Financing strategies for adaptation. Presentation for CANCC
Ukraine Crisis Webinar Series Session - V
1. Predicting Food Crop Production in Times of Crisis: The Case of Wheat in South Africa
Presented by
Dr. Greenwell Matchaya
Senior Researcher, ReSAKSS ESA Regional Coordinator,
IWMI/Pretoria
Based on Racine Ly, Greenwell Matchaya, and Khadim Dia, 2022. 17 Jan 2023
3. • With the numerous challenges related to agricultural trade that African countries
currently face,
• It is crucial for policymakers to be aware of potential/impending food production
disruptions
• Estimating agricultural production with greater precision is also generally important for
intervention planning
• Russia and Ukraine are key exporters of many agricultural products, including
sunflower oil and seed, wheat, barley, rapeseed and maize.
• Jointly, the two countries account for 27% of global wheat trade, and 23%, and 14% of
barley and maize trade respectively
• Furthermore, the two countries account for over 28% of the world’s nitrogen,
potassium and phosphorous fertilizer exports
Introduction
4. • The Russia-Ukraine war has therefore destabilized global food and other
agricultural value chains,
• This may worsen the longer and more intensely the war is fought.
• As net importers of both fertilizers and wheat, African countries are already
experiencing
• a rise in the prices of these commodities as well as their substitutes.
• Unfortunately, the fertilizer and wheat price increases will negatively impact
agricultural production in the
• current and coming seasons.
• Many more households may therefore need support in order to survive the resulting
and inevitable food price hikes.
5. • The focus on wheat is useful. Wheat is the second most important crop in South Africa
after corn.
• South Africa produces around 2Million MT of wheat and imports a further 1.7Million
MT.
• Consumption demand for wheat in South Africa is upward of 3.6 million MT annually
• The sales value is ~10% (R19 Billion)of total sales of all field crops
• So it’s a significant crop
• Importance of Accurate and Timely Statistics and the role of AI/Machine
Learning
• Under these conditions, more accurate and timely statistics on crop production are
critical
• To know how much production will be available nationally
• Which areas will produce less. This can facilitate development of interventions to
save livelihoods
6. Objectives
The objectives of this brief were:
• To forecast the quantity of wheat production in South Africa for the 2022
harvest
• To forecast the spatial distribution of wheat in South Africa in the 2022
harvest season and provide insights into implications
• Nevertheless, for a country with limited resources, and when time is of the essence,
• survey based methods for estimates may be inappropriate, infeasible or inefficient
• Using satellite data and artificial Intelligence and more specifically machine Learning can be useful in
these cases
• This method can also reduce the time and costs needed for calculating production levels from surveys
and field visits.
• Further, the inherent precision of the techniques contributes to the availability of better quality data
7. • Used the Africa Crop Production (AfCP) model to predict future production
• The model uses satellite remote sensing data as explanatory variables and machine
learning techniques as a predictive modelling framework,
• To provide production quantities before the harvesting period at the pixel level
• The remote sensing data makes it possible to uniquely examine different earthly objects
without having to see them physically.
• Remote sensing also enables the production of more extensive and better-quality data
over a shorter period.
• Machine learning makes it possible to extract the many hidden features in the vast
amount of remotely sensed data and make sense out of it.
Methodology
8. • The 2022 predictions are in line with expected
patterns- more production in Western cape,
Free State
• KZN production trails that of the WC and Free
State
• Northern Cape and North Western Provinces are
generally not heavily involved in wheat
production
Results
9. • Nationally, a total of around 1.8 million Mt
predicted for 2022
• Down from the 2Million realized in 2021
• Western Cape and Free State registered an
increase
• However all other provinces registered a
reduction
• More reduction in Eastern Cape, KwaZulu
Natal, and Mpumalanga
Production by province
10. • Within the North West Province,
there was a general decline
• the largest decline was in Dr Ruth
Mompati municipality (22%),
followed by Bojanala (20%.)
• In contrast, Gaetsewe
municipality registered the most
decline (42%) in Northern Cape,
but Siyanda and Frances Baard
municipalities saw an increase.
• The central Karoo saw the largest
decline (32%) in Western Cape,
but the West Coast, Eden and
Overberg saw an increase
Within Provinces Province
District
Municipality
2021
wheat
producti
on (MT)
2022
wheat
producti
on (MT)
production ratio (2022/2021)
North West
Ngaka Modiri
Molema
64049,8 58309,1 0,91
North West
Dr Ruth Segomotsi
Mompati
38682,3 29940,8 0,77
North West Bojanala 32198,2 25634 0,8
North West Dr Kenneth Kaunda 29259,2 24554 0,84
Northern Cape Namakwa 84751,4 91938,9 1,08
Northern Cape Pixley ka Seme 55615,9 53326,1 0,96
Northern Cape Frances Baard 35520,2 41222,2 1,16
Northern Cape Siyanda 32406,2 36406,8 1,12
Northern Cape
John Taolo
Gaetsewe
3502,19 2021,87 0,58
Western Cape West Coast 271847 308653 1,14
Western Cape Overberg 109065 115529 1,06
Western Cape Cape Winelands 94265,4 85324,7 0,91
Western Cape Eden 76451,9 76959,5 1,01
Western Cape City of Cape Town 10565 9068,85 0,86
Western Cape Central Karoo 7571,02 5118,48 0,68
11. • The Free state generally saw an
increase in production although
the Xhariep municipality
registered a12% decline
• The Eastern Cape however saw
the largest decline. The Buffalo
city saw a 58% decline while the
rest declined by at least 49%.
