Study Reveals Varying Accuracy of Cropland Maps in Sub-Saharan Africa

A recent study has evaluated and compared 11 publicly available land cover maps to assess their suitability for cropland classification and Earth observation (EO) based agriculture monitoring in Africa. The study, which encompassed a range of temporal availability, spatial resolutions, and classification approaches, aimed to help users determine the most suitable map for their needs and encourage future work to focus on resolving inconsistencies between maps and improving accuracy in low-accuracy regions.

Why this matters: Accurate cropland maps are crucial for informing agricultural policies, development investments, and food security monitoring in Africa, where half of the low-income and food-deficient countries are located. Improving the accuracy of these maps can have a significant impact on the lives of millions of people, particularly in regions wherefood security is a major concern.

Africa is a critical area for research on food security, with half of the low-income and food-deficient countries located on the continent. Satellite Earth observations provide an affordable, reliable, and timely source of information for assessing crop conditions and food production. However, EO-based monitoring systems require accurate cropland masks, which are typically derived from land cover maps.

The study selected 11 land cover maps that encompass a range of temporal availability (2009-2020), spatial resolutions (10 m px to 1000 m px), and classification approaches (tree-based to deep learning methods). The maps were evaluated using statistically rigorous reference datasets from 8 Sub-Saharan African countries.

The study found low consensus and varying performance metrics among the 11 cropland maps. WorldCover and Digital Earth Africa ranked highest in overall performance, with accuracy and precision scores correlating with spatial resolution and temporal mismatch.

The study's results will help users determine the most suitable map for their needs and encourage future work to focus on resolving inconsistencies between maps and improving accuracy in low-accuracy regions. This is crucial for informing agricultural policies, development investments, food and nutrition security monitoring, andclimate modelingin Africa.

The study compared 11 publicly available land cover maps across a range of temporal availability, spatial resolutions, and classification approaches, using reference datasets from 8 Sub-Saharan African countries. WorldCover and Digital Earth Africa ranked highest in overall performance, highlighting the importance of spatial resolution and temporal alignment in determining the accuracy of cropland maps for Earth observation based agriculture monitoring in Africa.