ニューラルネットワークを用いた閉鎖性海域における化学的酸素要求量の変化予測

Prediction of Chemical Oxygen Demand Variation in an Enclosed Sea Area using Neural Networks

Creators Yokota Motohiro

Creators Taira Yuichiro

Creators Morimoto Eiji

Creators Ezoe Satoru

Creators Ogawa Kazuo

This study develops a prediction method for a chemical oxygen demand in a fish farm using data obtained from experiments for bottom sediment improvement (environmental monitoring research) in the Katada Culture Farm in Ago Bay, Mie prefecture. Results show that the fluctuation of a chemical oxygen demand from the surface layer (a depth of 0.5m) to the bottom layer (a depth of 0.5m on the bottom face) can be estimated using a neural network whose inputs are water depth, water temperature, salinity, dissolved oxygen, pH, chlorophyll-a, hours of sunshine, and respective amounts of precipitation and mean air temperature. When the sensitivity analysis was carried out to clarify the contribution of each environmental factor for the chemical oxygen demand, it was affected considerably by weather conditions (hours of sunshine, precipitation, water temperature, etc.), salinity, and chlorophyll-a.

[ISSN]0370-9361