Digital Earth Africa National Land Cover and Crop Type Mapping

FrontierSI has been collaborating with the Food and Agriculture Organisation (FAO) and Digital Earth Africa (DE Africa) to develop a prototype automated Land Cover and Crop Mapping workflow to assist the Hand-In-Hand (HIH) Initiative.

The Office of the Chief Statistician at the FAO is leading the initiative, and its goal is to support the implementation of nationally-led programmes to accelerate the eradication of poverty, end hunger and malnutrition, and reduce inequalities. It aims to meet this goal by utilising existing and novel approaches, data sources, platforms, and analytical tools to drive agricultural innovation.

Land cover maps can be used to monitor the environment, identify land degradation trends, plan urban development, and more. To assess changes over time, land cover data must be produced using standardised methods and validation at regular intervals.

One of the challenges to achieving these HIH goals is the lack of available data, including limited resources to collect, analyse, and disseminate agriculture statistics regularly. A viable and effective solution is the use of Earth Observation (EO) to assist in the collection of timely and consistent data at local, regional and national scales. With this potential comes challenges such as data access, storage, pre-processing and analysis of large temporal datasets. Data requirements, for example, such as the size of imagery particularly when conducting large-scale national mapping exercises, may turn potential EO data users away.

DE Africa is a platform that provides reliable and timely access to analysis-ready EO data to support sustainable development. It removes the burden of data storage and pre-processing from users, provides a platform to perform intensive cloud processing, utilises pre-prepared analysis workflows, as well as supporting users to modify or develop workflows for their own needs. It hosts the Landsat and Sentinel surface reflectance time series datasets, Landsat surface temperature, elevation models, and a range of land and water products such as the Water Observations from Space and fractional cover maps.

To assist the FAO and the HIH initiative, FrontierSI has developed an adaptable land cover mapping workflow and tested it for two countries in Africa, Rwanda and Mozambique, taking input from users and adjusting the method to suit the environment and application needs of in-country users. The workflow is presented in Juypter notebooks that are designed to be run within the DE Africa Sandbox, a Python JupyterLab development environment. Upon project completion, the notebooks will be made freely available.

A critical component of the project is to build capacity in the selected countries to use and adapt the workflows. This is achieved on two levels, through webinars to raise awareness of the capability of EO data to support land cover and crop mapping and through technical workshops.

The first of the training sessions was held in November 2022. FAO introduced the project, DE Africa provided an overview of their platform, and FrontierSI summarised the Land Cover Mapping workflow developed thus far. The webinar was aimed at the end-users in Mozambique and Rwanda and was followed by an introductory session on the DE Africa platform run by the in-country DE Africa team. A Land Cover Mapping workshop focusing on the Mozambique workflow is scheduled to be held in December 2022. In this workshop participants will work through the training notebooks in an interactive environment to learn the machine-learning workflow for land cover using the DE Africa Sandbox. Further training on the workflow is planned for 2023, including a Rwanda-specific Land Cover Mapping Workshop.

This work was recently presented at the African Association of Remote Sensing of the Environment International Conference held in Rwanda. A presentation was given on behalf of the project team by Kenneth Mubea, Capacity Development Lead, and Joseph Tuyishimire, User Engagement Manager at DE Africa. The conference was a great success. Keep an eye out for the full conference proceedings published in the Earth Observations and Geospatial Science in Service of Sustainable Development Goals journal to read more about this work, or click the link above.

The project is now commencing its second phase that will explore the development of crop boundary and crop type mapping as well as crop yield estimation. Each of these components play an important role in reaching the HIH initiative goals and will assist in increasing the amount of timely and repeatable agriculture data and statistics in Africa. These workflows will again be designed within the DE Africa Sandbox environment and will be made available freely on completion.

Further capacity-building activities will be conducted toward the end of the project in 2023 via webinars and online technical workshops to assist end users in learning and adapting the crop mapping workflows for their specific needs.

To stay informed on this critically relevant project, or to learn more, send us a note at and let’s start the conversation.