Initiating of “environmental prediction science” for marine microbial ecology
May 2, 2018
Visualization of organic, inorganic and physically important factors from time-series modeling in normal and red-tide conditions
The surface waters surrouding Japan are one of the greatest biodiversity hotspot in the world. Ranking sixth in the world in the terms of ocean area, Japan is expected to properly “cultivate” this ocean in the near future. However, homeostatic collapse of ocean microbial ecosystems caused by rising seawater temperatures and coastal eutrophication due to urban industrialization and fertilizer inflows in recent years have resulted increase of red tides that potentially serious damage to fishery products. The homeostasis of the marine environment has been perturbed by various physical, chemical and biological factors related to ecosystem services.
RIKEN CSRS researchers have developed a method to visualize organic, inorganic and physically important factors for red tide prediction using machine learning for big data analysis of marine environments and time-series modeling. They propose an analytical strategy that involves obtaining big data in the aquatic environment and applying machine learning, factor mapping and time-series modeling to visualize the complicated relationships between factors while predicting future events. This strategy enhances environmental factor analysis methods, which are used to evaluate the samples from the natural environment from a variety of angles, to visualize the complicated systems in the natural environment.
Utilizing the results of this research, it is anticipated that it will become possible to forecast/provide early warning of environmental changes from fluctuations in key factors before an ecosystem becomes disturbed and control those key factors to improve the ecological homeostasis.
Science of the Total Environment doi:10.1016/j.scitotenv.2018.04.156
Oita, A., Tsuboi, Y., Date, Y., Oshima, T., Sakata, K., Yokoyama, A., Moriya, S. and Kikuchi, J.,
"Profiling physicochemical and planktonic features from discretely/continuously sampled surface water".
Environmental Metabolic Analysis Research Team