Introduction
The Intergovernmental Panel on Climate Change (IPCC) synthesis document of the Sixth Assessment Report (AR6) states that humans have “unequivocally caused global warming,” thus causing “widespread and rapid changes” in Earth and human systems. It also concludes that climate change adaptation and mitigation efforts have increased in recent years (IPCC et al., 2023), altering human systems and subsequently their effects on climate. Recent articles highlight this overall feedback loop by describing several human-Earth system interactions that demonstrate strong co-evolutionary relationships between human and Earth systems (Motesharrei et al., 2016) and arguing that incorporating human-Earth feedbacks in models will ultimately reduce climate uncertainties by capturing a more realistic range of climate futures (Beckage et al., 2020).
However, the dominant Earth modeling paradigm segregates human and Earth systems such that the human systems do not experience the impacts associated with the changing Earth (Figure 1). In the AR6 protocol, for example, human system models first generate scenarios of land use, anthropogenic emissions, and concentrations of various atmospheric constituents (IPCC et al., 2022; O’Neill et al., 2016) that are then passed to Earth system models (IPCC et al., 2021; Tebaldi et al., 2021), the outputs of which are then used to determine specific impacts on human and natural systems (IPCC, 2022). The scenarios do not include human responses to these impacts or to changing climate, limiting the accuracy of results and our ability to assess climate change adaptation and mitigation strategies.
[IMAGE OMITTED. SEE PDF]
Including climate impacts during scenario generation alters both the scenario and the Earth system response. For example, climate impacts on agriculture can affect crop prices, areas, yields, production, and revenue (Snyder et al., 2020) in scenarios, and change the estimated cost of climate change mitigation (Kyle et al., 2014). Climate warming has been shown to increase energy demand in scenarios due to increased building cooling (Clarke et al., 2018; Yalew et al., 2020), while climate change impacts to energy supply give mixed results for energy production across technologies (Cronin et al., 2018; Yalew et al., 2020). These impacts have been implemented in human system models as adjustments based on pre-calculated conditions and as such the resulting scenarios do not include the effects of changing conditions resulting from the adjustments (Cronin et al., 2018). For example, several studies have shown that land use change affects regional climate (e.g., Brovkin et al., 2013), including when the land change is driven by climate impacts on agriculture (Calvin, Bond-Lamberty, et al., 2019; Thornton et al., 2017). Changes in land use due to climate impacts can also affect terrestrial carbon stocks (Calvin, Bond-Lamberty, et al., 2019; Thornton et al., 2017) and how vegetation growth responds to temperature atmospheric CO2 concentration (Jones et al., 2018). These changing conditions can then change previously determined impacts, which completes the feedback loop. In addition, the benefits of specific climate change mitigation strategies may be overestimated if impacts are not accounted for. For example, carbon sequestration via forest expansion may be reduced by climate-driven fire increases (Jager et al., 2024). Without a closed feedback loop, however, the impacts and responses are not entirely consistent with each other.
The historical segregation of human and environmental research has translated into functional and conceptual mismatches between human and environmental models (Robinson et al., 2018) that hinder model implementation of human-environment feedbacks. For example, human system models generally have more detailed representations of agriculture than Earth system models, but less realistic representations of vegetation processes (Pongratz et al., 2018). This introduces error when applying human scenarios to Earth system models (Di Vittorio et al., 2014) and degrades the fidelity of radiative forcing targets and corresponding emissions between scenarios and Earth system model outcomes (Fredriksen et al., 2023; Jones et al., 2013). Development of human-relevant processes in Earth system models has increased recently (e.g., Laidlaw et al., 2019; Sinha et al., 2022; Zhou et al., 2020), but uncertainties are not well characterized and feedbacks with human systems generally are not present (Liu et al., 2017).
There is consensus on the importance of such feedbacks and the need for approaches that vary in complexity, computational intensity, geographic scale, and feedback type (Calvin & Bond-Lamberty, 2018; Motesharrei et al., 2016; Robinson et al., 2018; van Vuuren et al., 2012; Verburg et al., 2016). The integrated Earth system model (iESM) was one high-complexity approach that applied terrestrial productivity feedbacks from the Community Earth System Model (CESM) to land use decisions in the Global Change Analysis Model (GCAM) (Collins et al., 2015). Every 5 years GCAM generated CO2 emissions and land use for the next 5 years based on CESM land productivity changes over the previous 5 years (Figure 1). The iESM demonstrated the importance of including terrestrial feedbacks in GCAM (Calvin, Bond-Lamberty, et al., 2019; Jones et al., 2018; Thornton et al., 2017), but its implementation precluded further advances because it did not use the standard CESM coupling software that would have facilitated updates and expansion across all model components (e.g., land, atmosphere, ocean).
Here we present the human-Earth feedback configuration of E3SM version 2.1 and evaluate the effects of including terrestrial productivity feedbacks on model outcomes. While based on iESM, the novel advance here is the development of a human component within E3SM that is at the same functional level as the other model components. This enables straightforward use via standard E3SM tools, facilitates addition of human-Earth feedbacks through the E3SM coupling software to any E3SM component, and allows efficient updating of all components for use with human-Earth feedbacks. The human component is linked directly to the GCAM code repository, which enables efficient GCAM updates. To demonstrate the validity of this new model configuration, we address the following questions under the shared socio-economic pathway 2 and representative concentration pathway 4.5 scenario (SSP2-RCP45): (a) How does the inclusion of terrestrial productivity feedbacks alter the human scenario and the Earth system? (b) What are the relative contributions of agricultural and landscape carbon density feedbacks to the outcomes?
Models
The Energy Exascale Earth System Model
E3SM is an open-source Earth system model addressing science questions focused on the water cycle, cryosphere, and human-Earth system interactions (Leung et al., 2020), developed to meet U.S. Department of Energy mission needs and high-performance computing goals. Here we present the new human component that addresses human-Earth interactions in E3SM version 2.1, an extension of E3SM version 2 (Golaz et al., 2022) that improves simulation of the Atlantic Meridional Overturning circulation. This new E3SM component enables two-way, synchronous coupling between E3SM and the human systems model GCAM. The forthcoming E3SM version 3.1 will include this human component as a configurable option.
E3SM uses a flexible coupling architecture (henceforth referred to as the “coupler”) to integrate the atmosphere, ocean, sea ice, land ice, land, river, and human components (Figure 2). The E3SM coupler (based on Craig et al., 2012) controls the sequence, timing, and data exchange among the component models. It initializes the components, maintains a central clock, and calls each component when needed. The coupler facilitates data exchange through temporal aggregation (to maintain synchronous time steps) and spatial resolution conversion, thus allowing each component to operate on a distinct time step and resolution. The respective components are the E3SM atmosphere model (EAM), the Model for Prediction Across Scales (MPAS) for ocean (MPAS-O), sea ice (MPAS-SI), and land ice (MALI), the E3SM land model (ELM), the Model for Scale Adaptive River Transport (MOSART), and the new E3SM human component (EHC). The EHC includes GCAM and the software interface between GCAM and the E3SM coupler. Currently, the EHC can run by itself for testing, coupled with active ELM, or coupled with both ELM and EAM active. When EAM is active it utilizes data inputs in place of the ocean and sea ice components. Future configurations will include additional active components. E3SM with an active human component is limited to future simulations that start in 2015.
