河流封冻日期长期预报的MMFA模型

The Application of MMFA Model on Date Prediction of River Freeze －up

本文介绍了均值生成函数这一时间序列分析中的新概念，并经过主成分分析建立了时间序列模拟与预测的数学模型（记作ＭＭＦＡ）。该模型应用于黄河下游利津河段封河日期序列的模拟和预报。结果表明，１９５０－１９８１年度序列拟合的平均相对误差为９．５％，１９８１－１９９１年度序列预报的精度达到８５％以上。总之，ＭＭＦＡ模型应用于黄河下游封冻日期长期预报是完全可行的。同时指出了ＭＭＦＡ模型的实用性和存在的不足之处。

Mean Generation Function───A new concept in time series analysis is employed in thispaper， and an MMFA model about main factor analysis is presented too. It is showed thatMMFA model can simulate freeze－up dates in time series on Lijin reach of the Yellow River. and the simulated error and predicted precision are 9．5％ and 85％ respectively. The modelperformance and its wide applicability are presented at the end of this paper.