论文部分内容阅读
基于集合Kalman滤波及SCE-UA(shuffled complex evolution)算法发展了能够直接同化微波亮温的土壤湿度同化方案.该方案以陆面过程模式CLM3.0中的土壤水模型作为预报算子,以辐射传输模型作为观测算子.整个同化过程分为参数优化和土壤湿度同化两个阶段,利用SCE-UA算法优化辐射传输模型中难以确定的植被光学厚度参数和地表粗糙度参数,并利用优化参数作为观测算子的模型参数进行同化.通过人工理想试验表明该同化方案可以明显改善表层土壤湿度的模拟精度,并且对深层土壤湿度的模拟也有一定程度的改善;利用AMSR-E亮温(10.65GHz垂直极化)所进行的实际同化试验表明顶层(0～10cm)土壤湿度同化结果与观测的均方根误差(RMSE)由模拟的0.05052减小到0.03355,相对减小了33.6%,而较深层(10～50cm)平均减小了20.9%.这些同化试验显示该同化方案的合理性.
Based on the set Kalman filter and the shuffled complex evolution (SCE-UA) algorithm, a soil moisture assimilation scheme was developed to directly assimilate the microwave brightness temperature.The soil water model in the land surface process model CLM3.0 was used as a prediction operator, Transmission model as observation operator.The whole assimilation process is divided into two stages: parameter optimization and soil moisture assimilation. The parameters of vegetation optical thickness and surface roughness which are difficult to determine in radiative transfer model are optimized by SCE-UA algorithm. Observational operator model parameters assimilation.The results of artificial ideal experiments show that the assimilation scheme can significantly improve the simulation accuracy of surface soil moisture and to some extent improve the simulation of soil moisture in depth.Using AMSR-E brightness temperature (10.65GHz vertical Polarization) showed that the root mean square error (RMSE) of soil moisture assimilation and observation from the top (0 ~ 10cm) decreased from 0.05052 to 0.03355 in the top layer (a relative decrease of 33.6%), while the deeper 10 ~ 50cm) decreased by 20.9% on average.These assimilation tests showed the rationality of this assimilation scheme.