[IMAGE OMITTED. SEE PDF]
Currently, the EHC communicates only with EAM and ELM, and the other components respond through their respective couplings with these two components (Golaz et al., 2022). EAM version 2.1 operates on spectral element grids with 72 vertical layers (up to about 60 km), and separate subgrids for physical dynamics and air column parameterizations within each grid cell. The EAM emissions-driven CO2 dynamics and aerosol chemistry/dynamics are active by default with an active EHC. MPAS-O connects to EAM, has 60 vertical layers, and determines the fraction of each grid cell that EAM considers as land. MPAS-SI also connects to EAM and shares the MPAS-O horizontal grid. MALI and MOSART connect to ELM and are currently not active when the EHC is active. ELM version 2.1 also communicates with EAM and operates on finite element grids with no communication between grid points. It represents surface and subsurface water, biophysics, and energy, carbon, nitrogen, and phosphorus dynamics. The hierarchical sub-grid structure allows glaciers, lakes, wetlands, urban land, agricultural land, and other vegetation to share a grid cell. ELM has a two-layer canopy with vegetation, soil, snow aging, and aerosol deposition impacts on surface albedo. Subsurface processes include vertically resolved soil hydrology, biogeochemistry, and wetland and permafrost dynamics. There is one generic crop when coupling with GCAM, and the wildfire module is active.
The grid resolutions in this study are 0.925° × 1.25° for human and land (f09), circa 1° for the atmosphere (ne30pg2), and ranging from 30 to 60 km for the ocean and sea ice (oEC60to30v3). The EHC must have a grid that matches the ELM grid to be consistent with non-EHC simulations that are used for initialization and comparison. Currently only the f09 resolution is available for active EHC simulations. Proposed EHC-ELM resolutions include 0.125°, 0.5°, 0.925° × 1.25, and 1.925° × 2.5°.
The Global Change Analysis Model
GCAM is a dynamic-recursive model that includes detailed economic, energy, water, and land systems (Calvin, Patel, et al., 2019). It can be linked to a simplified climate model (Hartin et al., 2015) to explore climate change scenarios (e.g., IPCC et al., 2022a), but within E3SM this link is not used. Instead, the scenario is first defined by a GCAM standalone run, and then the corresponding GCAM configuration and some output data are applied within E3SM. GCAM version 6 (GCAM6, 2022) is currently configured with E3SM.
GCAM solves market prices for all energy, agricultural, and land markets in each 5-year period such that supply equals demand in all markets. The outputs represent the model state at the end of the period, which is the labeled output year. The primary drivers are region-specific annual population and initial gross domestic product, with prescribed rates of labor force participation and labor productivity growth (which affect gross domestic product over time). Various scenarios can be created by changing technology availability or cost, specifying radiative forcing targets (Calvin, Patel, et al., 2019), directly constraining emissions (Iyer et al., 2022), setting carbon or GHG prices, or by imposing market constraints as external boundary conditions (e.g., Calvin et al., 2014). All markets are regional (Figure S1 in Supporting Information S1), with distinct prices, and primary energy and agricultural commodities are traded among regions. Regional imports/exports are calibrated to historical trade patterns, and deviations from these patterns can occur over time due to shifting economics. GCAM is calibrated to 2015 over a 1975–2015 historical output range and can restart at any period given the state files of the previous periods. The primary outputs relevant to E3SM-GCAM coupling are anthropogenic emissions, which are determined at the regional level, and land allocation that is determined on a finer set of land units.
The GCAM land allocation module uses a profit-based land-sharing approach to determine land use/cover, agricultural and forestry production and consumption, land commodity prices, fertilizer use, agricultural water withdrawal and consumption, land carbon dynamics, and agricultural emissions (Wise et al., 2014). The land allocation module operates on a subgrid that is the intersection of 32 regions and 235 water basins (Figure S1 in Supporting Information S1), resulting in 384 distinct land units that are also used by the water module. Agricultural production is determined within these land units and aggregated to regional markets.
E3SM-GCAM
E3SM-GCAM coupling follows the scenario-based modeling framework as implemented for the IPCC assessment reports, but with the addition of a feedback from the Earth model back to the human model (Figure 1). Instead of GCAM providing a 100-year scenario that drives an independent E3SM simulation, every five years E3SM provides terrestrial productivity changes to the GCAM land allocation module and GCAM provides land allocation and CO2 emissions to E3SM. Challenges to this synchronous coupling include inherent model differences in resolution, temporal frequency, land characterization, initial/boundary conditions, and definitions. The CO2 emissions are passed one-way from GCAM to E3SM, thus limiting their main challenges to (a) downscaling region level data to a regular grid, (b) disaggregating the annual data to 30-min time steps, and (c) partitioning the data to different atmospheric levels.
The exchange of land data between GCAM and E3SM requires many steps and assumptions because of the differences in how each model characterizes the land surface. The conversion between GCAM land units and the E3SM regular grid is complicated by each model having its own distribution of unique land types that need to be reconciled (Table S1 in Supporting Information S1) on different distributions and definitions of overall land area. Furthermore, GCAM land types categorize an area while E3SM plant functional types (PFTs) define fractions of specific vegetation types within a grid cell. The transformation of data between GCAM's 41 land types and E3SM's 16 PFTs happens in both directions but is asymmetric because the data are different for each direction. For example, variations in climate in an E3SM grid cell uniquely affect the growth of each PFT in the cell. To address fundamental differences in model initial conditions, growth impacts are calculated as scaling values within E3SM relative to baseline values. These impacts to the crop PFT need to be distributed to several GCAM crops while the impacts to several tree PFTs need to be aggregated to a few GCAM forest types, in addition to being aggregated to GCAM land units. Each GCAM crop has an expected yield that increases over time to account for technological progress, and each GCAM forest type has a maximum potential carbon density. Each of these is scaled by the impacts calculated in E3SM, then GCAM projects the scenario for the next 5 years. The GCAM crop allocations are aggregated to one crop while the forest allocation and harvest are disaggregated to E3SM tree PFTs, and these allocations are downscaled to the E3SM grid. Ultimately, the coupling enables communication between two different models while providing a framework for advancing this communication as the models advance.
Implementation of E3SM-GCAM Coupling
EHC Within the E3SM Coupler Architecture
In contrast to the iESM, we designed the EHC to operate at the same level as the land, atmosphere, ocean, sea ice, land ice, and river components (Figure 2). The EHC thus exchanges information with the other components through the E3SM coupler and is the first component called in each model year to ensure that the scenario data (CO2 emissions and land use/cover) are available to EAM and ELM. The EHC calls GCAM once every 5 years to generate the scenario data and processes these data into inputs for EAM and ELM at the beginning of each E3SM model year. Each recipient component performs additional processing of these data to meet their structural and temporal needs. Currently, only terrestrial productivity is fed back to GCAM from ELM, and the coupler averages the requisite data over GCAM's 5-year run period before passing these data to the EHC.
ELM-GCAM (Land Model) Coupling
ELM-GCAM coupling is like the iESM coupling (Bond-Lamberty et al., 2014; Collins et al., 2015; Di Vittorio et al., 2014) with the main structural difference being that the E3SM coupler controls the EHC and its information exchange with ELM (rather than the land component controlling the EHC and the information being passed via files). Additionally, impacts to agricultural yield and potential carbon density have been implemented separately, and land conversion has been updated.
