A dual-branch typhoon-induced wave height forecasting network with tail-aware extreme value optimization
Bibliographic record
Abstract
Accurate forecasting of typhoon-induced wave height (WH), which supports timely evacuation and informed emergency responses, is essential for the effectiveness of early warning systems. Despite recent advances in deep learning for WH forecasting, a critical gap persists: current models often fail to reliably predict rare but catastrophic extreme WH under typhoon conditions due to data scarcity. To address this challenge, we propose a physics-guided multi-scale attention framework, named the typhoon-induced wave height network (TWHN), which adopts a dual-branch architecture that separately captures wind sea and swell features. Unlike architectures that rely on initial WH inputs, TWHN forecasts WH directly from historical wind fields, thereby reducing error accumulation and supporting predictions at future time steps. To enhance the representation of extreme WH events, we introduce a tail-aware extreme value optimization (TEVO) strategy, which integrates a progressive training scheme to shift model focus from global patterns to tail data and a quantile-aware hybrid loss to penalize underestimation of high-magnitude waves. Additionally, a feature distribution smoothing mechanism is employed to stabilize training in data-sparse regimes by mitigating feature dominance from frequent samples. The model is trained, validated, and tested on WH records from 1982 to 2022, using a reanalysis dataset that includes 1 060 typhoons in the Northwest Pacific. Evaluation based on regional fields and nearshore station comparisons suggests that TWHN maintains strong potential for forecasting high-impact typhoon wave events. This work may provide implications for the advancement of operational wave forecasting and the support of risk decision-making in response to typhoon-induced marine hazards.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".