Local Approximation法によるリアルタイムデータを用いた富栄養湖の溶存酸素の短期予測

概要

論文の詳細を見る
Lake Koyama, located in the eastern part of Tottori Prefecture, has various water environmental problems due to organic pollution and eutrophication. Anoxic water is especially generated in the deepest part of the pond in summer, and there is a strong suspicion that the progression of anoxia will cause further deterioration of the water environment. Therefore, predicting dissolved oxygen (DO) variations is an important subject to plan the preservation and improvement of water environments. In this study, we tried to predict in short-term the time transition of DO at the water surface and the lake bottom at the deepest site using a Local Approximation (LA) method of forecast technique developed in the chaos engineering field. We examined the presence of chaos in both DO time-series. As a result, a DO time-series in the surface didn't show low-dimensionality chaos in either summer or autumn, but could be predicted by the LA method three hours in advance. On the other hand, in summer a DO time-series at the bottom showed low-dimensionality chaos, but not in autumn. Also, DO in summer and autumn could be respectively predicted one hour and four hours in advance. In conclusion, the LA method that predicts time-series through the orbit information of historical data was effective for the short-term forecast of DO, regardless of the presence of chaoticity in its time-series data
日本雨水資源化システム学会の論文