ScholarMate
English简体中文繁體中文
客服热线:400-1616-289
登录注册
期刊论文

Machine learning modeling structures and framework for short-term forecasting and long-term projection of Streamflow

Tran, Trung Duc; Kim, Jongho*
Stochastic Environmental Research and Risk Assessment2024-2DOI: 10.1007/s00477-023-02621-y
3被引2阅读
引用收藏
分享

摘要

Reliable short-term forecasting and long-term projection of streamflow are essential. However, few research models for machine learning structures systematized for short- and long-term time-series predictions and frameworks for simultaneously optimizing hyperparameters are available. In this study, to improve the accuracy degradation phenomenon that occurs especially in long-term forecasting and maximize its efficiency, four long-short term memory (LSTM) model structures (msLSTM, srLSTM, spLSTM, and mjLSTM) with time-invariant or time-variant features are clearly systematized for the first time. We then compare three optimization methods and newly present a dual-loop framework for simultaneously optimizing various hyperparameters that are independent of each other. Finally, using a proposed model structure and dual-optimization framework, future changes in daily streamflow are evaluated. From the results obtained, we can conclude that: (i) the srLSTM structure is appropriate for short-term forecasting, specifically for predicting one to five steps ahead, while the mjLSTM structure is recommended for long-term projection; (ii) Bayesian optimization offers superior accuracy and efficiency compared with a Grid or Random search, and it is preferred for the design of the dual-loop optimization framework required to maximize model performance; and (iii) under the SSP5-8.5 climate change scenario, annual streamflow is projected to increase by approximately 10-20%, and monthly precipitation and streamflow are also expected to increase by at least 10% to a maximum of 40%. The comprehensive model structure and framework proposed in this study offer a promising solution for addressing both short-term forecasting and long-term projection problems using machine learning models.

学科与关键词

关键词
Modeling frameworkLSTM structureForecastingProjectionDual-optimizationMachine learning
相关
加载中…
  • Long lead-time daily and monthly streamflow forecasting using machine learning methods
    Cheng, M.; Fang, F.*; Kinouchi, T.; Navon, I. M.; Pain, C. C.Journal of Hydrology, 2020, 590: 125376.被引 161
  • Deep Learning Data-Intelligence Model Based on Adjusted Forecasting Window Scale: Application in Daily Streamflow Simulation
    Fu, Minglei; Fan, Tingchao; Ding, Zi'ang; Salih, Sinan Q.; Al-Ansari, Nadhir; Yaseen, Zaher Mundher*IEEE Access, 2020, 8: 32632-32651.被引 129
  • Streamflow and rainfall forecasting by two long short-term memory-based models
    Ni, Lingling; Wang, Dong*; Singh, Vijay P.; Wu, Jianfeng*; Wang, Yuankun; Tao, Yuwei; Zhang, JianyunJournal of Hydrology, 2020, 583: 124296.被引 197
查看全部 →

发表信息

论文状态
公开发表
期刊
Stochastic Environmental Research and Risk Assessment
发表日期
2024-02
卷期
第 38 卷, 第 2 期
页码
793-813
DOI
10.1007/s00477-023-02621-y
语言
ENGLISH
成果收录
SCIE
资助标注
Research Fund of University of Ulsan

作者

  • T

    Tran, Trung Duc (第一作者)

  • K

    Kim, Jongho (通讯)

登录解锁更多功能

  • 收藏这篇,随时在“我的 · 收藏”找回
  • 认领你的成果,生成推广素材
  • 关注作者获取最新成果
  • 加入项目组、协同写作
注册 / 登录

产品服务

  • 科研之友
  • 创新城
  • 科创云

服务支持

  • 帮助中心
  • 隐私政策
  • 服务条款

联系方式

在线客服:【】
客服热线:400-1616-289
电子邮箱:support@scholarmate.com

关注或下载科研之友

微信二维码
微信公众号
客户端下载二维码
下载客户端
科研成果科研人员科研机构科研动态爱瑞思软件

©2026 深圳市科研之友网络服务有限公司

公安备案图标粤公网安备 44030502000213
粤ICP备 16046710 号粤B2-20110417