 15% 88%
Other crops -4% -94% -6% 1% 5% 44%
Land (N=1121)
Perennial cropland 0% -59% -1% 2% 4% 21%
Annual cropland 0% -57% -1% 1% 4% 15%
Pasture 0% -69% -13% 2% 13% 60%
Forests 0% -62% -5% 3% 6% 53%
Other lands 0% -70% -13% 1% 14% 56%
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Results: The Baseline
Baseline, IMPACT projectionsCambio de area cultivo 2008 - 2030 Crops Projected
Change in
Price
Area
2030
(1000
Ha)
Cambio
2008 -
2030
(1000 Ha)
Tasa de
Cambio
(Area)
CACAO 25% 188 -3 -2%
COFFEE 30% 837 0 0%
PALM 92% 430 72 20%
PLANTAIN 36% 542 62 13%
OTHR_PERENNIAL 28% 184 6 4%
CASSAVA 39% 251 6 3%
MAIZE 38% 807 -3 0%
POTATO 26% 206 11 6%
RICE 25% 691 -8 -1%
SUGAR CANE 110% 445 106 31%
OTHR_ANNUAL 25% 158 -3 -2%
Total 4737 247
Escenario de referenciaCambio de uso de la tierra 2008 - 2030
Land use
category
2008
area
(Mha)
2030
area
(Mha)
Change
2008–
2030
(Mha)
Change
(%)
Perennial crops 1.9 2.1 0.1 7
Annual crops 2.4 2.5 0.1 5
Pasture 29.2 31.8 2.6 9
Forest 58.7 55.3 -3.4 -6
Other land
uses
22.1 22.7 0.5 2
Total 114.4 114.4
Projected changes in forest and pasture areas at municipality level, 2008–2030
Projected changes in sugarcane and palm area at municipality level, 2008–2030
Projected Changes in Carbon Stock
Land Use
Category
Soil
Organic
Carbon
2008
(Tg C)
Above
and
below
ground
Biomass
2008
(Tg C)
2008
Total
Carbon
stock
(Tg C)
Soil
Organic
Carbon
2030
(Tg C)
Above
and
below
ground
Biomass
2030
(Tg C)
2030
Total
Carbon
stock
(Tg C)
Net
Change
in
Carbon
Stock
(Tg C)
Cropland 530 52 583 557 54 612 29
Pasture 3242 246 3,488 3,481 272 3,753 266
Forest 6,133 5,919 12,052 5,690 5,588 11,279 -774
Other
Land
Uses
2,649 744 3,393 2,742 766 3,508 114
Total 12,554 6,962 19,516 12,470 6,682 19,152 -364
Crops Change in area
2008–2030
(1,000 ha)
Average
per ha GHG
emission in
2008
(Mg ha-1yr-1)
Average
per ha GHG
emission in
2030
(Mg ha-1yr-1)
Change in total
GHG emission
2008–2030
(Tg CO2eq)**
Pasture 2,598 1.7 1.8 85.1Perennial cropCacao -3 0.1* 0.1* 0Coffee 0 0.7* 0.7* 0
Palm 71 4.0 3.8 2.1Plantain 57 3.5 3.0 -0.5Sugarcane 107 1.7* 1.7* 1.8Other perennial 6 - - -
Annual crop
Cassava 6 2.0 1.9 -0.1
Maize -2 1.7 1.6 -0.8
Potato 11 3.5 3.5 0.3
Rice -7 7.2 6.9 -2.5
Other annual -3 - -Total 243 - 85.4
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Policy Simulations
Policy Scenarios
Land use policy scenario from identifies after
consultation with stakeholdersScenario 1 Reduction of pastureland by 10 million hectares
Scenario 3 Total halt to deforestation in the Amazon
Scenario 4 Total land allocated to palm production reaches a total
of 1.3 million hectares
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Scenario 1: Reduction of pastureland by 10 million hectares
Pasture reduction Zero deforestation Palm expansion
Crop Area 2030
Difference
from baseline
2030 Area 2030
Difference
from baseline
2030 Area 2030
Difference
from baseline
2030
Perennial crop 2,505 432 2,067 -6 2,309 236
Cacao 203 34 168 -1 43 -126
Coffee 948 140 806 -2 565 -243
Palm 535 110 423 -1 1,500 1,075
Plantain 593 98 493 -2 156 -339
Other crops 226 50 176 0 45 -131
