用变分法修正黑河试验数据的初步研究

为了对黑河试验观测资料进行订正,利用一个一维边界层垂直扩散模式和以拟牛顿极小化算法为基础的最优化方法建立了一个变分同化系统,其中,目标函数相对于控制变量的梯度用扰动法求取.并利用模式资料和黑河试验若干时刻的观测资料进行了一些数值试验.试验证明,将初始场变量与模式误差系数同时作为控制变量加以调整,能够有效调整初始场观测资料.并且,用变分同化方法能够较线性插值(时间、空间)方法更有效、合理的修正黑河试验观测资料场的种种缺陷.

A Research to Validate the Data from Heihe River Field Experiment by Variation

To modify the observational data from the Heihe River Field Experiment(HEIFE), a variational assimilation system is set up based on a vertical diffusion model in the boundary layer and a quasi-Newton minmization algorithm. The gradients of the cost function with respect to control varia- bles are calculated in the way of perturbation. The results of some experiments by using both model da- ta and some observation data in the HEIFE show that it can modify the initial fields(the atmosphere state) effciently, while making the initial fields and the coefficients of model error as control variables simultaneously. Compared with the linear interpolations(included in time and in space), variational as- similations can improve the defect in the data obtained in the HEIFE.