长江口外表层盐度预报方法初探

A PRELIMINARY INVESTIGATION ON THE PREDICTION METHOD OF SURFACE SALINITY OFF CHANGJIANG MOUTH

本文运用正交经验函数场分解法对长江口外表层盐度预报进行了初步探讨。采用了有较大实用意义的横序预报法。研究表明,用该方法对长江口外表层盐度进行预报,有较大的潜力和优越性。根据 1975年 4月至 1977年 6月,共 35个站位的断面观测盐度资料找出35个与时间无关的本征矢量,且获得一组仅和时间有关的新资料序。利用这组资料给出了十九阶自回归预报模式进行了试报。任何一月的35个站的盐度偏差均可表示成35个本征矢量的线性组合,系数可由回归预报模式给出。实际预报中,只需取前6—8个相应大本征值的本征矢量来进行组合,就有足够的精度。

The predictive way of two-dimensional fields of surface salinity off Changjiang(Yangtze River) mouth with empirical orthogonal functions, herein referred to aseigenvectors, is discussed. The eigenvectors are derived from monthly measurementsof sea-level salinity at 35 points off the area during April 1975 to June 1978. The35 eigenvectors only dapend on selected 35 stations, and they are independent ontime. The observed salinity deviation vectors (SD)_i (i=1,2,……,35) can be expressedas a linear combination of the 35 eigenvectors. The elemet Q_(ik) of the matrix Q isreferred to as the coefficient associated with the k-th eigenvector for the i-th observedsalinity deviation vector. The coefficient is only dependent on time. These coefficientsare used to find selfregression model of the prediction. It is shown that the empircalorthogonal function analysis has a larger potential and some advantages.