Annual crop 3,234 725 2,499 -11 2,465 -45
Cassava 330 83 244 -2 241 -5
Maize 1,039 249 785 -4 775 -14
Potato 242 45 196 0 197 0
Rice 898 226 669 -3 652 -20
Sugarcane 529 80 448 0 444 -5
Other crops 197 40 157 0 156 -1
Pasture 21,846 -10,000 30,899 -947 31,761 -85
Forests 59,479 4,161 56,651 1,332 55,289 -30
Other land uses 27,408 4,683 22,357 -368 22,649 -76
Total 114,473 - 114,473 - 114,473 -
Results (1,000 hectares)
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Scenario 2: Total halt to deforestation in the Amazon
Pasture reduction Zero deforestation Palm expansion
Crop Area 2030
Difference
from baseline
2030 Area 2030
Difference
from baseline
2030 Area 2030
Difference
from baseline
2030
Perennial crop 2,505 432 2,067 -6 2,309 236
Cacao 203 34 168 -1 43 -126
Coffee 948 140 806 -2 565 -243
Palm 535 110 423 -1 1,500 1,075
Plantain 593 98 493 -2 156 -339
Other crops 226 50 176 0 45 -131
Annual crop 3,234 725 2,499 -11 2,465 -45
Cassava 330 83 244 -2 241 -5
Maize 1,039 249 785 -4 775 -14
Potato 242 45 196 0 197 0
Rice 898 226 669 -3 652 -20
Sugarcane 529 80 448 0 444 -5
Other crops 197 40 157 0 156 -1
Pasture 21,846 -10,000 30,899 -947 31,761 -85
Forests 59,479 4,161 56,651 1,332 55,289 -30
Other land uses 27,408 4,683 22,357 -368 22,649 -76
Total 114,473 - 114,473 - 114,473 -
Results (1,000 hectares)
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Scenario 3: Total land allocated to palm production reaches a total of 1.5 million hectares
Pasture reduction Zero deforestation Palm expansion
Crop Area 2030
Difference
from baseline
2030 Area 2030
Difference
from baseline
2030 Area 2030
Difference
from baseline
2030
Perennial crop 2,505 432 2,067 -6 2,309 236
Cacao 203 34 168 -1 43 -126
Coffee 948 140 806 -2 565 -243
Palm 535 110 423 -1 1,500 1,075
Plantain 593 98 493 -2 156 -339
Other crops 226 50 176 0 45 -131
Annual crop 3,234 725 2,499 -11 2,465 -45
Cassava 330 83 244 -2 241 -5
Maize 1,039 249 785 -4 775 -14
Potato 242 45 196 0 197 0
Rice 898 226 669 -3 652 -20
Sugarcane 529 80 448 0 444 -5
Other crops 197 40 157 0 156 -1
Pasture 21,846 -10,000 30,899 -947 31,761 -85
Forests 59,479 4,161 56,651 1,332 55,289 -30
Other land uses 27,408 4,683 22,357 -368 22,649 -76
Total 114,473 - 114,473 - 114,473 -
Results (1,000 hectares)
Results
*Includes changes in SOC, Above and Below ground C caused by land use change.
**Changes in emissions from cropland and livestock caused by land use change. Exclude burning
**Changes in revenue from crop and meat production
Change in C
stock
(Tg C)
Change in
GHG emission
(Tg CO2eq)
Change in
total revenue
(US$ billion)
Reduce
pasture by
10 million
hectares
Cropland 144 26.7 56
Livestock -1,297 -184 -22
Forest 895 - -
Other 660 - -
Total 402 -157 34
Zero
deforestation
in the
Amazon
Cropland 4 -0.8 -1
Livestock -97 -16 -2
Forest 245 - -
Other -41 - -
Total 111 -17 -3
Increase area
allocated to
palm by 1.5
million
hectares
Cropland 18 24.5 -57
Livestock -10 -1 0
Forest -7 - -
Other -11 - -
Total -11 24 -57
Policy Comparison
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
THANK YOU
MUCHAS GRACIAS POR SUATENCION
Page 56