Every 5 years, changes in ELM terrestrial productivity are passed to GCAM, which then projects land use/cover for the next GCAM output year. These land data are downscaled to a regular grid, and then are interpolated annually to year-specific values, processed, and passed to ELM (Figure 3). Changes in terrestrial productivity are applied in GCAM as scaling values to agricultural yield and potential carbon density (except in 2015). The vegetation scaling values are applied to agricultural yields and vegetation potential carbon density and the soil scaling values are applied to soil potential carbon density. The coupler averages ELM heterotrophic respiration (HR), net primary productivity (NPP), and plant functional type area over 5 years and passes these gridded data to the EHC. The EHC removes outliers using a median absolute deviation method (Bond-Lamberty et al., 2014) and calculates average HR and NPP for each GCAM land type in each GCAM land unit by mapping ELM grid cells to GCAM land units (Figure S1 in Supporting Information S1) and mapping PFTs to GCAM land types (Tables S1 and S2 in Supporting Information S1). This aggregation is weighted by respective PFT, grid cell, land unit, and land type areas, giving each final GCAM land type distinct HR and NPP values. Scaling values for each land type in each land unit are calculated as:
[IMAGE OMITTED. SEE PDF]
The EHC employs four distinct modules to process GCAM land output before passing it to the coupler (Figure 3). The first module downscales the 5-year changes in GCAM crop, pasture, and forest area to the global land-use model (GLM) half-degree grid, interpolates these changes and the GCAM wood harvest to annual values for the current E3SM model year, and then updates the GLM input land state. Each model year the land state is updated to the beginning of the next model year and the wood harvest is calculated for the current model year. The second module is GLM (Di Vittorio et al., 2014; Hurtt et al., 2011), which downscales and separates the GCAM wood harvest into five categories and converts the land state to GLM output land types (crop, pasture, primary, secondary). GLM has been reconfigured with initial conditions based on LUH2 data (Hurtt et al., 2020). The third module translates the annual changes in GLM crop and pasture areas to changes in 16 ELM PFTs based on specified land conversion rules, with new crop PFT area set equal to the GLM crop area first (Di Vittorio et al., 2014). The reference for land area changes is the beginning of the current model year. An important change from iESM in land conversion is that pasture is tracked over time such that additional crops and pasture cannot be overlaid on existing pasture, which may increase conversion of forest or shrub PFTs to crop or grass/shrub PFTs. The land conversion assumptions can be tuned to match ELM and GCAM forest area changes for a given scenario. When abandoning cropland or pasture, if all potential vegetation limits have been reached then bare area fills the rest of the vegetated portion of the grid cell. Finally, the fourth module generates the PFT and wood harvest data in ELM format at the EHC/ELM resolution. These data are passed to the coupler as fractions of grid cell and written to a diagnostic land surface file.
E3SM runs for 5 years before terrestrial productivity is passed back to GCAM. Each year ELM distributes the annual land data to its 30-min time steps. The PFTs are interpolated between the two respective beginning-of-year land states and the wood harvest data are summed across the five GLM categories, evenly distributed in time, and applied to forest PFTs only. The coupler sums the required ELM outputs over the 5 years and then calculates the averages and passes them to the EHC.
EAM-GCAM (Atmosphere Model) Coupling
Atmospheric forcing for EAM is segregated by CO2 and non-CO2 constituents. GCAM's CO2 emissions are downscaled by type and time, passed through the coupler to EAM, and then interpolated to EAM time steps and vertical layers as necessary. E3SM-GCAM uses the existing EAM pathway for file-based inputs of CO2 emissions, but with routines specific to data received via the coupler.
Every five years GCAM projects CO2 emissions for its next output year for 53 surface sectors, one international shipping sector, and one aircraft sector. The EHC sums the surface sectors so that there are three CO2 types: surface, international shipping, and aircraft. These data are associated with 32 GCAM regions, and the EHC aggregates aircraft and international shipping to the globe prior to downscaling because they also occur over the ocean.
Annually, the EHC uses pattern downscaling to convert the GCAM data to a regular grid. It first interpolates these data to the current model year based on the previous and current GCAM output values. Next, the EHC calculates scaling factors for the three GCAM CO2 emission types (at the regional level for surface and at the global level for aircraft and international shipping) based on the GCAM year 2015 emissions outputs. The baseline data for scaling are gridded, 2014 emissions that include monthly values for eight surface sectors and 25 vertical layers of aircraft emissions (Feng et al., 2020) and that have been pre-processed to obtain monthly reference data for the three CO2 types, with two vertical levels for the aircraft data (8.845 km and below, and 9.455–14.95 km). This threshold was selected to group the largest emissions in the upper level and the relatively small and evenly distributed emissions in the lower level to facilitate redistribution of GCAM emissions to EAM vertical layers. The EHC first scales the global values of these reference data to match the GCAM 2015 global values, then the GCAM scaling factors are applied to the gridded reference data to obtain the monthly, gridded values for the current model year. The downscaled surface and international shipping data are summed as one surface type and passed to EAM along with the two levels of aircraft data.
EAM further processes the data it receives from the coupler each year. These data are interpolated to each 30-min EAM time step and the surface and international shipping data are input at the lowest atmosphere layer while the aircraft data are distributed to multiple vertical layers. Based on the gridded baseline data, the upper input level is inserted at 11 km and the lower input level is distributed evenly across the lower EAM layers.
EAM also requires scenario-specific, non-CO2 forcing data. Most of these data are provided by a standalone GCAM run, and the remaining data are obtained from corresponding CMIP6 forcing data. GCAM outputs aerosol emissions, CO2 emissions, and CO2, CH4, and N2O concentrations. Three steps are required to obtain EAM input files: (a) convert GCAM outputs to a standard processing format (GCAM2IAMC, 2021), (b) downscale to gridded Community Emissions Data System (CEDS) format (DOWNSCALE, 2024; Feng et al., 2020; Gidden et al., 2019), and (c) convert to EAM format. The first step creates files that are used by the downscaling software in step two to generate the standard CEDS format gridded data files. Step three creates 16 aerosol emission files (black carbon, SO4, particulate organic matter, SO2), two CO2 emission files (surface, aircraft), and one GHG concentration file (CO2, CH4, N2O, CFCs). Secondary organic aerosols, CFCs, chlorine, ozone, nitrate, HO2, and OH forcings are CMIP6 data and not updated by our standalone GCAM run.
Compilation and Configuration
GCAM must be run in standalone mode to generate scenario-specific input files for configuring and compiling E3SM-GCAM. If there is a radiative forcing target the GCAM carbon price path is required from this run in addition to the aerosol emissions and non-CO2 GHG concentrations described above. E3SM-GCAM is then set up with the scenario-specific standalone GCAM configuration plus the corresponding non-CO2 forcing data for EAM and, if necessary, the carbon price path. These boundary conditions ensure the validity of the configuration for a given scenario.
Given the required GCAM data, E3SM-GCAM is configured, compiled, and run using the standard E3SM process (E3SM documentation, 2019). A valid component set (compset) is selected to set up the model configuration. The compset defines the active components and specifies the default inputs for the associated scenario (e.g., SSP2-RCP45). Each component has its own configuration file for customizing a variety of variables ranging from baseline and mapping files to runtime and output options. Key EHC runtime variables include those that enable GCAM, agricultural yield scaling, carbon density scaling, CO2 emissions to EAM, and diagnostic outputs (Table S3 in Supporting Information S1). For example, terrestrial feedbacks can be disabled but still calculated and output, and then read as inputs for a subsequent simulation (during which the feedbacks will not be calculated). ELM and EAM have some variables that must be set appropriately in their configuration files to ensure proper integration with GCAM. While ELM always receives land cover change via GCAM when the EHC is active, the passing of runtime CO2 emissions from GCAM to EAM can be disabled and these data can be input from a file. Importantly, E3SM-GCAM simulations are CO2 emissions driven and thus ELM must run with active biogeochemistry when the EHC is active to provide EAM with land CO2 emissions.
We have generated a scientifically valid E3SM-GCAM compset for SSP2-RCP45 (O’Neill et al., 2016) that starts at 2015-01-01 with active EAM-ELM-EHC components, data ocean and sea ice, and the remaining components disabled. The prescribed ocean CO2 flux to the atmosphere, sea surface temperature, and sea ice extent are scenario-specific and based on CMIP6 data from CESM version 2 (ocean CO2 flux), and the Geophysical Fluid Dynamics Laboratory Earth system Model version 4 (GFDL-ESM4; sea surface temperature and ice extent). The initial conditions to start E3SM-GCAM in 2015 were determined by spinning the model up to an 1850 equilibrium state and then simulating years 1850 through 2014. This process used a similar configuration to that described above but with the EHC and CO2 fluxes disabled and corresponding CMIP6 data for sea surface temperature and sea ice extent (Bailey et al., 2011), GHG concentrations (including CO2; Meinshausen et al., 2017), and aerosol emissions (Hoesly et al., 2018; van Marle et al., 2017). A static, 1850-condition simulation ran for 200 years in accelerated spin-up mode and 400 years in non-accelerated spin-up mode until total land ecosystem carbon varied by less than 10 Pg C per century, then a historical simulation that included land change and historical forcings was performed through 2014. The baseline terrestrial productivity files were generated from years 2010–2014 of this historical simulation.
