Scaling of historical CO2 (esp. in 1.B.2)

[JGuetschow]

Historical CO2 emissions are scaled to match country reported data in the "CR" timeseries. For some sectors, especially 1.B.2 this leads to emissions estimates which are much higher than the original data because e.g. for the USA CDIAC reports very high emissions around 1950 and much lower emissions than in CRF data for the post 1990 period.

We need to develop a new matching strategy e.g. with a scaling factor that approaches 1 in the span of e.g. 20 years. This is however not easy to implement in the PRIMAP1 emissions module.

[JGuetschow]

See also here: https://twitter.com/robbie_andrew/status/1380049591236767745

Before changing the PRIMAP methodology we need to investigate the reason for the discrepancy. If it comes from different emissions factors we need to check whether these are consistent over time and which reasons are given for the choices.

[JGuetschow]

A summary of the long twitter thread linked above. Estimates for fugitive and flaring emissions vary strongly. According to the US EPA a major source of fugitive CO2 emissions from natural gas systems come from acid gas removal (AGR) units. CDIAC data only covers flaring of natural gas (which occurs in oil and natural gas production an processing) and thus does not include AGR CO2 emissions and other CO2 venting emissions. When scaling CDIAC to CRF we use flaring growth rates also for AGR emissions which overestimates historical emissions.

[JGuetschow]

A further problem is that CDIAC fugitive data also contains vented gas as it assumes that all gas vented or flared is oxidized.

[JGuetschow]

With the switch to CEDS data which is available on a country level we don't have to rely on growth rates for the extension of more recent data. The problem is reduced but not completely resolved.

CDIAC data for flaring is no longer used

[JGuetschow]

The new scaling algorithm has alleviated this problem, but it is nor completely solved. Countries affected are Belize, Morocco, Mexico, Turkey, USA.