Reproduce “DIAL: Dynamic and Informed Adjustment of the Native Stochastic Law for Reliable Weather Ensembles” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yucheng Wang, Haonan Guo, Tianhao Zhao, Di Wang, Yu Wu, Zengmao Wang, Bo Du
DIAL controls ensemble spread during forecasts from frozen FCN3, targeting reliability across variables, spatial scales, and lead times. An Estimator predicts signed spread needs, and a separate Pilot plus learned future-cost estimates guide budget-limited changes to native randomness. On held-out 2022 ERA5 data, its normalized multiscale score is 8.27% better than Native FCN3 and 1.57% better than KL-PG, while pooled reliability passes its registered threshold; results are specific to this evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.