Experimental Design
Scenario
We focus on SSP2-RCP45 to compare our results with iESM and CMIP6 simulations. E3SMv2.1 and GCAM6 perform similarly to the CMIP6 multi-model ensemble such that the global average surface temperature trajectory for E3SM-GCAM is very similar to the CMIP6 average trajectory for these scenarios (Figure S2 in Supporting Information S1; Tebaldi et al., 2021). GCAM6 land allocation is in line with CMIP6, except for having forest gain instead of forest loss (Table 1). Our SSP2-RCP45 scenario is, however, very different from the iESM because the models are different (e.g., Calvin, Bond-Lamberty, et al., 2019; Thornton et al., 2017).
Table 1 Global Land Allocation Scenario Comparison, Based on Human System Models
Model | Cropland | Pasture | Forest | Other |
E3SM-GCAM SSP2-RCP45 CONTROL | 3.158 | 0.749 (grazed) | 1.960 | −5.597 (shrub + ungrazed grass) |
CMIP6 SSP2 approximate range (Riahi et al., 2017, Figure 4) | 2.25 to 4 | −2 to 2 | −3 to −0.3 | −5.6 to 0.3 |
iESM RCP45 no feedbacks (Calvin, Patel, et al., 2019) | −0.509 | −6.090 (grazed + ungrazed) | 9.935 | −3.336 (grassland) |
GCAM is configured for SSP2 following CMIP6 protocols (Calvin et al., 2017), with a global carbon emissions price phase-in starting in 2025, and a land carbon price phase-in from 2040 to 2065 up to 25% of the global carbon price. This price pathway and the non-CO2 forcings for EAM are from a standalone GCAM target-finding simulation for SSP2-RCP45 or CMIP6 archives as described above.
We have tuned the land conversion assumptions to obtain a reasonable match between GCAM and ELM forest area changes for SSP2-RCP45 and use the same values for all simulations. Cropland expansion replaces non-crop PFTs such that non-forest PFTs are preferentially, but not exclusively, removed. Bare ground is replaced only if not enough non-bare PFT is available. Upon cropland abandonment non-crop and non-bare PFTs are added proportionally to their potential vegetation availability, with tree PFTs filled to potential before others are added. Pasture expansion replaces non-crop and non-bare PFTs proportionally to their current extent with grass. Upon pasture abandonment only tree PFTs are added proportionally to their potential vegetation availability.
Simulations
We perform four simulations for 2015–2100 (Table 2). Three include different levels of terrestrial productivity feedbacks: (a) agricultural yield feedbacks (AG_FDBK), (b) potential carbon density feedbacks (C_FDBK), and (c) both agricultural and carbon feedbacks (FULL_FDBK). The fourth simulation does not include terrestrial productivity feedbacks (CONTROL). Runtime GCAM CO2 emissions are passed to EAM in all simulations.
Table 2 E3SM-GCAM Simulations Described Here, Varying in the Degree of Terrestrial Feedback Information Being Passed From the E3SM Land Model (ELM) to the Human Systems GCAM Model
GCAM → ELM | GCAM → EAM | ELM → GCAM | ELM → GCAM | |
Name | Land use/cover | CO2 emissions | Feedback on agricultural yields | Feedback on potential carbon densities |
CONTROL | Yes | Yes | No | No |
AG_FDBK | Yes | Yes | Yes | No |
C_FDBK | Yes | Yes | No | Yes |
FULL_FDBK | Yes | Yes | Yes | Yes |
Verification and Expected Results
As both E3SM and GCAM have been evaluated independently, the main form of verification here was to ensure that the models were giving and receiving the correct data. Thus, we performed intensive testing to check values at all processing steps. Once we verified that the communication was correct we developed the scenario and experiment described above to demonstrate the efficacy of the two-way coupling. The land conversion assumption tuning was an expected process because it has been previously required for E3SM, and the land data fidelity between GCAM and E3SM has improved.
The CONTROL simulation emulates independent GCAM and E3SM runs, similar to the original IPCC protocol. We had to run an independent GCAM simulation to develop the scenario, and the CO2 emissions and land allocation of this independent run exactly match those of the CONTROL simulation. We also expected these drivers to match the range of CMIP6 SSP2 scenarios, including the RCP45 marker scenario (e.g., Riahi et al., 2017), and they match quite well except for forest allocation. This extra forest allocation was also expected, however, due to GCAM's propensity for using forest for mitigation (e.g., Calvin, Patel, et al., 2019).
We expected changes in carbon and climate due to the inclusion of terrestrial productivity feedbacks (FULL_FDBK) commensurate with the corresponding changes in land allocation. First of all, even though the models and scenario are quite different, we expected some similarity of the terrestrial productivity feedbacks in FULL_FDBK to those from the iESM. This is indeed the case, as presented in more detail below, which instills confidence in the coupled system because it shows that E3SM and the scaling routine are functioning as expected. The feedback pathway causes changes in GCAM crop productivity and/or potential land carbon density that can affect production and land value, thus changing profits, resource usage, and land allocation. This can alter energy production (via changes in bioenergy production) and emissions (via changes in land, water, or fertilizer use). For SSP2-RCP45 there are appropriate land allocations responses in GCAM that lead to expected magnitudes of carbon and climate responses in E3SM (see below). Overall, the new system is behaving as expected.
Coupled System Performance and Diagnostics
Performance
The EHC adds a disproportionate amount of time to simulations because it runs on a single processor while the other components wait. Approximately 48% of the EHC run time is GCAM, which runs once every 5 years. These metrics are based on 20 model years on the LCRC chrysalis machine with the coupler, ELM, EAM, and the data ocean/sea-ice components allocated 5,120 processors each. The CONTROL simulation cost 3,665 processor hours per simulated year with a throughput of 33.53 simulated years per day. The EHC constitutes 6.4% of the total run time in the CONTROL simulation and 7.5% in the FULL_FDBK simulation, indicating an approximate 1.33 simulated years per day decrease in throughput due to including the EHC in the CONTROL simulation. GCAM run times vary from year to year and are about 84% of the EHC in a GCAM-active year, the remainder of which includes processing of CO2 emissions, land use data, feedbacks, and setting the feedbacks in GCAM if they are enabled. In a non-GCAM year processing and setting of feedbacks are omitted from the workflow. GCAM requires a spinup the first time a scenario is run, which adds about 8 minutes to the EHC run time in the first model year.
Terrestrial Productivity Feedbacks
The terrestrial productivity scaling is distinct for each GCAM land unit and land type (Figure 4 and Figure S3 in Supporting Information S1). This results from a combination of geographic, PFT, and land type heterogeneity. ELM contributes geographic and PFT variability associated with environmental conditions and vegetation distribution, and GCAM contributes heterogeneous distributions of specific land types. Each PFT has its own response to changing environmental conditions in each ELM grid cell. Further differentiation of specific GCAM land types occurs at the land unit level due to the spatial distribution of these types (Figure S3 in Supporting Information S1).
[IMAGE OMITTED. SEE PDF]
Such detailed scaling is interesting both methodologically and scientifically. There is only one crop in ELM in this study, but each GCAM crop can have a unique scaling trajectory within a region because the spatial (dis-) aggregation and vegetation mapping generate distinct values. Based on observations of spatial heterogeneity of NEP and its interannual variability (Cui et al., 2020), these distinct values are more scientifically valid than just a few values that represent several land types across large areas. Furthermore, using 5-year averages to calculate scaling values facilitates model stability and validity. This averaging matches GCAM's projection period and reduces unrealistic shocks to land carbon density that could preclude GCAM from solving.
