Harmonising results conversion to IAMC format and Friendly Data

We have been developing a script to convert OSeMOSYS results into IAMC format. OSeMOSYS results are produced in individual CSV files in long-format with one file per result parameter. The script matches patterns in technology names using regex, and then performs an aggregation function, finally converting the aggregated results into IAMC format using pyam. The pattern matching and aggregation function are defined per IAMC variable using a configuration file passed to the script at runtime.

I have been monitoring the SENTINEL Friendly Data repository and am wondering at what point we should try to harmonise efforts for the results processing? There seem to be a number of overlaps in functionality, but it is not clear to me what the “entry point” is for Friendly Data? For example, would we need to write out OSeMOSYS results into a Friendly Data datapackage to use the IAMC conversion functionality?

Perhaps this could also be of utility for other modelling teams, and those more broadly who wish to participate in the model comparison exercise?

1 Like

Thanks for raising this question, @WillUsher!

Looking at the Friendly-Data repository and your prototype osemosys2iamc work, I see little benefit in adding a detour via the friendly-datapackage format coming from an OSeMOSYS model. Writing to the IAMC format is not that complicated, so duplicating a few lines of code and some functionality across two parallel repositories (one specific for OSeMOSYS, one more generic) is the lesser concern to having additional dependencies that need to be kept in sync.

1 Like