Data from Eitelberg, D.A., van Vliet, J. & Verburg, P.H. (2014). A review of global potentially available cropland estimates and their consequences for model-based assessments. Global Change Biology. 21(3): 1236-1248
Based on a synthesis of the assumptions underlying current global estimates of cropland availability we constructed maps of a high, a medium, and a low estimate that are representative of the range of estimates in the studies reviewed for this research.
The data used to produce the high, medium, and low cropland availability estimates shown here are outlined in Table 1. The constraints applied to each estimate are shown in table 2 (table 4 in the published manuscript). Land-cover data was derived from the Global Land Cover 2000 dataset (GLC2000). We reclassified the original 22 GLC2000 land-cover classes to match the 17 International Geosphere-Biosphere Programme (IGBP) land-cover classes according to the groupings described in Herold et al. (2008) because these classes best matched the land-cover classes described in the literature of cropland availability estimates. These IGBP land-cover classes were then used, in addition to biophysical and institutional constraints, to define areas to be excluded from the high, medium, and low estimates that were produced. We used the World Database on Protected Areas and areas designated as protected prior to or including the year 2000 were identified. From these protected areas, those designated with the International Union for Conservation of Nature (IUCN) codes V and VI were not regarded as a limit to cropland expansion since cropland can exist in these areas. We then apply a 15% reduction in area to the three numerical estimates to account for portions of cell areas that are not used for cropland. Studies have shown that cropland estimates based on raster cells often overestimate the true amount of cropland because they don’t account for infrastructure, settlements, and other areas that are unsuitable for crops at the sub-pixel level (Young, 1999; Fritz et al., 2013). Verburg et al. (2009), through an analysis of cropland across Europe, found that between 3% and 20% of the main (large-scale) cropland areas were occupied by infrastructure, buildings, or other non-productive landscape features.
The maps are in the equal area projection Eckert IV with a resolution of 9.25 x 9.25 km (corresponding to 5 arcminutes). Data are provided in ASCII format.