An Open-Source Framework for Crop Type Mapping in Africa.
Food security in Africa is a serious issue. It is estimated that more than 250 million Africans suffer from undernourishment, with an additional 399 million people that are moderately food insecure. Fifty-eight million children under five exhibit stunted growth due to malnourishment. Earth Observation (EO) has been identified as an important enabler in addressing food security, both in Africa and across the world. While EO data is a powerful tool in support of improving food security, it is limited by the lack of a systematic, end-to-end approaches based on a scalable and sustainable open data infrastructure. This project leveraged the Digital Earth Africa (DE Africa) platform to create an end-to-end workflow to identify crop types from satellite data over central Zambia, driven by the African GEO community and stakeholders across the continent.
Project partners included Tetra Tech, Digital Earth Africa and the Regional Centre for Mapping of Resources for Development (RCMRD).
DE Africa is a continental-scale, not-for-profit initiative focused on improving access to EO across sectors in Africa. The solution used DE Africa data and platforms to produce an end-to-end workflow that could identify crop types from satellite data. The workflow was developed in partnership with the Regional Centre for Mapping of Resources for Development (RCMRD) for a priority use case focused on crop type mapping in Zambia.
The workflow allows users to:
- Develop their own field sampling approach
- Collect ground truth data using the newly developed Enabling Crop Analytics at Scale (ECAAS) ODK toolkit
- Upload ground truth data directly into the DE Africa platform
- Explore features of importance for crop type separation
- Run and assess the accuracy of machine learning algorithms to create crop type maps for their area of interest
This framework can be easily adapted, built upon and is open and freely available to all users. The workflow is well documented through a series of Jupyter notebooks that will be used as part of capacity development efforts. Universal data inputs layers useful for crop type mapping were identified, and DE Africa will work to provide these as continental-scale services, allowing users to scale and extend the workflow to other areas.
The project created an open-source framework that allows for future innovation in the development of EO solutions for increasing food security, which can be extended and scaled to other regions in Africa. In partnership with AGRYHMET, one of the DE Africa Implementing Partners, extension of the workflow into Niger is being tested.
Contact us at email@example.com or Technical Lead Caitlin Adams, firstname.lastname@example.org, to find out more.