类比合成方法在干旱区内陆河径流量预报中的应用

类比合成算法是一种多维模式搜索法，它具有适用范围广、对资料要求低等优点，可用于单变量及多变量时间序列的延拓预测。通过介绍类比合成算法，并把它应用于塔里木河源流叶尔羌河、和田河年、月径流量预报，其中重点分析了模式长度和合成预报的模式个数等因素对预报结果的影响。通过实测径流资料对预报结果的检验和分析表明，类比合成算法可以较好地挖掘径流序列中隐藏的信息，在中长期水文预报中是一种行之有效的计算方法。

Application of the Analog Complexing Algorithm in the Prediction of Runoff Volume in Arid Areas

Hydrological prediction is an important part in hydrology, and is useful in water resources regulation, flood control, drought-relief activities and many other fields. Analog complexing algorithm (ACA) is a multidimensional pattern search method that can be used to predict the development of one-dimensional or multidimensional temporal series. ACA can be used to automatically select several similar patterns (analog patterns) related to a given pattern (reference pattern) for a given pattern similarity measurement, and a prediction pattern for the reference pattern can be developed by using their known continuations. It can be applied in various fields by using the limited observation data. An algorithm is briefly introduced, and it is applied in predicting the annual and monthly runoff volumes of the Yarkant River and Hotan River, two of the sources streams of the Tarim River in Xinjiang. The influences of the pattern length, number of analog patterns and pattern similarity measurements on the predicted results are analyzed. The results show that the pattern length affects significantly the predicted results, especially for the annual runoff volumes. The predicted results of the monthly runoff volumes in dry season is more ideal than that in flood season. The comparison and analysis of the predicted and measured results of runoff volumes reveal that the analog complexing algorithm is feasible and effective in predicting runoff volumes.