Winter wheat yield estimation based on MODIS EVI

Before Terra was launched, NOAA AVHRR was mainly adopted in land cover and vegetation research for global and regional scale. The Terra-MODIS has two vegetation indexes: the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI). In order to compare the result assessed in crops with these two kinds of vegetation indexes, MODIS NDVI and MODIS EVI were used to monitor growth condition and forecast yield of winter wheat in USA. The NDVI and EVI values and the yields for many years were carried on correlation analysis using the specific breeding time of the regional crop and the one linear method was adopted to set up the regression equation to estimate the yield of that year. The results show that the MODIS-EVI has obvious advantages over the MODIS NDVI. The correlation coefficient of the regression equation was more than 0.7 by using the EVI, but the correlation coefficient of the regression equation was lower and varied greatly by using the NDVI. So the MODIS EVI was used to estimate yield of winter wheat in USA(for 18 states of winter wheat's main product). The estimated yield in 2004 was 52.89Bushel/Acre and improved by 3.49% compared with that in 2003. The estimated production in 2004 was 1334675000 Bushel and reduced by 12.92% compared with that in 2003. For verifying its accuracy, the estimation result was compared with the forecast result of State Statistics Bureau of USA on June 1 2004, it was found that the yield estimation error of the forecast result of State Statistics Bureau of USA was 3.05% and the production error was -2.56%, however the yield estimation error was 2.62% and the production error was -1.77% in this paper, and forecasting time can be advanced for about two weeks. It is obvious that the MODIS-EVI can monitor and forecast crops more effectively and can improve the predicted accuracy.