https://geoserver.iiasa.ac.at/geoserver/transparent-monitoring/wms?tiled=true

This map is widely considered to be the most accurate land cover map of CDI, and hence the most accurate cocoa map.

The Cote d'Ivoire BNETD 2020 Land Cover Map was produced by the Ivorian Government through a national institution, the Center for Geographic Information and Digital from the National Study Office Techniques and Development (BNETD-CIGN), with technical and financial support from the European Union. The methodology used to produce the map was transparent, participatory and in line with international standards.

Cocoa 2020 (BNEDT)

Cocoa farming maps for Côte d'Ivoire and Ghana at a resolution of 10m. The maps are based on Sentinel-2 images taken between January 2019 and December 2021, which are interpreted with an ensemble of convolutional neural networks (CNN), each trained on a large corpus of geo-referenced ground truth cocoa farms and non-cocoa sites, including more than 100,000 farms. Each model predicts a binary cocoa map and the final map is computed by taking the average, indicating the degree of agreement among the models. Additionally, we compute the best threshold at 0.65 according to a held-out validation set, i.e. all values above 65 % in the probability map are classified as cocoa and all others as non-cocoa areas.

https://nk.users.earthengine.app/view/cocoa-map

Cocoa 2019-2021 (ETH)

This map appears to differ from all 3 other maps in terms of spatial distribution, with an almost uniform distribution of cocoa present across the growing region.

https://www.sciencedirect.com/science/article/pii/S1470160X21005288#ec-research-data

Côte d'Ivoire and Ghana are the largest producers of cocoa in the world. In recent decades the cultivation of this crop has led to the loss of vast tracts of forest areas in both countries. Efficient and accurate methods for remotely identifying cocoa plantations are essential to the implementation of sustainable cocoa practices and for the periodic and effective monitoring of forests. In this study, a method for cocoa plantation identification was developed based on a multi-temporal stack of Sentinel-1 and Sentinel-2 images and a multi-feature Random Forest (RF) algorithm. The Normalized Difference Vegetation Index (NDVI) and second-order texture features were assessed for their importance in an RF classification, and their optimal combination was used as input variables for the RF model to identify cocoa plantations in both countries. The RF model-based cocoa map achieved 82.89% producer's and 62.22% user's accuracy, detecting 3.69 million hectares (Mha) and 2.15 Mha of cocoa plantations for Côte d'Ivoire and Ghana, respectively. The results demonstrate that a combination of an RF model and multi-feature classification can distinguish cocoa plantations from other land cover/use, effectively reducing feature dimensions and improving classification efficiency. The results also highlight that cocoa farms largely encroach into protected areas (PAs), as 20% of the detected cocoa plantation area is located in PAs, and almost 70% of the PAs in the study area house cocoa plantations.

https://doi.pangaea.de/10.1594/PANGAEA.917473

Cocoa 2019 (JMU, JRC, SU)

McKinsey acquired Vivid Economics in March 2021. The map is no longer available.

Object-based classification using the following data sources: Sentinel 1/2. Eight land use categories were retained. Cocoa was then extracted from the map for visualisation here. No field-based validation has taken place.

This map was acquired through direct contact with Vivid Economics. 

Cocoa 2018-2019 (Vivid Economics)

Cocoa maps

https://ies-ows.jrc.ec.europa.eu/iforce/gfc2020/wms.py

Bourgoin, C., Ameztoy, I., Verhegghen, A., Desclée, B., Carboni, S., Bastin, J., Beuchle, R., Brink, A., Defourny, P., Delhez, B., Fritz, S., Gond, V., Herold, M., Lamarche, C., Mansuy, N., Mollicone, D., Oom, D., Peedell, S., San-Miguel, J., Colditz, R. and Achard, F., Mapping Global Forest Cover of the Year 2020 to Support the EU Regulation on Deforestation-free Supply Chains, Publications Office of the European Union, Luxembourg, 2024, doi:10.2760/262532, JRC136960.

Global forest maps are produced from a combination of existing global spatial layers, e.g. on land cover, land use and tree height. The map on global forest cover (GFC 2020) aims at representing the state of standing forest cover by 31 December 2020. The global land cover from the ESA World Cover project serves as one baseline layer to define the extent of tree cover for year 2020 at 10 m spatial resolution. The meta data and approach used to produce the GFC 2020 map are available in the JRC data catalogue (JRC, 2024) and described in a technical report released in March 2024 (Bourgoin et al 2024). A comparative study between GFC 2020 and a national map was carried out over Ivory Coast (Verhegghen et al, 2024).The GFC 2020 map may be revised if new information, additional large-scale data layers or revised global spatial data layers will be made available for year 2020. The viewer shows both the GFC 2020 map and a preliminary version of the global map of forest types for year 2020 (GFT 2020), in line with the definition of forest degradation as set out in the EUDR. A consolidated version of GFT 2020 map will be released in 2025. Global maps of forest cover and forest types for year 2020 (GFC 2020 and GFT 2020) have no legal value per se but may serve as a tool to comply with the Regulation on deforestation-free supply chains. More specifically, it may help to assess the risk of deforestation (GFC 2020) ) or the risk of forest degradation (GFT 2020), without dispensing operators and Member States from the duty to carry out due diligence.

The global map of forest cover was created by combining available global datasets (wall-to-wall or global in their scope) on tree cover, tree height, land cover and land use into a single harmonized globally-consistent representation of where forests existed in 2020.



https://forest-observatory.ec.europa.eu/forest/rmap