本文针对复杂的年径流序列,提出首先识别并处理跳跃项,然后再进行趋势项、周期项的识别和提取,接着对剩余序列建立随机项模型,最终将趋势项模型、周期项模型和随机项模型耦合,建立复杂年径流预测耦合模型的新思路。并对某流域的年径流序列进行分析和预测,其预报结果具有一定的精度,表明模型是有效的,对其它复杂径流序列的研究具有一定的借鉴和指导作用。 For complex annual runoff sequences, we propose to identify and process the jump items first, then to identify and extract the trend items and periodic items, and then to establish the stochastic item models for the remaining sequences. Finally, the trend item model, the periodic item model and the random item Model coupling to establish a new model of complex annual runoff prediction coupled model. The runoff sequence of a river basin is analyzed and predicted. The prediction results have certain accuracy, which shows that the model is effective and has certain reference and guidance to the study of other complex runoff sequences.