Initialization, Calibration, Model Testing🔗
C-ROADS initializes and calibrates to available historical data, primarily provided by the following sources: Greenhouse Gas Emissions
- Global Carbon Budget (2025) (CO2 Energy Emissions, Land Use Change Emissions, and Ocean and Land Carbon Sinks)
- PRIMAP v2.7 (2025) (Non-CO2 GHG Emissions only)
- Houghton and Nassikas (2017) (CO2 Land Use only)
Land Areas
- Land Use Harmonization (LUH2) data (Hurtt et al., 2018)
GHG Concentrations, Temperature Change, Sea Level Rise
- National Oceanic and Atmospheric Administration (NOAA) concentrations (2026) and effective radiative forcings (2025)
- Goddard Institute for Space Studies (GISS) Forcings in GISS Climate Model: Greenhouse Gas Mixing Ratios (2012)
- GISS Global Mean Estimates based on Land and Ocean Data 1880-2025 (2026)
- HadCRUT5 1850-2025 (2026)
- University of Colorado Sea Level Research Group (2018)
C-ROADS calibrates to projected values provided by the following sources:
- Network for Greening the Financial System (2024)
- GCAM 6.0 (U.S.)
- MESSAGEix-GLOBIOM 1.1-M-R12 (IIASA)
- REMIND-MAgPIE 3.2-4.6 (Germany)
- SSP Version 2.0 scenarios (2018 - Available at: https://tntcat.iiasa.ac.at/SspDb)
- Netherlands Environmental Assessment Agency (PBL). Integrated Model to Assess the Global Environment (IMAGE): Detlef van Vuuren, David Gernaat, Elke Stehfest
- International Institute for Applied Systems Analysis (IIASA). Model for Energy Supply Strategy Alternatives and their General Environmental Impact - GLobal BIOsphere Management (MESSAGE-GLOBIOM): Keywan Riahi, Oliver Fricko, Petr Havlik
- National Institute for Environmental Studies (NIES). Asia-Pacific Integrated Model (AIM): Shinichiro Fujimori
- Pacific Northwest National Laboratory (PNNL). Global Change Assessment Model (GCAM): Kate Calvin and Jae Edmonds
- Potsdam Institute for Climate Impact Research (PIK). REMIND-MAGPIE: Elmar Kriegler, Alexander Popp, Nico Bauer
- European Institute on Economics and the Environment (EIEE). World Induced Technical Change Hybrid-GLobal BIOsphere Management (WITCH-GLOBIOM): Massimo Tavoni, Johannes Emmerling
Importing as data variables, C-ROADS assesses the GHG concentrations and temperature change projections given various emissions projections for model validation. Accordingly, there are necessary files, generated from data models, which must accompany the model. We test the model against the NGFS net emissions projections for their 6 scenarios. Reliably, for each scenario, the model captures the key dynamics of the NGFS models.
Although outdated now, we ran comparable assessments against all of the Shared Socioeconomic Pathway (SSP) of the IPCC's AR5 scenarios. Comparisons were against the output of 6 models for 5 SSP scenarios, each with up to 6 radiative forcing options, i.e., 1.9, 2.6, 3.4, 4.5, 6.0, and Baseline. Reliably, for each SSP storyline and RF level, the model captures the key dynamics of the SSP models.
Carbon Cycle Calibration🔗
Four carbon cycle constants are calibrated together: the sensitivity of dissolved inorganic carbon in the ocean to atmospheric CO2, the ocean mixing flow rate, the CO2 fertilization (biostimulation) coefficient, and the rate at which CO2 fertilization saturates. They are fitted to:
- Atmospheric CO2 concentration, 1850-2024 (GISS to 1978 and NOAA from 1979)
- The ocean carbon sink and the net land carbon flux (land use emissions minus the land sink) from the Global Carbon Budget (2025)
- The Revelle (buffer) factor of the surface ocean (Jiang et al., 2023)
The fit also keeps the four standard carbon cycle feedbacks, the responses of ocean and land carbon uptake to CO2 and to warming, within the range of the CMIP6 Earth system models (Arora et al., 2020). It limits drift too: with CO2 and temperature held at their 1850 levels, the ocean should take up little carbon, so that the historical ocean sink comes from rising CO2 rather than from the ocean's starting state being out of balance.
Land Calibration🔗
Land use change is calibrated based on the Land Use Harmonization (LUH2) data prepared for the Climate Research Program Coupled Model Intercomparison Project (CMIP6). Our output for each land type strongly aligns with historic data. However, our projections suggest more farmland and less forest than do the LUH projections and those of the NGFS models. The differences are due to our accounting for the temperature effect on reducing crop yield, which translates to more farmland expansion to meet food demands. The other models do not account for that feedback.