Emulating maize yields from global gridded crop models using statistical estimates

Joint Program Report
Emulating maize yields from global gridded crop models using statistical estimates
Blanc, E. and B. Sultan (2015)
Joint Program Report Series, 39 p.

Report 279 [Download]

Abstract/Summary:

This study estimates statistical models emulating maize yield responses to changes in temperature and precipitation simulated by global gridded crop models. We use the unique and newly-released Inter Sectoral Impact Model Intercomparison Project Fast Track ensemble of global gridded crop model simulations to build a panel of annual maize yields simulations from five crop models and corresponding monthly weather variables for over a century. This dataset is then used to estimate statistical relationships between yields and weather variables for each crop model. The statistical models are able to closely replicate both in- and out-of-sample maize yields projected by the crop models. This study therefore provides simple tools to predict gridded changes in maize yields due to climate change at the global level. By emulating crop yields for several models, the tools will be useful for climate change impact assessments and facilitate evaluation of crop model uncertainty.

Citation:

Blanc, E. and B. Sultan (2015): Emulating maize yields from global gridded crop models using statistical estimates. Joint Program Report Series Report 279, 39 p. (http://globalchange.mit.edu/publication/16202)
  • Joint Program Report
Emulating maize yields from global gridded crop models using statistical estimates

Blanc, E. and B. Sultan

Report 

279
39 p.
2016

Abstract/Summary: 

This study estimates statistical models emulating maize yield responses to changes in temperature and precipitation simulated by global gridded crop models. We use the unique and newly-released Inter Sectoral Impact Model Intercomparison Project Fast Track ensemble of global gridded crop model simulations to build a panel of annual maize yields simulations from five crop models and corresponding monthly weather variables for over a century. This dataset is then used to estimate statistical relationships between yields and weather variables for each crop model. The statistical models are able to closely replicate both in- and out-of-sample maize yields projected by the crop models. This study therefore provides simple tools to predict gridded changes in maize yields due to climate change at the global level. By emulating crop yields for several models, the tools will be useful for climate change impact assessments and facilitate evaluation of crop model uncertainty.