Zimbabwe maize yield, with lead times up to one year. This information could be a valuable component of famine early-warning systems for the region.

Rosenzweig (1995) provided a global perspective on the effects of ENSO on crop yields. Her preliminary work indicated that yields of five major crops in about thirty countries appear to be correlated with ENSO variations. For most of the crops, yields tended to be lower in years with anomalously warm equatorial Pacific sea surface temperature. One-third of the world's production of maize, wheat, sorghum, soybeans, and rice was found to be correlated with phases of ENSO, although for many regions the signal was small.

Investigations of links between agricultural production and ENSO are just beginning for the United States. Adams et al. (1995) used the historical temperature and precipitation data for the southeastern United States, together with a model of crop yield, to estimate that climate variations associated with ENSO could decrease crop yield in the southeast by up to 15 percent. The value of improving an ENSO forecast from correlations at the level of roughly 0.6 to a level of 0.8 at lead times of six months in advance of planting decisions was estimated to be more than $100 million per year for the southern agricultural sector alone.

Skill for predictions outside the tropics is likely to be much lower than skill for the tropics because of the greater variability of the climate around the mean. Nevertheless, experimental prediction activities for the United States have begun (Ji et al. 1994a, Kerr 1994). Near the end of TOGA, the National Meteorological Center began to use coupled ocean-atmosphere forecasts, together with several statistical techniques, to arrive at an official consensus forecast with lead times up to two seasons in advance.