GCAM v8.2 SSP3 Harmonization Issue

[vishva-GT]

Hello GCAM Team,

I have been running GCAM v8.2 SSP scenarios and noticed an issue related to historical harmonization in SSP3 that I wanted to ask about.

While comparing energy cost outputs across SSP1–5, I observed that several biomass and synthetic-fuel-related technologies in SSP3 do not align with the other SSPs in the historical period (prior to 2020), even though I expected all scenarios to share identical historical calibration, as the other scenarios do.

Examples where SSP3 diverges in the past include:

• FT biofuels
• FT biofuels CCS level 1
• FT biofuels CCS level 2
• BTL with hydrogen
• Cellulosic ethanol
• Cellulosic ethanol CCS level 1
• Cellulosic ethanol CCS level 2
• Biomass gasification
• Some electricity-related prices

The divergence appears specifically when SSP3 includes:

<Value name="bio">../input/gcamdata/xml/en_transformation_low.xml</Value>

When I remove this file from the SSP3 configuration, the historical mismatch largely disappears and the outputs become harmonized with the other SSPs.

My understanding was that SSP scenarios should differ primarily in future assumptions, while historical years should remain calibrated consistently across scenarios.

I wanted to ask:

1. Is en_transformation_low.xml intended to modify historical calibration values?
2. Is there a recommended way to preserve historical harmonization while still applying the SSP3 low-bio assumptions in future years?

I also noticed extremely large values (~59000 2024$/GJ equivalent) for “FT biofuels CCS level 1/2” after 2025 and was wondering whether this technology is intended as a non-deployed/backstop pathway rather than a realistic production technology.

Any clarification would be greatly appreciated. I have attached pictures for FT biofuels cost before and after removing that file showing how SSP3 goes from unharmonized to harmonized. Also I have included screenshots from the Model Interface, showing that the data for SSP3 is different compared to the other scenarios.

Thanks,
Vishva