validation for total and cumulative land emissions from the Global Carbon Budget, including all bookkeeping models
Details
The historical series pooled under Emissions|CO2|Land|+|Land-use Change are all LUC-scale CO2
SOURCES and are broadly comparable to MAgPIE's +|Land-use Change. Verified against the ingested
validation data (not the parent publications), they form a single positive cloud (Mt CO2/yr, World,
2000-2010): bookkeeping BLUE/OSCAR/GCB/H&C ~2600-6600 and the national-statistics series
FAO_EmisLUC/EDGAR_LU/PRIMAPhist ~2800-5400 - the latter numerically indistinguishable from the
bookkeeping cloud. They differ mainly on PEAT; a second, conceptual axis (INDIRECT/Grassi) matters in
principle but is NOT represented in the ingested data (see Axis 2):
| source | peat | nature (as ingested) |
| BLUE, OSCAR, H&C2023 | excl | bookkeeping ELUC (direct, ex-peat) |
| GCB (this fn) | excl | = mean(BLUE, OSCAR, H&C2023) |
| Gasser et al 2020 | excl | OSCAR bookkeeping |
| FAO_EmisLUC | incl | FAOSTAT net LULUCF ("Land Use total"), a source |
| EDGAR_LU | incl | EDGAR LULUCF CO2, a source |
| PRIMAPhist | incl | PRIMAP-hist CAT5 (LUCF) CO2, a source |
Axis 1 - PEAT. Bookkeeping estimates exclude peat drainage/fire; the national-statistics series include
it (drained organic soils). MAgPIE's Land-use Change INCLUDES peat (separable via its +|Peatland
child), so the peat-clean bookkeeping comparison uses Land-use Change net of that peat child (magpie4's
...|Land-use Change|Excl Peatland line, where present, provides this directly). The peat term
(~1-1.5 Gt CO2/yr) is small relative to the source magnitude and does not move the inventories out of
the LUC cloud.
Axis 2 - INDIRECT (Grassi) - NOT represented in the ingested data. In principle the bookkeeping-vs-NGHGI
gap (~5 Gt CO2/yr; Grassi et al. 2021, doi:10.1038/s41558-021-01033-6) arises because country inventories,
reporting the sink-inclusive NET LULUCF over a large managed-land area, embed the environmental sink and
sit far BELOW bookkeeping ELUC - which would make net Emissions|CO2|Land the matching counterpart.
BUT none of the FAO_EmisLUC/EDGAR_LU/PRIMAPhist series ingested here is that Grassi-adjusted NGHGI net:
FAO is FAOSTAT's net "Land Use total" (its forest sink included but modest, so still a ~4 Gt SOURCE),
EDGAR/PRIMAP are LULUCF CO2 series - all positive, LUC-scale, none carrying a net-flux or Indirect sink
series. They must therefore be compared to MAgPIE's +|Land-use Change like the bookkeeping sources,
NOT to net Land. The sink-inclusive NGHGI-net quantity Grassi contrasts with bookkeeping is absent from
this validation cloud, so MAgPIE's own +|Indirect (its Grassi managed-land sink ~-5.6 Gt CO2/yr,
i52_land_carbon_sink) has no inventory counterpart here to validate against.
GCB note - the GCB column ingested here is the bookkeeping-model MEAN and is EX-peat: GCB == mean(BLUE, OSCAR, H&C2023) exactly (World 2020: 4298 == mean(5607, 4724, 2564)). The published GCB ELUC additionally adds peat drainage/fire (a separate GCB.xlsx column) not folded into the ingested net. GCB's peat column exists in the workbook but is intentionally not ingested (it could not be an additive child of Land-use Change, since GCB's net excludes it).
Do NOT benchmark net Land against GCB: the only Indirect / net Emissions|CO2|Land series here is
GCB's, where Indirect = the GCB terrestrial sink S_LAND over ALL land (World 2020: -11403 Mt CO2/yr,
~the whole-biosphere sink), NOT the managed-land Grassi quantity MAgPIE reports.