Artificial Neural Network Modeling of Bird Behavior and Reactions to Environmental Parameters in Wajiro Tidal Flat Reclamations

摘要

Reclamation projects give an impression that a severe environmental damage is certain. However, the extent of damage cannot be measured by a rule of thumb. In this paper, an analysis of the environmental conditions focusing on the population of certain bird species was performed. It is only natural to surmise that bird population is decreasing due to man-made structures, much more destroying a natural habitat, though, at this stage of the study the degree as to how much the population has changed remains unknown. A model of the biological brain, known as artificial neural networks (ANN), which is the main essence of this research, might open doors to a more rigorous investigation of the environmental changes that are occurring due to tidal flat reclamations. This network is able to train itself from input parameter values and thus can predict values of desired output variables. A specific type of ANN algorithm used for calculation is the backpropagation algorithm, which is also known as the generalized delta rule. Input parameters like air temperature, daylight hours, and tidal flat organisms that birds feed were chosen. Sensitivity analysis was performed to identify the birds' behavioral patterns and their reaction to the state variables.