Spatially explicit information about the differences between crop yield potential or water-limited yield potential and actual farm yields (so-called yield gaps) in current agricultural systems is essential to direct sustainable intensification of agriculture. Nevertheless, reliable location-specific information required to estimate yield gaps is only available for a limited amount of locations. In order to scale up local-specific estimates of yield gaps we developed the Global Yield Gap Atlas Extrapolation Domain (GYGA-ED) approach, that makes use of a matrix of three categorical variables (growing degree days, aridity index, temperature seasonality) with a focus on areas where crops are grown (Van Wart et al, 2013).

The suitability of the GYGA-ED zones to upscale yield gaps has been tested in three regions/countries: US Corn Belt, Germany, and Burkina Faso. In each climate zone several weather station datasets were collected. With the collected weather data potential and water-limited yields were simulated with the crop growth models Hybrid-maize (Yang et al., 2004), WOFOST (Van Diepen et al., 1989), and CERES-Wheat (Ritchie et al., 1985). Per climate zone the simulated yields were compared. By using an allowed difference in the simulated yields of maximally 10% we concluded that the weather conditions within a climate zone can be represented by data from a single station and thus the GYGA-ED approach can be used to upscale yield gaps.

Knowing the extrapolation domain of weather station data allows for optimized search for or selection of actual weather data for use in modelling crop growth at large scales. This simplifies upscaling as well as provides a framework for the extrapolation of results of previous research.