暂态模型LET-KF数据同化地表温度序列分析

LET-KF Data Assimilation for Land Surface Temperature Series Analysis Based on Transient Model

针对传统地表温度序列分析方法预测精度不高的问题,提出一种基于暂态模型补充预测的局部集合扩展卡尔曼(LET-KF)数据同化地表温度序列分析方法。首先,对地表温度模型进行研究,根据植被和泥土覆盖率,设计地表温度暂态模型;然后,为提高卡尔曼滤波算法性能,并结合地表温度暂态模型预测输出特点,设计局部集合变换卡尔曼数据同化算法。通过与泰安周边地表温度卫星观测值和实测值数据同化实验表明,所提算法有效提高了对地表温度预测的精度。

In order to solve the problem of low prediction accuracy in traditional land surface temperatureseries analysis, here presented the transient model based local ensemble transform kalman Filter（LETKF）data assimilation for surface temperature series analysis algorithm. Firstly, it studied the landsurface temperature model, and proposed the land surface temperature short-time model according to thevegetation and soil cover rate; Then, in order to improve the performance of kalman filter algorithm, heredesigned the local ensemble transform kalman filter based data assimilation algorithm, which combinedwith the output characteristics of the surface temperature short-time prediction model. Through the dataassimilation experiment with Tanan land surface temperature observed by satellite and the actualmeasurement, the proposed algorithm effEC-LAStively improve the prediction accuracy of land surfacetemperature.