In this study, we used convolutional neural networks (CNNs)—which are well-known deep learning models suitable for image data processing—to estimate the temporal and spatial distribution of chlorophyll-a in a bay. The training data required the construction of a deep learning model acquired from the satellite ocean color and hydrodynamic model. Chlorophyll-a, total suspended sediment (TSS), visibility, and colored dissolved organic matter (CDOM) were extracted from the satellite ocean color data, and water level, currents, temperature, and salinity were generated from the hydrodynamic model. We developed CNN Model I—which estimates the concentration of chlorophyll-a using a 48 × 27 sized overall image—and CNN Model II—which uses a 7 × 7 segmented image. Because the CNN Model II conducts estimation using only data around the points of interest, the quantity of training data is more than 300 times larger than that of CNN Model I. Consequently, it was possible to extract and analyze the inherent patterns in the training data, improving the predictive ability of the deep learning model. The average root mean square error (RMSE), calculated by applying CNN Model II, was 0.191, and when the prediction was good, the coefficient of determination (R²) exceeded 0.91. Finally, we performed a sensitivity analysis, which revealed that CDOM is the most influential variable in estimating the spatiotemporal distribution of chlorophyll-a.

This study investigates the optimal particle density for accurate wave simulation in non-breaking and breaking wave conditions using Smoothed Particle Hydrodynamics (SPH). Simulations were conducted using various particle densities, or particles per wavelength, to evaluate their impact on simulation accuracy and computational efficiency. In tests involving regular, non-breaking waves, particle densities in the range of np per L = 560 to 800 were found to provide satisfactory accuracy and computational efficiency. Increasing the particle density beyond np per L = 800 for more complex setups involving breaking waves did not significantly improve accuracy but resulted in much higher computational costs. Overall, the range of np per L = 560 to 800 offers the best balance between accuracy and efficiency, making it a practical choice for simulations with limited resources. This study underscores the trade-offs between accuracy and computational demand, providing guidance on the selection of particle density for various wave simulation scenarios.