Bernhard Ahrens
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Bernhard Ahrens has a total of 225 co-authors. Here are the top co-authors by number of shared publications:
- Markus Reichstein
Biogeochemical Integration Max Planck Institute for Biogeochemistry Jena Germany
40 shared publications Marion Schrumpf
21 shared publications- Thomas Wutzler
Max Planck Institute for Biogeochemistry, Hans-Knöll-Straße 10, 07745 Jena, Germany
16 shared publications - Lin Yu
Department of Earth System Sciences, University of Hamburg, Hamburg, Germany
12 shared publications Martin Jung
Department of Biogeochemical Integration, Max Planck Institute for Biogeochemistry, Jena, Germany
11 shared publications- Sönke Zaehle
Biogeochemical Signals Department, Max Planck Institute for Biogeochemistry, Jena, Germany
10 shared publications
Latest Publications
- Open Access
- 27 May 2026
- Published by Copernicus GmbH in Geoscientific Model Development
- Vol. 19 (10) , 4467-4496
H2CM (v1.0): hybrid modeling of global water–carbon cycles constrained by atmospheric and land observations
We present the Hybrid Hydrological Carbon Cycle Model (H2CM) – a global model that couples the terrestrial water and carbon cycles by integrating a process-informed deep learning approach with observational constraints for the water and carbon cycles. H2CM extends the hybrid hydrological model with vegetation (H2MV) to represent key terrestrial carbon fluxes, including gross primary productivity (GPP), autotrophic and heterotrophic respiration at daily resolution and 1° spatial scale. H2CM uses neural networks to learn and predict ecosystem properties governing water and carbon fluxes, such as carbon and water use efficiencies and basal respiration rate. H2CM provides a “reanalysis” of recent land water-carbon cycle variations by combining multiple observational constraints synergistically: on top of hydrological and vegetation data constraints on terrestrial water storage variations, snow water equivalent, evapotranspiration, runoff and fraction of photosynthetically active radiation, the carbon cycle is informed by an observation-based GPP product, and net ecosystem exchange (NEE) from satellite and in situ based atmospheric CO2 inversion datasets. H2CM reproduces the seasonal and interannual dynamics of carbon fluxes well. H2CM outperforms both purely data-driven models as well as state-of-the-art process-based model ensembles in capturing NEE seasonality, especially in challenging regions such as the South American tropics and Southern Africa. Moreover, H2CM reveals emergent spatial patterns in precipitation use efficiency, light use efficiency, and water-carbon coupling, consistent with empirical ecological understanding. Notably, we show that H2CM learns to represent the rain pulse effect on respiration in dry regions, which is often not well reproduced by global models. H2CM represents a key step toward a new generation of data-driven diagnostic land surface models, with planned extensions to include the energy cycle.
- 27 May 2026
- Published by Copernicus GmbH in Geoscientific Model Development
- Vol. 19 (10) , 4467-4496
Open AccessH2CM (v1.0): hybrid modeling of global water–carbon cycles constrained by atmospheric and land observations
We present the Hybrid Hydrological Carbon Cycle Model (H2CM) – a global model that couples the terrestrial water and carbon cycles by integrating a process-informed deep learning approach with observational constraints for the water and carbon cycles. H2CM extends the hybrid hydrological model with vegetation (H2MV) to represent key terrestrial carbon fluxes, including gross primary productivity (GPP), autotrophic and heterotrophic respiration at daily resolution and 1° spatial scale. H2CM uses neural networks to learn and predict ecosystem properties governing water and carbon fluxes, such as carbon and water use efficiencies and basal respiration rate. H2CM provides a “reanalysis” of recent land water-carbon cycle variations by combining multiple observational constraints synergistically: on top of hydrological and vegetation data constraints on terrestrial water storage variations, snow water equivalent, evapotranspiration, runoff and fraction of photosynthetically active radiation, the carbon cycle is informed by an observation-based GPP product, and net ecosystem exchange (NEE) from satellite and in situ based atmospheric CO2 inversion datasets. H2CM reproduces the seasonal and interannual dynamics of carbon fluxes well. H2CM outperforms both purely data-driven models as well as state-of-the-art process-based model ensembles in capturing NEE seasonality, especially in challenging regions such as the South American tropics and Southern Africa. Moreover, H2CM reveals emergent spatial patterns in precipitation use efficiency, light use efficiency, and water-carbon coupling, consistent with empirical ecological understanding. Notably, we show that H2CM learns to represent the rain pulse effect on respiration in dry regions, which is often not well reproduced by global models. H2CM represents a key step toward a new generation of data-driven diagnostic land surface models, with planned extensions to include the energy cycle.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Estimating carbon dynamics using H2CM: a hybrid global carbon-water cycle model
Process-based models (PBMs) and machine learning (ML) offer complementary strengths for representing the coupled carbon-water cycle. PBMs enforce physical principles and provide interpretable diagnostics but rely on incomplete process knowledge, many priors, and very limited use of expanding Earth observations, leading to substantial inter-model spread. ML leverages observations to uncover complex patterns and reduce reliance on assumptions, but can violate physical constraints and extrapolate poorly. Hybrid modeling combines both, uniting ML’s flexibility with PBMs’ interpretability and process consistency.We present H2CM, a hybrid carbon-water cycle model that merges process‑informed deep learning with direct learning from observations (Baghirov et al., 2025; https://doi.org/10.5194/egusphere-2025-3123). H2CM simulates carbon fluxes—gross primary productivity (GPP), autotrophic respiration, and heterotrophic respiration—and water storages (soil moisture, groundwater, snow) and fluxes (evapotranspiration, runoff). The model is informed by carbon observations—GPP, net ecosystem exchange (NEE) from satellite- and in situ–based inversions, and fAPAR—and by water-cycle observations—evapotranspiration, runoff, terrestrial water storage, and snow. H2CM runs daily at 1° spatial resolution.H2CM outperforms both purely data-driven approaches and state-of-the-art PBMs in reproducing seasonal NEE, particularly in wet and dry tropics, and it captures the rain‑pulse respiration response in drylands that many models miss. Its estimates of global NEE interannual variability align more closely with satellite- and in situ–based inversion products than do PBM estimates. Finally, we disentangle photosynthetic versus respiratory controls and quantify how different regions (e.g., wet vs. dry tropics) contribute to global variability in land–atmosphere carbon exchange.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessEstimating carbon dynamics using H2CM: a hybrid global carbon-water cycle model
