A new cropland area database by country circa 2020

We describe a new dataset of cropland area circa the year 2020, with global coverage and with data for 221 countries and territories and 34 regional aggregates. Data are generated from geospatial information on the agreement-disagreement characteristics of six open-access high-resolution cropland maps derived from remote sensing. The cropland area mapping (CAM) aggregation dataset provides information on (i) mean cropland area and its uncertainty, (ii) cropland area by six distinct cropland agreement classes, and (iii) cropland area by specific combinations of underlying land cover product. The results indicated that world cropland area is 1500 +/- 400 Mha (mean and 95 % confidence interval), with a relative uncertainty of 25 % that increased across regions. It was 50 % in Central Asia (40 +/- 20 Mha), South America (180 +/- 80 Mha), and Southern Europe (40 +/- 20 Mha) and up to 40 % in Australia and New Zealand (50 +/- 20 Mha), Southeastern Asia (80 +/- 30 Mha), and Southern Africa (16 +/- 6 Mha). Conversely, cropland area was estimated with better precision, i.e., smaller uncertainties in the range 10 %-25 % in Southern Asia (230 +/- 30 Mha), Northern America (200 +/- 40 Mha), Northern Africa (40 +/- 10 Mha), and Eastern Europe and Western Europe (40 +/- 10 Mha). The new data can be used to investigate the coherence of information across the six underlying products, as well as to explore important disagreement features. Overall, 70 % or more of the estimated mean cropland area globally and by region corresponded to good agreement of underlying land cover maps - four or more. Conversely, in Africa cropland area estimates found significant disagreement, highlighting mapping difficulties in complex landscapes. Finally, the new cropland area data were consistent with FAOSTAT (FAO, 2023) in 15 out of 18 world regions, as well as for 114 out of 182 countries with a cropland area above 10 kha. By helping to highlight features of cropland characteristics and underlying causes for agreement-disagreement across land cover products, the CAM aggregation dataset may be used as a reference for the quality of country statistics and may help guide future mapping efforts towards improved agricultural monitoring. Data are publicly available at https://doi.org/10.5281/zenodo.7987515 (Tubiello et al., 2023a).