Long-term Trends of Water quality data: Case Study of a Wastewater treatment plant in Southern Taiwan

• Main Authors: Chen, Jun-Lun, 陳俊綸
• Other Authors: Lee, Chih-Sheng
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/j4v2bq

碩士 === 崑山科技大學 === 環境工程研究所 === 103 === Recently, due to the effects of global warming and climate change, people cannot forecast the weather change efficiently. However, the regional micro-climate change has caused sudden heavy rainfalls that lead to drought is becoming a trend. This trend has not only affected the water usage of civilians, but also destroyed the stability of environmental ecology. Under this vicious cycle, the development of environmental ecology will be affected. If there is a warning mode that can alert us about disastrous events so we could have precaution, accidents can be prevented. This is a case study of a waste water treatment plant in Southern Taiwan. The database is from 2008 to 2013, after using Gauss weights allocation to remove system error of the varieties of water quality test data, then by using standard deviation, skewness and autocorrelation of statistical parameters to find the leading indicators of the time series of 100, 250, 500 and 1000 days. After that, observing the varieties of water quality under unstable and dramatic changes one by one and finding if the leading indicators of standard deviation, skewness and autocorrelation have phenomenon of positive correlation feedback as a warning signal. Through research analysis, when water quality diversity range is not large or recovery rate is fast after water quality had changed, it is determined as stable, thus harder to detect any warning signals; however when water quality diversity range is large or recovery rate is slow, it is easier to detect warning signals. In addition, analysis results shows that it is easier to observe if the leading indicators of standard deviation, skewness and autocorrelation have phenomenon of positive correlation feedback, best days amongst the leading indicators are both 250 and 500 days.