Research map / Video / Video segmentation and tracking
Digital Typhoon: Long-term Satellite Image Dataset for the Spatio-Temporal Modeling of Tropical Cyclones
Asanobu Kitamoto, Jared Hwang, Bastien Vuillod, Lucas Gautier, Yingtao Tian, Tarin Clanuwat
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| crop image around typhoon center instead of resizing (ablation) | ResNet18 · Pressure regression | RMSE (hPa): 10.47 -> 10.06 | Improved |
| crop instead of full 512x512 image (ablation) | ResNet18 · Pressure regression | RMSE (hPa): 10.51 -> 10.06 | Improved |
| crop instead of resizing (ablation) | ResNet50 · Pressure regression | RMSE (hPa): 11.63 -> 10.09 | Improved |
| crop instead of resizing (ablation) | ResNet18 · Wind regression | RMSE (kt): 10.09 -> 9.25 | Improved |
| crop instead of full image (ablation) | ResNet50 · Wind regression | RMSE (kt): 10.05 -> 9.13 | Improved |
| use ResNet50 instead of ResNet18 (ablation) | ResNet · Pressure regression, cropped | RMSE (hPa): 10.06 -> 10.09 | No clear effect |
| use ResNet50 instead of ResNet18 (ablation) | ResNet · Pressure regression, resized | RMSE (hPa): 10.47 -> 11.63 | Hurt |
| use ViT instead of VGG (ablation) | classification model · Grade classification, resized | accuracy (%): 67.2 -> 62.0 | Hurt |
| use VGG instead of ResNet18 (ablation) | classification model · Grade classification, cropped | accuracy (%): 67.8 -> 69.0 | Improved |
| crop instead of resizing (ablation) | ViT · Grade classification | accuracy (%): 62.0 -> 64.9 | Improved |
| increase forecast horizon from 1 to 12 hours (ablation) | ResNet18 on ConvLSTM images · Pressure forecasting | RMSE (hPa): 10.24 -> 14.69 | Hurt |
| train reanalysis model on first generation, test on second generation versus training on second (ablation) | ResNet18 · Reanalysis (208 sequences) | RMSE (hPa): 11.05 -> 12.80 | Hurt |
Closest papers by meaning
- Long-Term Typhoon Trajectory Prediction: A Physics-Conditioned Approach Without Reanalysis Data (ICLR 2024 · similarity 0.86)
- VQLTI: Long-Term Tropical Cyclone Intensity Forecasting with Physical Constraints (AAAI 2025 · similarity 0.83)
- Data-driven Precipitation Nowcasting Using Satellite Imagery (AAAI 2025 · similarity 0.83)
- IDOL: Meeting Diverse Distribution Shifts with Prior Physics for Tropical Cyclone Multi-Task Estimation (NeurIPS 2025 · similarity 0.83)
- DisasterM3: A Remote Sensing Vision-Language Dataset for Disaster Damage Assessment and Response (NeurIPS 2025 · similarity 0.83)
- TC-Diffuser: Bi-Condition Multi-Modal Diffusion for Tropical Cyclone Forecasting (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
