小波变换对中长期入库径流预测的适应性研究

Research on the Adaptability of Wavelet Transform in Medium Long-term Inflow Runoff Forecasting

通过典型水库入库径流预测实例,分析了小波周期分析模型预测效果与径流系列长度、时段长及流域尺度之间的关系。初步结论为:①小波周期分析模型依据时间序列长度不能过短,但也不是愈长愈好,当时间序列长度超过一定限时,模型的预测效果将不会再改善;②月、旬两种时段长对比结果表明,时段长越短,径流序列奇异性将越显著,小波周期分析模型的预测效果越差;③小波周期分析模型对流域尺度反应灵敏,流域面积越小,径流序列变化就愈不规则,模型的预测效果愈差。

This paper analyses the relations of forecasting result using wavelet periodic analysis model with the length of runoff series.time interval as well as the drainage basin scale from a typical reservoir inflow runoff forecasting example. The initial conclusions can be gotten as follows： ① The length of time series should neither be too long nor too short in the wavelet periodic analysis model. Because when the length of time series exceeds a certain limit, the forecasting result of the model will have no further improvement; ② Through the contrast between month and range time interval, we can get that when time interval becomes short;the singularity of runoff series would become more sahence, and the forecasting result the model becomes worse; ③The wavelet periodic analysis model is sensitive to the basin scale.The smaller the basin area is. the more irregular the runoff series change as a result,the forecasting result of the model is worse.