신경망 모형을 이용한 낙동강 중·하류 하도의 클로로필a 추정에 관한 연구

Abstract

A study for estimation of chlorophyll-a in a mid-lower reach of the Nakdong River using a neural network

ByeongGwon Park

Department of Civil Engineering, The Graduate School, Pukyong National University

Abstract

Eutrophication has been continued by the discharge of untreated nitrogen and phosphorus from industrial complexes of Daegu and Gumi cities in the upstream area of the mid-lower Nakdong River. Prediction of algae is currently performed by measuring Chlorophyll-a and the number of cells of Cyanobacteria that are main indications of occurrence of aigal blooms. The purpose of the study is to introduce a neural network model generalizing complicated mathematical model and to seek water quality factors estimating Chlorophlly-a that can be an indicator of eutrophication of water system. The measurement stations are Namgang, Jeokpo, Cheongam and Chilseo located on the mid-lower Nakdong River and the daily data of DO, water temperature, TN, TP, TOC, pH, tubidity are inputted as the training factors. Chlorophlly-a is selected as the estimation variable of the neural network. As a result, the outcome of the combination of training factors of DO, water temperature, TN, TP, pH measured by automated water quality monitoring station in Namgang, Jeokpo and Cheongam is superior to the combinations of the other training factors. Also, the combination of training factors of DO, water temperature, TOC, pH by automated water quality monitoring station in Chilseo estimates Chlorophlly-a best. The result of this study suggests that Chlorophyll-a be estimated by a neural network model using the real-time data of automated water quality monitoring.

Keywords: neural network, chlorophll-a, aigal bloom, eutrophication, automated water quality monitoring station.