Simulating Global Climate Change and Variability in Economic Models

Roy Darwin

Seminar at the Climate Impacts Group, Center for Science in the Earth System, Joint Institute

of the Atmosphere and Ocean, University of Washington, Seattle, WA, October 13, 2003

The views expressed herein do not necessarily reflect

those of the U.S. Department of Agriculture

Outline

• Future Agricultural Resources Model

• Climate Variability

• Inter-Annual Variability

FARM’s Environmental Component

Climate

Growing Season Water Runoff

LandCharacteristics

Water Supply

FARM’s Economic Component

Land Cover

FARM’s Economic ComponentFARM’s Environmental Component

Land Water

Labor CapitalProduction Possibilities

PopulationProduction

ConsumerPreferences

Foreign Region1

Foreign Region n

Income

Other Regions

Supply/Demand

Supply/Demand

Trade

Payments

Payments

Topics Analyzed with FARM

• Impacts of greenhouse gas emissions on agriculture and forestry

• Costs of sea level rise

• Effects of trade deregulation and population growth on tropical forests

• Costs of protecting global ecosystem diversity

• Impacts of technological advance in agriculture on land use

Database Components

• Land cover

• Agro-ecological zones

• Production

• Production distribution

Land Cover Component• The land-cover component organizes data on land

cover characteristics• The main data source is a 1-km resolution global

land cover characteristics database• These data are organized by 0.5 grid and second

order political unit• The original category codes were normalized,

disaggregated, and reaggregated into 10 general land-cover categories

Percent GrasslandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent TundraDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Coniferous ForestlandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Nonconiferous ForestlandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Mixed ForestlandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Scattered TreesDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent ShrublandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Built-Up LandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Other LandDerived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001.

Percent Cropland in 1997Derived from: U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. , 2001; Food and Agriculture Organization of the United Nations. FAOSTAT Agriculture Data. 2001; Döll, P. and S. Siebert. A digital global map of irrigated areas. 2000; Siebert,S. and P. Döll. A digital global map of irrigated areas—An update for Latin America and Europe . 2001.

Percent Irrigated Land in 1997Derived from: Döll, P. and S. Siebert. A digital global map of irrigated areas. 2000; Siebert, S. and P. Döll. A digital global map of irrigated areas—An update for Latin America and Europe . 2001; U.S. Geological Survey. EROS Data Center. Global Land Cover Characteristics Data Base Version 2.0. 2001; Food and Agriculture Organization of the United Nations. FAOSTAT Agriculture Data. 2001.

Agro-Ecological Zone Component

• Agro-ecological zones (AEZs) replace land classes in the current version of FARM

• The AEZ component constructs AEZs based on length of growing season and thermal regime

• Growing season and thermal regime are calculated from meteorological data with a soil temperature and moisture algorithm

• The meteorological data are monthly temperature and precipitation at 0.5-degree grid, 1901-1998

Rainfed Agro-Ecological Zones in 1997Derived from: University of East Anglia. Climate Research Unit. CRU05 0.5 Degree 1901-1995 Monthly Climate Time-Series. East Anglia, Great Britain.

Rainfed Thermal Regimes in 1997Derived from: University of East Anglia. Climate Research Unit. CRU05 0.5 Degree 1901-1995 Monthly Climate Time-Series. East Anglia, Great Britain.

Production Component

• The production component organizes price and quantity data for agricultural and forestry commodities

• It tracks production of 173 crops by country• It tracks 37 primary livestock commodities• It tracks 6 categories of timber products• It also tracks 18 categories of live animals

Production Distribution Component

• The production distribution component links production to land covers by AEZ

• Commodities are distributed to appropriate land covers

• Average production for each AEZ in the appropriate land cover is estimated with regression analysis

• Commodities are distributed by AEZ and calibrated to official levels

0

500

1000

1500A

EZ

1

AE

Z2

AE

Z3

AE

Z4

AE

Z5

AE

Z6

TR6TR5

TR4TR3

TR2TR1

Estimated Average Wheat Production by AEZxTR (mt/1000ha)

Wheat Production in 1997 (mt/1000ha land) Interpolated from country or state level data

Paddy Rice Production in 1997 (mt/1000ha land) Interpolated from country or state level data

Distributing Livestock and Forest Products

• Many grasslands or forestlands do not provide livestock or forest products because people do not live there

• A particular livestock animal, livestock commodity, or forest product may be associated with more than one land cover

• Consistency between live animals and livestock commodities must be maintained

Population Density in 1998 (no/1000ha land)Derived form Oak Ridge National Laboratory. LandScan Global Population 1998 Database, Oak Ridge, Tennessee.

