【摘要】 对广西春播期降水量序列利用 3种时间序列分析法 ,通过相关系数与符号相关结合选取因子 ,建立了 6种回归方程 ,试报结果表明 :多因子场序相关相似法有较可信的结果 ;在逐步回归预测方法中 ,多因子逐步回归优于单因子场逐步回归 ;集成预报的结果明显好于单一方法的结果。而在组合计算预报因子中 ,用 5 0 0 h Pa高度场、海平面气压场的滑动块状平均场作为基本因子 ,不仅符合降水发生所应具有的天气形势的系统性和稳定性 ,给自动化编程也带来方便。更多还原

【Abstract】 The precipitation sequence during the spring sowing of Guangxi is analyzed with three time sequence analysis methods, and 6 kinds of regression equations are set up by selecting factors through the correlation combination of correlate parameters with the symbols. The examination results show that: the multi-factor field sequence correlation similitude method results in comparatively reliable forecast results; In the stepwise regression method, the multi-factor stepwise regression is superiors to single-factor field stepwise regression; Results of integration forecast is much better than that of single forecast method. While in the group calculation forecast factors, the use of 500hPa height field and the slider average field of sea-level pressure field as the basic factor not only complies with the systematicality and stability of the weather situation occurring during the rainfall, but also makes the automatic programming very convenient.更多还原