The main goal of the DataBio project is to show the benefits of Big Data technologies in the raw material production from agriculture, forestry and fishery/aquaculture for the bio-economy industry to produce food, energy and bio-materials responsibly and sustainably. DataBio proposes to deploy a state of the art, big data platform on top of the existing partners’ infrastructure and solutions – the Big DATABIO Platform. Achieved impacts are measured against anticipations. In this webinar, we present the impact of the DataBio project and of its big data platform after three years of implementation, and we illustrate some of the novel breakthroughs on practical cases of artificial intelligence applications that are meaningfully boosting crop monitoring businesses with a global potential.
Schema on read is obsolete. Welcome metaprogramming..pdf
BDV Webinar Series - Ephrem - Big Data Breakthroughs for Global Bio-economy Business
1. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
1
This project has received funding from
the European Union’s Horizon 2020
research and innovation programme
under grant agreement No 732064
This project is part
of BDV PPP
BIG DATA BREAKTHROUGHS FOR GLOBAL BIO-ECONOMY BUSINESSES
DR. Ephrem Habyarimana
Chief Scientist & Research Scientist
CREA Research Center for Cereal and Industrial Crops
Italy
BDVe Webinar
December 16, 2019
3. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
3
Estimate: 7.7 billion people as of April 2019.
World population is forecast to grow to 9 billion
by 2050.
1 in 6 is already hungry and food production must
increase by 70-100% if it is to feed this growing
population (Ammann, K., 2012, Chapter 27)
No single solutions will solve this problem but Big
Data technologies can help to increase
agricultural productivity and save people from
hunger and other crises in a sustainable manner.
WORLD POPULATION AND ASSOCIATED ISSUES
4. DATABIO’S TARGET: CONTRIBUTION TO INCREASING
EUROPEAN AND WORLD’S PRODUCTIVITY
DA TA BIO targets to demonstrate that an increase of bioeconomy
productivity of 20% is possible.
STRATEGY:
Engaging European ICT and Earth Observation Industries in innovative
Big Data Technologies to boost the main European bioeconomy sectors
of Agriculture, Fisheries, and Forestry.
4
5. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
5
Flows and expected outcomes
AGRICULTURE
(13 pilots)
FORESTRY
(8 pilots)
FISHERY
(6 pilots)
Big Data Sources
and Big Data Types
Structured and unstructured data
Spatio-temporal data
Machine generated data
Image/sensor data
Geospatial data
Genomics data
Data
Management
Collection
Preparation
Curation
Linking
Access
Data
Processing
Batch
Interactive
Streaming
Real-time
Data Analytics
Classification
Clustering
Regression
Deep learning
Optimization
Simulation
Raw material production
for Food and Energy
Biomaterials
Responsible
production
Sustainability
Data Visualization and User Interaction
1D, 2D, 3D + temporal
Virtual and Augmented Reality
Validation
6. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
6
Project at a glance
2017 2018 2019
27
7. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
7
A Sample of Key Breakthroughs in the DataBio’s
Agricultural Pilots
8. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
8
PILOT 1 [A1.1] PRECISION AGRICULTURE IN OLIVES, FRUITS, GRAPES
PILOT SUMMARY
• Smart Farming pilot focusing on the
exploitation of heterogeneous data,
facts and scientific knowledge to
facilitate decisions and their
application in the field,
• Sustainable farming practices
through the provision of irrigation,
fertilization and pest/disease
management advices,
• The farmer directly benefits from
big-data technologies and advisory
services by better managing the
natural resources, optimizing the
use of agricultural inputs and
increasing farm yields.
