Skip to content

Latest commit

 

History

History
24 lines (18 loc) · 1.9 KB

README.md

File metadata and controls

24 lines (18 loc) · 1.9 KB
Raw

Setup environment

Follow Dockerfile to setup a container environment for NVIDIA GPUs.

Prepare forecast data from GraphCast

Setup a Google Cloud Storage key named gcs_key.json and put it at GraphCast_src/

Execute python3 GraphCast_src/graphcast_runner.py --resolution .25 --pressure_levels 13 --autoregressive_steps 8 --test_year_start 1977 --test_year_end 2016 and python3 GraphCast_src/graphcast_runner.py --resolution .25 --pressure_levels 13 --autoregressive_steps 1 --test_year_start 1977 --test_year_end 2016 to generate 48h and 6h GraphCast forecast data need for training the diffusion model.

Prepare GraphCast weights and parameters

Visit: https://console.cloud.google.com/storage/browser/dm_graphcast and download GraphCast_operational - ERA5-HRES 1979-2021 - resolution 0.25 - pressure levels 13 - mesh 2to6 - precipitation output only.npz under params folder and the whole stats folder.

Train diffusion model

Execute python3 DiffDA/train_conditional_graphcast.py and pass in required arguments (hyperparameters, ERA5, forecast data path, etc.)

Run data assimilation

  • Execute python3 DiffDA/inference_data_assimilation.py --num_autoregressive_steps=1 ... to run single step data assimilation
  • Execute python3 DiffDA/inference_data_assimilation.py --num_autoregressive_steps=n ... (n > 2) to run autoregressive data assimilation
  • Execute python3 DiffDA/inference_data_assimilation_gc.py --num_autoregressive_steps=n ... to run (autoregressive) GraphCast forecast on single step assimilated data

Implementation detail

  • GraphCast_src/graphcast/normalization.py: ddpm and repaint algorithm for inference
  • DiffDA/train_conditional_graphcast.py: training diffusion model with GraphCast as backbone
  • DiffDA/inference_data_assimilation.py: run single step & autoregressive data assimilation
  • DiffDA/inference_data_assimilation_gc.py: run GraphCast forecast on single step assimilated data