Zoning of reservoir water temperature field based on K-means clustering algorithm

Abstract

Study region: Dongqing Reservoir located in Guizhou, China. Study focus: Zoning the RWTF (reservoir water temperature field) is of great significance and is an effective way to analyze RWTF’s nature. In practice, the conditions of RWTF fluctuate greatly as time goes on, which leads to the existing RWTF zoning methods can’t give a steady zoning result. In consequence, this paper creates a kind of zoning method to study the properties of RWTF in Dongqing Reservoir, which has two main steps: firstly, numerical simulation is used to obtain the whole data of RWTF, and then the K-means clustering algorithm is executed based on the numerical simulation results. New hydrological insights for the region: This paper proved that the zoning method developed in this paper, which combined numerical simulation and unsupervised machine learning, can be effectively applied and divided the RWTF into four zones without using experimental parameters in Dongqing Reservoir. Moreover, on the base of the four zones’ spatial borders, the influencing factors of water temperature in each zone of Dongqing Reservoir were able to be found, which would be of value in further research of RWTF.

Keywords

• Reservoir water temperature field
• Unsupervised machine learning
• K-means clustering algorithm
• Numerical simulation
• Zoning method