• In Gauteng, the largest decline
was in the City of Johannesburg,
while the least decline was sin
Sedibeng municipality
Province
District
Municipality
2021 wheat
production (MT)
2022 wheat production (MT)
Wheat
production ratio
(2022/2021)
Eastern Cape Cacadu 44656,11 22876,1 0,51
Eastern Cape Chris Hani 43860,85 22425,9 0,51
Eastern Cape Joe Gqabi 33596,64 16896,1 0,5
Eastern Cape Amathole 30409,15 13911,8 0,46
Eastern Cape O.R. Tambo 17688,94 8395,26 0,47
Eastern Cape Alfred Nzo 15374,96 7889,43 0,51
Eastern Cape Buffalo City 4063,81 1723,22 0,42
Eastern Cape
Nelson
Mandela Bay
2115,19 1001,43 0,47
Free State
Thabo
Mofutsanyane
170240,53 182930 1,07
Free State Lejweleputswa 130248,06 136636 1,05
Free State Fezile Dabi 110797,8 120263 1,09
Free State Xhariep 86528,24 76167,1 0,88
Free State Mangaung 25708,98 26876,7 1,05
Gauteng
City of
Tshwane
8698,8 6089,84 0,7
Gauteng West Rand 6597,22 4833,72 0,73
Gauteng Sedibeng 5220,8 4193,79 0,8
Gauteng Ekurhuleni 4694,26 3237,05 0,69
Gauteng
City of
Johannesburg
1661,03 897,74 0,54
12. • All municipalities in KwaZulu
Natal, Limpopo and
Mpumalanga registered
declines in wheat production
when compared to 2021
• The worst decline in KZN was in
Umkhanyakude (49%)
• Mpumalanga’s Ehlanzeni
municipality (40%), and
Limpopo’s Vhembe (31%) were
some of the municipalities with
the worst decline
Province District Municipality
2021
wheat
productio
n (MT)
2022 wheat
production (MT)
Wheat production ratio
(2022/2021)
KwaZulu-Natal Zululand 21452,2 12423,4 0,58
KwaZulu-Natal Sisonke 15850,9 9875,3 0,62
KwaZulu-Natal Uthukela 14652,4 9043,82 0,62
KwaZulu-Natal Umgungundlovu 14227,6 8719,67 0,61
KwaZulu-Natal Umkhanyakude 15511,2 7900,99 0,51
KwaZulu-Natal Umzinyathi 12668,3 7343,12 0,58
KwaZulu-Natal Uthungulu 11022,2 6011,76 0,55
KwaZulu-Natal Amajuba 9862,35 5829,75 0,59
KwaZulu-Natal Ugu 8750,5 4850,64 0,55
KwaZulu-Natal iLembe 4959,81 2905,44 0,59
KwaZulu-Natal eThekwini 2323,38 1261,28 0,54
Limpopo Waterberg 64804,9 56212,1 0,87
Limpopo Sekhukhune 23657,9 19122,5 0,81
Limpopo Capricorn 22824,1 19061,3 0,84
Limpopo Mopani 19843,3 15202,9 0,77
Limpopo Vhembe 15568,7 10714,3 0,69
Mpumalanga Gert Sibande 48711 32487 0,67
Mpumalanga Nkangala 25720,1 18000,5 0,7
Mpumalanga Ehlanzeni 21307,3 12772,2 0,6
13. • Huge positive and negative
differences in rainfall patterns for
example <-20 and >20 mm
appear to correlate negatively
with wheat production .
• Larger anomalies observed in
North West, Kwazulu Natal,
Limpopo, parts of Western
Cape. Many of these have low
production
Relationships with biophysical elements
Rainfall Anomalies and Wheat
production patterns
14. • Huge anomalies observed in
parts of Limpopo,
Mpumalanga, Western Cape
(Rising temperature)
• Temperatures have deviated
negatively over parts of North
West, Northern Cape
• Some of the spatial variations
in wheat production may be
explained by this spatial
variation in LST anomalies
Land Surface Temperature Anomalies and Wheat production
15. • Modest changes in NDVI
appear to favor better wheat
production
• Areas of the Northern Cape
and the North West have larger
NDVI and are also those where
wheat production is weakest
Normalized difference vegetation index (NDVI) Anomalies and Predicted Wheat Production
16. Key Messages
• The Russia-Ukraine war presents a challenge to global food security and
household resilience
• Especially in those countries that depend on international trade for
agricultural inputs and food.
• Predicting future agricultural production is critical for anticipating and crafting
timely interventions that may limit the negative effects emanating from the war.
• Remote sensing and artificial intelligence techniques can play a key role in
provide data needed for decision making, timeously
• Using these techniques, it was possible to predict the impending production
decline in wheat production for South Africa long before its harvest.
• The predictions are consistent with actual production ranges observed
17. • Equipped with such information, policy-makers can leverage on such information to
develop mechanisms that increase consumer access to local production
• Which can minimize the threat emanating from the disruption of global wheat
supply chains for those households in areas with declining production levels.
• This information can also motivate for planted area expansion and can make the
government and stakeholders interested in food security, more prepare.
• It may be useful to encourage farmers to expand areas planted to wheat in the next
season to limit the extent of future global price increases
• Good agricultural water management can also boost production in water scarce
parts of South Africa.
• This would dampen the observed negative correlations between rainfall anomalies
and predicted wheat production.
18. • Another input affected by the war is inorganic fertilizer.
• This will continue to be a limiting factor in wheat production as long as value
chain disruption persist
• To develop food system resilience, agricultural research in South Africa should
consider placing further emphasis on wheat yield improvement programs that
aim to produce wheat varieties with low input requirements
• Targeting low fertilizer and water requirements in breeding will reduce the
input costs in wheat production while also increasing production, productivity,
and profitability.