ccafs science meeting a.2 jerry nelson - global futures

INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE

From Global Futures to Strategic Foresight

Gerald Nelson

Senior Research Fellow , IFPRI

Theme Leader, CRP2 and CRP7

Moving Beyond Norman

Borlaug

What is the Global Futures Project?

Develop methods, tools, and a consistent system to help the CGIAR answer the following question

If an investor provides an additional $x million to the CGIAR, how should it be spent to

provide the greatest return on investment?

• Financial ROI

• Reduction in poverty

• Improvements in sustainability

Unprecedented Collaboration

Enhance modeling tools

• IMPACT model

• HarvestChoice

• DSSAT

Use expertise at centers and elsewhere

• IFPRI, IRRI, ICRISAT, CIMMYT, ILRI, ICRAF, CIP, CIAT, others to be added

• DSSAT crop model experts

CRP7 – to support a climate change add-on

How to evaluate potential technological improvements: The Delphi method

Ask the experts; aka the Delphi approach

With an additional $20

million, what productivity

improvements can you

come up with?

3 % per

year for 20

years 2 % per year

for 15 years

Nothing! We need more money!

How to evaluate potential technological improvements: The virtual crop method

1. Ask the experts for details on what they can accomplish

What specific changes in

plant phenotype are

relatively easy to

implement to improve

drought tolerance?

Heat shock proteins for increased protection

Reduce partitioning photosynthate into grain

Early planting and

morning flowering to

avoid pollen sterility


2. Convert these responses into crop model genetic coefficients

CERES Maize Model Coefficient Brief description

P1 Degree days (base 8°C) from emergence to end of

juvenile phase

P2 Photoperiod sensitivity coefficient (0-1.0)

P5 Degree days (base 8°C) from silking to physiological

maturity

G2 Potential kernel number

G3 Potential kernel growth rate mg/(kernel d)

PHINT Degree days required for a leaf tip to emerge

(phyllochron interval) (°C d)


3. Use crop modeling software on HPC to ‘grow’ the virtual variety everywhere and evaluate performance relative to existing varieties

GCM with SRES A1B average yield (mt per hectare)

DSSAT generic maize varieties

2000 yield 4.2

CSIRO, 2050 4.1

MIROC, 2050 3.7

DSSAT specific varieties

2000 yield 5.4

CSIRO, 2050 5.4

MIROC, 2050 4.9

DSSAT virtual varieties

2000 yield 5.5

CSIRO, 2050 5.6

MIROC, 2050 5.2

Supply/ demand

interactions

FPU level yield and area

scenarios FPU

boundaries SPAM crop distributions

DSSAT yield scenarios

Planting months

Climatic conditions

Soils Management

practices

Incorporating productivity effects: Combine biophysical and socioeconomic

Virtual crop

activities

Socioeconomic

modeling


WHY SCENARIOS?

The future is an uncertain place

Challenges in Modeling Climate Change

10

Average temperature change, 2 modeling groups, scenario A2

Yield Effects, Rainfed Maize, CSIRO A1B (% change 2000 climate to 2050 climate)

Yield Effects, Rainfed Maize, MIROC A1B (% change 2000 climate to 2050 climate)

Challenges in Modeling Socioeconomics: Identifying Plausible Futures

Optimistic

• High GDP and low population growth

Baseline

• Medium GDP and medium population growth

Pessimistic

• Low GDP and high population growth

Climate change scenario effects on prices differ (price increase (%), 2010 – 2050, Baseline economy and demography)

Minimum and maximum

effect from four climate

scenarios

January Global Futures Meeting Proof of concept test

Investment - $10 million

Promising technology choices

• Drought tolerance

• Herbicide resistance

ROI, Drought Tolerant Groundnut (Proof of Concept Only)

Welfare and returns on

investment

Climate change scenarios

No climate

change

MIROC

369 A1B

MIROC

369 B1

CSIRO

369 A1B

CSIRO

369 B1

Changes in producer surplus (NPV,

m US$) -3,876 -4,275 -3,790 -4,540 -4,698

Changes in consumer surplus

(NPV, m US$) 10,443 11,338 10,082 11,997 12,507

Net welfare change (NPV, m US$) 6,567 7,063 6,292 7,457 7,809

Cost (NPV, m US$) 15 15 15 15 15

Benefit-cost ratio 448 482 430 509 533

Net benefits (NPV, m US$) 6,553 7,048 6,277 7,443 7,795

IRR (%) 54 55 53 55 56

Welfare Effects: Drought-tolerant groundnut (Proof of concept only)

