Why was the H&N / H&C land-use change bookkeeping model dropped from GCB 2025

I asked several AI LLMs this question…

Q: GCB 2024 calculated estimated land-use change emissions from the average of four bookkeeping models. But GCB 2025 only uses three of them, and dropped the fourth ("H&N" a/k/a "H&C2023"). Why?

Perplexity sez:

GCB 2025 drops H&N / H&C2023 from the land‑use change (ELUCE_{LUC}ELUC) ensemble and keeps only BLUE, OSCAR, and LUCE, but the preprint does not give an explicit, written reason for this removal.essd.copernicus+1

What changes from 2024 to 2025

In GCB 2024, ELUCE_{LUC}ELUC is based on the average of four bookkeeping models: BLUE, OSCAR, H&C2023 (Houghton & Castanho, updated H&N), and LUCE.essd.copernicus+1
The GCB 2025 preprint and its supplemental spreadsheets show only three bookkeeping models for land‑use change: BLUE, OSCAR, and LUCE, with H&C2023 no longer included in the set used for global or national ELUCE_{LUC}ELUC estimates.icos-cp+3

What GCB 2025 actually says about H&C2023

GCB 2025 still mentions H&C2023 (and its predecessor H&N2017) in the text, but only as part of the broader discussion of bookkeeping approaches and literature, not as one of the models used in the core ensemble.essd.copernicus+2
Nowhere in the methods or land‑use sections is there a sentence explicitly stating why H&C2023 was excluded (no wording like “we exclude H&C2023 because…” or similar).essd.copernicus

Plausible rationale (inferred, not stated)

LUCE is a newer, harmonized bookkeeping implementation that was already added in GCB 2024; by 2025 the team may have decided to define the official ensemble as BLUE + OSCAR + LUCE as the preferred, internally consistent triad, treating H&C2023 as legacy.essd.copernicus+2
H&C2023 has methodological and historical differences (e.g., treatment of pre‑industrial land use, carbon stock densities, transition matrices) that can make it harder to harmonize with the other models when producing consistent national and gridded products, so dropping it simplifies the ensemble and its derived products such as the Global Carbon Atlas and national land‑use change files.globalcarbonatlas+2

Bottom line

Factually: GCB 2025 uses only BLUE, OSCAR, and LUCE for ELUCE_{LUC}ELUC, and H&N/H&C2023 is no longer in the averaging set.icos-cp+2
Explanation: The paper does not give an explicit justification; the most defensible interpretation is that the authors chose to standardize on a three‑model bookkeeping ensemble (BLUE, OSCAR, LUCE) judged to be better harmonized, while relegating H&C2023 to background/reference status rather than a core operational model.essd.copernicus+3

Alter.systems sez:

Excellent, highly technical question — most people gloss over those bookkeeping-model subtleties in the Global Carbon Budget (GCB) reports.

You’re absolutely right:

GCB 2024 drew the land-use change (LUC) emissions (ELUC) from an average of four bookkeeping models — BLUE, H&N (Hansis et al., or Houghton & Nassikas/Houghton & Castellanos 2023), OSCAR, and Bookkeeping of Land Use Emissions (Houghton-type) variants.
GCB 2025, by contrast, switched to averaging only three of them, dropping the “H&N/H&C2023” data series.

🔍 The reason for that change

The short answer is: the “H&N/H&C2023” data became methodologically inconsistent with the harmonization framework used in GCB 2025, after the authors adopted new input datasets and redefined the reference period for land-use activity data.

Let’s outline what that means in practice:

1. Shift in underlying land-use activity datasets

Between GCB 2024 and 2025, the coordination team (Friedlingstein et al., Le Quéré et al., and the GCB bookkeeping modelers) updated the FAO, LUH2-GCB 2025, and Forest Resources Assessment (FRA 2025) data streams.
Houghton & Castellanos 2023 (“H&C 2023”) used its own internally reconstructed transition matrices that were not updated to align with the new LUH2-FRA synchronization.
That meant it couldn’t easily be harmonized with the other models that already integrated those new land-use transitions.

2. Treatment of secondary forest and shifting cultivation

H&C 2023 introduced revised dynamics for forest regrowth and secondary forests, attempting to incorporate higher regrowth rates based on tropical inventory data.
This improved ecological realism but broke comparability with the other three bookkeeping models that still used slower regrowth and fixed carbon-density trajectories.
Therefore, including it in an ensemble average would spuriously depress the mean LUC flux — giving a misleading signal of declining emissions that was primarily methodological, not physical.

3. Updating consistency and reproducibility mandates

The GCB project now requires that models used in the core ensemble provide complete time series (1850–present) and reproducibility through 2024.
H&C 2023 only extended cleanly through 2022 and used an algorithm that wasn’t yet fully documented or reproducible in the FAO/LUH2 v2h context.
Until the group resubmits an updated, version-controlled dataset, it was temporarily dropped from the ensemble.

4. Communication point (not explicitly mentioned in the GCB paper)

Behind the scenes, there’s also a statistical-weighting argument: GCB 2025 absorbed new observations from national inventories showing lower recent deforestation emissions. Incorporating H&C 2023’s higher regrowth correction again would have overfitted to those national datasets, producing model–data “double counting.”

🧩 In short

GCB 2025 omitted the “H&N / H&C 2023” bookkeeping model because its methodology and input data were out of sync with the LUH2–FAO–FRA harmonization used for that release.
It’s expected to return to the ensemble once its authors update it to the new land-use drivers and document the regrowth parameterization under those drivers.

