Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Probabilistic Rainfall Downscaling: Joint Generalized Neural Models with Censored Spatial Gaussian Copula
David Huk, Rilwan A. Adewoyin, Ritabrata Dutta
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 censored latent Gaussian copula with joint generalised neural marginals (Cens-JGNM) instead of ConvCNP | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | Energy score: 17.0146 -> 15.7253 | Improved |
| use Cens-JGNM instead of VAE-GAN | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | Variogram score: 9,091,811 -> 6,862,187 | Improved |
| use Cens-JGNM instead of ConvCNP on marginals | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | CRPS: 3.2678 -> 3.5631 | Hurt |
| use Cens-JGNM instead of VAE-GAN | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | CRPS: 4.1146 -> 3.5631 | Improved |
| use Cens-JGNM instead of deterministic IFS | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | RMSE: 3.756 -> 3.8194 | Hurt |
| use Cens-JGNM instead of deterministic IFS | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | MAB: 2.118 -> 2.1438 | Hurt |
| use Cens-JGNM instead of ConvCNP | probabilistic rainfall downscaling · UK rainfall downscaling (ERA5 to E-OBS) | RMSE: 3.9314 -> 3.8194 | Improved |
| train on 10 years instead of 4 years of data (ablation) | Cens-JGNM · UK rainfall downscaling (training-data length study) | Energy score: 16.0627 -> 15.9226 | No clear effect |
| train on 10 years instead of 4 years of data (ablation) | Cens-JGNM · UK rainfall downscaling (training-data length study) | RMSE: 4.0362 -> 3.9724 | No clear effect |
| train on 8 years instead of 6 years of data (ablation) | Cens-JGNM · UK rainfall downscaling (training-data length study) | CRPS: 3.4322 -> 3.6996 | No clear effect |
| fix topography weight a=0 instead of jointly optimising a (ablation) | copula distance matrix · UK rainfall downscaling (copula topography weight) | Energy score: 15.7852 -> 15.7253 | Improved |
| add topography weight a=0.005 to the copula distance (ablation) | copula distance matrix · UK rainfall downscaling (copula topography weight) | Energy score: 15.7253 -> 15.7394 | Hurt |
Closest papers by meaning
- Quantile-Regression-Ensemble: A Deep Learning Algorithm for Downscaling Extreme Precipitation (AAAI 2024 · similarity 0.85)
- Precipitation Downscaling with Spatiotemporal Video Diffusion (NeurIPS 2024 · similarity 0.85)
- Data-driven Precipitation Nowcasting Using Satellite Imagery (AAAI 2025 · similarity 0.85)
- Fourier Amplitude and Correlation Loss: Beyond Using L2 Loss for Skillful Precipitation Nowcasting (NeurIPS 2024 · similarity 0.84)
- PostCast: Generalizable Postprocessing for Precipitation Nowcasting via Unsupervised Blurriness Modeling (ICLR 2025 · similarity 0.84)
- PrecipDiff: Leveraging Image Diffusion Models to Enhance Satellite-Based Precipitation Observations (AAAI 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.
