Fig. 5: Comparative analysis of prediction errors in sound speed for March and September 2015 using our model based on WOA data.
From: Leveraging satellite observations and machine learning for underwater sound speed estimation
Top panels: A, B show depth-based heatmaps of prediction errors for March and September 2015, with the color spectrum from blue to red indicating error frequency, and gray areas representing fewer than 50 predictions. Bottom Panel: C presents the Mean Absolute Error (MAE) as a function of depth for March (blue) and September (red) 2015.