Effects of Terrestrial Productivity Feedbacks on the Human System
Including terrestrial productivity feedbacks incorporates climate change impacts on vegetation into the scenario, which changes land allocation projections. At the global level FULL_FDBK has an increasing vegetation productivity trend and a decreasing soil carbon trend for aggregated land types, with considerable inter-period variability (Figure 5). The largest changes in land allocation due to including feedbacks are in total cropland (−5% in 2075) and unmanaged forest (+3.5% in 2100), with some increase in total forest by the end of the century (Figure 6). Most notably, the inter-period variability of the feedbacks translates into land allocation, particularly for cropland. Increases in cropland allocation (Figure 6e) are aligned with decreases in crop productivity (Figure 5a) driven by varying environmental conditions (and vice versa). Feedbacks increase the rate of forest expansion after land carbon valuation starts in 2040, and forest expansion continues longer than without feedbacks due to greater carbon accumulation potential with feedbacks. Inter-period variability in non-crop land types is likely driven by the variability in cropland allocation in the FULL_FDBK simulation.
[IMAGE OMITTED. SEE PDF]
[IMAGE OMITTED. SEE PDF]
These results are different from the corresponding iESM experiment due to scenario and feedback differences. The iESM RCP4.5 control scenario had a very large increase in forest area (>5 M km2) and an ∼1 M km2 loss in crop area by 2100 that were both enhanced by feedbacks (Calvin, Bond-Lamberty, et al., 2019; Thornton et al., 2017). The effects of feedbacks in iESM were largely driven by the forest response because of the dominant forest expansion signal. By 2100, E3SM-GCAM SSP2-RCP45 CONTROL has a much smaller increase in forest area (≤2 M km2) and an increase in crop area (∼3 M km2), which contribute to different feedback dynamics even though the iESM and E3SM-GCAM feedback signals have similar magnitudes (with the exception of shrubland) (Figure 7; Thornton et al., 2017). The E3SM-GCAM response to feedbacks is thus driven by the dominant cropland expansion signal and its interaction with the productivity trend and inter-period variability. This indicates that the effects of incorporating terrestrial feedbacks and their corresponding impacts are scenario and feedback dependent, which means that the effects are also model dependent because the scenario is generated by the human system model and the feedbacks are generated by the Earth system model.
[IMAGE OMITTED. SEE PDF]
Including feedbacks generally decreases crop prices, although there is considerable variability across crop type, time, and region (Figure 7). Rice, soybean, and wheat see greater than 5% median decreases in 2100, with two regions having greater than 20% decreases in rice price. The reasons for price decreases vary because of the complex interactions between crop productivity, cropland allocation, trade, and land value. Crop productivity is generally higher with feedbacks (Figure 5), which allows for less global cropland to meet demand (Figure 6), but the change in cropland distribution can increase or decrease regional yields and production, which affect prices, demand, and trade. Including feedbacks has little effect on bioenergy crop prices, which corresponds with their limited effect on global bioenergy cropland allocation, and as a result there are negligible effects on global CO2 emissions and concentrations.
The projected cropland and forest area changes are passed to ELM with good fidelity, although model differences in spatial grid, resolution, and land data alter the projected area changes (Figure S4 in Supporting Information S1). Changes in ELM cropland area follow GCAM cropland changes relatively well: there is only a 7% difference between ELM's and GCAM's crop decreases in 2100 due to feedbacks. Cumulative differences between GCAM and ELM forest changes are on the order of only 1% of ELM forest area but lead to a 153% greater increase in global forest area with feedbacks in ELM over that projected by GCAM (Figure S4b in Supporting Information S1). This discrepancy results from complex interactions between land conversion assumptions, land area change, existing land type, potential land type, and model differences including grid specification and land type definition and distribution. For example, ELM has higher absolute values of land type area, including 67% more forest area than GCAM, which affects data fidelity because values are passed as fractions of grid cell rather than area. These inconsistencies are a persistent problem when coupling human and Earth system models. We have tuned the land conversion assumptions and revised some code to increase the fidelity of forest changes between the models, but without an explicit, comprehensive characterization of land use and land cover for both models there will always be some inconsistency.
Effects of Terrestrial Productivity Feedbacks on the Earth System
Including terrestrial productivity feedbacks changes the SSP2-RCP45 land allocation, which in turn alters carbon and climate projections at local and global levels. By 2100, global terrestrial ecosystem carbon is reduced by 15 PgC with feedbacks (−11% of the 140 PgC increase in CONTROL from 2015 to 2100) due to ongoing land cover change that outweighs overall forest expansion (Figure 8). For comparison, the CMIP6 multi-model mean increase in land carbon from 2015 to 2100 for SSP2-RCP45 is ∼185 PgC (Liddicoat et al., 2021). Global wood harvest is also affected by changes in land allocation, primarily after land carbon valuation is initiated in 2040 (Figure S5 in Supporting Information S1). Local differences in terrestrial ecosystem carbon by 2090 are statistically significant over most land, vary considerably across space, and exceed ±600 gC/m2 in many places (Figure S6 in Supporting Information S1).
[IMAGE OMITTED. SEE PDF]
Including feedbacks affects climate primarily at the local to regional levels, as a result of the changes in land allocation. These changes in climate are comparable to related estimates from other studies, but an ensemble is required to fully evaluate these results. Based on temporal variability, differences in climate metrics are statistically significant locally for a relatively small portion of land. Average near surface temperature changes over land range from −0.76 to +0.94°C during 2071–2090 (Figure S7 in Supporting Information S1), which is expected based on previous estimates of land change effects on surface temperature (e.g., Brovkin et al., 2013). Feedbacks can also change rainfall extremes by more than ±50% (Figure 9) and temperature extremes by several degrees, with high spatial heterogeneity. Maximum daily and 5-day rainfall between 2071 and 2090 can increase in places where annual average rainfall decreases (Figure 9 and Figure S8 in Supporting Information S1). Maximum daily maximum temperature during 2071–2090 may shift by ±2°C and minimum daily minimum temperature by −4 to +6°C (Figure S9 in Supporting Information S1). These changes in climate extreme metrics due to human-Earth feedbacks are comparable to or greater than mean CMIP6 estimated changes due to climate change for SSP2-RCP45 (Chen et al., 2020) (Figure 9 and Figures S8 and S10 in Supporting Information S1).
[IMAGE OMITTED. SEE PDF]
Influence of Agricultural Yield Versus Carbon Density Feedbacks on Outcomes
Impacts to agricultural yields drive the feedback responses in our SSP2-RCP45 scenario. GCAM projected land allocation for AG_FDBK matches well with FULL_FDBK, while C_FDBK is like CONTROL, through 2080 (Figure 6). After 2080, C_FDBK increases in importance as expanding forests reach sufficient growth for interactions between productivity feedbacks and carbon values to increase forest expansion rates. Correspondingly, AG_FDBK global terrestrial ecosystem carbon follows FULL_FDBK (Figure 8a), but regional carbon and climate effects for the C_FDBK simulation can be nearly as high as those for the AG_FDBK and FULL_FDBK simulations because land change occurs in all scenarios (Figures S6 and S11 in Supporting Information S1). The carbon density feedbacks contribute to land change emissions starting around 2070 (Figure 8b), likely due to increasing land carbon values from 2040 to 2065 that drive forest expansion later in the century. This contrasts with the iESM responses that appear to be driven by carbon density impacts on forest, likely due to the dominance of forest expansion in the iESM RCP4.5 scenario (Calvin, Bond-Lamberty, et al., 2019; Thornton et al., 2017). Importantly, the agricultural and carbon feedbacks combine non-linearly to produce the full feedback effects in the Earth system.
Conclusions and Future Work
Our simulations demonstrate the efficacy of a novel, two-way human-Earth coupling in a global Earth system model that successfully incorporates synchronous terrestrial productivity feedbacks between the Earth and human components. These feedbacks alter the scenario by applying climate change impacts to agricultural yield and potential carbon densities, which in turn alters the Earth system projection. The inclusion of terrestrial productivity feedbacks affects global land carbon, regional near surface temperature, and regional climate extremes. This demonstrates the importance of including human-Earth feedbacks when assessing potential land-based climate change mitigation and adaptation strategies.
