基于多种空间滤波方案的海温资料同化实验

为解决海洋资料同化中空间滤波方案选择的主观性问题,基于最优插值算法,借助现场资料分析系统(ISAS),分别试验高斯函数(Gauss)、二阶自回归函数(SOAR)以及贝塞尔函数(Bessel)作为拟合函数的空间滤波方案,同化全球实时海洋观测计划(Argo)温度剖面资料,并从多个角度比较分析场的质量,以期为寻找最优空间滤波方案提供借鉴。实验表明,Bessel函数方案同化结果的均方根误差最小,SOAR函数方案的均方根误差最大,但是在个别层次上,Bessel函数方案的误差大于其他2种方案;在多个典型层次上,Bessel函数方案的分析场能够分辨更多中小尺度的信息,这与Bessel函数本身可以揭示不同波谱能量误差分布的特性有关。结合海面动力地形数据和Argo剖面的分布状况分析,几个显著的中尺度冷暖中心分布是符合物理海洋学常识的。需要指出的是,在网格分辨率未提高的前提下,这些中小尺度的信号需要借助其他辅助资料和分析方法进行甄别。

Experiments of sea temperation data assimilation based on various space filtering scheme

To select spatial filtering scheme in the ocean data assimilation,with the aid of in-situ analysis system(ISAS) and the optimal interpolation algorithm, this paper performed three different schemes of space filtering: Gauss function, second-order autoregressive(SOAR) function and Bessel function by assimilating the array for real-time geostrophic ocean graphy(Argo) profile temperature data. In order to find the best scheme for space filtering, they evaluated the quality of analyzed temperature from different aspects. Experiment results show that the scheme based on Bessel function owns a minimum root mean square error and the one based on SOAR function has the maximum root mean square error, although there are a few levels in which Bessel Scheme results in more errors than the other two schemes. In several typical levels, the Bessel Scheme can distinguish more information ranging from large scale to meso and sub-meso scales, which maybe benefits from its advantage of revealing the characteristics of error distribution of the different spectral energy with different scales. Further work should be done to assess whether the meso and sub-meso scale information is useful signal or noisy signal by using more objective analysis method.