High resolution, annual maps of the characteristics of smallholder-dominated croplands at national scales (original) (raw)

Mapping the changing characteristics of Africa's smallholder-dominated agricultural systems, including the sizes and numbers of fields, is crucial for understanding food security and a range of other socioeconomic and environmental concerns. However, accurately mapping these systems is difficult because of 1) the spatial and temporal mismatch between satellite sensors and smallholder fields, and 2) the lack of high-quality labels needed to train and assess machine learning classifiers. We developed an approach designed to address these two problems, which we used to map Ghana's annual croplands for the year 2018. To overcome the first problem, we converted daily, high resolution CubeSat (PlanetScope) imagery into two cloud-free seasonal composites covering a single agricultural year. To address the second problem, we created a labelling platform that rigorously assesses and minimizes label error, and used it to iteratively train a Random Forests classifier with active learning, which identifies the most informative training sample based on prediction uncertainty. Minimizing label errors improved model F1 scores by up to 25%. Active learning increased F1 scores by an average of 9.1% between first and last training iterations, and 2.3% more than models trained with randomly selected labels. We used the resulting 3.7 m map of cropland probabilities within a segmentation algorithm to delineate crop field boundaries. Based on an independent map reference sample (n=1,207), the cropland probability and field boundary maps have respective overall accuracies of 88% and 86.7%, user's accuracies for the cropland class of 61.2% and 78.9%, and producer's accuracies of 67.3% and 58.2%. Using the map reference sample to calculate an unbiased area estimate from the field boundary map, we found that cropland covers 17.1% (15.4-18.9%) of Ghana. Using the most accurately digitized labels to calculate and correct for biases in the segmented field boundaries map, we further estimated the average size (1.73 ha) and total number (1,662,281) of crop fields in Ghana. Our results demonstrate an adaptable and transferrable approach for mapping the characteristics of croplands on an annual basis and over national extents, with several features that effectively mitigate the errors inherent in remote sensing of smallholder-dominated agriculture.