This highly depends on what data you have to support one or the other method. With crop models, you will be able to generate a crop yield estimate based on only climate variables, for which you need to have suitable downscaled climate projections (see www.agmip.org for uncertainty of crop models). For the ricardian method you will need additional information from farm-level surveys to build the statistical model. As a result, the ricardian model will deliver a more robust information that also considers economic factors. However, as soon as process-based feed-back mechanisms are involved, a crop model will play out its advantages. New varieties, new management techniques and future price development remain unconsidered in both approaches.

Dear Dr. Muhamad, Depending on the purpose we can say which one is better.

I would agree with Claas, and add that the choice also depends on the time frame and type of projections you are interested in. Ricardian models tend to provide equilibrium solutions, while dynamic crop models are better at assessing timelines of change along a future trajectory. I would add that dynamic models are very well suited--and have been used for years to this end--to describe new varieties as well as changes in management practices. Ricardian models are better suited at capturing (implictly and indirectly) effects of economic change.

Best Regards