use explicit gdx_refprices in convGDX2MIF, ignore LDV variables with NA unit

[orichters]

• needs adaptation in remind such that the gdx_refprices is passed to convGDX2MIF.
• I was a bit unsure whether to rename gdx_ref to gdx_refpolicycosts in order to be consistent, but I'm a bit afraid of breaking stuff… @LaviniaBaumstark / @Renato-Rodrigues, would you prefer that?
• accept NA in some LDV units (see variables with unit NA #408), as suggested by @robertpietzcker
• improve explanations in reportPrices, move transport aggregation before automated additions to avoid repetition, caused by adding fegat to fix reportPrices reports NA for fegat in transport #405
• also ask to select file if colorScenConf is called not with "", but with NULL

[bs538]

Thanks Oliver!
If it can be done without breaking anything, I would be in favour of renaming, as it's easier to understand and more consistent with the scenario_config.csv (gdx_ref used for fixing in GAMS and prices, and gdx_ref_policycosts only for the policycosts)

[orichters]

Ok, so if gdx_refpolicycost is not specified, it defaults to gdx_ref, which allows to be backward-compatible. In REMIND, the situation should never happen that gdx_refpolicycost does not exist while gdx_ref is there because if path_gdx_refpolicycost is empty, it falls back automatically to path_gdx_ref.

[orichters]

I now set gdx_refpolicycost = gdx_ref in the function description, so you can overwrite it as you like, but if it is not provided, it defaults to gdx_ref:

> z <- function(x = 3, y = x) return(paste("x:", x, "/ y:", y))
> z(3, 5)
[1] "x: 3 / y: 5"
> z(3)
[1] "x: 3 / y: 3"

[LaviniaBaumstark]

there is no gdx_refargument anymore

[orichters]

that is not correct, we still need gdx_ref…

[LaviniaBaumstark]

why do we need gdx_ref still for price correction?

[orichters]

Because prices are Moving Averages, and if cm_startyear = 2025, and then the policy scenario calculates its 2020 price values, this takes into account the 2025 value which differs from its reference scenario, and so the reported prices would differ in 2020 even though they should not. See the long debates in #402

[LaviniaBaumstark]

ah, thanks

[LaviniaBaumstark]

but this it not used in default reporting in REMIND, right?

[orichters]

This is used for all non-SSP2 scenarios at the moment, as long as this remind issue is not fixed. Please discuss that with @soergel.

[bs538]

Sounds good to me. I understand the user workflow for calculating policy costs is still the same after this, right?

• selecting refpolicycost in scenario config OR manually through ex-post output.R -> policyCosts
• (if rerunning reporting): automatically using right reference

[orichters]

Sounds good to me. I understand the user workflow for calculating policy costs is still the same after this, right?

* selecting refpolicycost in scenario config OR manually through ex-post output.R -> policyCosts

* (if rerunning reporting): automatically using right reference

Yes, that remains unchanged. I thought about renaming the gdx_ref function parameter in reportPolicyCosts as well to gdx_ref_policycost, but then I thought it is unnecessary, may cause scripts to fail and the title of the function sufficiently clearly suggests that we are talking about policy costs here, I guess.

[orichters]

For the record: Had a chat with @Renato-Rodrigues, he is fine with the changes in reportPrices.