In Cameroon, as well as other major cocoa producing countries in West and Central Africa, cocoa production is an important land use and livelihood strategy. The growing global demand of cocoa beans and by-products has sparked intensive and expansive land use strategies to increase cocoa production and related national export earnings. Thus, remnant tropical forests are threatened by annual, yet unquantified, deforestation due to cocoa farm expansion. In these countries, cocoa (Theobroma cacao) farms constitute plantation systems that range from full-sun to various magnitudes of multi-strata canopy trees. Unfortunately, the changing climate (or climate crisis) has reported negative impacts on cocoa production, and this is expected expand with predicted extreme dry seasons.
In cocoa agroforests with stratified canopy structure, appropriate shade management is important to ensure resilience to changing climates. Yet, information on the spatial and temporal variation in shade cover, which can guide farm management, is largely scarce.
Air- or space-borne remotes sensing (RS) data, unlike in-situ point estimates, provide relatively efficient and large-scale assessment of vegetation structure. Synthetic Aperture Radar (SAR) data, unlike multi-spectral optical RS data, provide all-weather, day and night images of target ground features; such data are, thus, especially reliable for environmental monitoring in tropical regions.
This is a presentation of thesis (research) that provides new contribution on the application of Sentinel-1A SAR data for delineating and mapping cocoa agroforests in landscapes with heterogeneous vegetation. We applied the grey level co-occurrence matrix (GLCM) textures for discriminating different vegetation types. We estimated the spatial variation of canopy closure in cocoa production landscapes. The spatial quantification of cocoa agroforests is vital for the sustainable management of remnant forests and multi-use cocoa production landscapes. Information as such will better inform management decisions and support climate change mitigation mechanism REDD+ implementation.
Spatial Discrimination of Land Uses in Multi-use Tropical Landscapes
1. FORSIT
Towards optimized spatial quantification of cocoa agroforests in
multi-use tropical landscapes: an application in Cameroon
Frederick Nkeumoe Numbisi
19th November, 2019
2. How manytrees for a chocolate
fix?
(Ake Mamo and Philippe Vaast, 2014 – ICRAF Blog)
Cocoa production - a source of income and
means to send me to school!!!
Why is the cocoa tree (Theobroma cacao) an
important perennialcrop?
2
3. Data Source: FAOSTAT 2019 (Accessed on Nov 4th, 2019)
60-70% of global cocoa bean is
produced in tropical sub-Saharan Africa
3
4. 60-70% of global cocoa bean is
produced in tropical sub-Saharan Africa
Data Source: FAOSTAT data 2019 (Accessed on Nov. 4th, 2019)
4
6. Schroth et al ., 2016, Vulnerability to climate change of cocoa in West Africa
Southward retraction of climatically suitable
areas is predicted by 2050
A lot of climatically unsuitable (<20%)
areas across the belt except for Cameroon.
Savannah transition areas increasingly
vulnerable
What prospects for cocoa production in the
African belt?
6
7. Increasing rate of expansion of cocoa harvest
area in Cameroon?
Data Source: FAOSTAT data 2019 (Accessed on Nov. 4th, 2019)
7
Cameroon:
Increasing cocoa harvest
area
8. What is the contribution of farm expansion to
production records?
Data Source: FAOSTAT data 2019 (Accessed on Nov. 4th, 2019)
Cameroon:
Increasing cocoa harvest area
8
9. What is the actual area of cocoa land on the
ground?
Data Source: FAOSTAT 2019 (Accessed on Nov 4th, 2019)
9
10. Diverse tree species
retained or planted
Similar canopy to
transition forests
Unsuccessfuluse of
multi-spectral data
(Ordway et al., 2017)
10
How to map cocoa agroforests from transition
forests?
Challenges
11. Ce – Ceiba pentandra
Ir – Milicia excelca
Ci – Citrus spp
Mu – Musa spp
Co – Theobroma cacao
Ta – Erythrophleum suaveolens
11
Which horizontal pattern of tree/crop to adopt?
Challenges
12. Land tenure constraints
Diversifying farm income and
source of nutrition
Changing microclimate –
related
production/investment
losses
12
How are cocoa farmers adapting to changing climates?
Challenges
13. 13
What was our focus? Changes in the configuration of cocoa
trees
14. What was our focus? Delineating cocoa agroforests
14
Ground truth data
28. Canopy gap fraction prediction
Spatialcorrelation
Random Forest
regression
Spatialcorrelation
Canopy gap
maps
SAR
backscatter
Ground
truth data
Ground
truth data
SAR
backscatter
28
Random Forest
regression
Mapping canopy closure
30. Results
long scale (51 m)
spatial correlation
Forest Landscape
30
Mapping canopy closure
31. 31
Main findings
Cocoa trees in agroforests were managed at much smaller spatial
distances, in a random pattern, and without regard of distances to
non-cocoa trees
Texture images from multi-seasonal Sentinel-1A SAR backscatter
provided reliable information to discriminate cocoa agroforests from
transition forests
Canopy gap fraction was spatially correlation at scales that reflect the
distribution of dominant canopy trees in the cocoa production
landscapes