AgMIP Crop Teams – Intercomparison and Improvement of Crop Models for Response to Climatic Factors.

The Agricultural Model Intercomparison and Improvement Project (AgMIP) has the goal of intercomparing, testing, and improving crop models for their responses to climate change factors of temperature, CO2, water availability, as well as crop management and soils effects.  AgMIP crop modeling teams evaluated sensitivity of 28 wheat models, 23 maize models, and 14 rice models to CO₂, temperature, and rainfall, simulated at each of four sentinel sites differing in yield level, temperature, and water supply.  An important finding is that the models were less certain at elevated CO₂ or at above ambient temperatures.  Compared to observed yields at four sentinel sites per crop, the ensemble mean of the models more accurately predicted observed yields than any given single model.  This was true for wheat, maize, and rice model ensembles.  An ensemble of 5 to 8 models was needed to reduce the error of prediction to a reasonable 14%.  The required number of models in the ensemble increased with rising temperature and CO₂.  For the model ensembles, yield response to doubled CO₂ (360 to 720 ppm) was 7.5% for maize, 20.6 to 26.2% for rice, and 20-30% for wheat.  Maize models varied considerably from no response to substantial response, while individual rice and wheat models also differed considerably.  The models predicted declining yield with rising temperature (wheat by 6% per ^oC, maize by 4.5 to 7.8% per ^oC, rice by 5.3 to 8.3% per ^oC).  Crop life cycle was shortened with rising temperature and yield reduction was greater for already warmer sites.  The maize model ensemble response to temperature or CO₂ was similar whether calibrated or non-calibrated for site-specific productivity.  The teams are now testing correctness of these temperature and CO₂ responses against observed response data.  Other crop modeling teams are underway for sugarcane, potato, sorghum, groundnut, canola, and bioenergy crops.