O3as Project

Motivation

Monitoring and projecting stratospheric ozone is mandated by UN Environment to safeguard a healthy planet. Regularly many climate models project future climate and ozone change, producing huge amounts of data that have to be analysed for key metrics. Those key metrics help policy makers to judge if measures implemented to protect the stratospheric ozone layer are working.

Project goals

To provide a framework to efficiently explore ozone projections, including the calculation of key metrics:

  • Improve the existing workflow and provide a reliable tool for scientists to perform analysis in a more efficient manner

  • Ensure reproducibility of results

  • Simplify data access and the use

  • Publish high-level data to citizens

Problem to solve

  • A typical workflow of today has many manual steps

  • Full processing from raw data takes hours

  • The code is not always accessible or well maintained

Consequences:
=> Plots are not easy to rebuild for various inputs
=> Possible inconsistency in results
=> No-way to assess the results by non-specialists

O3as solution

O3as service general structure

O3as service general structure

  • Climate Models data (10’s TB) are collected in one place (Large Scale Data Facility, LSDF at KIT)

  • The data are reduced to the parameters of interest and homogenized at HPC by the means of o3skim component

  • The reduced data (100’s MB) can be accessed with the REST API in seconds (o3api component)

  • A user may do final processing and plotting by leveraging the WebApp (o3webapp to come)

All components are open source (GPLv3), documented, implemented with continuous integration and delivery (CI/CD) based on Jenkins (JePL), and dockerised: o3skim is run via udocker in HPC, o3api and o3webapp are in the cloud (Kubernetes cluster).