Scholarlink

<p>Sub-seasonal to seasonal (S2S) precipitation forecasting has long been regarded as a “forecasting desert” due to limited skill beyond seven lead days, undermining downstream hydrological forecasts. However, the higher predictability of streamflow compared to precipitation, and its disproportionate improvement relative to precipitation forecast, have often been overlooked. This study integrates a distributed hydrological model with a probabilistic statistical model to enhance S2S flood forecast by assimilating statistical hydroclimate relationships. The ensemble approach is validated at 24 hydrological stations across Pearl River Basin with complex hydrology. Its modest forecasts show mean Nash–Sutcliffe Efficiency (NSE) scores ranging from 0.36 to 0.16 for weeks 2 to 6, and a 15% improvement in Continuous Ranked Probability Score Skill (CRPSS) compared to hydrological model alone. This study underscores the value of integrating physical and statistical models to improve S2S streamflow prediction, offering a practical pathway to enhance forecast skill in flood-prone basins.</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

A hybrid framework for sub-seasonal to seasonal streamflow prediction: integrating numerical and statistical models

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月28日

查看全文
  • Lingfeng Li,
  • Huan Wu,
  • Lulu Jiang,
  • Yiwen Mei,
  • John S. Kimball,
  • Lorenzo Alfieri,
  • Zhijun Huang,
  • Ying Hu,
  • Sirong Chen,
  • Shaorou Dong,
  • Yaming Hu,
  • Wei Wu

摘要

Sub-seasonal to seasonal (S2S) precipitation forecasting has long been regarded as a “forecasting desert” due to limited skill beyond seven lead days, undermining downstream hydrological forecasts. However, the higher predictability of streamflow compared to precipitation, and its disproportionate improvement relative to precipitation forecast, have often been overlooked. This study integrates a distributed hydrological model with a probabilistic statistical model to enhance S2S flood forecast by assimilating statistical hydroclimate relationships. The ensemble approach is validated at 24 hydrological stations across Pearl River Basin with complex hydrology. Its modest forecasts show mean Nash–Sutcliffe Efficiency (NSE) scores ranging from 0.36 to 0.16 for weeks 2 to 6, and a 15% improvement in Continuous Ranked Probability Score Skill (CRPSS) compared to hydrological model alone. This study underscores the value of integrating physical and statistical models to improve S2S streamflow prediction, offering a practical pathway to enhance forecast skill in flood-prone basins.