Mathematical Modeling of Fish Behavior to Hydraulic Environments in Agricultural Canal

鱼类行为是基本指标之一，不仅用于评估开发活动对河流生态系统的影响后的水环境，而且还用于估计河流改善的功能方面。与过去的生态学研究试图通过自然栖息地的现场测量来解释鱼类的偏好不同，我们的研究旨在开发一种借助数值模拟技术进行此类评估的方法。我们研究的最终目的是通过对鱼群模型的数值模拟和水工建筑物周围的观测水力学数据来评估水工环境中的鱼类偏好。本课题采用水槽试验和数值模拟相结合的方法，研究了不同水槽条件下水渠中鱼类行为的数学模型。首先，通过水槽试验对JAP…的鱼群行为进行了表征更多的青竹鱼(大米)。在实验中引入了几个特征变量来量化鱼群的全球结构和典型特征：最近邻距离、极化和广度。利用混沌工程领域提出的分维、回归映射和关联维来表征特定环境下鱼类个体的神经应激行为。其次，研究了牛顿动力学模型和行为模式模型来模拟水箱中的鱼群行为。两种模型的比较表明，行为模式模型能更准确地再现真实鱼群的特征变量。最后，进行了明渠水力试验，确定了水深、流速、屏蔽系数等水力环境的偏好曲线。较少

Fish behavior is one of the essential indicators, not only for evaluating the water environments following the impact of development activities on river ecosystems, but also for estimating the functional aspects of river improvement. In contrast to the past ecological studies that have tried to explain the preferences of fish from field measurements in a natural habitat, our study is directed toward developing a methodology for such evaluations with the aid of numerical simulation techniques. The final goal of our study is to evaluate the fish preference in a hydraulic environment by numerical simulations with a fish school model and the observed hydraulic data around the hydraulic structures. In this project, mathematical models for predicting the fish behavior to various hydraulic environments in agricultural canals have been examined by using water tank experiments and numerical simulations.First, the water tank experiments were executed to characterize the schooling behavior of Jap … More anese Medaka fish (Oryzias latipes). Several characteristic variables were introduced to quantify the global structure and the typical characteristics of a fish school in the experiments : the nearest neighbor distance, the polarization and the expanse. The fractal dimension, the return map and the correlation dimension, which are the parameters proposed in the field of chaos engineering, were also used to characterize the nervous strain of fish individuals in the given environment.Next, the Newtonian dynamics model and the behavioral pattern model were investigated to simulate the schooling behavior obtained in the water tank. The comparison of both models showed that the behavioral pattern model could reproduce the characteristic variables of real fish school more accurately. The artificial neural network model could also simulate the schooling behavior obtained in the water tank experimentsFinally, the hydraulic experiments using an open channel were carried out in order to determine the preference curves for the hydraulic environments such as the water depth, the flow velocity and the shield factor. Less