同化MODIS温度产品估算地表水热通量

Estimation of sensible and latent heat flux by assimilating MODIS LST products

基于集合卡尔曼滤波和通用陆面模型(CLM1.0)发展了一个地表温度的同化系统。这个系统同化了MODIS温度产品,并将MODIS的叶面积指数引入CLM模型中,主要用于改进地表水热通量的估算精度。将CLM输出的地表温度与MODIS地表温度建立关系,并作为同化系统的观测算子。将MODIS地表温度与实测地表温度进行了比较,将其均方差(Root Mean Square Error,RMSE)作为观测误差。选取3个美国通量网站点(Blackhill、Bondville、Brookings)作为实验数据,结果表明:同化结果中地表温度、显热通量的估算精度均有提高。其中Blackhill站的估算精度改进最大,均方差由81.5W·m-2减小到58.4W·m-2,Bondville站均方差由47.0W·m-2减小到31.8W·m-2,Brookings站均方差由46.5W·m-2减小到45.1W·m-2。潜热通量估算精度在Bondville站均方差由88.6W·m-2减小到57.7W·m-2,Blackhill站均方差由53.4W·m-2减小到47.2W·m-2。总之,结合陆面过程模型同化MODIS温度产品估算地表水热通量是可行的。

In this paper, a land surface temperature data assimilation scheme is developed based on Ensemble Kalman Filter （EnKF） and Common Land Model version 1.0 （CLM）, which is mainly used to improve the estimation of the sensible and latent heat fluxes by assimilating MODIS land surface temperature （LST） products. Leaf area index （LAI） is also updated dynamically by MODIS LAI products. In this study, the relationship between the MODIS LST and the CLM surface temperature is determined and taken as the observation operator of the assimilation scheme. Meanwhile, the MODIS LST is compared with the ground-measured surface temperature, and the Root Mean Square Error （RMSE） is taken as the observation error. The scheme is tested and validated based on measurements in three observation stations （B1ackhill, Bondville and Brookings） of Ameriflux. Results indicate that data assimilation method improves the estimation of surface temperature and sensible heat flux. The RMSE of sensible heat flux reduced from 81.5W·m^-2to 58.4W·m^-2 at the Blackhill site, from 47.0W·m^-2 to 31.SW·m^-2 at the Bondville site, from 46.5W·m^-2 to 45.1W·m^-2 at the Brookings site. The RMSE of latent heat fluxes reduced from 88.6W·m^-2 to 57.7W·m^-2 at the Bondville site, from 53.4W·m^-2 to 47.2W·m^-2 at the Blackhill site. In addition, it is a practical way to improve the estimation of sensible and latent heat flux by assimilating MODIS LST into land surface model.