CGIAR PLATFORM FOR BIG DATA IN AGRICULTURE · data in agriculture: building a global agricultural...
Transcript of CGIAR PLATFORM FOR BIG DATA IN AGRICULTURE · data in agriculture: building a global agricultural...
CGIAR PLATFORM FOR BIG
DATA IN AGRICULTURE:
BUILDING A GLOBAL
AGRICULTURAL DATA AND
OPEN INNOVATION
ECOSYSTEM FOR IMPACT
CGIAR Big Data Platform – An overview
Brian King – CIAT
MODULES
rganiORGANIZESUPPORT AND
IMPROVE DATA
GENERATION, ACCESS,
AND MANAGEMENT
ConveneCONVENE
COLLABORATE AND
CONVENE AROUND BIG
DATA AND
AGRICULTURAL
DEVELOPMENT
InspireINSPIRE
DEMONSTRATING THE
POWER OF BIG DATA
ANALYTICS THROUGH
INSPIRING AND
INNOVATIVE PROJECTS
170,000 publications & 27,000 datasets discoverable
GARDIAN also includes guidance, workflows and tools to make CGIAR data
assets open and Findable, Accessible, Interoperable, Reusable (FAIR).
ORGANIZE
SUPPORT AND IMPROVE DATA GENERATION, ACCESS, AND MANAGEMENT IN CGIAR
DIGITAL TRANSFORMATION | THROUGH DATA STANDARDS AND SHARING
Collaborative GARDIAN Labs (CG Labs) offers researchers a secure
analytic environment to find data and collaborate on
analyses within the GARDIAN ecosystem, integrating single sign-on and
Globus, a service enabling secure data sharing.
CGIAR EXPERT FINDER
Enables discovery of CGIAR research expertise organized by people, geographies, Centers, funding agencies, and
publications.
ANNUAL CONVENTION
1,800 attendees +
8,000 online viewers, to date
Approx. 65% external to CGIAR,
catalysing alliances with stakeholders
across the technology and agricultural
sectors
CONVENE
COLLABORATE AND CONVENE AROUND BIG DATA AND AGRICULTURAL DEVELOPMENT
DIGITAL TRANSFORMATION | THROUGH ALLIANCES AND COMMUNITIES OF PRACTICE
SHARED SERVICES
Empowering CGIAR and its
community to deliver on the potential of
big data to bring results for smallholder
agriculture.
COMMUNITIES OF PRACTICE
3.5K experts across six communities
with approx. 80% external to CGIAR
THE INSPIRE CHALLENGE
Revealing
Food
Systems
Sensing & Renewing
EcosystemsSustaining Farm
Income
Measuring &
Building Resilience
DIGITAL TRANSFORMATION THROUGH DIGITAL INNOVATION
2.6M USD in total grants
awarded
600,000 USDraised --and growing-- from
external investors
11 CGIAR Centers & CRPs
partnered with
31 expert industry partners
500 Inspire Challenge
applications
18 grants awarded to
innovative projects
3 Rapid Response grants
awarded to mitigate
COVID-19 impacts
INSPIREDEMONSTRATING THE POWER OF BIG DATA ANALYTICS THROUGH INSPIRING AND INNOVATIVE PROJECTS
Show me the data!
Medha Devare – IFPRI
Jawoo Koo - IFPRI
FUTURE FARMS…
BIG DATA-DRIVEN SOLUTIONS.
Find data Visualize
Interpret Analyze
Aggregate
https://www.openaire.eu/how-to-make-your-data-fair
RESEARCH DATA | OPEN BY DEFAULT
….
DIGITAL TOOLS TO
COLLECT STANDARDS-
COMPLIANT DATA
WORKFLOWS TO
COMPLY WITH
STANDARDS
TOOLS FOR
DATA ANALYSIS,
VISUALIZATION
WIDE DATA
DISCOVERY, USE
DATA
DISCOVERY
PORTAL
nutrition women
FCDO
USAID - DDL
data.gov.in
World Bank
IFPRI
ILRI
CIP
AgroFIMS
Agronomy Field Information
Management System
Generate standardized field books
to collect agronomic data that is…
https://apps.cipotato.org/hidapagrofims
“I want to be able to explore data visually!”