Regions Target

countries

D Kcals per $

invested

D Malnourished

children per $

invested

D At Risk of

Hunger per $

invested (million)

ESA

Malawi 1.9 -1,285 -6.1

Tanzania 0.6 -672 -3.6

Uganda 1.7 -1,824 0.0

WCA

Burkina Faso 3.6 -1,666 -1.8

Ghana 3.1 -904 0.0

Mali 2.5 -935 0.0

Nigeria 4.4 -13,604 -4.9

Senegal 5.6 -973 0.0

SSEA

India 0.9 -8,840 -32.0

Indonesia 1.2 -1,429 0.0

Myanmar 3.0 -971 -6.9

Vietnam 1.0 -511 -1.4

Tasks remaining/for next phase

Revise and resubmit results

Test with more types of virtual cultivars

Address issues such as

• Ruminants

• Land use

• GHG emissions

• New climate data

• Non-tradable goods

• Improvements to and new crop models


FROM GLOBAL FUTURES TO STRATEGIC FORESIGHT

Moving Beyond Norman Borlaug

The Borlaug Paradigm

Borlaug key insight - Do trial and error approach with LOTS of trials

• Limited collection of data other than yield

Exploit mega environments

• Regions with similar agronomic characteristics

• Do trials where mandate crop is currently important

Led to

• Breeders are key

• Physiologists are not

What has changed?

Benefits of Borlaug approach fully exploited -> costs are rising

Mega environments are changing

• Climate change, resource availability, demand

Information technology revolution

• Computing and data storage steadily cheaper

Genetics revolution

• Fundamental understanding of biological processes

New system needed to recognize

and exploit these changes

The importance and implications of quantitative modeling for strategic foresight

What are models

• Reduced form quantification of biological/socioeconomic processes

• Calibrated with real world data

Why model

• When interactions become too complex to understand intuitively

• When costs of modeling are less than the benefits

Insights for the CGIAR

Institutionalize model use and development

Design data collection efforts to support model improvements

Employ people who can contribute to improved models

Develop systems that make it easy for others to

• Use the models

• Contribute to model improvement

New Approach with Two Elements: Coordinating Unit

• Develop methodologies and tools needed to conduct integrated assessments of potential research outputs

• Place those tools in an integrated suite of biophysical and socioeconomic models

• Ensure that models are evaluated based on the science behind the components, including uncertainty

• Ensure that the models are available as global public goods (open source utilizing GPL licenses)

• Support multidisciplinary teams Develop guidelines, protocols and modeling expertise to

complement that of each center for both socioeconomic and biophysical production system models

New Approach with Two Elements: Multidisciplinary Center-based Teams

Link to experimentalists to provide in-depth, state of the art knowledge about mandate crops, animals, and systems

Identify promising options for technology enhancements Adapt/improve production/system-specific models to simulate

• existing plant varieties and livestock breeds in targeted ecosystems • new varieties and breeds in those and new ecosystems, taking into

account existing and plausible future socioeconomic and natural resource conditions

Help design critical experiments and data collection protocols to • Ensure adequacy and availability of data for mandate systems • Contribute data for a global database of agronomic and breeder

trial data for evaluating and improving models, that facilitate analyses from household to global of technology, policy, and climate changes

Outputs

Strategic foresight quantitative modeling tools

Ex ante evaluation of promising technologies

Outreach – Food Security Futures Conference

• first scheduled tentatively April 15-19, 2013

Capacity building


DO WE REALLY NEED MORE DATA?

How much irrigated area in India?

Intl. Water Management Inst.

113 M ha (net)

Government of India

57-62 M ha

Source: Thenkabail 2009


COMPARING LAND COVER DATA IN AFRICA

Globcover 2005 – (300m)

GLC2000 2000 – (1km)

MODIS 2001 – (5km)

Africover 1999-20 01 – (30m)

Globcover

MODIS

GLC2000

Africover

Kenya

Zhe Guo, HarvestChoice

2011 (unpublished).”

MODIS

GLC2000 Globcover

Uganda

Rwanda



Africover

MODIS

GLC2000 Globcover

Tanzania



Africover

Ethiopia

MODIS

GLC2000 Globcover



Thanks!

ccafs science meeting a.2 jerry nelson - global futures

Technology