Is it OK to have a large C source in the mid 20th Century?

[rosiealice]

The simulation that Jessie analysed to make this plot

has a large source in the mid 20th C, which is outside the range of TRENDY. I want to understand how it differs from TRENDY, what evidence we have IRL that TRENDY is accurate for this time period, and whether that evienc contradicts FATES outputs.

Charlie's simulations show that rangeland clearing is driving much of the source.

The GCB NBP 'constraint' has most models being almost neutral at the beginning, with CanESM gaining ±10PG before all becoming sources:

But the GCB plots have the LUC fluxes (from bookeeping?) being notably higher than the 'land sink', suggesting that their assessment is that the land is actually a source pre-1980-90ish.

The data from the GCB website, downloaded into this file
https://docs.google.com/spreadsheets/d/1csQx6wnh1gClzSAwpo1wnYuYDAPI_h-ZGvXPy8xhvfE/edit?gid=0#gid=0
look like this

Which seems to inducate that the yellow-light green inthe above plot collectively become a net sink shortly after 1960.

From this paper we get the breakdown of TRENDY into components, and the 'benchmark' which starts in 1960 is after the part where it assessed the land to be a slight source.
https://agupubs.onlinelibrary.wiley.com/doi/pdfdirect/10.1029/2024GB008102
we get this plot

For the NH we have

FATES

Randerson et al

I have not got my thoughts fully together on this, but at least these are some basic plots to compare with...

[benmsanderson]

We need to compare directly with the TRENDY output - I'm downloading S0,S1,S2,S3 for all models here:
/datalake/NS9560K/bensan/trendy

[rosiealice]

Some more plots/comments from @JessicaNeedham (from Teams)

"FATES total is a sink in 1850 but then as soon as we turn on transient land use the global becomes a source, mostly driven by the NH. The SH is bouncing between source and sink, but mostly also a souce"

"And there is the last 100 years of the post-AD spin up compared with GCP 1750-1850

I am interpreting this as land use driving NBP to be more negative in FATES than what is predicted by other TRENDY models
"

[rosiealice]

It would be good to have, for FATES and TRENDY, decadal plots of FCO2/NBP, so we could see, for example, what FATES is doing with its carbon in the 1960's vs what TRENDY is doing...

[kjetilaas]

Thanks for all of this, @rosiealice, @benmsanderson and @JessicaNeedham.

I think this is also a key figure to compare with from CGB, showing how non of the models capture the high-latitude (>30N) sink seen in the inversions:

Would it be possible to plot this from your run, @JessicaNeedham?

[JessicaNeedham]

[JessicaNeedham]

We do have a NH lank sink, but I'm not sure if this is comparable to the figure above @kjetilaas . Do I need to subtract LU emissions?

[benmsanderson]

Alright- first pass at processing the TRENDY data. S3 simulations do seem to show a mean source in the mid 20thC, switching to a net uptake in the 70s/80s.

[benmsanderson]

global mean CSVs here:
https://github.com/benmsanderson/trendy-nbp.git

[benmsanderson]

Looking at @JessicaNeedham 's plot at the top of the thread - it looks like the TRENDY distribution there is S2, which shows a multi-model mean sink in 1900. S3 in the plots above shows about mean zero NBP in 1900, becoming a net source in 1950 and a net sink again after 1980.

[JessicaNeedham]

Ok, I think I must be plotting GCB wrong. What is NBP from the GCB data? I was just plotting the column 'GCB' from the terrestrial sink sheet from data downloaded from here: https://globalcarbonbudget.org/archive/

[rosiealice]

@JessicaNeedham the data @benmsanderson got are from here https://mdosullivan.github.io/GCB/

[benmsanderson]

In the XLS from https://globalcarbonbudget.org/download/2358/?tmstv=1762816387, it's S2 data by design (climate and co2 effect only - no land use) - they infer land use from the bookkeeping models. The apples-to-apples comparison here is with S3.

Clean CSV files here:
https://github.com/benmsanderson/trendy-nbp/tree/main/pivot_tables

[JessicaNeedham]

ok, thanks @benmsanderson - I will remake the comparison plots using the correct TRENDY data

[benmsanderson]

*just updated this to be in PgC/yr

In the XLS from https://globalcarbonbudget.org/download/2358/?tmstv=1762816387, it's S2 data by design (climate and co2 effect only - no land use) - they infer land use from the bookkeeping models. The apples-to-apples comparison here is with S3.

Clean CSV files here: https://github.com/benmsanderson/trendy-nbp/tree/main/pivot_tables

[maritsandstad]

For reference, we assume you got the data from here: https://mdosullivan.github.io/GCB/ @benmsanderson ? Just querying for nbp?