Analyze Code and U-Net Models For Capturing Subgrid Cold Pool Dynamics with U-Net: Insights fromLarge Eddy Simulation for Storm-Resolving Modeling

The ZIP files include the code and the trained U-Net models used to capture subgrid cold pool dynamics. The code for reproducing the figures can be found in draw_function.py, while the code related to deep learning model explainability is provided in explain.py. The trained models are located in the same folder with the prefix following the experimental setup. Those interested in the original CRM output data used to train the U-Net models may contact the authors directly.