Handle warnings in convertEdgeBuildings

[fbenke-pik]

The convert function convertEdgeBuildings currently produces two possibly related warnings

WARNING: 
~~~~~ calcOutput('IOEdgeBuildings', subtype = X), with X in (output_EDGE, output_EDGE_buildings) produces negative values, set to 0

~~~~~ WARNING: 
~~~~~ Weight sum is 0, so cannot normalize and will return 0 for some aggregation targets. This changes the total sum of the magpie object! If this is really intended set zeroWeight = "allow", or "setNA" to return NA.

The latter warning causes data to be swallowed during regional aggregation. This might not be a problem (warning can be disabled then), but this should be checked by a sector specialist.

[ricardarosemann]

I've investigated a bit and found the cause of these warnings:

• The first warning is irrelevant, as the negative values reported there (coming from calcOutput("IOEdgeBuildings", subtype = "output_EDGE", aggregate = FALSE)) are filtered out later anyways. I moved the check to a later position - if this warning comes up again (once my changes are merged that is), it should be taken seriously.
• The second one is due to a discrepancy between IEA data and edgebuildings projections: The IEA does not see any FE demand supplied by heat for IRL in 2021 and 2022, while edgebuildings does. Quantities are rather small (I think), at order of magnitude 1E-4, 1E-5. But since edgebuildings results should align with IEA data, we probably should investigate further.

[ricardarosemann]

The reason for the second issue probably is that the IEA data read in by convertEdgeBuildings is already the latest data set, while the input data used in the most recent edgebuildings results is quite old (from February) so that we apparently do not use the latest IEA data yet - at least the FE data fed into edgebuildings only goes up to 2020. And edgebuildings then projects some heat for IRL in 2021 and 2022.

[ricardarosemann]

I checked on the issue again with the new EdgeBuildings 2.5 input data: The issue persists, probably because we crop the years 2021 and 2022 somewhere in calcFEbyEUEC in the mredgebuildings version used for that data. I believe I already merged the potential fix in mredgebuildings though, so hopefully, this will be resolved soon.
In addition, there are now deviations in MEX and HUN for traditional biomass, but I guess that is only because GDP data going into edgebuildings on the one hand and into calcIOEdgeBuildings on the other are not consistent.

[robinhasse]

there are now deviations in MEX and HUN for traditional biomass, but I guess that is only because GDP data going into edgebuildings on the one hand and into calcIOEdgeBuildings on the other are not consistent.

I hope so, too. Though I'd expect HUN to be above the limit GDP/cap in either version. But let's see.

[ricardarosemann]

there are now deviations in MEX and HUN for traditional biomass, but I guess that is only because GDP data going into edgebuildings on the one hand and into calcIOEdgeBuildings on the other are not consistent.

I hope so, too. Though I'd expect HUN to be above the limit GDP/cap in either version. But let's see.

I did a quick test with the edgebuildings results including the uvalue changes and there the warning is gone. Let's check again once we have that version in mrremind and if all goes well we can close this issue then.