As part of the supported E3SM architecture, the E3SM-GCAM coupling can be expanded and improved without having to rebuild the human-Earth coupling. For example, feedbacks such as temperature effects on heating/cooling demand or water availability limitations on water use can be added using this new infrastructure. Furthermore, land change inconsistencies persist that can be reduced by expanding the information passed from GCAM to ELM rather than implementing new coupling. For example, passing all land type changes will reduce differences in non-crop land types such as forest.
E3SM-GCAM is an advance toward more comprehensive and sophisticated human-Earth feedback analyses, and represents a unique capability for the current generation of Earth System Models (IPCC et al., 2021). While the results presented here are scenario- and model-dependent, we intend to implement additional scenarios and hope to inspire the research community to engage in similar efforts to incorporate human-environment feedbacks at various model scales and complexities.
Acknowledgments
This research was supported as part of the E3SM project, funded by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research (BER) through the Earth System Model Development (ESMD) program area. Additional support was provided by BER's Multi-Sector Dynamics program area. Developmental and experimental simulations were performed using the BER ESMD program's Compy computing cluster located at Pacific Northwest National Laboratory and Chrysalis computing cluster located at Argonne National Laboratory. Work at Lawrence Berkeley National Laboratory was performed under contract DEAC02-05CH11231, and work at Battelle-operated Pacific Northwest National Laboratory was performed under contract DE-AC05-76RL01830. Dr. Katherine Calvin is currently detailed to the National Aeronautics and Space Administration. Dr. Calvin's contributions to this article occurred prior to her detail. The views expressed are her own and do not necessarily represent the views of the National Aeronautics and Space Administration or the United States Government. We acknowledge Dr. Louise Chini's (Department of Geographical Sciences, University of Maryland) contributions to updating some of the EHC reference data, and Dr. Ruby Leung's (Pacific Northwest National Laboratory) insightful review of the manuscript. We also thank the anonymous reviewers for helping us improve the manuscript.
Data Availability Statement
E3SM-GCAM version 2.1 code is available at zenodo and github (E3SM-GCAM, 2024). Model data for this paper, Supporting Information S1, and diagnostic scripts are available at zenodo and github (Di Vittorio & Sinha, 2025). Full model outputs are publicly archived at the National Energy Research Scientific Computing Center (NERSC, 2024).
Bailey, D., Holland, M., Hunke, E., Lipscomb, B., Briegleb, B., Bitz, C., & Schramm, J. (2011). Community Ice CodE (CICE) user’s guide version 4.0. National Center for Atmospheric Research. Retrieved from https://www2.cesm.ucar.edu/models/ccsm4.0/cice/ice_usrdoc.pdf
Beckage, B., Lacasse, K., Winter, J. M., Gross, L. J., Fefferman, N., Hoffman, F. M., et al. (2020). The Earth has humans, so why don’t our climate models? Climatic Change, 163(1), 181–188. https://doi.org/10.1007/s10584‐020‐02897‐x
Bond‐Lamberty, B., Calvin, K., Jones, A. D., Mao, J., Patel, P., Shi, X., et al. (2014). Coupling earth system and integrated assessment models: The problem of steady state. Geoscientific Model Development Discussions (GMDD), 7, 1499–1524. https://doi.org/10.5194/gmdd‐7‐1499‐2014
Brovkin, V., Boysen, L., Arora, V. K., Boisier, J. P., Cadule, P., Chini, L., et al. (2013). Effect of anthropogenic land‐use and land‐cover changes on climate and land carbon storage in CMIP5 projections for the twenty‐first century. Journal of Climate, C4MIP collection, 26(18), 6859–6881. https://doi.org/10.1175/JCLI‐D‐12‐00623.1
Calvin, K., & Bond‐Lamberty, B. (2018). Integrated human‐earth system modeling—State of the science and future directions. Environmental Research Letters, 13(6), 063006. https://doi.org/10.1088/1748‐9326/aac642
Calvin, K., Bond‐Lamberty, B., Clarke, L., Edmonds, J., Eom, J., Hartin, C., et al. (2017). The SSP4: A world of deepening inequality. Global Environmental Change, 42, 284–296. https://doi.org/10.1016/j.gloenvcha.2016.06.010
Calvin, K., Bond‐Lamberty, B., Jones, A., Shi, X., Di Vittorio, A., & Thornton, P. (2019). Characteristics of human‐climate feedbacks differ at different radiative forcing levels. Global and Planetary Change, 180, 126–135. https://doi.org/10.1016/j.gloplacha.2019.06.003
Calvin, K., Patel, P., Clarke, L., Asrar, G., Bond‐Lamberty, B., Cui, R. Y., et al. (2019). GCAM v5.1: Representing the linkages between energy, water, land, climate, and economic systems. Geoscientific Model Development, 12(2), 677–698. https://doi.org/10.5194/gmd‐12‐677‐2019
Calvin, K., Wise, M., Kyle, P., Patel, P., Clarke, L., & Edmonds, J. (2014). Trade‐offs of different land and bioenergy policies on the path to achieving climate targets. Climatic Change, 123(3–4), 691–704. https://doi.org/10.1007/s10584‐013‐0897‐y
Chen, H., Sun, J., Lin, W., & Xu, H. (2020). Comparison of CMIP6 and CMIP5 models in simulating climate extremes. Science Bulletin, 65(17), 1415–1418. https://doi.org/10.1016/j.scib.2020.05.015
Clarke, L., Eom, J., Marten, E. H., Horowitz, R., Kyle, P., Link, R., et al. (2018). Effects of long‐term climate change on global building energy expenditures. Energy Economics, 72, 667–677. https://doi.org/10.1016/j.eneco.2018.01.003
Collins, W. D., Craig, A. P., Truesdale, J., Di Vittorio, A. V., Jones, A. D., Bond‐Lamberty, B., et al. (2015). The integrated Earth System Model (iESM): Formulation and functionality. Geoscientific Model Development, 8(7), 2203–2219. https://doi.org/10.5194/gmd‐8‐2203‐2015
Craig, A. P., Vertenstein, M., & Jacob, R. (2012). A new flexible coupler for earth system modeling developed for CCSM4 and CESM1. The International Journal of High Performance Computing Applications, 26(1), 31–42. https://doi.org/10.1177/1094342011428141
Cronin, J., Anandarajah, G., & Dessens, O. (2018). Climate change impacts on the energy system: A review of trends and gaps. Climatic Change, 151(2), 79–93. https://doi.org/10.1007/s10584‐018‐2265‐4
Cui, E., Bian, C., Luo, Y., Niu, S., Wang, Y., & Xia, J. (2020). Spatial variations in terrestrial net ecosystem productivity and its local indicators. Biogeosciences, 17(23), 6237–6246. https://doi.org/10.5194/bg‐17‐6237‐2020
Di Vittorio, A. V., Chini, L. P., Bond‐Lamberty, B., Mao, J., Shi, X., Truesdale, J., et al. (2014). From land use to land cover: Restoring the afforestation signal in a coupled integrated assessment ‐ Earth system model and the implications for CMIP5 RCP simulations. Biogeosciences, 11(22), 6435–6450. https://doi.org/10.5194/bg‐11‐6435‐2014
Di Vittorio, A. V., & Sinha, E. (2025). Data, supplemental information, and diagnostic scripts for this paper [Dataset]. GitHub. https://doi.org/10.5281/zenodo.15642038
