Default C2H6 emissions set to Tszompa-Sosa et al. (2017) emissions

[Twize]

Name and Institution (Required)

Name: Tyler Wizenberg
Institution: University of Toronto

Confirm you have reviewed the following documentation

• Support guidelines
• User manuals
• Current and past Github issues

Description of your issue or question

Hi!

I have found that the default emissions used for Ethane (C2H6) in HEMCO is the Tszompa-Sosa et al. (2017) emissions, instead of CEDSv2. The Tszompa-Sosa et al. emissions are only provided for 2010, and thus, there is no inter-annual variability in the emissions which is quite problematic for longer simulations. Additionally, outside of North America, there is no seasonal variability in the C2H6 emissions, so it is basically constant and fixed emissions across all years. I have done two runs using HEMCO Standalone, one with the Tszompa-Sosa et al. emissions, and another with CEDSv2, and I have attached some quick plots below.

Asian emissions (2003-2021):
Tszompa-Sosa emissions -
HEMCO_Asian_C2H6_Emissions_Timeseries.pdf
CEDSv2 -
HEMCO_Asian_C2H6_Emissions_Timeseries_wCEDS.pdf

European emissions (2003-2021):
Tszompa-Sosa emissions -
HEMCO_European_C2H6_Emissions_Timeseries.pdf
CEDSv2 -
HEMCO_European_C2H6_Emissions_Timeseries_wCEDS.pdf

North American emissions (2003-2021):
Tszompa-Sosa emissions -
HEMCO_NA_C2H6_Emissions_Timeseries.pdf
CEDSv2
HEMCO_NA_C2H6_Emissions_Timeseries_wCEDS.pdf

I believe that the Tszompa-Sosa emissions should be provided as an optional switch for those who are interested in using them, but should not be the default emissions. Having the emissions fixed like this will impact the simulated trends in C2H6, but also in other species which are products of C2H6. There also appears to be a very large discrepancy in terms of the magnitude of the emissions of the two inventories in all regions, but particularly over North America.

[Twize]

I just want to quickly bump this post since I haven't heard anything back yet. Maybe someone from the Emissions Working Group would able to chime in? @eamarais @jaegle

[eamarais]

Hi @Twize, the Tszompa-Sosa emissions were added to address a large underestimate in the available-at-the-time regional and global emissions. Only because I don't have time currently to look over the plots you generated in detail, can you summarise whether Tszompa-Sosa is yielding higher/lower emissions than the other inventories you tested and by how much? Ball park numbers are fine.

[Twize]

Hi @eamarais, ah I understand. Sure I can summarize the comparisons briefly below.

So in my plots I'm only comparing HEMCO SA runs with the Tszompa-Sosa emissions, and then CEDSv2. The Tszompa-Sosa emissions are still broadly higher than the CEDSv2 emissions by a relatively significant margin.

In Asia, the Tszompa-Sosa emissions are a constant ~110 kg/s across all years (2003-2021), while the CEDSv2 emissions are generally in the range of 50-75 kg/s but display seasonal variability as well as trends in the emissions across the years.

In Europe, the Tszompa-Sosa emissions are also constant (i.e., no seasonal or inter-annual variability) with a value of ~60 kg/s, while CEDSv2 is generally in the range of 10 to 20 kg/s, with seasonal variability and a clear decreasing trend in the emissions after about 2010 or so.

Lastly in North America, the Tszompa-Sosa emissions do show seasonal variability in the emissions (e.g., higher emissions in winter, lower emissions in summer, as expected), but no inter-annual variability since the inventory is just for 2010. The Tszompa-Sosa emissions values range between ~50 to 56 kg/s or so. For CEDSv2, the emissions are quite a bit lower, in the range of 10-14 kg/s in 2003 in my simulation, to 7-9 kg/s in 2020.

I am a bit surprised at the discrepancy in the magnitude of the emissions between the Tszompa-Sosa inventory and CEDSv2. I have done some brief comparisons of C48 GCHP simulations using the Tszompa-Sosa emissions against ground-based FTIR measurements from the NDACC network, and the magnitude of the emissions don't actually seem too high, but the main problem is that there is no seasonal variability in the emissions outside of North America. There is also no increasing trend in the emissions as has been observed over the last decade or so (after ~2010). Maybe scaling up the CEDSv2 emissions to match the magnitude of the Tszompa-Sosa emissions in 2010 could be a viable ad-hoc approach?

[eamarais]

@Twize, your ad-hoc suggestion sounds reasonable. Either that, or scale Tszompa-Sosa by the annual (and perhaps also seasonal) scale factors from CEDSv2.

The much lower emissions in widely used regional and global emissions inventories is a relatively old(ish) problem. It was why the Tszompa-Sosa and now much older Xiao inventories were created. Xiao is now only used for propane (C3H8). I forget the full details that led to identification of the problem. Comparison to aircraft observations, perhaps? I assume Tszompa-Sosa and Xiao discuss their diagnosis of the issue in their work. The citations for these works are in the GEOS-Chem narrative description (https://geoschem.github.io/narrative.html).

[Twize]

@Twize, your ad-hoc suggestion sounds reasonable. Either that, or scale Tszompa-Sosa by the annual (and perhaps also seasonal) scale factors from CEDSv2.

Yeah, that suggestion would work as well and probably produce similar results.

The much lower emissions in widely used regional and global emissions inventories is a relatively old(ish) problem. It was why the Tszompa-Sosa and now much older Xiao inventories were created. Xiao is now only used for propane (C3H8). I forget the full details that led to identification of the problem. Comparison to aircraft observations, perhaps? I assume Tszompa-Sosa and Xiao discuss their diagnosis of the issue in their work. The citations for these works are in the GEOS-Chem narrative description (https://geoschem.github.io/narrative.html).

I just had another quick look at the Tszompa-Sosa paper, and it seems that yeah the basically found that the 2011 US NEI inventory was underestimating airplane/ground-based observations by ~40%, which necessitated upscaling the emissions by that approximate factor to match the observations. From what I understand, outside of the US the emissions were directly estimated from GOSAT CH4 observations by applying emission factors to back-out the C2H6 emissions.