Regional yield estimation for spring maize with multi-temporal remotely sensed data in Junchuan, China

Establishing timely and high accurate models for crop yield estimation is of great significance for crop management and as well as decision makers. The arm of this study is to gain an approach of the method, depending on crop growth model and entropy method, to estimate spring maize yield with multi-temporal remotely sensed Landsat TM/ETM+ data at main growth and development stages of spring maize. The experiment had been conducted in Junchuan Farm of Northeast China. In this paper, the combined weights of the single-temporal estimation models were calculated by applying the entropy method (EM), and a combination forecasting (CF) model was developed. In order to improve the rationality of CF-EM and the accuracy of yield estimation, especially to follow the law of crop growth, the combination forecasting of combined weights method (CF-CM) was developed. The results showed that the yield estimation model based on CF-CM could increase the precision of the yield estimation model based on single-temporal remote images, the correlation coefficient was remarkably improved, and the value was increased by 0.09. The combined weights in the CF-CM were proposed for selecting the favorable coefficient of the subjective weight and objective weight, and that was of great importance for some key aspects: supplying usefulness information, how to raise maize yield and selecting key temporal satellite images to estimate maize yield. The CF-CM model discussed in this paper is feasible and effective to estimate spring maize yield.