jasong-ovo / int_mpas_onnx Goto Github PK

If you use a CPU environment, please run:

pip install -r requirements_cpu.txt

After the above steps are finished, please check inference_cpu.py for an example of making a 12-min weather forecast on CPU with the 12-min model.

For example, running the following command, one can get the 12-min forecast in the output_data folder:

python inference_cpu.py

Also, inference_iterative.py shows an example to generate per-12-min forecast within 6 hour.

Data should be converted to two numpy .npy file. The first .npy is for node which has a shape (1, 256002,221) and the second .npy is for edge with a shape of (1, 768000, 55). The channel order of node is theta rho qv w, and the height level order is 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54(, 55 for w). Edge vairiable is u with level order 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55.

The model has 5 parts (~5.5GB totally) from Google drive.

The 12-min model (ft12minxxx): xxx The 1-hour model (ft1hxxx): xxx put the 12-min model into onnx_models/12min and 1h model into onnx_models/1h and change model parts in inference_cpu.py and inference_iterative.py.