As agriculture shifts from subsistence to commercial levels in Africa, UAVs will increasingly provide the granularity and agility required to monitor agricultural operations and performance in smallholder conditions, and hence the ability to provide advisories that capitalize on heterogeneity, rather than avoid it. The International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), along with other world-leading institutions is heralding this new era of decentralized information streams in Mali and Nigeria, generating unprecedented datasets that reveal the enormous variability encountered in African cropping systems. UAV technology will help decrease the cost of crop performance monitoring over space and time. Embedded into mobile data streams, imagery will assist agricultural value chain actors optimize their businesses and support the development of technologically advanced rural jobs attractive for the youth.

1. UAVs – Granular Remote Sensing
Solutions to Leverage Smallholder
Agriculture
A. Tukur, H.A. Ajeigbe, P.C.S. Traore (ICRISAT),
M. Badamasi (BUK/CDA), I.Y. Tudunwada
(NASRDA), D. Annerose (MANOBI)

2. STARS aims at learning, identifying opportunities and
challenges, and testing hypotheses around the
potential exploitation of RS technology in improving
the productivity of crop-based smallholder
production systems and the livelihoods of
smallholder farmers in SSA and SA.

3. the ISABELA
concept
Hypothesis 1: lack of transparent
land tenure information systems = key deterrent
to sustainable investment in land resources by
smallholders, communities and the local private
sector. Disempowers them in both their current
internal transactions and in transactions with
urban and international investors
Hypothesis 2: inability to cost-effectively recognize cultivated species in smallholder fields
significantly protracts the prediction and valuation of seasonal agricultural
production, prevents equitable access to physical and financial inputs and
markets by smallholders and agro-dealers
Value Proposition 1: provide a sustainable,
subscription-based rural land tenure
information service supported by very high-
resolution satellite imagery
Value Proposition 2: develop digital libraries and algorithms for
smallholder crop recognition at scale

4. ISABELA
research
sites
 2 contrasted dryland agriculture transects with intensive study sites at
Kofa (Bebeji LGA, Kano State, Nigeria) and Sukumba (Koutiala District,
Sikasso Region, Mali)
 Very heterogeneous smallholder production systems

5. Sukumba, Mali
Average field size of
1.45 ha with 95%
pure crops, but
estimated 30% of
field boundaries
change every year
and many trees in
fields
The problem: smallholder systems are
(very) heterogeneous…
Kofa, Nigeria
Fairly stable
‘boundaries’, but
average field size of
0.22 ha, with only
5% pure crops (50%
of fields have 3
crops or more at
any time)
3km3km

6.  Above: proportion of cloud-free satellite images retrieved for one of the STARS study
sites in 2014. The decrease in frequency during the Jul-Sep period is conspicuous
 Crop growth and management are very time-sensitive, non-linear processes
 2 systems deployed in STARS-ISABELA:
 SenseFly eBee multispectral data (NIR/R/G): 6-10 overpasses @ 10cm, on-farm
 GEO-X8000 Tetracam 6-band: two weekly passes, on-station
Why UAVs?

8.  Support precision agriculturalists
(smallholder farmers) with very high
resolution information derived from
satellite and UAVs
 Create conditions for the emergence of
imagery value chains
Specific objectives :
1. Assess crop condition using multi-temporal
UAV and satellite data across the cropping
season
2. Delineate parcels and generate land tenure
information
3. Prototype mobile agricultural advisory
services supported by imagery
Objectives

9. In situ data
Since 2014:
Yearly land use /
land cover using
JECAM protocols
(5,000+ fields for
Nigeria)
Yearly above-ground
biomass and grain
yield (ML: 1,350
quadrats; NG: 615
quadrats)
 Bi-weekly crop development
& growth (same quadrats as above)
 Hourly Tmin, Tmax, RH, SRAD from 1 auto.
weather station at Bebeji distric HQs
 + socio-economic data through VP1, matched
with parcel databases (270,000+ parcels for Bebeji LGA,
Nigeria and 60,000+ for Molobala sub-division, Mali)
In-situ data
collection

10. Mali example (2014) -
plant growth
ground measurements
48 fields over 30km2
x 5/6 fertility plots
x 5 quadrats
x 5 plants
x 3 variables
x 3 visits
59,400 records
Variables measured
Biweekly:
light interception / LAI
f-cover
plant height
chlorophyll content
BBCH devpt. stage
End-of-season:
FW, DW for veg. & repr.
unprecedented datasets in
the making

11. UAV
operations
 70% Lateral
Overlap
 75% Longitudinal
Overlap
 286m ATO for
10cm Resolution

12. UAV image
processing
geotagging
alignment
mosaic

13.  NDVI (DG)
 Plant height (UAV)
Peanut (R2=0.92)
Cotton (R2=0.51)
Maize (R2=0.15)
Millet (R2=0.85)
Sorghum (R2=0.62)
 Robust relationships between NDVI and f-cover for
DG (R2=0.8) and UAV (R2=0.5), but species-dependent
 Potential of NDVI for biomass estimation, and to a
lesser extent for crop type mapping improved through
environmental stratification
 Biomass maps reveal variability within species and
within catena class from simple to double or more
UAV results
(Mali)

14. From parcel delineation to
mobile advisory services

15. Learnings and conclusion

16. AMEDD
B. Sogoba, O. Dembele, S. Coulibaly,
G. Dembele, B. Sissoko, D. Sanou, O.
Diabate, N. Dembele, N. Dembele
GERSDA M. Djire
UCL
P. Defourny, G. Chome, X. Blaes
WUR
B. Boekelo, J. Davidse, F. de
Schaetzen, W. van Ommeren, A.G.T.
Schut
ITC
R. de By, L. Eelderink, M.H. Stroeven,
J.W.F. Timmerman
NASRDA
I.Y. Tudunwada, S. Nannim, G. El-
Hassan, R. O. Hyat,R. D. Mohammed,
A. Abubakar
CDA
I.Y. Badamasi, S. Momale, A.I.
Muhammad, I. A. Na Abdu, Y. S.
Kurawa
IER
K.B. Traore, S. Toure, S. Guindo, K. Ba
ICRISAT
H.A. Ajeigbe, A. Whitbread, M. B. Vabi, I.
I. Angarawai, A. H. Inuwa, A. Kunihya, A.
Adinoyi, J. Jonah, H. Peter, O.M. Ndiaye,
F. Sagounta, I. Kassogue, G. Poda, M.
Diancoumba, T. Dembele, K. Toure, E.
Niare
MANOBI
D. Annerose, C. Champenois
UdeS A. Safia, K. Goita
District and village authorities and producers in
Kofa (Bebeji, Kano, Nigeria)
Sukumba (Koutiala, Sikasso, Mali)
BMGF
K. Schneider, S. Wood
acknowledgements