基于模拟光谱数据的冬小麦播期遥感监测最佳时相研究

The Best Phase Selection for Monitoring Winter Wheat Sowing Time Using Remote Sensing Data Based on Simulated Spectral Data

为利用高时空分辨率的航天数据对区域冬小麦播期实现尽早监测,对冬小麦播期的不同遥感监测时相精度进行了分析。首先利用耦合作物模型和辐射传输模型模拟不同播期冬小麦从播种至返青的冠层光谱反射率,分析不同播期的冠层光谱响应差异,选取对不同播种日期敏感的波段。然后,根据敏感波段的冠层光谱,选择训练样本并计算不同播期之间的J-M距离,初步判断出光谱可分性较好的时相。最后,对不同的播期进一步进行判别分析,判定未知类别样本的所属类别。根据正确分类的精度,在华北平原北部选择播期监测的最佳时相为12月中旬,精度达到89.5%。

The stable availability of remote sensing data with high spatial and high temporal resolution makes it possible to detect sowing time in early growth stage. In this study,the best phase of satellite imaging to extract winter wheat sowing time was chosen by data mining on various simulation experiments. Firstly by coupling of the crop model and radiative transfer model,the canopy spectral reflectance dynamics of winter wheat with different sowing time from sowing to turn green stage were simulated,and sensitive bands to detect different sowing date was selected. Secondly,based on sensitive bands,the J-M distances of the canopy spectral reflectance under different sowing dates were calculated,and the better spectral-distinguished phase was selected according to J-M distance. Lastly,the discriminant analysis on the canopy spectral reflectance of winter wheat under different sowing dates was used,according to the accuracy of sowing time classification,the best sowing time monitoring phase was chosen,with the accuracy of 89. 5%. The result of our study shows that the best phase for winter wheat sowing time monitoring using remote sensing data in northern plain of North China is about December 19.