Comparison of scenario reduction approaches for reservoir inflow timeseries generated by a Bayesian Neural Network
Ja-Ho Koo, Edo Abraham, Andreja Jonoski, Dimitri P Solomatine · 2026
Read Comparison of scenario reduction approaches for reservoir inflow timeseries generated by a Bayesian Neural Network by Ja-Ho Koo, Edo Abraham, Andreja Jonoski, Dimitri P Solomatine (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0350095.txt.
Description
Ja-Ho Koo and Dimitri P Solomatine led a study of flood planning for dams. Forecast tools spawn thousands of paths a river might take. No team can plan for them all. So the set must shrink, but stay true. The team grew paths with a learning model that states its own doubt. Then they tried four ways to prune the set.
The model caught the flow's rhythm and scored well short-term. It still missed peak floods, for lack of past data. One pruning method, based on energy distance, kept the set's stats most true. Plain distance methods kept the wild flood paths best. Those paths matter most for safety. The team lays out the trade-offs for dam crews.
Catalog record
| CREATOR | Ja-Ho Koo, Edo Abraham, Andreja Jonoski, Dimitri P Solomatine |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60499 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |