WOFOST模型同化时序HJ CCD数据反演叶面积指数

Retrieving LAI by assimilating time series HJ CCD data with WOFOST

摘要: 为增强作物叶面积指数遥感反演的机理性并提高反演精度，在深入分析作物长势模型WOFOST机理的基础上，采用最小二乘法作为同化算法，以生长季内获取的时序HJ CCD遥感数据作为外部数据源，反演冬小麦叶面积指数进行长势监测和估产应用。以河北省玉田县为试验区，以三要素法和实测LAI作为基准，模型模拟产量和LAI作为反演精度的度量指标，成熟期LAI估算误差由模型同化前的14.95%降至同化后的9.97%，产量误差由同化前的18.17%降为同化后的15.89%。叶面积指数的同化结果与实测数据具有较好的拟合度，表明该方法的具有一定可行性，为作物生长模型区域化应用提供了参考。

Abstract: For the accuracy improvement of retrieving Leaf Area Index(LAI) from remote sensing images, the least square method was used as the algorithm to assimilate time series HJ CCD images with WOFOST based on its fundamental principals analysis. The calibrated crop growth model was then used for crop growth monitoring and yield estimation in Yutian, Hebei province. The results showed that the error of yield and LAI compared with their site specific counterparts were improved by 2.28% and 4.98% respectively, which shows the feasibility of the presented method and provides a new choice for spatialising crop growth model at regional scale.