DOWNSCALE. (2024). Software to downscale and grid IAMC data for use by Earth system models. Modified for current application. Retrieved from https://github.com/daleihao/emissions_downscaling/tree/daleihao/post_SSP_update
E3SM documentation. (2019). Model user’s guide. Retrieved from https://acme‐climate.atlassian.net/wiki/spaces/DOC/pages/924385312/Model+User+s+Guide
E3SM‐GCAM software. (2024). E3SM‐GCAM v2.1 [Software]. Zenodo (TBD) and Github. https://doi.org/10.5281/zenodo.15159486
Feng, L. F., Smith, S. J., Braun, C., Crippa, M., Gidden, M. J., Hoesly, R., et al. (2020). The generation of gridded emissions data for CMIP6. Geoscientific Model Development, 13(2), 461–482. https://doi.org/10.5194/gmd‐13‐461‐2020
Fredriksen, H.‐B., Smith, C. J., Modak, A., & Rugenstein, M. (2023). 21st century scenario forcing increases more for CMIP6 than CMIP5 models. Geophysical Research Letters, 50(6), e2023GL102916. https://doi.org/10.1029/2023GL102916
GCAM2IAMC. (2021). Software to process GCAM outputs to IAMC format. Retrieved from https://github.com/JGCRI/GCAM_to_IAMC_Aggregation
GCAM6. (2022). Global change Analysis model, version 6. Retrieved from https://github.com/JGCRI/gcam‐core/releases/tag/gcam‐v6.0
Gidden, M. J., Riahi, K., Smith, S. J., Fujimori, S., Luderer, G., Kriegler, E., et al. (2019). Global emissions pathways under different socioeconomic scenarios for use in CMIP6: A dataset of harmonized emissions trajectories through the end of the century. Geoscientific Model Development, 12(4), 1443–1475. https://doi.org/10.5194/gmd‐12‐1443‐2019
Golaz, J.‐C., Van Roekel, L. P., Zheng, X., Roberts, A. F., Wolfe, J. D., Lin, W., et al. (2022). The DOE E3SM model version 2: Overview of the physical model and initial model evaluation. Journal of Advances in Modeling Earth Systems, 14(12), e2022MS003156. https://doi.org/10.1029/2022ms003156
Hartin, C. A., Patel, P., Schwarber, A., Link, R. P., & Bond‐Lamberty, B. P. (2015). A simple object‐oriented and open‐source model for scientific and policy analyses of the global climate system ‐ Hector v1.0. Geoscientific Model Development, 8(4), 939–955. https://doi.org/10.5194/gmd‐8‐939‐2015
Hoesly, R. M., Smith, S. J., Feng, L., Klimont, Z., Janssens‐Maenhout, G., Pitkanen, T., et al. (2018). Historical (1750–2014) anthropogenic emissions of reactive gases and aerosols from the Community Emissions Data System (CEDS). Geoscientific Model Development, 11(1), 369–408. https://doi.org/10.5194/gmd‐11‐369‐2018
Hurtt, G. C., Chini, L., Sahajpal, R., Frolking, S., Bodirsky, B. L., Calvin, K., et al. (2020). Harmonization of global land use change and management for the period 850‐2100 (LUH2) for CMIP6. Geoscientific Model Development, 13(11), 5425–5464. https://doi.org/10.5194/gmd‐13‐5425‐2020
Hurtt, G. C., Chini, L. P., Frolking, S., Betts, R. A., Feddema, J., Fischer, G., et al. (2011). Harmonization of land‐use scenarios for the period 1500–2100: 600 years of global gridded annual land‐use transitions, wood harvest, and resulting secondary lands. Climatic Change, 109(1–2), 117–161. https://doi.org/10.1007/s10584‐011‐0153‐2
IPCC. (2022). H.‐O. Pörtner, D. C. Roberts, M. Tignor, E. S. Poloczanska, K. Mintenbeck, A. Alegría, et al. (eds.), Climate change 2022: Impacts, adaptation and vulnerability. Contribution of working group II to the sixth assessment report of the intergovernmental panel on climate change (3056). Cambridge University Press. Cambridge University Press. https://doi.org/10.1017/9781009325844
IPCC, Guivarch, C., Kriegler, E., Portugal‐Pereira, J., Bosetti, V., Edmonds, J., et al. (2022). Annex III: Scenarios and modelling methods. In P. R. Shukla, J. Skea, R. Slade, A. AlKhourdajie, R. vanDiemen, D. McCollum, et al. (Eds.), Climate change 2022: Mitigation of climate change. Contribution of working group III to the sixth assessment report of the intergovernmental panel on climate change. Cambridge University Press. https://doi.org/10.1017/9781009157926.022
IPCC, Gutiérrez, J. M., & Tréguier, A.‐M. (2021). Annex II: Models. In V. Masson‐Delmotte, P. Zhai, A. Pirani, S. L. Connors, C. Péan, S. Berger, et al. (Eds.), Climate change 2021: The physical science basis. Contribution of working group I to the sixth assessment report of the intergovernmental panel on climate change (pp. 2087–2138). Cambridge University Press. https://doi.org/10.1017/9781009157896.016
IPCC, Lee, H., & Romero, J. (2023). Summary for policymakers. In H. Lee & J. Romero (Eds.), Climate change 2023: Synthesis report. Contribution of working groups I, II and III to the sixth assessment report of the intergovernmental panel on climate change (pp. 1–34). IPCC. https://doi.org/10.59327/IPCC/AR6‐9789291691647.001
Iyer, G., Ou, Y., Edmonds, J., Fawcett, A. A., Hultman, N., McFarland, J., et al. (2022). Ratcheting of climate pledges needed to limit peak global warming. Nature Climate Change, 12, 1129–1135. https://doi.org/10.1038/s41558‐022‐01508‐0
Jager, F., Schwaab, S. J., Quilcaille, Y., Windisch, M., Doelman, J., Frank, S., et al. (2024). Fire weather compromises forestation‐reliant climate mitigation pathways. Earth System Dynamics, 15(4), 1055–1071. https://doi.org/10.5194/esd‐15‐1055‐2024
Jones, A. D., Calvin, K. V., Shi, X., Di Vittorio, A. V., Bond‐Lamberty, B., Thornton, P. E., & Collins, W. D. (2018). Quantifying human‐mediated carbon cycle feedbacks. Geophysical Research Letters, 45(11), 11370–11379. https://doi.org/10.1029/2018GL079350
Jones, C., Robertson, R., Arora, V., Friedlingstein, P., Shevliakova, E., Bopp, L., et al. (2013). Twenty‐first‐century compatible CO2 emissions and airborne fraction simulated by CMIP5 Earth system models under four representative concentration pathways. Journal of Climate, 26(13), 4398–4413. https://doi.org/10.1175/JCLI‐D‐12‐00554.1
Kyle, P., Müller, C., Calvin, K., & Thomson, A. (2014). Meeting the radiative forcing targets of the representative concentration pathways in a world with agricultural climate impacts. Earth's Future, 2, 83–98. https://doi.org/10.1002/2013EF000199
Laidlaw, E. K., O'Neill, B. C., & Harp, R. D. (2019). The use of the Community Earth System Model in human dimensions climate research and applications. WIREs Climate Change, 10(3), e582. https://doi.org/10.1002/wcc.582
Leung, L. R., Bader, D. C., Taylor, M. A., & McCoy, R. B. (2020). An introduction to the E3SM special collection: Goals, science drivers, development, and analysis. Journal of Advances in Modeling Earth Systems, 12(11), e2019MS001821. https://doi.org/10.1029/2019MS001821
Liddicoat, S. K., Wiltshire, A. J., Jones, C. D., Arora, V. K., Brovkin, V., Cadule, P., et al. (2021). Compatible fossil fuel CO2 emissions in the CMIP5 Earth system models’ historical and shared socioeconomic pathway experiments of the twenty‐first century. Journal of Climate, 34(8), 2853–2875. https://doi.org/10.1175/JCLI‐D‐19‐0991.1
Liu, S., Bond‐Lamberty, B., Boysen, L. R., Ford, J. D., Fox, A., Gallo, K., et al. (2017). Grand challenges in understanding the interplay of climate and land changes. Earth Interactions, 21(2), 1–43. https://doi.org/10.1175/EI‐D‐16‐0012.1
