BiST-SA-LSTM: A Deep Learning Framework for End-to-End Prediction of Mesoscale Eddy Distribution in Ocean
Mesoscale eddies play a critical role in sea navigation and route planning, yet traditional prediction methods have often overlooked their spatial relationships, relying on indirect approaches to capture their distribution across extensive maps. To address this limitation, we present BiST-SA-LSTM, a...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Journal of Marine Science and Engineering |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2077-1312/13/1/52 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Mesoscale eddies play a critical role in sea navigation and route planning, yet traditional prediction methods have often overlooked their spatial relationships, relying on indirect approaches to capture their distribution across extensive maps. To address this limitation, we present BiST-SA-LSTM, an end-to-end prediction framework that combines Bidirectional Spatial Temporal LSTM and Self-Attention mechanisms. Utilizing data sourced from the South China Sea and its surrounding regions, which are renowned for their intricate maritime dynamics, our methodology outperforms similar models across a range of evaluation metrics and visual assessments. This is particularly evident in our ability to provide accurate long-term forecasts that extend for up to 10 days. Furthermore, integrating sea surface variables enhances forecasting accuracy, contributing to advancements in oceanic physics. |
|---|---|
| ISSN: | 2077-1312 |