paper
2023
Hybrid physics-ML & parameterization
Explainability, uncertainty & physics New model or method
CNN / U-NetPhysics-informed / physics-constrainedColumbia University (LEAP, Lamont-Doherty) Other
2023-06-23
026
Jun 2023
Sara Shamekh, Pierre Gentine
preprint
Columbia University (LEAP, Lamont-Doherty)
Physically constrained neural network with a 2D latent space predicts boundary-layer turbulent fluxes from LES data, outperforming the eddy-diffusivity mass-flux scheme.
Hybrid physics-ML & parameterization
Learning Atmospheric Boundary Layer Turbulence
In one sentence. Physically constrained neural network with a 2D latent space predicts boundary-layer turbulent fluxes from LES data, outperforming the eddy-diffusivity mass-flux scheme.
Labels
- Topic
- Hybrid physics-ML & parameterization, Explainability, uncertainty & physics
- Type
- New model or method
- Method
- CNN / U-Net, Physics-informed / physics-constrained
- Domain
- Local / station
- Data
- Other
- Status
- Research
- Identifiers
- DOI:10.22541/essoar.168748456.60017486/v1 OpenAlex:W4381743464
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.