Cattle Inventory in 1997 (head/1000 ha land) Interpolated from country and state level data

Beef, Veal, and Cattle Hide Production in 1997 (mt/1000ha land) Interpolated from country and state level data

Cow Milk Production in 1997 (mt/1000ha land) Interpolated from country and state level data

Conifer Sawlog Production in 1997 (m3/1000ha land) Interpolated from country and U.S. regional level data

Nonconifer Sawlog Production in 1997 (m3/1000ha land) Interpolated from country and U.S. regional level data

Climate Change and World Crop Production

-3

-2.5

-2

-1.5

-1

-0.5

0

0.5

0 2 4 6

Change in Mean Global Temperature (degrees C)

Perc

ent C

hang

e

FARM Estimates Summary EstimatesUpper 0.95 Limit Lower 0.95 LimitUpper 0.67 Limit Upper 0.33 Limit

%CQ = -0.223 T, adjR2 = 0.832 (-6.656)

Climate Change and Average World Crop Price

-1

0

1

2

3

4

5

6

0 2 4 6

Change in Mean Global Temperature (degrees C)

Perc

ent C

hang

e

FARM Estimates Summary EstimatesUpper 0.95 Limit Lower 0.95 LimitLower 0.67 Limit

%CP = 0.577 T, adjR2 = 0.930 (9.809)

Climate Change and World Per-Capita Welfare

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0 2 4 6

Change in Mean Global Temperature (degrees C)

Perc

ent C

hang

e

FARM Estimates Summary EstimatesUpper 0.95 Limit Lower 0.95 LimitUpper 0.67 Limit Upper 0.33 Limit

%W = -0.015 T, adjR2 = 0.541 (-3.431)

Climate Change and Per-Capita Welfare in Japan

-0.1

0

0.1

0.2

0.3

0 2 4 6

Change in Mean Global Temperature (degrees C)

Perc

ent C

hang

e

FARM Estimates Summary EstimatesUpper 0.95 Limit Lower 0.95 LimitLower 0.67 Limit Lower 0.33 Limit

%W = 0.018 T, adjR2 = 0.644 (4.225)

Climate Change and Per-Capita Welfare in

Southeast Asia

-2

-1.5

-1

-0.5

0

0.5

0 2 4 6

Change in Mean Global Temperature (degrees C)

Perc

ent C

hang

e

FARM Estimates Summary EstimatesUpper 0.95 Limit Lower 0.95 LimitUpper 0.67 Limit

%W = -0.157 T, adjR2 = 0.935 (-9.694)

Estimating Impacts of Meteorological Phenomena on Length of Growing Seasons in the 20th Century

• Yi = 0 + 1T + 2ENSOj + 3ENSOj*Dj + where Yi is a vector of irrigated or rainfed growing season lengths, T is a trend vector equal to (1,96), ENSOj is either a SOI or CTI vector, Dj is a dummy variable (= 0 when ENSOj 0, but = 1 when ENSOj 0), and is a vector of error terms with a finite variance

Estimated Average Change in Length of Rainfed GrowingSeason, 1902-1997: Derived from Regression Models Using the CTI to Capture ENSO Cycles: Days and Confidence

Estimated Average Impact of El Niño on Length of Rainfed Growing Season, 1902-1997: Derived from Regression ModelsUsing the CTI to Capture ENSO Cycles: Days and Confidence

Estimated Average Impact of La Niña on Length of Rainfed Growing Season, 1902-1997: Derived from Regression ModelsUsing the CTI to Capture ENSO Cycles: Days and Confidence

Current Cropland in Low-Income Countries Affected by Climate-Induced Changes in Growing Season

Region

Length of Growing Season Longer Shorter (1000 ha)

Sub-Saharan Africa 11,581 52,279

North Africa 1,820 8,171

Asia 21,146 51,688

Latin America & Caribbean 539 8,668

New Independent States 11,758 2,461

Total 46,844 123,267Estimated with results from regression models with the Cold Tongue Index for theEl Niño/Southern Oscillation cycle.

Impact of Climate Change on Cropland Growing Seasons in Selected Low-Income Countries, 1902-1997

Estimated with results from regression models with the Cold Tongue Index for theEl Niño/Southern Oscillation cycle.

Food Distribution Gap in Various Low-Income Countries Projected for 2012 (kg grain/person)

Estimated with ERS’s Food Security Assessment model

Estimated with ERS’s Food Security Assessment model

Estimated Contribution of Climate-Induced Changes in Average Growing Season to 2012 Projections of Food Distribution Gaps in Selected Low Income Countries (percent)