A dataset of yearly probabilistic crop type maps for the EU from 1990 to 2018

Abstract

We provide an ensemble of probabilistic crop type maps for the entire European Union, mapping the shares of 25 different crop types at 1km resolution for all years from 1990 to 2018. We generate the maps using a recently developed approach based on a model of the data generating process from field- to region: essentially, we link knowledge about which crops farmers are more likely to grow under certain environmental conditions with data on regional crop acreages. Consequently, the resulting crop share estimates are consistent with regional crop area statistics while considering spatial heterogeneity. To reflect estimation uncertainty of the provided crop shares, we sample 100 maps per year and EU country, each being coherent with regional or national information. This ensemble of maps allows users to sample from potentially different but similarly likely spatial crop type distributions and thereby adequately reflect uncertainty in their applications, for example, in an environmental model. We additionally provide maps with only the most likely crop shares for users mainly interested in point estimates. The maps provide researchers, policy makers, and other stakeholders information to evaluate the impact of changing political, economic, and environmental conditions over three decades on agricultural production in the European Union.