Scholarlink

<p>The Great Lakes are vital freshwater resources for both the United States and Canada. Improved water level predictions can aid flood risk management, optimize water resource allocation, and support ecological conservation. This study aims to eliminate the lag effects of traditional machine learning models using phase space reconstruction (PSR). Predictions were conducted using historical monthly mean water level datasets of Lake Ontario from 1918 to 2023, divided into training (1918–2002) and testing (2003–2023) datasets. The results revealed that PSR-RF outperforms standard random forest, KNN, and LSTM models across all metrics, including Correlation Coefficient (0.999), Nash–Sutcliffe Efficiency (0.998), Root Mean Squared Error (0.014), Coefficient of Determination (0.998), and regression equation slope and intercept (𝑦 = 0.98𝑥 + 1.484).</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 Approach Combining Phase Space Reconstruction with Random Forest, KNN, and LSTM for Reducing Prediction Lag in Lake Water Level Forecasting

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Bo Yuan

摘要

The Great Lakes are vital freshwater resources for both the United States and Canada. Improved water level predictions can aid flood risk management, optimize water resource allocation, and support ecological conservation. This study aims to eliminate the lag effects of traditional machine learning models using phase space reconstruction (PSR). Predictions were conducted using historical monthly mean water level datasets of Lake Ontario from 1918 to 2023, divided into training (1918–2002) and testing (2003–2023) datasets. The results revealed that PSR-RF outperforms standard random forest, KNN, and LSTM models across all metrics, including Correlation Coefficient (0.999), Nash–Sutcliffe Efficiency (0.998), Root Mean Squared Error (0.014), Coefficient of Determination (0.998), and regression equation slope and intercept (𝑦 = 0.98𝑥 + 1.484).