Research map / Probabilistic and scientific modeling / Dynamical systems
Conditional Local Convolution for Spatio-Temporal Meteorological Forecasting
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace the baseline spatial-temporal model with CLCRN | WeatherBench forecasting · Temperature, Cloud cover, Humidity, Wind (12h) | MAE and RMSE: Temperature MAE 1.2551±0.0080 -> 1.1688±0.0457; Cloud cover MAE 1.5925±0.0023 -> 1.4906±0.0037; Humidity MAE 0.5007±0.00 | Helped |
| Use horizon maps instead of logarithmic maps (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.2638±0.1554 -> 1.1688±0.0457; Cloud cover 1.5599±0.0019 -> 1.4906±0.0037; Humidity 0.4663±0.0082 -> 0.4531 | Helped |
| Use recurrent CLCRN rather than attention-based CLCSTN (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.3325±0.2204 -> 1.1688±0.0457; Cloud cover 1.7483±0.0044 -> 1.4906±0.0037; Humidity 0.5691±0.0385 -> 0.4531 | Helped |
| Use only angle scaling rather than the full kernel (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.1205±0.2031 vs 1.1688±0.0457; Cloud cover 1.4919±0.0019 vs 1.4906±0.0037; Humidity 0.4538±0.0064 vs 0.4531 | Hurt |
| Use only distance scaling rather than the full kernel (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: 16.5671±0.0002; 2.7308±0.0002; 1.3443±0.0001; 4.0531±0.0000 | Hurt |
| Use only the MLP kernel rather than the full kernel (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.8815±0.1934 -> 1.1688±0.0457; Cloud cover 1.9047±0.0023 -> 1.4906±0.0037; Humidity 0.6208±0.1074 -> 0.4531 | Hurt |
| Add distance scaling to the MLP kernel (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.4505±0.2248; Cloud cover 1.8743±0.0038; Humidity 0.6289±0.0711; Wind 2.5454±0.3485 | Hurt |
| Add angle scaling to the MLP kernel (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.1205±0.2031; Cloud cover 1.4919±0.0019; Humidity 0.4538±0.0064; Wind 1.2991±0.0344 | Mixed |
| Combine angle and distance scaling without the MLP (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: Temperature 1.4986±0.0872; Cloud cover 1.6907±0.0219; Humidity 0.5378±0.0363; Wind 1.8932±0.2488 | Hurt |
| Add distance scaling to MLP + Angle (ablation) | CLCRN · Temperature, Cloud cover, Humidity, Wind (12h) | MAE: MLP + Angle: 1.1205±0.2031, 1.4919±0.0019, 0.4538±0.0064, 1.2991±0.0344; full: 1.1688±0.0457, 1.4906±0.0037, 0.4531±0.00 | Mixed |
| Increase the number of graph-convolution layers (ablation) | CLCRN · WeatherBench | MAE | Hurt |
| Increase neighbor count K (ablation) | CLCRN · WeatherBench | MAE | Mixed |
Closest papers by meaning
- On the Integration of Spatial-Temporal Knowledge: A Lightweight Approach to Atmospheric Time Series Forecasting (NeurIPS 2025 · similarity 0.87)
- ST-FiT: Inductive Spatial-Temporal Forecasting with Limited Training Data (AAAI 2025 · similarity 0.87)
- Graph-based Forecasting with Missing Data through Spatiotemporal Downsampling (ICML 2024 · similarity 0.87)
- Expand and Compress: Exploring Tuning Principles for Continual Spatio-Temporal Graph Forecasting (ICLR 2025 · similarity 0.86)
- Mesh Interpolation Graph Network for Dynamic and Spatially Irregular Global Weather Forecasting (NeurIPS 2025 · similarity 0.86)
- ST-ReP: Learning Predictive Representations Efficiently for Spatial-Temporal Forecasting (AAAI 2025 · similarity 0.86)
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.