|Title||Geostatistical interpolation and aggregation of crop growth model outputs|
|Event||ESA 14, 2016-09-05/2016-09-09|
Soil Geography and Landscape
ISRIC - World Soil Information
|Publication type||Abstract in scientific journal or proceedings|
|Abstract||Many mechanistic crop growth models require daily meteorological data; consequently, model simulations can only be obtained for locations close to weather stations with long-term records. Those simulations deliver potential yields as point data.
A widely used approach for aggregating (estimating total production per country from the simulated yields at points) is based on agro-ecological Climate Zones (CZ), e.g. the Global Yield Gap Atlas (www.yieldgap.org). A geostatistical approach that exploits the spatial correlation of simulated yields at points and its correlation with external environmental factors offers additional features to the CZ approach: yield predictions adjusted to conditions on every single location, quantification of the uncertainty of the predictions, and quantification of uncertainty of aggregated country production. As a case study, we interpolate and aggregate potential yields of millet in West Africa. We compare the results of the geostatistical approach with those of the CZ approach.