File size: 2,396 Bytes
9b62a9b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
{
  "model_name": "Spherical Fourier Neural Operators",
  "model_type": "sfno",
  "architectures": [
    "SFNO"
  ],
  "framework": "PyTorch",
  "domain": "climate-and-atmosphere",
  "task": "global-weather-forecasting",
  "implementation": {
    "entry_point": "model/sfno.py",
    "scope": "configurable wrapper around torch_harmonics.examples.models.sfno.SphericalFourierNeuralOperator for deterministic single-state 6-hour forecasting"
  },
  "architecture": {
    "family": "Spherical Fourier Neural Operator",
    "spectral_operator": "spherical harmonic transform",
    "input_format": "B C H W",
    "prediction": "one atmospheric state to the next 6-hour state; longer forecasts use autoregressive rollout",
    "activation": "GELU",
    "normalization": "InstanceNorm",
    "repository_default_config": {
      "purpose": "connectivity validation with synthetic ERA5 data",
      "img_size": [
        32,
        64
      ],
      "scale_factor": 2,
      "in_channels": 6,
      "out_channels": 6,
      "embed_dim": 16,
      "num_layers": 2,
      "use_mlp": true,
      "mlp_ratio": 2.0,
      "drop_rate": 0.0,
      "drop_path_rate": 0.0,
      "hard_thresholding_fraction": 1.0,
      "residual_prediction": false,
      "positional_embedding": "none",
      "bias": false
    },
    "paper_reference_config": {
      "grid_resolution_degrees": 0.25,
      "grid_size": [
        721,
        1440
      ],
      "channels": "26 or 73",
      "embed_dim": 256,
      "num_layers": "approximately 8",
      "note": "the weather model's exact internal spectral downsampling factor is not disclosed"
    }
  },
  "data": {
    "dataset": "ERA5",
    "variables": [
      "10m_u_component_of_wind",
      "10m_v_component_of_wind",
      "2m_temperature",
      "mean_sea_level_pressure",
      "geopotential_500",
      "temperature_850"
    ],
    "time_step_hours": 6,
    "input_length": 1,
    "output_length": 1,
    "channels": 6,
    "spatial_size": [
      32,
      64
    ],
    "storage_format": "HDF5 fields with T C H W layout",
    "train_years": [
      1951,
      1952
    ],
    "validation_years": [
      1953
    ],
    "test_years": [
      1954
    ]
  },
  "configuration_sources": [
    "README.md",
    "conf/config.yaml",
    "model/sfno.py",
    "scripts/train.py",
    "scripts/inference.py",
    "scripts/fake_data.py",
    "configuration.json"
  ]
}