Deep Learning Prediction of Fish Spawning Habitats Under Regulated River Flows

Abstract

The river ecosystem acts as a necessary breeding ground for fish species depending on the presence of suitable flow, water quality, and habitat conditions for their spawning process. Nevertheless, regulation of the river flow through construction of dams, water abstraction, and hydrological modifications has disturbed the natural spawning habitats of fishes along with their breeding process. In this paper, a deep learning-aided ecological evaluation technique is introduced for the prediction of fish spawning habitats in the case of regulated river flow. The suggested method considers various environmental parameters such as flow velocity, water depth, temperature, dissolved oxygen, substrate properties, and habitat structure for determining the suitable spawning sites. A model based on deep learning predictions was used as an auxiliary technique to detect patterns related to the presence of spawning sites in relation to the characteristics of their habitats. It is clear from the findings that flow conditions have a notable effect on the presence of fish spawning grounds, with stability in flow areas and favorable substrate conditions making the habitat more suitable. This analysis proved the ability of deep learning techniques to predict spawning grounds while taking into account multiple factors related to the environment