客观预报中多时刻因子的应用及其效果

Application of Multi-time Factors in Objective Forecast

分析了在客观预报中单时刻因子与预报量之间存在着时间尺度上的不匹配问题,提出使用多时刻因子的必要性和可能性。对比试验表明,多时刻因子可改善大多数预报方程的质量,有效率达90.0%;多时刻因子使预报误差明显减小,预报方程的残差平方和平均减少23.8%。残差平方和的减少在各种场合有所不同,但呈现出一定的规律性。同时又指出,在多时刻因子中并不是包含的时刻越多越好,以防出现“维数灾”现象。

The problem of no match on time scale between single-time factors and predictor in objective forecast is analyzed, and the necessity and the possibility of using multi-time factors are put forward. Contrast experiment shows that multi-time factors can improve the quality of prediction equations and make effective rate reach 90.0 percents and reduce sum of residual squares 23.8 percents meanly. The reduction appears some regularity in spite of it's variation in various situation. It is also indicated that it is not benefit to contain more time number in multi-time factors in order to avoid “dimension disaster”.