Scholarlink

<p>Winter precipitation forecasting at sufficient lead times has several benefits, including aiding water allocation decisions and supporting individual water users’ decision-making. This study evaluates the performance of ensemble-mean cool-season (December through March) precipitation forecasts from individual models within the North American Multi-Model Ensemble (NMME) over the Colorado River basin and California’s Sacramento–San Joaquin–Tulare basins (hereafter referred to as the SST). These ensemble-mean forecasts are compared to a newly developed statistical forecasting model, using the rain-gauge-based Parameter-elevation Regressions on Independent Slopes Model (PRISM) as the reference product. While NMME models effectively capture the spatial pattern of mean precipitation, they struggle to predict year-to-year variability and extremes. Forecast skill is higher in the Colorado basin than in the SST Basin. Anomaly correlations between ensemble-mean NMME forecasts and observations vary by model and basin, with GEM5-NEMO and GFDL-SPEAR showing relatively higher skill. Performance in forecasting droughts and wet/dry years remains inconsistent across models, with most models missing key events such as the 2023 and 2017 wet years and the 2022 drought. Simple statistical models using key atmospheric–oceanic predictors outperformed the more complex ensemble-mean NMME dynamical models in both basins. The most effective predictors were the Oceanic Niño Index, Tropical South Atlantic sea surface temperatures, and the Quasi-Biennial Oscillation for the SST Basin, and the North Atlantic Oscillation and Tropical North Atlantic sea surface temperatures for the Colorado basin.</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

Cool-season precipitation forecast evaluation over the headwaters of Central Valley and the Colorado River basin

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月11日

查看全文
  • Yu-Chuan Tien,
  • Mekonnen Gebremichael

摘要

Winter precipitation forecasting at sufficient lead times has several benefits, including aiding water allocation decisions and supporting individual water users’ decision-making. This study evaluates the performance of ensemble-mean cool-season (December through March) precipitation forecasts from individual models within the North American Multi-Model Ensemble (NMME) over the Colorado River basin and California’s Sacramento–San Joaquin–Tulare basins (hereafter referred to as the SST). These ensemble-mean forecasts are compared to a newly developed statistical forecasting model, using the rain-gauge-based Parameter-elevation Regressions on Independent Slopes Model (PRISM) as the reference product. While NMME models effectively capture the spatial pattern of mean precipitation, they struggle to predict year-to-year variability and extremes. Forecast skill is higher in the Colorado basin than in the SST Basin. Anomaly correlations between ensemble-mean NMME forecasts and observations vary by model and basin, with GEM5-NEMO and GFDL-SPEAR showing relatively higher skill. Performance in forecasting droughts and wet/dry years remains inconsistent across models, with most models missing key events such as the 2023 and 2017 wet years and the 2022 drought. Simple statistical models using key atmospheric–oceanic predictors outperformed the more complex ensemble-mean NMME dynamical models in both basins. The most effective predictors were the Oceanic Niño Index, Tropical South Atlantic sea surface temperatures, and the Quasi-Biennial Oscillation for the SST Basin, and the North Atlantic Oscillation and Tropical North Atlantic sea surface temperatures for the Colorado basin.