准确预测区域尺度的小麦成熟期，指挥麦收机械化作业有序开展，具有十分重要的社会和经济效益。该文针对目前区域冬小麦成熟期预测中时效性差、缺乏空间分布以及缺少定量描述等突出问题，选择华北地区河北、河南和山东3省冬小麦为研究对象，首先基于S-G滤波后的2013年冬小麦生育期时间序列MODIS LAI，采用动态阈值法获取抽穗期具体日期，即叶面积指数（LAI）达到峰值时的具体日期；然后基于由2008－2012年农业气象资料与地面气象资料构建的抽穗-成熟期有效积温模型和总辐射模型，逐个栅格单元计算MODIS LAI获取的抽穗期具体日期到当前日期的积温、太阳辐射总量，并结合全球多模式集合预报（THORPEX Interactive Grand Global Ensemble，TIGGE）资料对当前日期（5月10号至6月8号）之后的16 d冬小麦成熟期进行逐日动态预测以得到全部区域的成熟期预测值；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，结果表明：冬小麦成熟期预测值与观测值的决定系数R2为0.92，均方根误差RMSE约为3 d，两者具有良好的相关性。该研究方法对其他大区域的农作物成熟期预测具有借鉴价值。%Knowing the proper time to harvest crops is a major step towards successful farming because it helps to avoid the negative effects of bad weather and improves the quality and quantity of the crops. Thus, it is of significant importance to predict the maturity period of the crops for improving the production benefits and decreasing the loss. At present, researchers have developed models to predict the maturity period, including meteorological statistical model, crop-growing model, and remote sensing monitoring model, etc. But those models have limitations in timeliness, regional promotion or have complex implementation process. Therefore, in this research, we aimed at improving maturity period prediction of winter wheat using winter wheat planting regions in Hebei, Henan and Shandong province as study areas. Firstly we retrieved the heading stage from S-G filtered MODIS LAI in 2013 when wheat LAI reached the peak using dynamic threshold method. Then, in order to obtain the maturity period forecasts values for 1 km by 1 km winter wheat grids, effective accumulated temperatures and total radiation distribution from heading to maturity have been collected through historical agro-meteorological observational data and ground meteorological data from 2008 to 2012 by the Thiessen polygons method. It assumed that a Thiessen polygon has uniform varieties of winter wheat, uniform effective accumulated temperature and total solar radiation. Effective accumulated temperatures and total radiation model from heading to maturity were built based on the effective accumulated temperatures and total radiation distribution, then each grid cell was calculated for the accumulated temperature of the date when LAI reached its peak and total solar radiation. After that, in order to obtain the predicting data of all the regions, we predicted the maturity period of the winter wheat for each day for 16 days after the present time combining with TIGGE (THORPEX Interactive Grand Global Ensemble) and ground meteorological data. When the effective accumulated temperatures and total radiation of a grid cell met the requirements of effective accumulated temperatures and total radiation from heading to maturity, the winter wheat would reach maturity date. Finally, we used the heading period data of agricultural meteorological station to verify the data of maturity period. The results showed that the correlation coefficientR2 and the root mean square error (RMSE) between observed date and predicted date for the heading were 0.89 and 3.62 days, respectively. TheR2 and RMSE between predicting date and observed date for the maturity was 0.92 and 2.89 days respectively. Predicting errors of maturity which was extracted from MODIS LAI haven’t increased much more, it turned out that predicting accuracy for maturity based on meteorological data was higher than the maturity date based on remote sensing data, but the prediction of maturity based on remote sensing data fitted large scale region. So the prediction which used both remote sensing data and meteorological data could obtain the satisfactory results. The method provided a reference of crop maturity data for other agricultural regions. So the method was easy to be used in larger scale, and also serves as a simplified model. Besides, the method solved the existing problems of poor timeliness and lacking spatial distribution, thus it helps a lot to predict the maturity period of the crops.
展开▼
