Wheat Yield Forecasting at County Scale Based on Time Series Vegetation Temperature Condition Index

Selecting the drought monitoring results of remotely sensed vegetation temperature condition index (VTCI) for winter wheat at the ten-day intervals from 2008 to 2016 in the Guanzhong Plain, the weights of drought impact on wheat yields at the 4 main growth stages were determined by applying the best weighting method. Linear regression analysis was employed to study the correlation between the weighted VTCIs and wheat yields of counties, and the yield prediction was carried out at 1-, 2-and 3-ten day intervals between 2008 and 2016 by using the monitored VTCIs and forecasted ones by the autoregressive integrated moving average models. The results show that the weights of drought impact on wheat yields at the turning green stage, the elongation stage, the heading-filling stage and the dough stage are 0.035, 0.489, 0.427 and 0.049 respectively based on the best combination weighting approach of the improved analytic hierarchy method and the entropy method. There is a significant correlation between the weighted VTCIs and the ground-measured yields published in the related statistical yearbooks, indicating the accuracy of the estimated yields is high. The forecasted yield accuracies are quite high and decreased with the increase of the forecasting intervals.