Scholarlink

<p>The Songhua River basin (SHRB), situated in northeast China, serves as a critical commercial grain production base that significantly contributes to national food security. Seasonal precipitation anomalies over SHRB (hereafter SHRBP) during the flood season (May-September, MJJAS) exert dominant impacts on agricultural productivity and regional socioeconomic stability, underscoring the imperative for skillful seasonal forecasting to support disaster mitigation strategies. This study develops a novel Dynamic-Statistical Prediction Model (DSPM) incorporating time-scale decomposition to enhance MJJAS SHRBP prediction skill. The model integrates observed sea surface temperature (SST) indices with dynamical forecasts from the National Centers for Environmental Prediction Climate Forecast System, version 2 (CFSv2) initialized in March. Key findings reveal: (1) Multiscale Variability Characteristics: MJJAS SHRBP exhibits pronounced interannual (standard deviation = 50.1&#xa0;mm) and interdecadal (50.4&#xa0;mm) variability components of comparable magnitude. (2) Physical Mechanisms: on the interannual scale, winter-spring tropical Indian Ocean SST anomalies and CFSv2-predicted East Asian summer monsoon (EASM) circulation patterns demonstrate significant linkages with MJJAS SHRBP variability; on the interdecadal scale, MJJAS precipitation fluctuations show robust associations with preceding winter-spring Pacific horseshoe-mode SST anomalies, North Atlantic SST anomalies, and CFSv2-predicted MJJAS EASM circulation. (3) Model Performance: the DSPM employs multiple regression to separately model interannual and interdecadal components before synthesizing total precipitation predictions. Comparative evaluation against operational models demonstrates DSPM’s superior skill during 2017–2024 independent validation. Comparative analysis demonstrates that the DSPM achieves marked improvement in predicting the MJJAS SHRBP compared with both the CFSv2 and European Centre for Medium-Range Weather Forecasts (ECMWF) model. During the independent validation, the DSPM predictions exhibit superior performance metrics, with a temporal correlation coefficient (TCC) of 0.72 and a relative root mean square error (RRMSE) of 15.1% against observations. In contrast, the CFSv2 model yields substantially lower skill with a negative TCC (–0.31) and higher RRMSE (22.3%), while the ECMWF model shows marginally better but still suboptimal performance (TCC = − 0.18; RRMSE = 23.9%). The demonstrated skill improvements highlight DSPM’s potential value for agricultural planning and water resource management in this crucial grain-producing region.</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 skillful dynamical-statistical prediction model with time-scale decomposition for precipitation over Songhua river basin in China during May-September

  • 期刊论文

  • 发表日期：

    2025年5月14日

查看全文
  • Yuxin Zhao,
  • Junhu Zhao,
  • Yongsheng Li,
  • Xiaolong Jia,
  • Zongjian Ke,
  • Han Zhang,
  • Baizhu Shen

摘要

The Songhua River basin (SHRB), situated in northeast China, serves as a critical commercial grain production base that significantly contributes to national food security. Seasonal precipitation anomalies over SHRB (hereafter SHRBP) during the flood season (May-September, MJJAS) exert dominant impacts on agricultural productivity and regional socioeconomic stability, underscoring the imperative for skillful seasonal forecasting to support disaster mitigation strategies. This study develops a novel Dynamic-Statistical Prediction Model (DSPM) incorporating time-scale decomposition to enhance MJJAS SHRBP prediction skill. The model integrates observed sea surface temperature (SST) indices with dynamical forecasts from the National Centers for Environmental Prediction Climate Forecast System, version 2 (CFSv2) initialized in March. Key findings reveal: (1) Multiscale Variability Characteristics: MJJAS SHRBP exhibits pronounced interannual (standard deviation = 50.1 mm) and interdecadal (50.4 mm) variability components of comparable magnitude. (2) Physical Mechanisms: on the interannual scale, winter-spring tropical Indian Ocean SST anomalies and CFSv2-predicted East Asian summer monsoon (EASM) circulation patterns demonstrate significant linkages with MJJAS SHRBP variability; on the interdecadal scale, MJJAS precipitation fluctuations show robust associations with preceding winter-spring Pacific horseshoe-mode SST anomalies, North Atlantic SST anomalies, and CFSv2-predicted MJJAS EASM circulation. (3) Model Performance: the DSPM employs multiple regression to separately model interannual and interdecadal components before synthesizing total precipitation predictions. Comparative evaluation against operational models demonstrates DSPM’s superior skill during 2017–2024 independent validation. Comparative analysis demonstrates that the DSPM achieves marked improvement in predicting the MJJAS SHRBP compared with both the CFSv2 and European Centre for Medium-Range Weather Forecasts (ECMWF) model. During the independent validation, the DSPM predictions exhibit superior performance metrics, with a temporal correlation coefficient (TCC) of 0.72 and a relative root mean square error (RRMSE) of 15.1% against observations. In contrast, the CFSv2 model yields substantially lower skill with a negative TCC (–0.31) and higher RRMSE (22.3%), while the ECMWF model shows marginally better but still suboptimal performance (TCC = − 0.18; RRMSE = 23.9%). The demonstrated skill improvements highlight DSPM’s potential value for agricultural planning and water resource management in this crucial grain-producing region.