Scholarlink

<p>River streamflow forecasting is important for managing and controlling the water resource system. This study developed new hybrid models, namely wavelet packet first-order response surface (WPFORS) and wavelet packet quadratic response surface (WPQRS), using the wavelet packet decomposition (WPD) technique with the first-order response surface (FORS) and quadratic response surface (QRS) models. This study also focuses on forecasting the performance of three traditional models: multiple linear regression (MLR), FORS, and QRS. The wavelet packet decomposition technique is used to remove noise from hydrological data. Daily streamflow data from the 2005 to 2013 monsoon season (1st July to 31st September) for the Chenab river basin in Pakistan is used in the study. To check the forecasting performance of the observed models, the criteria used are root mean square error (RMSE), mean square error (MSE), Nash Sutcliffe coefficient of efficiency (NSE), and mean absolute error (MAE). The results showed that WPFORS and WPQRS achieved better forecasting accuracy than traditional models such as MLR, QRS, and FORS. In addition, the overall performance of WPQRS is better than that of the WPFORS model for 1-d ahead forecasting of streamflow data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Examining the Applicability of Wavelet Packet Decomposition on Statistical Models for River Stage Forecasting

  • 期刊论文

  • 发表日期：

    2026年5月16日

查看全文
  • Adnan Bashir,
  • Muhammad Ahmed Shehzad

摘要

River streamflow forecasting is important for managing and controlling the water resource system. This study developed new hybrid models, namely wavelet packet first-order response surface (WPFORS) and wavelet packet quadratic response surface (WPQRS), using the wavelet packet decomposition (WPD) technique with the first-order response surface (FORS) and quadratic response surface (QRS) models. This study also focuses on forecasting the performance of three traditional models: multiple linear regression (MLR), FORS, and QRS. The wavelet packet decomposition technique is used to remove noise from hydrological data. Daily streamflow data from the 2005 to 2013 monsoon season (1st July to 31st September) for the Chenab river basin in Pakistan is used in the study. To check the forecasting performance of the observed models, the criteria used are root mean square error (RMSE), mean square error (MSE), Nash Sutcliffe coefficient of efficiency (NSE), and mean absolute error (MAE). The results showed that WPFORS and WPQRS achieved better forecasting accuracy than traditional models such as MLR, QRS, and FORS. In addition, the overall performance of WPQRS is better than that of the WPFORS model for 1-d ahead forecasting of streamflow data.