Generating global crop distribution maps: From census to grid

• Published in: Agricultural systems Vol. 127; pp. 53 - 60
• Main Authors: You, Liangzhi, Wood, Stanley, Wood-Sichra, Ulrike, Wu, Wenbin
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2014
• Subjects: Agricultural production; Crop suitability; Cross entropy; Farming system; Global; Spatial allocation; Crop suitability; Farming system; Cross entropy; Spatial allocation; Agricultural production; Global
• ISSN: 0308-521X; 1873-2267
• DOI: 10.1016/j.agsy.2014.01.002

•Describe an improved spatial crop allocation model.•Estimate crop production for 20 major crops across a global 5arc minute grid.•Summarize factors that give rise to systematic differences of the global datasets. We describe a new crop allocation model that adds further methodological and data enhancements to the available crop downscaling modeling. The model comprises the estimates of crop area, yield and production for 20 major crops under four rainfed and irrigated production systems across a global 5arc minute grid. The new model builds on prior work by the authors (and published in this journal) in developing regional downscaled databases for Latin America and the Caribbean (LAC) and sub-Saharan Africa (SSA) and encompasses notions of comparative advantage and potential economic worth as factors influencing the geographic distribution of crop production. This is done through a downscaling approach that accounts for spatial variation in the biophysical conditions influencing the productivity of individual crops within the cropland extent, and that uses crop prices to weigh the gross revenue potential of alternate crops when considering how to prioritize the allocation of specific crops to individual grid cells. The proposed methodology also allows for the inclusion of partial, existing sources of evidence and feedback on local crop distribution patterns through the use of spatial allocation priors that are then subjected to an entropy-based optimization procedure that imposes a range of consistency and aggregation constraints. We compare the global datasets and summarize factors that give rise to systematic differences amongst them and how such differences might influence the fitness for purpose of each dataset. We conclude with some recommendations on priorities for further work in improving the reliability, utility and periodic repeatability of generating crop production distribution data.