Ticket #455 (closed task: fixed)

Opened 14 years ago

Last modified 14 years ago

Granule summarisation from CSML

Reported by: ko23 Owned by: ko23
Priority: blocker Milestone: MOLES 1.3
Component: MOLES Version:
Keywords: Cc: domlowe


Identify what needs to be summarised up from CSML, and identify where it should go. This includes parameters, and so which CF cell methods should be promoted.

Change History

comment:1 Changed 14 years ago by domlowe

Bearing in mind there are no cell methods in CSML.

I've raised a ticket (458) about this.

comment:2 Changed 14 years ago by domlowe

As well as what we want to summarise I think we also need to think about the process for doing this. Getting stuff from CSML is fairly straightforward, but then what do we do with it? Is it going into an existing MOLES document, a text file (for processing later) or is it forming the basis of a new MOLES record?

comment:3 Changed 14 years ago by ko23

  • Status changed from new to closed
  • Resolution set to fixed

Dom and Kev had a meeting about creating MOLES from CSML. Main points were:

  1. Collections of CSML records could be summarised to provide some input for MOLES, but complete MOLES records cannot be created from CSML alone.
  1. CSML is strictly about the DGs, although the overlap extends for single DG DEs.
  1. DG summary can be produced OK, with the following points:
  • dgCoverage - OK, but providing resolutions may be problematic;
  • should add descSection to MOLES DG summary;
  • name OK, but abbreviation will usually be the same;
  • parameter value, std names, etc OK, but some technical challenges (independent parameters are a bit buried in CSML);
  • access control is no problem.
  1. DE Summary

There are some decisions to be made regarding summarisation, but simple summarisation is probably best for discovery. For example, a DE consists of 3 DGs with parameter value ranges of: 0-10; 5-15; 17-23, and this should be summarised to a single 0-23 value.

However, the range may not be available. e.g. for large datasets it may be impractical to calculate, in which case augmentation of the record may be needed. Only practical investigation will show how much of a problem this may be.

The final conclusion was that a fully valid MOLES record could not be produced, but a "MOLES skeleton" can be.

This will be done by writing a python tool that uses the CSML parser to read multiple CSML documents and then write out the MOLES XML skeleton to a file.

Note: See TracTickets for help on using tickets.