ISP Meeting, Ouagadougou, 23 Oct 2012

Making Climate Information More Relevant to Smallholder Farmers

James Hansen, CCAFS Theme 2 LeaderIRI, Columbia University, New York

Prerequisites to benefitting from an information service

• Credibility

• Salience

• Legitimacy

• Access

• Understanding

• Capacity to respond

WG 1

WG 3

WG 2,4

WG 5

}Information product}Information service

Delivery system

}Users

Salience: What kind of information do farmers need?

• Types of climate information:

– Historic observations

– Monitored

– Predictive, all lead times ≤ ~20 years

• Some generalizations:

– Downscaled, locally-relevant

– Tailored to types & timing of decisions

– “Value-added” climate information: impacts on agriculture, advisories

– Capacity to understand and act on complex information

Anecdote 1: RCOFs for farmer decision-making?

“Weather-within-climate”

Probabilistic information needed for risk management

Capacity development through training, dialog with trusted advisors(http://www.wmo.int/pages/prog/wcp/wcasp/clips/outlooks/climate_forecasts.html)

climate community

“users”

applications

“…a hub for activation

and coordination of

regional climate

forecasting and

applications activities

into informal networks”

Basher et al. (Ed) (2001). Coping with Climate: A Way Forward. Summary and Proposals for Action. Palisades, New York: IRI.

• Owned, designed, convened by providers

• Spatial scale

• Seasonal rainfall total

• Probabilistic: Tercile format, often lost before reaching users

• Capacity development through stakeholder meeting participation

Anecdote 2: Early doubts about value of seasonal forecasts to farmers

• Error accumulates from:

– SSTs to regional rainfall

– Regional to local rainfall

– Local rainfall to crop yield

• Therefore prediction of climate impacts on farms is not feasible.

• Given the inherent uncertainty, poor farmers can’t bear the risk of a wrong forecast.

Barrett, 1998. AJAE 80:1109-12

• Depends on time horizon of decision

• Generalizations about increasing lead time:

– Decisions more context- and farmer-specific

– Information becomes more uncertain, hence more complex

– Therefore the scope of services needed increases

• “Weather-within-climate:”

– Timing of season onset, length

– Seasonal total = frequency × intensity. Frequency more predictable.

– Dry, wet spell length distributions

Elements of salience: Time scale

HOURS DAYS WEEKS MONTHS YEARS DECADES …

WEATHER CLIMATE

• Tillage

• Sowing

• Irrigation

• Crop protection

• Harvest

• Changing farming or livelihood system

• Major capital investment

• Migration

• Family succession

• Land allocation

• Crop selection

• Household labor allocation, seasonal migration

• Technology selection

• Financing for inputs

• Contract farming

Elements of salience: Spatial scale

Correlation of observed (85 stations) vs. predicted rainfall in Ceará, NE Brazil, as a function of spatial scale. Gong, Barnston, Ward, 2003. J. Climate 16:3059-71.

Co

rre

latio

n

Scale

Elements of salience: Communicating uncertainty

• Relate measurements to farmers’ experience

Elements of salience: : Communicating uncertainty

• Relate measurements to farmers’ experience

• Convert series to relative frequency, then probability

Oct-Dec rainfall (mm)

Year

s wi

th a

t lea

st th

is m

uch

rain

Elements of salience: Communicating uncertainty

• Relate measurements to farmers’ experience

• Convert series to relative frequency, then probability

• Explanation & repetition

?

Elements of salience: Communicating uncertainty

• Relate measurements to farmers’ experience

• Convert series to relative frequency, then probability

• Explanation & repetition

• Compare with e.g., El Niño years to convey forecast as a shifted distribution

Elements of salience: Communicating uncertainty

• Relate measurements to farmers’ experience

• Convert series to relative frequency, then probability

• Explanation & repetition

• Compare with e.g., El Niño years to convey forecast as a shifted distribution

• Explore management implications

Elements of salience: Translation to impacts on agriculture

• Example: Integrate seasonal forecasts into yield predictions

• Reduces uncertainty, more early in growing season

• Before planting, forecasts potentially more accurate for crop yield than for seasonal rainfall

Traditional sorghum, Dori, Burkina Faso. Mishra et al., 2008. Agric. For. Meteorol. 148:1798-1814.

Correlations of Jun-Sep rainfall, and observed, de-trended wheat yields with May GCM output, prior to planting, Qld., Australia. Hansen et al., 2004. Agric. For. Meteorol. 127:77-92

200 0 200 400 km

Correlation< 0.34 (n.s.)0.34 - 0.450.45 - 0.500.50 - 0.550.55 - 0.600.60 - 0.65 > 0.65

Rain

Yield

Elements of salience: Translation to management guidance

• At weather time scale, broadly-relevant advisories for time-sensitive decisions (sowing, irrigation, pest and disease control)

• At climate time scale, caution about top-down recommendations:– Decisions more farmer-

specific

– Uncertainty is greater

• Combine sources of expertise

• Involve trusted advisors

• Dialog with experts

• Farmer-to-farmer discussion

Institutional arrangements for salience?

• Limitations of supply-driven climate services

• Expanding the boundary to agricultural research and development

• Expanding the boundaries to give farmers a voice

CLIMATE SERVICE

NMS(climate)

User (farmer)

INFORMATION

CLIMATE SERVICE

NMS(climate)

User (farmer)

VALUE-ADDEDINFORMATION

NARES(agricultur

e)PARTNERSHIP

CLIMATE SERVICE

NMS(climate)

Co-owner(fa

rmer)

NARES(agriculture

)

PARTNERSHIP

Salience and historic data

• Local decision-making depends on local information.

• Many promising opportunities to adapt to climate variability and change depend on historic data, are constrained by gaps.

• In Africa, feasible to blend station and satellite rainfall data => complete 30-year, 5-10 km grid daily record. Extending to other agriculturally-important variables.

• Meteorological data policy – Is it time to consider change?STATION BLENDED SATELLITE