This is scripts for 4D-Var using a neural network surrogate model obtained by machine learning for Lorenz 96 model.
-
Misc
- README.md: this file
- LICENSE: license file
-
For Experiments
- lorenz96.py: Lorenz 96 model
- net.py: neural network model
- train_1step.py: for one-step learing
- train_10step.py: for ten-step learing
- 4dvar.py: for 4D-Var experiment without observation error
- 4dvar_err.py: for 4D-Var experiment with observation error
- 4dvar_phy.py: for 4D-Var experiment with manually constructed adjoint model
-
For Analysis and Visualization
- run_err.py
- learning_curve_1step.py
- graph_loss.py
- graph_4dvar.py
- accuracy_4dvar.py
- lorenz96.rb
- learning.rb
- cost.rb
- 4dvar.rb
- 4dvar_rmse.rb