Estimation of fish distribution by means of neural network using monitoring data by fishing boats

Abstract

It is important to understand and predict fish behavior to assess the impacts of coastal development on ecosystem or to conduct appropriate fishery management. In order to collect information for understanding fish behavior, we measured the environmental factors and the distribution of fish simultaneously by using fishing boats. The memory-type sensors were attached to the fishing gear of the small trawling boats in Ise Bay, through which the water temperature, dissolved oxygen, and depth were measured as well as position information by GPS. At the same time, fish catch of each haul by trawling were recorded to grasp the fish distribution. The obtained data provides much more information for temporal and spatial distribution of water qualities than conventional monitoring. We tried to predict resource distributions by artificial neural network using the obtained data. It is found that the environmental factor affecting the behavior of conger eel varies with respect to the months. The developed neural network could predict the horizontal distribution of the conger eel fairly well.