Lessons from Climate Modeling on the Design and Use of Ensembles for Crop Modeling
By Yurong Yu, AgMIP Intern
Daniel Wallach, lead author of Climatic Change
An important recent development in crop modeling is the use of model ensembles. The recently published article “Lessons from Climate Modeling on the Design and Use of Ensembles for Crop Modeling” in Climatic Change by Wallach et al., identified questions and approaches related to crop model ensembles, based on the experience of the climate modeling community. The benefits of this method not only include better estimates of uncertainty, but also improved predictions and closer collaboration within the modeling community. However there are numerous open questions about the best way to create and analyze such ensembles.
Climate modeling and crop modeling are two fields in which multiple groups have developed different models to represent complicated systems. Combining these models in a multi-model ensemble however has a longer history in climate modeling than in crop modeling, and so the latter has much to learn from the former.
Outstanding questions include those about the creation of both multi-model ensembles and single model ensembles, that sample parameter or input variability. Exploration is also needed into how the choice of models, the number of models, and possible weighting of models affect the quality of the mean or median as a predictor. Consideration of the output variables will also be important since improvement may not impact all outputs equally.
To access the full article please visit the Climatic Change website here.
This and other publications are available via the AgMIP Publications Database