Research map / Probabilistic and scientific modeling / Differential equations and PDEs
DeepPrim: a Physics-Driven 3D Short-term Weather Forecaster via Primitive Equation Learning
Jiawei Chen, Weiqi Chen, Rong Hu, Peiyuan Liu, Haifan Zhang, Liang Sun
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 |
|---|---|---|---|
| add 3D modules (cross-pressure self-attention and pressure-level derivatives) (ablation) | DeepPrim · ERA5 global forecasting | RMSE: average degradation of 12.3% when removed | Improved |
| add temporal variation and 3D spatial derivative of initial conditions to initialization network (ablation) | DeepPrim · ERA5 global forecasting | RMSE | Improved |
| add the 3D spatial derivative term in initialization (ablation) | DeepPrim · ERA5 global forecasting | RMSE | Improved |
| add source-sink network (ablation) | DeepPrim · ERA5 global forecasting | RMSE | Improved |
| train on only five key output variables instead of all variables | DeepPrim · ERA5 5.625 deg | RMSE: DeepPrim† generally outperforms full-output-variable DeepPrim | Improved |
| use DeepPrim instead of ClimODE | DeepPrim · ERA5 5.625 deg | RMSE: outperforming ClimODE by 38.6% | Improved |
| use DeepPrim instead of pre-trained ClimaX | DeepPrim · ERA5 5.625 deg | RMSE: by 10.7% on average | Improved |
| use DeepPrim instead of pre-trained ClimaX | DeepPrim · ERA5 1.40625 deg | RMSE: by 36.11% on average | Improved |
| use DeepPrim instead of GraphCast | DeepPrim · ERA5 0.25 deg z500 | RMSE: 44.8 vs 41.0 | Hurt |
| use DeepPrim instead of second-best baseline | DeepPrim · regional forecasting (3 regions) | RMSE: by 35.8% on average | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- ClimODE — ClimODE: Climate and Weather Forecasting with Physics-informed Neural ODEs (ICLR 2024) · better on 6 of 6 shared results
- ClimaX — ClimaX: A foundation model for weather and climate (ICML 2023) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ClimODE: Climate and Weather Forecasting with Physics-informed Neural ODEs (ICLR 2024 · similarity 0.85)
- OmniCast: A Masked Latent Diffusion Model for Weather Forecasting Across Time Scales (NeurIPS 2025 · similarity 0.84)
- DiffDA: a Diffusion model for weather-scale Data Assimilation (ICML 2024 · similarity 0.84)
- Scaling Storm-Resolving Atmospheric AI Simulation to the Entire Planet (NeurIPS 2026 · similarity 0.84)
- PINP: Physics-Informed Neural Predictor with latent estimation of fluid flows (ICLR 2025 · similarity 0.84)
- HR-Extreme: A High-Resolution Dataset for Extreme Weather Forecasting (ICLR 2025 · similarity 0.84)
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.
