基于时间序列HJ-1 A/B卫星数据的冬小麦成熟期预测

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

准确、及时地掌握大区域尺度的冬小麦成熟期信息能够为农业机械调度、优化农作物收割顺序提供重要的参考依据。以华北平原中部冬小麦为研究对象,首先使用2013年研究区冬小麦生育期内HJ-1 A/B CCD时间序列影像,通过线性插值构建像元尺度上逐日的时间序列NDVI,随后采用上包络线S-G滤波方法重构时间序列NDVI,通过动态阈值法逐像元提取冬小麦抽穗期;然后以抽穗至成熟期的有效积温模型为判别依据,利用欧洲中期天气预报中心(ECMWF)提供的日平均气温预报数据,实现未来10 d冬小麦成熟期的动态预测;最后采用农业气象站点的成熟期观测值对预测结果进行验证,重点对比分析了从不同成熟期预报起始时间点获得的冬小麦成熟期精度,以确定最优的预报起始时间点。结果表明:当预报时效小于等于10 d时,成熟期预测精度趋于稳定,因此,综合考虑确定提前10天对预测冬小麦成熟期在时效和精度上最优,平均误差为3 d。该方法为地块尺度的区域农作物成熟期预测提供了可参考的技术途径。

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.