Infrared spectroscopy - bringing soil health information to smallholder farmers
Jan 15, 2016
Infrared spectroscopy - bringing soil health information to smallholder farmers
Using only light to analyse soils
Scan a soil sample in 30 seconds Submit to online
spectral prediction app & get predicted soil properties
Spectral shape relates to basic soil properties
• Mineral composition• Iron oxides• Organic matter• Water (hydration,
hygroscopic, free)• Carbonates• Soluble salts• Particle size distribution
Functional properties
How does it work?
How do we use it?Mapping 3D soil properties for targeting soil fertility management strategies in Ethiopia
pH SOC
Africa Soil Information Service
www.africasoils.net
How do we use it?
Enabling cost-effective soil advisory services to farmers
Establish a rural soil lab for $50,000
•IAMM, Mozambique
•AfSIS, Sotuba, Mali
•AfSIS, Salien, Tanzania
•AfSIS, Chitedze, Malawi
•CNLS, Nairobi, Kenya
•CNRA, Abidjan, Cote D’Ivoire
•KARI, Nairobi, Kenya
•ICRAF, Yaounde, Cameroon
•Obafemi Awolowo University, Ibadan, Nigeria
•IAR, Zaria, Nigeria
•ATA, Addis Ababa, Ethiopia (6)
•IITA, Ibadan, Nigeria
•IITA, Yaounde, Cameroon
•IER, Arusha, Tanzania
•FMARD, Nigeria
•CNLS, Nairobi, Kenya
•BLGG, Kenya (mobile)
Who uses it?
Who else uses it?
Governments from Ghana, Nigeria and Tanzania signed up for national soil surveillance systems
Trained 717 (171 female) scientists/technicians in land/soil health surveillance field or laboratory techniques in past 12 months
2nd hands-on soil spectroscopy training course
Piloted farm soil monitoring in World Bank LSMS Ethiopia
IR analytical services to 18 CGIAR projects
Training 47 counties in Kenya with ChromAfrica
Piloting farm advisory service with One Acre Fund
Capacity building
ICRAF Soil-Plant Spectral Diagnostics Lab received 500 visitors per year for the past three years, over half of whom have received training
Conducted two hands-on soil spectroscopy training courses (each with 50 participants from 10 African countries)
In-country training in support of the spectral lab network
IR analytical services to 18 CGIAR projects
Helped private soil labs establish soil spectral analytical services: Soil Cares Initiative (mobile lab) and Crop Nutrition Services Ltd
Piloting farm advisory service with One Acre Fund to provide services to 25,000 farmers in eastern Africa
Training 47 counties in Kenya with ChromAfrica Ltd
Request from Karnataka State Government of India to help analyze 300,000 soil samples in 3 years to provide farm soil health cards
Partnerships
Improving measurements of agricultural productivity by combining household
level and soil fertility data
Coupling farm soil health measurement with household panel surveys
Using Central Statistics Agency sampling frame
Oromiya region of Ethiopia
Trained and supervised
19 enumerators and 2 supervisors in soil sample collection
4 lab technicians in soil sample processing (from Forest Research Centre, Ambo University, Hawasa Research Centre and Yabello Research Centre)
3 staff from the LSMS project and CSA attended 3-day soil infrared spectroscopy training course at ICRAF’s Soil-Plant Spectral Diagnostics Laboratory
Field guide for soil sampling
Composite soil sampling and coning
Soil sampling is tied to crop cut areas, but whole fields are also sampled
Next steps
Soil fertility status reports for Woredas Poverty – soil fertility relationships Extend pilot to another country Recommendations on standardizing the
approach within LSMS
Assessment of hydrological, financial and social risk around the supply of groundwater to
Wajir town
• ICRAF: Jan de Leeuw, Eike Luedeling, Todd Rosenstock, Keith Shepherd• Acacia Water, the Netherlands (hydrological risk assessment)• CETRAD Nanyuki, Kenya and University College London, U.K. (Social
risk assessment)
The decision problem
Research for impact
Most science never supports any decisions, even though decision-
makers are hungry for information.
http://www.mynamesnotmommy.com
Most research does not answer questions that are critical for decisions, or it is not readily available when it is needed.
Tailor research specifically to address particular decisions
Decision making under uncertainty
Identify all uncertainties in the decision of interest
Make probabilistic projections of likely decision outcomes
http://www.relationshipeconomics.net
Engage directly with decision makers
This is the core of WLE’s decision analysis procedures
Identify uncertain variables with high ‘information values’ (these
are priorities for measurements)
Why a quantitative model?
Popular approaches to assessing a large list and use of soft “scoring methods” that require a subject matter expert to pick a value on some scale for each of several factors. These usually introduce errors.
Two common and significant sources of error in expert forecasting and evaluation tasks: • overconfidence (where experts are right far less
often than their perceived confidence would indicate) and
• Inconsistency
WLE’s decision analysis approach
Optimize the decisionUse preferences of decision makers to determine best
decision
Ye sIs there significant value
to more information? Yes
No
Compute the value of additional informationDetermine where and how much measurement effort is
needed
Model the current state of uncertainty of all relevant factors Initially use calibrated estimates and then actual
measurements
Define the decision(s) – Identify relevant variablesSet up the ‘Business Case’ for the decision.
Measure where the information value is
high – Reduce uncertainty using proven empirical
methods
Outputs
Replenishment
Irrigation growth
Initial irrigated area
Water use per hectare
Aquifer size
Natural water use
Importance threshold
Identification of high-value variables
Probabilistic impact projections
Aquifer size after 70 years of abstraction (% of original)
Advantages and applications
Inclusion of uncertain variables allows truly holistic impact assessments
High information value variables are almost always not those typically measured
Probabilistic impact projections can often provide sufficient certainty about decision outcomes
Possible applications include:• Quantitative and probabilistic impact pathways• Ex-ante impact projections• Counterfactuals for impact assessments• Definition of priority variables for impact monitoring
What did we do?
What did we find
Value of information analysis
What did we achieve?