Author(s): Wei Huang; Shinichiro Yano; Jianmin Zhang
Linked Author(s): Shinichiro Yano, Jianmin Zhang
Keywords: River metabolism; Least square support vector machine; River health; Artificial neural network; Reaeration coefficient
Abstract: In this paper, we employ the least square support vector machine (LSSVM) to study the river metabolism, which is one of the most integrative ecosystem functions, highly sensitive to many anthropogenic and natural stressors and thereby often used to assess the impairment or health state of river ecosystem. With the data from the continual field measurement, the discharge, travel time of water mass, dissolved oxygen concentration and water temperature are selected as the main variables for LSSVM. From the results, it can be found that the LSSVM can be used successfully in predicting the metabolism rate. Moreover, compared to other artificial intelligence tools like back propagation artificial neural networks (BP＿ANN), it seems that LSSVM can perform better with higher accuracy and shorter time computation. Thus, it may be considered as an alternative method to estimate the metabolism rate and to assess the river ecosystem health in river study.
Year: 2011