Context

To understand how the global cropland and field size paper came to be, there are three key events that led to this research and the subsequent publication: 

1 Highlighting the Spatial Disagreement in Land Cover Maps

As part of his previous job at the Joint Research Centre (JRC) of the European Commission in Ispra, Italy, Steffen Fritz was responsible for taking the individual regional land cover maps being produced for the year 2000 by expert teams around the world and creating a single global land cover product for the environmental baseline year 2000 (GLC2000 - Fritz et al., 2003). As part of his work, Steffen then compared the GLC2000 with other global land cover products such as that produced by the team in Boston using data from the MODIS sensor (Friedl et al. 2010) and he noticed large spatial discrepancies between the maps in some locations. Papers showing these discrepancies and how to deal with them from a user perspective were subsequently published (Fritz et al., 2005; 2008). This led to further work comparing three different global land cover products (GLC2000, MODIS, GlobCover) that highlighted the continued uncertainty in our knowledge of land cover, particularly in cropland and forest cover (Fritz et al., 2011). For example, we showed that there is a discrepancy in total cropland of between 325 to almost 400 Million hectares when comparing GLC2000 with MODIS and GlobCover, respectively. Hence the global characterization of cropland is problematic and needs to be improved.

2 Geo-Wiki and the Crowdsourcing of Cropland and Agricultural Field Size Data  

As part of his job at the JRC, Steffen also compared global land cover maps with freely available very high resolution satellite imagery available through Google Earth. Based on this visual comparison, he also realized that the global land cover maps are wrong in certain locations and that Google Earth could be used to help improve these products. Out of this realization, the Geo-Wiki project was born (Fritz et al., 2009; 2012). In the original version of Geo-Wiki, users (i.e. members of the 'crowd') were asked to determine how good the GLC2000, MODIS and GlobCover land cover products were in relation to what they could see on the satellite imagery (Figure 1).