1. Title, Journal and Authors
Title: Research on ANN-based Pre-warning Water Bloom Model of LiuHai Lake in Beijing
Journal: Procedia Environmental Science 2 (2010) 625-635
Authors: Weihua ZENG*, Qilong SONG, Hengchen LIU, Tao WANG
School of Environment, Beijing Normal University, Beijing 100875 China
In recent years, water bloom flooded in many lakes in China. In order to solve the problem, many researchers started to study by using various modeling methods. In this paper, modeling method was also used to diagnose and alert the outbreak of water bloom.
In this paper, a feedforward back-propagation network(BP network) was established for the pre-warning model for water bloom. Modeling areas are LiuHai lake and its surrounding lakes. According to the result of correlation analysis, the reason that has great relations with water bloom, including water temperature, pH value, DO, TN, transparency, daily average flow of TieLing Floodgate and concentration of chl-a on the day was selected an input data. Considering the recovery period, the indicator of the outbreak that is concentration of chl-a 7 days later was selected an output data.
This method was induced using neurons delivering input signals to other layer’s neurons. Error were calculated with difference of expected values and neuron’s output signal. 2/3 samples are used as training samples and 1/3 samples are used for validation of the pre-warning model. Errors of the results of pre-warning model were 14%, 23%, 12%, 29%, 16% and 7%. Options of class and range of pre-warning of water bloom were determined and they consisted of red, yellow and green. Proper management measures will be taken in accordance with the alert state.
In one sample analysis, we had the conclusion. 1) Hydrodynamic condition is the main factor for water bloom. 2) If water bloom erupts, emergency water may relax the tension of water bloom. 3) Flow velocity is decisive factor but temperature is not necessary factor for water bloom. 4) The eutrophication is not the inevitable factor.
Dae Seong Jeong / Intern student
Phone : +82-10-2003-7860
E-mail : email@example.com