3 Pilot Sites….
…and 4 Data
Dimensions
DATABIO PARTNERS INVOLVED
9. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
9
PILOT RESULTS
avgcostofspraying
(euros/ha)
avgcostofirrigation
(euros/ha)
Nitrogen
use(kg/ha)
0
500
1000
1500
Olives Grapes Peaches
2017 2018 2019
0
1000
2000
3000
4000
Olives Grapes Peaches
2017 2018 2019
0
100
200
300
Olives Peaches
2017 2018 2019
Pilot 1 [A1.1] Precision Agriculture in Olives, Fruits, Grapes
12.40%
40%
30%
2.84%
33.76%
N/A
3.58%
42.87%
71.89%
0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00%
Crop protection
Irrigation
Fertilization
Input Cost Reduction
Peaches Grapes Olives
Target:
36%
N/A
9%
30%
30%
30%
5%
3.5%
15%
Rainfall (mm)
10. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
10
28/11/2017 - The Greek Minister
of Digital Policy, Telecommunications
and Media Dr. Nikos Pappas visits
DataBio’s pilot site on peaches and
gets informed about the benefits
gained through the adoption of big-
data technologies.
This visit served as a stepping stone,
as the ministry launched a 28M euros
call for covering the whole Greek
territory with agro-climate sensors,
thus, enabling the provision of Smart
Farming services to all Greek farmers
http://www.ypaithros.gr/en/yannis-olive-grove-
reduction-by-30-in-production-costs-and-
parallel-increase-of-sales/
SUCCESS STORIES
Chalkidiki Pilot
Veroia Pilot
Pilot 1 [A1.1] Precision Agriculture in Olives, Fruits, Grapes
Stimagka Pilot
http://www.gaiasense.gr/en/a-greek-innovation-
gaiasense-evolves
11. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
11
Pilot 6 [B1.2] Cereal, Biomass and Cotton Crops_2
12. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
12
Pilot 6 [B1.2] Cereal, Biomass and Cotton Crops_2
fieldremote
eye farm
Big
data
Increase Profits, minimize
environmental footprint
PILOT SUMMARY
• Smart farming pilot focusing on the exploitation of
heterogeneous data, facts and scientific knowledge
to facilitate decisions and their application in the
field,
• The pilot will promote sustainable farming practices
through the provision of irrigation advices,
• Evapotranspiration monitoring is being explored in
order to provide useful information on the farm’s
water availability,
• The farmer will directly benefit from the provided
big-data technologies and advisory services by better
managing the natural resources.
1 pilot site (Kileler, Greece)
1 targeted crop (cotton)
1 advisory service (irrigation)
4 data sources
Data fusion
Advice generation and extrapolation
Decision Support
DATABIO PARTNERS INVOLVED
13. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
13
PILOT RESULTS
avg cost of irrigation
(euros/ha)
2670
2379
1881
0
500
1000
1500
2000
2500
3000
Cotton
2017 2018 2019
Pilot 6 [B1.2] Cereal, Biomass and Cotton Crops_2
0% 10% 20% 30% 40% 50% 60%
Cotton
Irrigation Cost Reduction
Target Measured
14. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
14
Pilot A2.1. Big data management in greenhouse ecosystems
15. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
15
GENOMIC PREDICTION & SELECTION DATA SCIENCE
The High-level Challenge in Genomic Selection:
Incorporating MAS for yield into practical breeding programmes.
For ~30 years of QTL mapping, this was not yet possible.
16. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
16
PRO G RESS IN BREED IN G RESPO N SE TO SELECTIO N
Genomic selection is a gold standard approach for
estimating breeding values
17. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
17
IT CAN BE INFERRED: ONLY 4 WAYS TO INCREASE RATE OF RESPONSE:
TARGET BREEDING METHOD
1. Increase variation Wide crosses / mutation
2. Increase precision Test more plots per line
3. Select harder Test more lines
4. Reduce time Out of season nurseries GS
In GS: Genomic prediction-driven intercrosses => shorter generation intervals
ΔR = ihσg/t
18. http://itema.cereteth.gr/
IOT, CROSSING BLOCKS, AND PHENOTYPING FACILITIES IN ITALY & GREECE (1/2)
Big data:
1) Phenotypes + biochemistry
2) IoT data => Glasshouse
environmental characteristics
19. 19
IOT, CROSSING BLOCKS, AND PHENOTYPING FACILITIES IN ITALY & GREECE (2/2)
BIG DATA:
1) Phenotypes +
Multi/Hyperspectral data
2) IoT data
20. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
20
HIGH-THROUGHPUT GENOTYPING: DDRADSEQ, GBS SNPS, WHOLE-GENOME RE-SEQUENCING
Nucleic Acids Info:
1. Genomic data
2. Transcriptomic data
21. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
21
Single/multi-trait & single/multi-environment predictive analytics
CV1: prediction of lines that have not
been evaluated in any glasshouse/field
trials
CV2: prediction of lines that have been
evaluated in some but NOT all target
environments (fields, glasshouses).
