Abstract: Using the data of per unit corn yield in 1961-2006,the daily average temperature and precipitation during the growing seasons of the fourteen representative stations in the corn growing areas of USA,the monthly average sea surface temperature of the West Pacific Ocean and the data of the circulation at the 500hPa height of the Northern Hemisphere,three forecasting models for USA corn yield based on the ground meteorological data,the ocean surface temperature and the circulation data respectively were built.The per unit corn yield in the USA in 1995-2004 was forecasted and the models were validated.The results showed that the average accuracy for ten years was over 95% for each model,but with relative high fluctuation.The integrative forecasting model was introduced to improve the forecast accuracy based on the historic accuracy of each model by a weighted method.The validation of the forecast accuracy for the integrative model was over 95% in 2005-2006.On the whole,the integrative model could meet the needs of operational service.

Key words: United States of America(USA), United States of America(USA), Corn yield, Operational forecast