Table 3 Comparison of the mean absolute distance errors (km) predicted by multiple deep-learning models. Bold values highlight the best performance.
From: Prediction of surface drifter trajectories in the South China sea using deep learning
6 h | 12 h | 18 h | 24 h | |||||
|---|---|---|---|---|---|---|---|---|
RMSE | MAE | RMSE | MAE | RMSE | MAE | RMSE | MAE | |
RNN | 0.1423 | 12.7105 | 0.1552 | 14.3501 | 0.2237 | 20.5022 | 0.2493 | 22.7264 |
LSTM | 0.1249 | 10.9243 | 0.1588 | 13.7308 | 0.2138 | 18.1094 | 0.2512 | 21.6609 |
GRU | 0.1914 | 16.9772 | 0.2451 | 22.2429 | 0.2540 | 22.2866 | 0.2903 | 24.8279 |
Transformer | 0.1640 | 14.2004 | 0.2508 | 24.2498 | 0.3423 | 31.8949 | 0.2830 | 24.7458 |
Informer | 0.1046 | 8.6226 | 0.1016 | 8.4038 | 0.1418 | 11.7772 | 0.1962 | 15.5273 |
