A Medium and Long-Term Runoff Forecast Method Based on Massive Meteorological Data and Machine Learning Algorithms

Yujie Li,Jing Wei,Bin Xu,Yueping Xu,Dong Wang,Bo Li,Huaping Huang

doi:10.3390/w13091308

Abstract

Accurate and reliable predictors selection and model construction are the key to medium and long-term runoff forecast. In this study, 130 climate indexes are utilized as the primary forecast factors. Partial Mutual Information (PMI), Recursive Feature Elimination (RFE) and Classification and Regression Tree (CART) are respectively employed as the typical algorithms of Filter, Wrapper and Embedded based on Feature Selection (FS) to obtain three final forecast schemes. Random Forest (RF) and Extreme Gradient Boosting (XGB) are respectively constructed as the representative models of Bagging and Boosting based on Ensemble Learning (EL) to realize the forecast of the three types of forecast lead time which contains monthly, seasonal and annual runoff sequences of the Three Gorges Reservoir in the Yangtze River Basin. This study aims to summarize and compare the applicability and accuracy of different FS methods and EL models in medium and long-term runoff forecast. The results show the following: (1) RFE method shows the best forecast performance in all different models and different forecast lead time. (2) RF and XGB models are suitable for medium and long-term runoff forecast but XGB presents the better forecast skills both in calibration and validation. (3) With the increase of the runoff magnitudes, the accuracy and reliability of forecast are improved. However, it is still difficult to establish accurate and reliable forecasts only large-scale climate indexes used. We conclude that the theoretical framework based on Machine Learning could be useful to water managers who focus on medium and long-term runoff forecast.

Highlights

Refined medium and long-term runoff forecast technology is an important basis for the design, construction, and operation management of water conservancy and hydropower projects [1,2,3]
We summarize the Filter, Wrapper, and Embedded theories from Feature Selection (FS) based on Machine Learning (ML) and apply Partial Mutual Information (PMI), Recursive Feature Elimination (RFE), Classification and Regression Tree (CART) as the typical algorithms of the above three theories to select forecast factors suitable for medium and long-term runoff forecast
We summarize three typical methods of Feature Selection theory based on Machine Learning in the selection of forecast factors and employ PMI, RFE, and CART as typical algorithms of Filter, Wrapper, and Embedded, respectively, to obtain three sets of forecast factor schemes

Summary

Introduction

Refined medium and long-term runoff forecast technology is an important basis for the design, construction, and operation management of water conservancy and hydropower projects [1,2,3]. It is a basic key technology to realize the scientific allocation of water resources and improve the utilization efficiency of water resources and has important supporting significance for the dispatching management and optimal allocation of water resources [4,5,6]. The practice of water resources regulation, such as unified regulation of basin water quantity, inter-basin water transfer, and joint regulation of super-large cascade reservoirs, has continued to advance, putting forward higher requirements for the forecast accuracy and forecast scale of medium and long-term runoff forecast. In the unified regulation and management of water quantity in China’s Yangtze River, Yellow River, Pearl River and Huai River basins, medium and long-term runoff forecast are important scientific and theoretical bases for water resource allocation and regulation schemes, and the reliability, stability, and extensibility of the forecast results play a significant value [11,12]

Objectives

Methods

Findings

Conclusion