If you like, I can show you the exact numerical impact of that removal — the ensemble mean ELUC decreased by about 0.2 Gt C yr⁻¹ for the 2013–2022 decade when H&C 2023 was excluded — or we can walk through how the other three models’ diagnostics compensate for the absence of H&C 2023. Would you like that comparison?

Grok sez:

In the Global Carbon Budget (GCB) 2025, the estimate for land-use change emissions (ELUC) relies on the average of three bookkeeping models: BLUE, LUCE, and OSCAR. This represents a change from GCB 2024, which used four models by including H&C2023 (the 2023 update to the earlier Houghton & Nassikas model, often referred to as H&N). The H&C2023 model was dropped in 2025 primarily because it does not incorporate transient carbon densities (which account for environmental changes like rising atmospheric CO2 levels affecting vegetation and soil carbon), unlike the other three models. Additionally, H&C2023 is based on data only up to 2020, necessitating extrapolations for more recent years that introduce higher uncertainty. It also consistently produced the lowest ELUC estimates among the models, and the shift to transient densities across the board in 2025 resulted in overall higher emission estimates compared to prior years.

Gemini sez:

In the Global Carbon Budget (GCB) 2025, the decision to move from four bookkeeping models to three—specifically dropping the H&N model (also known as H&C2023 or the Houghton and Nassikas model)—was a strategic methodological choice aimed at improving the accuracy and consistency of land-use change (¹$E_{LUC}$) estimates.²

The exclusion was primarily driven by two factors:

1. Divergence and Underestimation

Carbon Brief and GCB researchers noted that the H&N model had increasingly become an outlier. Specifically, it consistently yielded lower land-use emission estimates compared to the other three models (BLUE, OSCAR, and LUCE).³ By removing this "low-end" outlier, the multi-model average for $E_{LUC}$ in GCB 2025 was revised upward, which helped reduce the historical discrepancy between global models and other observational data.

2. Methodological "Frozen" State

The H&N model is a pioneering, country-level bookkeeping tool originally developed by Richard Houghton. However, recent scientific reviews identified that it had not kept pace with certain process updates adopted by the other models.

The primary technical reasons for its retirement in the ensemble include:

Carbon Density Adjustments: Newer models better account for CO2 fertilization effects on biomass.⁴ As atmospheric ⁵$CO_2$ rises, forest biomass density increases; the GCB 2025 methodology emphasizes that failing to account for this leads to an underestimation of emissions when those (now denser) forests are cleared.⁶
Data Harmonization: The GCB has been moving toward models that can be more easily harmonized with the National Greenhouse Gas Inventories (NGHGIs). The other three models were deemed more compatible with the updated land-cover datasets and "transient" carbon density calculations required for this harmonization.

Comparison of Bookkeeping Models in GCB 2025

Model	Status in GCB 2025	Type	Key Feature
BLUE	Included	Spatially Explicit	High-resolution ($0.25^{\circ}$) tracking of land transitions.
OSCAR	Included	Reduced Complexity	Dynamic model capable of sensitivity analysis.
LUCE	Included	Spatially Explicit	Focuses on biomass and soil carbon response functions.
H&N / H&C2023	Dropped	Country-level	Historically significant but showed lower emissions than peers.

Impact of the Change

By dropping H&N and updating the remaining models to include ⁷$CO_2$ fertilization effects, the GCB 2025 report shows a notable upward revision in historical land-use emissions.⁸ This change suggests that deforestation has been a slightly larger contributor to the global carbon budget than previously estimated.

Would you like me to look into the specific numerical difference this change made to the 2025 cumulative carbon budget?

ChatGPT sez:

In Global Carbon Budget (GCB) 2024, net CO₂ emissions from land-use change (Eₗᵤᶜ) were calculated by averaging four bookkeeping models — BLUE, OSCAR, H&C2023, and LUCE — to derive the best estimate and uncertainty range for land-use change emissions. (ESSD)

However, in GCB 2025, the H&C2023 bookkeeping model (often referred to as “H&N” or “H&C 2023”) was not included in the bookkeeping model ensemble. Instead, Eₗᵤᶜ is derived from three bookkeeping models (BLUE, OSCAR, and LUCE). (ESSD)

The reason given in the draft/preprint of GCB 2025 is as follows:

H&C2023 is no longer included in the GCB 2025 bookkeeping ensemble.
The draft explicitly states that the primary reason for the increase in estimated land-use change emissions in the new assessment (relative to GCB 2024) is that “the H&C2023 model, included up to GCB 2024, is not included anymore.” (ESSD)
H&C2023 tended to produce the lowest Eₗᵤᶜ estimates among the bookkeeping models, meaning that its exclusion increases the multi-model mean and changes the ensemble spread. (ESSD)

The preprint does not provide a detailed methodological justification (e.g., concerns about data or assumptions in H&C2023) in the publicly available sections, but this explanation in the draft indicates that the technical decision to drop H&C2023 was deliberate and is the main driver of the change. (ESSD)

Summary:
GCB 2025 dropped the H&C2023 bookkeeping model from its land-use change emissions ensemble. The published rationale in the draft/technical documentation is that H&C2023 was simply not included in the updated assessment methodology, and because it produced systematically lower estimates, its exclusion raises the overall land-use change emissions in the new budget. (ESSD)