Search

Accurate predictions of extreme water temperatures are criticalto understanding the variability of the marine environment and reducing marine disasters maximized by global warming. In this study, we propose a two-pathway framework with separated spatial and temporal encoders for accurate prediction of water temperature, especially extremely high water temperature, through effective spatiotemporal representation learning. The spatial and temporal encoder networks based on the Transformer's self-attention mechanism performs the task of predicting the water temperature time series at the 16 coastal locations around the Korean Peninsula for the seven consecutive days ahead at daily intervals with various combinations of patch embedding methods, positional embedding for spatial features. Comparative experiments with conventional deep convolutional and recurrent networks are also conducted for comparison. By comparing and assessing these results, the proposed two-pathway framework can improve the predictability of extremely high coastal water temperature by better capturing spatiotemporal interrelationships and long-range dependencies from open ocean and regional sea, and further determines the optimal architectural details of self-attention-based spatial and temporal encoders. Furthermore, to examine the explainability of the proposed model and its consistency with domain knowledge, spatial and temporal attention maps are visualized and analyzed that represents weights for spatiotemporal input sequences that are more relevant to predict for future predictions. • Development of two-pathway spatiotemporal representation learning framework for consecutive multi-step-ahead SST prediction. • Spatiotemporal representation learning to capture the temporally correlated regional to local spatial dependencies. • Notable predictability of extreme coastal SST for consecutive 7-day ahead spatiotemporal forecasts. • Visualization of attention map to secure domain knowledge-based validity and explainability of model predictions. • Discovering spatiotemporal teleconnections of SST variability and its consistency with ocean dynamics. [ABSTRACT FROM AUTHOR]

This study proposes a spatio-temporal geographical information-guided neural network to predict multi-step ahead space-time series of ocean waves. The network is designed to learn the ocean wave dynamics from external atmospheric forcing and internal wave processes. It also captures the nonlinear relationships in multiple input and at the spatial and temporal levels and shares their dependencies. The model learns these dependencies through a convoluted encoded latent feature, while a decoder predicts the space-time series of ocean waves from the latent representations. The model is trained on 35 years of a state-of-the-art global reanalysis dataset produced at 1-hour temporal and 25 km spatial resolutions around the Korean Peninsula. It is evaluated by predicting the same resolution's multi-step ahead space-time series of ocean waves for a 48-hour forecast lead time for the 5 years not used for training. We conducted an ablation experiment to determine the optimal model architecture, input variable, and training period. The prediction results are compared and analyzed with the in-situ ocean wave measurements at the 18 observation stations. We consider the prediction results according to the start time of prediction with the in-situ measurements and hindcast results to examine the performance on the high waves that caused wave-induced disaster. • STG-OceanWaveNet is developed to accurately predict waves by emulating wave dynamics. • The RMSE of 48-hours ahead SWH and MWP is 0.4 m and 0.5 sec. for the 5 years. • Compared with in-situ measurements, the RMSE for SWH and MWP are 0.35 m and 0.55 sec. • The deep learning approach is feasible to predict ocean waves with reliable accuracy. [ABSTRACT FROM AUTHOR]