Development of an AI-Based Predictive Model and Correlation Analysis of Marine Environmental Variables for Sebastes schlegelii

Key Points

• The aim is to analyze the correlation between catch volume and marine environmental variables, and to predict catch using AI.
• Analyzed monthly catch data of Sebastes schlegelii from 2009 to 2024.
• Evaluated lag-based correlations using cross-correlation functions among 15 marine environmental variables.
• Utilized Long Short-Term Memory (LSTM) model to predict catch based on correlated environmental factors.
• Assessed the accuracy of the LSTM model based on different input data types.
• High correlation found between catch and environmental variables including chemical oxygen demand, chlorophyll-a, temperature, and salinity.
• The model showed the highest accuracy when using selected environmental measurement data along with catch data.

Abstract