“I want to be able to work with my team to find, securely share, and analyze data.”
Users must have
Globus accounts:
https://www.globus.org/
via gmail or
institutional account
https://gardian.bigdata.cgiar.org/labs.php
Oganization-wide secure data sharing
and discovery services
Multi-faceted collaboration spaces at
customizable levels (project teams, groups,
labs, etc.)
Role-based access and Single Sign-On
Data sharing & discovery
CollaborationSecurity
Management
Index and search internal data to enable secured
web-based and API access to organizational assets
Private Data
COLLABORATIVE GARDIAN LABS | AN ANALYTICS PIPELINE
… and Open Source
Note: Some CC0 and
CCBY datasets are
locked down and
cannot be
downloaded yet
CG LABS | DOWNSCALED CMIP6 DATASET (7 TB)
Click 1: Select CMIP6 Model
Click 2: Pin region
Click 3: Select time period
Click 4: Download data as GeoJSON or CSV
C. Porter et al., Univ. of Florida, 2019C. Porter et al., Univ. of Florida, 2019
The data variables map to the
ICASA Data Dictionary, and
Agronomy Ontology and other
ontologies. AgMIP translation
tools can therefore translate
the data for several widely-
used crop simulation models.
Ava’s GARDIAN search reveals
37 field experiments with
several crops.
COMING SOON | TO CG LABS…
“I need to recommend sustainable intensification practices for West Africa.”
Data shmata…
Profitability of fertilizer use in Africa
A case-study in data re-use
Camila Bonilla Cedrez – CIAT
Jordan Chamberlin – CIMMYT
Robert Hijmans – UC Davis
Fertilizer use in Africa
Why is fertilizer use low ?
low crop response (poor soil / climate)?
unfavorable prices?
uncertainty (that can perhaps be insured)?
We need site-specific
crop response to fertilizer model
and input and output price data
108 data sets
760 locations
12,000 observations
Findable and Accessible data
Interoperable?
Aggregated (raw) data
N response
Reuse: Machine learning predictions
P response
Reuse: Price data
Maximum profitability
• Old data can be used for new science
• Empirical models will gain importance as more
data becomes available
• Interoperability for data aggregation
will become increasingly important
Toward operational multi-scale
investment prioritisation
Julian Ramirez Villegas – CIAT
Outline
1. Climate change –a wicked problem
2. What investments? Where? –data is
here to help
• Smallholder Adaptation Atlas
• Country CSA Investment Planning
• … and diving even deeper
3. Powered by CGLabs
Models of crops,
livestock and
socioeconomics
Impacts
Vulnerability
Adaptation
options
Investment
costs
What investments? Where? –data is here to help
Smallholder Adaptation Atlas
To support investment prioritization by
the global commission on adaptation
Smallholder Adaptation Atlas
Data to support decisions
12Hazard
variables
5Climate
scenarios
33Crops
Heat stress
occurrence (future)
continental coverage
> 100k individual layers
180Adaptation
options
AGRA Countries
Inorganic
inputs
Cons. Ag.
R&D
Country CSA Investment Planning
Frequency: Every other yearIntensity: severe (80/100)
System: rainfed maize-beanSensitivity: high (75/100)
Hazard characteristics
System characteristics
Risk scoring for selected hazard
Fre
qu
en
cy
Impact
Mild drought
Severe flooding
Lightning
Severe drought
Moderate heat stress on livestock
Waterlogging
Moderate heat stress
on crops
Earthquakes
Pollution
Understand climate hazards
Assess and prioritize investments
Powered by CGLabs
Easy
to use
Customized and
customizable
Fast, reliable
tech support
CGIAR’s
data ready
Fast for
downloading
non-CGIAR data
Easily
scalable
Replicability, use in
teaching and
debugging / testing
Can share with
anyone, at minimal
transaction cost
Notebooks are
useful for capacity
building
Streamlining
analysis
Why CGLabs?
bigdata.cgiar.org