MetNet

A neural network that forecasts precipitation up to 8 hours into the future at the high spatial resolution of 1 km² and at the temporal resolution of 2 minutes with a latency in the order of second.
https://github.com/openclimatefix/metnet

Category: Atmosphere
Sub Category: Meteorological Observation and Forecast

Keywords

pytorch

Keywords from Contributors

nowcasting solar forecasting-models graph-neural-networks weather gan compression pytorch-lightning pytorch-implementation nowcasting-precipitation

Last synced: about 16 hours ago
JSON representation

Repository metadata

PyTorch Implementation of Google Research's MetNet and MetNet-2

• Host: GitHub
• URL: https://github.com/openclimatefix/metnet
• Owner: openclimatefix
• License: mit
• Created: 2021-09-02T11:20:11.000Z (over 3 years ago)
• Default Branch: main
• Last Pushed: 2025-05-05T17:22:10.000Z (12 days ago)
• Last Synced: 2025-05-09T20:03:58.039Z (8 days ago)
• Topics: pytorch
• Language: Python
• Homepage:
• Size: 185 KB
• Stars: 265
• Watchers: 5
• Forks: 54
• Open Issues: 33
• Releases: 43
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

README.md

pip install -r requirements.txt
pip install -e .

from datasets import load_dataset

dataset = load_dataset("openclimatefix/goes-mrms")

from metnet import MetNet, MetNet2
model = MetNet().from_pretrained("openclimatefix/metnet")
model = MetNet2().from_pretrained("openclimatefix/metnet-2")

from metnet import MetNet
import torch
import torch.nn.functional as F

model = MetNet(
        hidden_dim=32,
        forecast_steps=24,
        input_channels=16,
        output_channels=12,
        sat_channels=12,
        input_size=32,
        )
# MetNet expects original HxW to be 4x the input size
x = torch.randn((2, 12, 16, 128, 128))
out = []
for lead_time in range(24):
        out.append(model(x, lead_time))
out = torch.stack(out, dim=1)
# MetNet creates predictions for the center 1/4th
y = torch.randn((2, 24, 12, 8, 8))
F.mse_loss(out, y).backward()

from metnet import MetNet2
import torch
import torch.nn.functional as F

model = MetNet2(
        forecast_steps=8,
        input_size=64,
        num_input_timesteps=6,
        upsampler_channels=128,
        lstm_channels=32,
        encoder_channels=64,
        center_crop_size=16,
        )
# MetNet expects original HxW to be 4x the input size
x = torch.randn((2, 6, 12, 256, 256))
out = []
for lead_time in range(8):
        out.append(model(x, lead_time))
out = torch.stack(out, dim=1)
y = torch.rand((2,8,12,64,64))
F.mse_loss(out, y).backward()

Owner metadata

• Name: Open Climate Fix
• Login: openclimatefix
• Email: [email protected]
• Kind: organization
• Description: Using open science to mitigate climate change
• Website: openclimatefix.org
• Location: London
• Twitter: openclimatefix
• Company:
• Icon url: https://avatars.githubusercontent.com/u/48357542?v=4
• Repositories: 88
• Last ynced at: 2024-04-15T07:32:15.529Z
• Profile URL: https://github.com/openclimatefix

GitHub Events

Total

• Create event: 1
• Release event: 1
• Issues event: 2
• Watch event: 29
• Issue comment event: 17
• Push event: 41
• Pull request event: 1
• Pull request review event: 3
• Pull request review comment event: 11
• Fork event: 5

Last Year

• Create event: 1
• Release event: 1
• Issues event: 2
• Watch event: 29
• Issue comment event: 17
• Push event: 41
• Pull request event: 1
• Pull request review event: 3
• Pull request review comment event: 11
• Fork event: 5

Committers metadata

Last synced: 8 days ago

Committer domains:

• openclimatefix.org: 1
• bieker.tech: 1

Issue and Pull Request metadata

Last synced: 1 day ago

Package metadata

• Total packages: 1
• Total downloads:
  • pypi: 629 last-month
• Total dependent packages: 0
• Total dependent repositories: 2
• Total versions: 42
• Total maintainers: 2

Dependencies

Score: 14.3447884859858