Rain Amounts and Temporal Distribution: The Thousand-Pound Gorilla in Projecting Crop Yields in Response to Climate Change.

Assessments of the impact of climate change on crop yield almost invariably focus on possible changes in temperature and atmospheric carbon dioxide concentration.  While each of these variables play a role in influencing crop yield, the impacts are usually small compared to the possibilities that might result from changes in rainfall patterns.  Throughout agricultural history, and certainly in recent years, variations in rainfall have had the major influence on crop productivity.  Currently in the U.S., large geographical and annual variations in crop yield are associated with variations in rainfall.  Uncertainty in predicting future rain patterns overshadows attempts to project possible crop responses to climate change.  There are several sources of uncertainty in assessing possible change in crop water balances.  It is sometimes argued that atmospheric vapor pressure will be increasing due to increased temperatures resulting in greater rates of soil evaporation and crop transpiration.  However, thus far vapor pressure deficit has remained stable and this may not be a critical variable.  Regional climate models have generally predicted increased annual precipitation over the United Sates, which would seemingly be beneficial to crops.  However, much of the increased precipitation is predicted to occur in the autumn and winter, which probably would add little to the crop water balance.  The summer rainfall, on the other hand, is predicted to be decreased for virtually all of the major cropping regions of the U.S.  To add to the uncertainty, changes in the intervals between rains and the amounts of individual rain events are unresolved, although current trends indicate the most severe rains storms are increasing in both their frequency and intensity.  Until the rain amounts and temporal distribution in future climates are resolved, predictions of crop response to climate change in the U.S. are virtually not resolvable.