Project 7 (TWP1): Facilitating the distribution and analysis of CMIP5 and related projects

Project lead: Sebastien Denvil

Post-doctoral researcher: Guillaume Levavasseur

Project Start/End: October 2013 – September 2014

Data analysis of large databases like CMIP5 quickly becomes complex and time-consuming. The aim of this task is to improve support for CMIP5 analysis at IPSL and work has been done along the three following sub-tasks:

  • Clarify and facilitate data access

At the beginning of the project, CMIP5 IPSL files were divided into 3 storage areas due to the tape archive system replacement at TGCC. In order to make easier data access and to ensure long term archive of the data, the files were merged at TGCC-CCRT, duplicate on tapes and made available on the distributed data platform ESGF from TGCC (figure 1).

  • Improving support through errata and documentation

In order to clearly identify the produced data and their access, a huge documentation work was conducted to list all known errors in IPSL CMIP5 data. Users can now easily find information about all issues as short and complete descriptions, the affected files, the corresponding version numbers, and the issue status. They can also send a request about downloaded data to know if they have the latest file version, if modifications occur and their history and the links to the corresponding description. Also, we use the IPSL-CM5 errata as a proof of concept to promote such a tool into the ESGF platform and to design a more controlled versioning system during the publication process. In relation to that goal we are now part of the ESGF Publication Working Team where we represent the French ESGF users community and we provide an essential support to them. (

  • A CMIP5 analysis software stack for IPSL community

Another goal was to enable easy and reliable simulation analyses. We developed an ecosystem of CMIP5 analysis software stack, allowing in particular users to find and list the available CMIP5 aggregations at IPSL in a fast and researcher-friendly way. These tools were developed and tested following a study on Pattern Scaling (figure 2).

Position offer and results (January 2017 update)