GPM IMERG Half-Hourly Virtualizarr Data Pipeline (shorthand: gpmimerghh-vdp)

This repo uses the Virtualizarr Data Pipelines (VDP) github template repository intended to help users create and manage Virtualizarr/Icechunk stores on AWS in a consistent, scalable way.

Read the virtualizarr-data-pipelines (VDP) documentation to learn more about the cloud architecture. Then head to the design doc to learn about how VDP was used for this dataset.

Local development

Set up the development environment

./scripts/setup.sh

Run unit tests

uv run pytest

Unit tests run against local fixtures and require no AWS access.

Run integration tests

Integration tests hit real S3 buckets and require valid AWS credentials and the ICECHUNK_BUCKET / ICECHUNK_PREFIX environment variables to be set:

# set AWS credentials in your environment
uv run pytest -m integration

Cloud Deployment

NASA Earthdata credentials

GPM IMERG files require NASA Earthdata authentication. Store the credentials in AWS Secrets Manager rather than as plain text in .env:

Create the secret (once, before deploying):

aws secretsmanager create-secret \
  --name "<your-stack-name>/earthdata-credentials" \
  --description "NASA Earthdata credentials for the virtualizarr pipeline" \
  --secret-string '{"username":"<your-username>","password":"<your-password>"}' \
  --region <your-region>

Then set EARTHDATA_SECRET_ARN in your .env file to the ARN returned by the command above:

EARTHDATA_SECRET_ARN=arn:aws:secretsmanager:<region>:<account-id>:secret:<your-stack-name>/earthdata-credentials-<suffix>

The Lambda functions fetch the secret and set the credentials as environment variables for the Earthdata S3 credential provider.

Icechunk bucket settings

There are two bucket-related settings that control where the Icechunk store is written:

• ICECHUNK_BUCKET_NAME — the name of a new S3 bucket CDK will create for the store (default: icechunk-outuput).
• ICECHUNK_BUCKET — the name of an existing S3 bucket to use instead. When this is set, CDK will reference the bucket rather than create it.

If you already have a bucket (e.g. nasa-eodc-public), set ICECHUNK_BUCKET=nasa-eodc-public in your .env file to avoid the already exists error on deploy.

You can also set ICECHUNK_PREFIX for any additional path to the icechunk store.

Project commands 🔨

To set up the development environment

./scripts/setup.sh

Run unit tests

uv run pytest

Unit tests run against local fixtures and require no AWS access.

Run integration tests

Integration tests hit real S3 buckets and require valid AWS credentials and the ICECHUNK_BUCKET / ICECHUNK_PREFIX environment variables to be set:

# set AWS credentials in your environment
uv run pytest -m integration

Review your infrastructure before deploying

uv run --env-file .env.sample cdk synth

EARTHDATA_SECRET_ARN=arn:aws:secretsmanager::㊙️/earthdata-credentials-

The Lambda functions fetch the secret and set the credentials as environment variables for the Earthdata S3 credential provider.

### Create the .env file for other custom settings

```bash
cp .env.example .env

Modify .env as needed to customize the settings made available in cdk/settings.py.

Icechunk bucket settings

There are two bucket-related settings that control where the Icechunk store is written:

• ICECHUNK_BUCKET_NAME — the name of a new S3 bucket CDK will create for the store (default: icechunk-outuput).
• ICECHUNK_BUCKET — the name of an existing S3 bucket to use instead. When this is set, CDK will reference the bucket rather than create it.

If you already have a bucket (e.g. nasa-eodc-public), set ICECHUNK_BUCKET=nasa-eodc-public in your .env file to avoid the already exists error on deploy.

You can also set ICECHUNK_PREFIX for any additional path to the icechunk store.

Review your infrastructure before deploying

uv run --env-file .env.sample cdk synth

Deploy the CDK infrastructure.

uv run --env-file .env.sample cdk deploy