Star 0
Version 0.1.2
FLEXPART-IFS Atmospheric Dispersion Workflow
Maintainers
nina.burgdorfer[at]meteoswiss.ch
Support level
Community
Category
Data Access Data Processing Data Visualization
Technology
Ansible Playbook

FLEXPART-IFS Atmospheric Dispersion Workflow

This blueprint deploys a ready-to-use FLEXPART-IFS atmospheric transport and dispersion workflow on a virtual machine in the European Weather Cloud (EWC).

The reference deployment has been validated on the ECMWF side of the EWC using EWCCLI. Deployment on the EUMETSAT side has not yet been tested.

FLEXPART is a Lagrangian atmospheric transport and dispersion model used to simulate the transport and dispersion of air masses and atmospheric tracers. This workflow runs FLEXPART-IFS using meteorological input from the ECMWF Integrated Forecasting System (IFS).

The complete processing chain is handled by the deployed workflow:

ECMWF MARS -> flexprep -> FLEXPART-IFS -> Pyflexplot

IFS forecast data are retrieved directly from ECMWF MARS and prepared for FLEXPART with flexprep:

https://gitlab.phaidra.org/flexpart/flexprep

FLEXPART-IFS then performs the atmospheric transport and dispersion simulation:

https://github.com/MeteoSwiss/flexpart

The resulting FLEXPART output is processed with Pyflexplot to generate PDF visualisations of the simulated dispersion:

https://github.com/MeteoSwiss-APN/pyflexplot

Workflow orchestration is provided by the MeteoSwiss flex-container-orchestrator:

https://github.com/MeteoSwiss/flex-container-orchestrator

After deployment, simulations can be started from the command line using the flexpart-workflow command.

For example, the following command selects the latest available IFS forecast cycle, runs a 24-hour simulation, and uses BEZ (Beznau) as the release location:

flexpart-workflow --latest --step 24 --location BEZ

Users can also select an explicit forecast date and time instead of the latest available cycle, allowing simulations to be reproduced for a known IFS forecast cycle.

The reference deployment has been validated on Ubuntu 24.04 with 8 vCPU, 16 GiB RAM, and 80 GB storage. This configuration is the recommended baseline for the tested workflow.

The blueprint is compatible with deployment through EWCCLI. It has been tested using an ECMWF-side EWCCLI profile. Users should follow the deployment instructions in the blueprint repository for the currently supported EWCCLI version and configuration.

If the native ECMWF MARS client is not already available on the virtual machine, the blueprint applies the official ECMWF Data Flavour 2.0.2 before installing the FLEXPART-IFS workflow.

Access to ECMWF MARS is user-specific. MARS credentials and data entitlements are not included in the blueprint. Users must provide their own credentials and have permission to retrieve the requested IFS data.

Organisation-specific network configuration, including VPNs, proxies, or SSH bastions, is deliberately kept outside the blueprint. Such configuration should be provided through the user's local SSH or Ansible environment when required.

Source code, deployment instructions, prerequisites, and usage examples are available in the blueprint repository:

https://github.com/MeteoSwiss/flexpart-ifs-ewc-blueprint

Other
Deployable
EWCCLI-compatible