Mall, R. K., Gupta, A., & Sonkar, G. (2017). 2 ‐ Effect of climate change on agricultural crops, In S. K. Dubey, A. Pandey, & R. S. Sangwan (Eds.), Current developments in biotechnology and bioengineering: Crop modification, nutrition, and food production. https://doi.org/10.1016/B978‐0‐444‐63661‐4.00002‐5
Meinshausen, M., Vogel, E., Nauels, A., Lorbacher, K., Meinshausen, N., Etheridge, D. M., et al. (2017). Historical greenhouse gas concentrations for climate modelling. Geoscientific Model Development, 10(5), 2057–2116. https://doi.org/10.5194/gmd‐10‐2057‐2017
Motesharrei, S., Rivas, J., Kalnay, E., Asrar, J. R., Busalacchi, A. J., Cahalan, R. F., et al. (2016). Modeling sustainability: Population, inequality, consumption, and bidirectional coupling of the Earth and Human Systems. National Science Review, 3, 470–494. https://doi.org/10.1093/nsr/nww081
NERSC. (2024). Full model outputs for the E3SM‐GCAM v2.1 simulations in this paper [Dataset]. National Energy Research Scientific Computing Center. Retrieved from https://portal.nersc.gov/archive/home/e/esinha/www/E3SM_GCAM_JAMES_2024
O’Neill, B. C., Tebaldi, C., van Vuuren, D. P., Eyring, V., Friedlingstein, P., Hurtt, G., et al. (2016). The scenario model intercomparison project (ScenarioMIP) for CMIP6. Geoscientific Model Development, 9, 3461–3482. https://doi.org/10.5194/gmd‐9‐3461‐2016
Pongratz, J., Dolman, H., Don, A., Erb, K.‐H., Fuchs, R., Herold, M., et al. (2018). Models meet data: Challenges and opportunities in implementing land management in Earth system models. Global Change Biology, 24(4), 1470–1487. https://doi.org/10.1111/gcb.13988
Rezaei, E. E., Webber, H., Asseng, S., Boote, K., Durand, J. L., Ewart, F., et al. (2023). Climate change impacts on crop yields. Nature Reviews Earth & Environment, 4(12), 831–846. https://doi.org/10.1038/s43017‐023‐00491‐0
Riahi, K., van Vuuren, D. P., Kriegler, E., Edmonds, J., O’Neill, B. C., Fujimori, S., et al. (2017). The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Global Environmental Change, 42, 153–168. https://doi.org/10.1016/j.gloenvcha.2016.05.009
Robinson, D. T., Di Vittorio, A., Alexander, P., Arneth, A., Barton, C. M., Brown, D. G., et al. (2018). Modelling feedbacks between human and natural processes in the land system. Earth System Dynamics, 9(2), 895–914. https://doi.org/10.5194/esd‐9‐895‐2018
Sinha, E., Calvin, K. V., Bond‐Lamberty, B., Drewniak, B. A., Ricciuto, D. M., Sargsyan, K., et al. (2022). Modeling perennial bioenergy crops in the E3SM land model (ELMv2). Journal of Advances in Modeling Earth Systems, 15(1), e2022MS003171. https://doi.org/10.1029/2022ms003171
Snyder, A., Calvin, K., Clarke, L., Edmonds, J., Kyle, P., Narayan, K., et al. (2020). The domestic and international implications of future climate for U.S. agriculture in GCAM. PLoS One, 15(8), e0237918. https://doi.org/10.1371/journal.pone.0237918
Tebaldi, C., Debeire, K., Eyring, V., Fischer, E., Fyfe, J., Friedlingstein, P., et al. (2021). Climate model projections from the scenario model intercomparison project (ScenarioMIP) of CMIP6. Earth System Dynamics, 12(1), 253–293. https://doi.org/10.5194/esd‐12‐253‐2021
Thornton, P. E., Calvin, K., Jones, A. D., Di Vittorio, A. V., Bond‐Lamberty, B., Chini, L., et al. (2017). Biospheric feedback effects in a synchronously coupled model of human and Earth systems. Nature Climate Change, 7, 496–500. https://doi.org/10.1038/nclimate3310
van Marle, M. J. E., Kloster, S., Magi, B. I., Marlon, J. R., Daniau, A.‐L., Field, R. D., et al. (2017). Historic global biomass burning emissions for CMIP6 (BB4CMIP) based on merging satellite observations with proxies and fire models (1750–2015). Geoscientific Model Development, 10(9), 3329–3357. https://doi.org/10.5194/gmd‐10‐3329‐2017
van Vuuren, D. P., Bayer, L. B., Chuwah, C., Ganzeveld, L., Hazeleger, W., van den Hurk, B., et al. (2012). A comprehensive view on climate change: Coupling of earth system and integrated assessment models. Environmental Research Letters, 7(2), 024012. https://doi.org/10.1088/1748‐9326/7/2/024012
Verburg, P. H., Dearing, J. A., Dyke, J. G., van der Leeuw, S., Seitzinger, S., Steffen, W., & Syvitski, J. (2016). Methods and approaches to modelling the Anthropocene. Global Environmental Change, 39, 328–340. https://doi.org/10.1016/j.gloenvcha.2015.08.007
Wang, R., Bowling, L. C., & Cherkauer, K. A. (2016). Estimation of the effects of climate variability on crop yield in the Midwest USA. Agricultural and Forest Meteorology, 216, 141–156. https://doi.org/10.1016/j.agrformet.2015.10.001
Wise, M., Calvin, K., Kyle, P., Luckow, P., & Edmonds, J. (2014). Economic and physical modeling of land use in GCAM 3.0 and an application to agricultural productivity, land, and terrestrial carbon. Climate Change Economics, 5(2), 1450003. https://doi.org/10.1142/S2010007814500031
Yalew, S. G., van Vliet, M. T. H., Gernaat, D. E. H. J., Ludwig, F., Miara, A., Park, C., et al. (2020). Impacts of climate change on energy system in global and regional scenarios. Nature Energy, 5(10), 794–802. https://doi.org/10.1038/s41560‐020‐0664‐z
Zhou, T., Leung, R. L., Leng, G., Voisin, N., & Li, H.‐Y. (2020). Global irrigation characteristics and effects simulated by fully coupled land surface, river, and water management models in E3SM. Journal of Advances in Modeling Earth Systems, 12(10), e2020M5002069. https://doi.org/10.1029/2020M5002069
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2025. This work is published under http://creativecommons.org/licenses/by-nc/4.0/ (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Modeling human‐environment feedbacks is critical for assessing the effectiveness of climate change mitigation and adaptation strategies under a changing climate. The Energy Exascale Earth System Model (E3SM) now includes a human component, with the Global Change Analysis Model (GCAM) at its core, that is synchronously coupled with the land and atmosphere components through the E3SM coupling software. Terrestrial productivity is passed from E3SM to GCAM to make climate‐responsive land use and CO2 emission projections for the next 5‐year period, which are interpolated and passed to E3SM annually. Key variables affected by the incorporation of these feedbacks include land use/cover change, crop prices, terrestrial carbon, local surface temperature, and climate extremes. Regional differences are more pronounced than global differences because the effects are driven primarily by differences in land use. This novel system enables a new type of scenario development and provides a powerful modeling framework that facilitates the addition of other feedbacks between these models. This system has the potential to explore how human responses to climate change impacts in a variety of sectors, including heating/cooling energy demand, water management, and energy production, may alter emissions trajectories and Earth system changes.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details






1 Climate and Ecosystem Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA
2 Atmosoheric, Climate, & Earth Sciences Division, Pacific Northwest National Laboratory, Richland, WA, USA
3 Pacific Northwest National Laboratory, Joint Global Change Research Institute, College Park, MD, USA