A collection of challenges and baseline models for the SEVIR weather dataset.

• Radar Nowcasting
• Others coming soon

The challenges in this repo are based on the SEVIR weather dataset. This dataset is made of up sequences of weather imagery sampled and aligned across radar and satellite. It was constucted as a benchmark dataset to support algorithm development in meterology. For a detailed tutorial on this dataset, see the SEVIR tutorial.

SEVIR is currently available for download from the AWS Open Data registry. In total, the dataset is approximately 1TB in size, however smaller samples of the full dataset are provided for selected challenges (see s3://sevir/data/processed/). To construct larger datasets, you can download SEVIR using one of the following methods:

If you have AWS CLI, you can download SEVIR using the

aws s3 sync --no-sign-request s3://sevir .

To download only a specific modalitiy, e.g. vil, you can instead run

aws s3 cp --no-sign-request s3://sevir/CATALOG.csv CATALOG.csv
aws s3 sync --no-sign-request s3://sevir/data/vil .

Using the python boto3 modules (conda install boto3) you can obtain SEVIR data by first connecting to the S3 bucket

import boto3
from botocore.handlers import disable_signing
resource = boto3.resource('s3')
resource.meta.client.meta.events.register('choose-signer.s3.*', disable_signing)
bucket=resource.Bucket('sevir')

Then, get a list of files using

objs=bucket.objects.filter(Prefix='')
print([o.key for o in objs])

Finally, download files of interest from this list, e.g.

bucket.download_file('CATALOG.csv','/home/data/SEVIR/CATALOG.csv')
bucket.download_file('data/vil/2017/SEVIR_VIL_STORMEVENTS_2017_0701_1231.h5','/home/data/SEVIR/data/vil/2017/SEVIR_VIL_STORMEVENTS_2017_0701_1231.h5')
#... etc