SEVERAL SCENARIOS OF BREEDING INTEREST ARE MODELLED
CV3: Montecarlo Cross-validation:
Repeated holdout (e.g., 70% TRN : 30%
TST)
22. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
22
GENOMIC SELECTION FOR HEALTH-PROMOTING PRODUCTS TO MANUFACTURE SPECIALTY FOODS (1/2)
Habyarimana et al., 2019
23. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
23
GEN O M IC SELECTIO N FO R HEALTH -PRO M O TIN G PRO D U CTS TO MAN U FACTU RE
SPECIALTY FO O DS (2/2)
Habyarimana et al., 2019
24. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
24
ACCURACY ACHIEVED IN VEGETABLES (CULTIVATED POTATO)
Habyarimana et al., 2017
Agronomic and trading traits in cultivated potato
25. IN OUR OWN EVALUATION: GENOMIC SELECTION VS. PHENOTYPIC SELECTION
25
Empiricial results (KPIs): GS can sustain plant breeding programmes,
allowing increased genetic gain per unit time and cost:
26. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
26
Pilot B1.3. Cereal and Biomass Crops
27. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
27
Research
challenges
27
Main challenge: Being able to use high-resolution satellite imageries to
predict sorghum biomass yields early within season, and with high precision
to avoid Stakeholders’ aversion.
Current yields forecasting approaches (background info): Field surveys,
Censuses, Coarser spatial (250-1000m) resolution satellites (e.g., MODIS,
SPOT-VEGETATION); all of which are undependable and/or costly.
Our project was therefore designed to address these shortcomings.
Research context: Biomass sorghum crop monitoring using fAPAR data
derived from Sentinel-2 satellite constellations.
Funds: This work was carried out in the framework of DataBio project
(2017-2019; www.databio.eu), EU H2020 research and innovation
programme.
28. BIG BIG DATA COLLECTION IN BIOMASS SORGHUM: REAL-TIME
STREAMS + NEAR REAL-TIME SATELLITE CONSTELLATION IMAGERIES
28
29. Results (1/3): We accurately predicted yields early in the cropping
season
• Meaningful Achievements: we accurately predicted yields up to 6 months ahead of harvest
30. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
30
Results 2/3: Our technology’s hallmark : Reliability, high accuracy
Model SMAPE
(%)
MAPE
(%)
MAE
(t ha-1
)
R2
LM 0.74 0.99 10.47 0.47
bartMachine 0.18 0.16 2.32 0.51
Bayesglm 0.74 0.98 10.34 0.48
xgbTree 0.44 0.36 4.07 0.62
Bayesian machine learning was the most powerful machine learning algorithm
(Habyarimana et al., 2019b)
SMAPE, MAPE, MAE, R2, symmetrical mean absolute percentage error, mean absolute percentage error, mean absolute error, and coefficient of determination, respectively.
LM, bart-Machine, bayesglm, xgbTree, respectively, simple linear model, Bayesian additive regres-sion trees (bartMachine method), Bayesian generalized linear model
(bayesglm method), and eXtreme Gradient boosting (xgbTree method).
30
31. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
31
Results 3/3: Towards remote fingerprinting of sorghum phenology and types
Cropping season: April – November 2017
fAPARindex:scale0-1
31
32. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
32
PILOT B1.4 CEREALS AND BIOMASS CROPS_4
33. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
33
PILOT B1.4 CEREALS AND BIOMASS CROPS_4, CZECH REPUBLIC
• Delineation of yield potential zones based on the time-
series analysis of EO data (Landsat 8, Sentinel-2) for whole
acreage of pilot farm (8,300 ha), updated every year
• Since spring 2019, the variable rate technology for
application of nitrogen fertilizers was implemented over
3,000 ha by addressing yield potential heterogeneity within
the fields.
