Hi, I'm following the project as I am keen to see where this goes. I'm using Tabular Datapackages to hold input (and eventually) output data for OSeMOSYS with data conversions facilities provided by otoole. See a working prototype here.
In a project at University of Oxford we hard linked multiple models together using a Python package we developed called smif. You might be interested in the approach we took for dealing with the metadata. While we didn't get as far as using Tabular DataPackages, we stored all metadata in YAML and had a reasonable structure to record units, spatial and temporal dimensions for model inputs, outputs and so on. In addition, we built an architecture in smif which converted this hard linked model into a linear workflow which could be sent to a scheduler, including automated (but customisable) conversions of spatial and temporal scales. It might be worth a look!
Hi, I'm following the project as I am keen to see where this goes. I'm using Tabular Datapackages to hold input (and eventually) output data for OSeMOSYS with data conversions facilities provided by otoole. See a working prototype here.
In a project at University of Oxford we hard linked multiple models together using a Python package we developed called smif. You might be interested in the approach we took for dealing with the metadata. While we didn't get as far as using Tabular DataPackages, we stored all metadata in YAML and had a reasonable structure to record units, spatial and temporal dimensions for model inputs, outputs and so on. In addition, we built an architecture in smif which converted this hard linked model into a linear workflow which could be sent to a scheduler, including automated (but customisable) conversions of spatial and temporal scales. It might be worth a look!
An application of smif can be seen in NISMOD2.