Prediction of Maturity Data for Winter Wheat Based on Time Series of HJ-1 A/B CCD Images

Abstract:

Accurately and timely regional maturity date of winter wheat provides an important information and reference for commanding farm machinery and optimizing the crop harvesting order. This study took main winter wheat planting area of the North China Plain as a case study. Firstly, the daily NDVI time series have been obtained through linear interpolation using time series of HJ-1 A/B CCD images from the winter wheat’s booting to filling stages, the upper envelope Savitzky-Golay (S-G) filtered method was used to reconstruct the time series of NDVI, and then the heading stage was extracted from the NDVI profile using dynamic threshold method. Then, the resulting heading date and the effective accumulated temperature model of heading-maturity stage, combining with the average daily temperature forecast data from European Centre for medium range weather forecasts (ECMWF) was used to conduct realtime dynamic prediction maturity date of winter wheat. Finally, the prediction results were validated by observed maturity date in agricultural meteorological stations, and we test the optimal starting time for maturity date through comparing four schemes with different prediction starting date. The results showed the high accuracy of prediction maturity date when we conducted the prediction ahead of 10 d. The RMSE of predicted maturity date was about 3 d. These results also showed that predicting winter wheat maturity date at the 30 m resolution was promising and can be used for operational crop maturity monitoring and forecasting in the near future.