• Base N fertilizing – before sowing of spring barley and maize
• 1st N application in top-dressign of winter cereals (winter wheat,
winter barley)
• testing of combination of yield potential zoning with current
crop status monitoring by Sentinel-2 for VRA of crop growth
regulators in spring barely
34. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
34
Pilot 10 [C1.1] Insurance
35. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
35
Systemic risk
detection through
agro-climatic
measurements
Vegetation-index continuous
monitoring at parcel level
Crop models generated
through statistical or ML
methodologies used as
baseline
Detection of anomalies and
Reporting after ~2 weeks
Pilot 10 [C1.1] Insurance (Greece)
PILOT SUMMARY
• EO-based pilot dedicated for the agriculture
insurance market that eliminates the need for on-
the-spot checks for damage assessment and
promotes rapid pay outs,
• Collaboration with INTERAMERICAN, a high profile
insurance company in Greece,
• Focus on annual crops (tomato, rice, maize,
cotton) and regions (Thessaly, Evros) with
significant economic footprint on the Greek agri-
food sector,
• Comes as a response to specific climate-related
systemic perils (flood, drought, high/low
temperature).
DATABIO PARTNERS INVOLVED
36. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
36
Pilot C2.1. CAP Support
37. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
37
New innovativeapproach
developed by
ROMANIA – CAP SUPPORT MONITORING
C developments
improvements v 4 y
Version v.05
Thisdocument is part of aproject that hasreceivedfunding
from the European ’ Horizon 2020 researchand innovationprogramme
under agreementNo 732064. It is the property ofthe DataBioconsortium and shall not be distributed or
reproduced without the formal approval of the DataBioManagementCommittee. Find usatwww.databio.eu.
3
7
38. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
38
Forthe 2018 and 2019 agricultural years,
Terrasigna also monitored the CAP
declarations for the entire agriculturalarea
of Romania.
Thetotal surveyed areaexceeds
9milion ha, corresponding tomore than
6milion plots of various sizesandshapes.
ThenecessaryEarth Observation (EO)data
required multiple Sentinel-2 scenes
projected in 2 UTMzones.
21%of the total number of plots within the
test areashave surfacesbelow 1ha.
Thisdocument is part of aproject that hasreceivedfunding
from the European ’ Horizon 2020 researchand innovationprogramme
under agreementNo 732064. It is the property ofthe DataBioconsortium and shall not be distributed or
reproduced without the formal approval of the DataBioManagementCommittee. Find usatwww.databio.eu.
3
8
39. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
39
Validation using reference
data
97.28% 32crops,
Based on 16.000+
plots, 60.000+ ha
Thisdocument is part of aproject that hasreceivedfunding
from the European ’ Horizon 2020 researchand innovationprogramme
under agreementNo 732064. It is the property ofthe DataBioconsortium and shall not be distributed or
reproduced without the formal approval of the DataBioManagementCommittee. Find usatwww.databio.eu.
3
9
40. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
40
Independent validationresults
(validated against VHRimagery,
Sentinel-2 and other data
sources)
98.3% correct estimations
for 8crop categories
Increased performance for
larger plots
T j v
E ’ H z v
7 4. I y b b
v M C . F . b . .
4
0
41. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
41
• H y , v
• S by v
• H z k
• F v
•
CONCLUSIONS & PERSPECTIVES
Thisdocument is part of aproject that hasreceivedfunding
from the European ’ Horizon 2020 researchand innovationprogramme
under agreementNo 732064. It is the property ofthe DataBioconsortium and shall not be distributed or
reproduced without the formal approval of the DataBioManagementCommittee. Find usatwww.databio.eu.
4
1
42. POTENTIAL:
Services at a Global
Level
42
Advisory Services
without border
Predictive, Descriptive, Recommendation Analytics
Visualization
[Historical + Current] data;
[Crop + Disease] models
Georeferenced (Commercial) Fields
43. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
reproduced without the formal approval of the DataBio Management Committee. Find us at www.databio.eu.
43
Thank you for your attention!
ephrem.habyarimana@crea.gov.it