Process-based models (PBMs) and machine learning (ML) offer complementary strengths for representing the coupled carbon-water cycle. PBMs enforce physical principles and provide interpretable diagnostics but rely on incomplete process knowledge, many priors, and very limited use of expanding Earth observations, leading to substantial inter-model spread. ML leverages observations to uncover complex patterns and reduce reliance on assumptions, but can violate physical constraints and extrapolate poorly. Hybrid modeling combines both, uniting ML’s flexibility with PBMs’ interpretability and process consistency.We present H2CM, a hybrid carbon-water cycle model that merges process‑informed deep learning with direct learning from observations (Baghirov et al., 2025; https://doi.org/10.5194/egusphere-2025-3123). H2CM simulates carbon fluxes—gross primary productivity (GPP), autotrophic respiration, and heterotrophic respiration—and water storages (soil moisture, groundwater, snow) and fluxes (evapotranspiration, runoff). The model is informed by carbon observations—GPP, net ecosystem exchange (NEE) from satellite- and in situ–based inversions, and fAPAR—and by water-cycle observations—evapotranspiration, runoff, terrestrial water storage, and snow. H2CM runs daily at 1° spatial resolution.H2CM outperforms both purely data-driven approaches and state-of-the-art PBMs in reproducing seasonal NEE, particularly in wet and dry tropics, and it captures the rain‑pulse respiration response in drylands that many models miss. Its estimates of global NEE interannual variability align more closely with satellite- and in situ–based inversion products than do PBM estimates. Finally, we disentangle photosynthetic versus respiratory controls and quantify how different regions (e.g., wet vs. dry tropics) contribute to global variability in land–atmosphere carbon exchange.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Spatial hybrid modeling of soil organic carbon processes: testing common assumptions using multivariate, dynamic data with simple models
The land surface accounts for a large share of variability in the global carbon cycle. Although increasing atmospheric CO₂ concentrations have led to higher net primary production and increased land carbon stocks, vegetation carbon stocks appear largely constant, implying that changes in land carbon are primarily driven by soil organic carbon (SOC). As SOC represents the largest active carbon pool, its dynamics are critical for land–atmosphere feedbacks. However, strong spatial heterogeneity and measurement limitations result in sparse and mostly static SOC data, complicating the identification of dominant processes.Recent studies address this limitation by assimilating soil carbon models to spatial SOC and covariate datasets using neural networks (hybrid modeling). The resulting spatial parameter fields are then interpreted in terms of underlying mechanisms. These approaches typically rely on three key assumptions: steady-state conditions, adequate process representation by the assimilated SOC model, and the sufficiency of bulk SOC data to infer processes. In this study, we explicitly test these assumptions.We use the Europe-wide LUCAS dataset, which provides spatially resolved physical and chemical soil data at multiple time points. A subset of the dataset includes SOC subfractions, including mineral-associated organic carbon and microbial biomass carbon. Several simple SOC models were assimilated in their steady-state form in the hybrid framework, while accounting for differences in model flexibility. This allowed exclusion of specific modeling assumptions. Comparisons across time steps were used to assess the validity of the steady-state assumption. In addition, first results obtained with a dynamic SOC model are presented.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessSpatial hybrid modeling of soil organic carbon processes: testing common assumptions using multivariate, dynamic data with simple models
The land surface accounts for a large share of variability in the global carbon cycle. Although increasing atmospheric CO₂ concentrations have led to higher net primary production and increased land carbon stocks, vegetation carbon stocks appear largely constant, implying that changes in land carbon are primarily driven by soil organic carbon (SOC). As SOC represents the largest active carbon pool, its dynamics are critical for land–atmosphere feedbacks. However, strong spatial heterogeneity and measurement limitations result in sparse and mostly static SOC data, complicating the identification of dominant processes.Recent studies address this limitation by assimilating soil carbon models to spatial SOC and covariate datasets using neural networks (hybrid modeling). The resulting spatial parameter fields are then interpreted in terms of underlying mechanisms. These approaches typically rely on three key assumptions: steady-state conditions, adequate process representation by the assimilated SOC model, and the sufficiency of bulk SOC data to infer processes. In this study, we explicitly test these assumptions.We use the Europe-wide LUCAS dataset, which provides spatially resolved physical and chemical soil data at multiple time points. A subset of the dataset includes SOC subfractions, including mineral-associated organic carbon and microbial biomass carbon. Several simple SOC models were assimilated in their steady-state form in the hybrid framework, while accounting for differences in model flexibility. This allowed exclusion of specific modeling assumptions. Comparisons across time steps were used to assess the validity of the steady-state assumption. In addition, first results obtained with a dynamic SOC model are presented.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Crowd-sensed plants as living sensors of urban climate and soils: High-resolution insights from 326 European cities
Human populations are increasingly concentrated in cities, creating some of Earth’s most modified ecosystems. Yet, spatially explicit, observation-based assessments of urban climates and especially soils remain scarce. This limits evidence-based planning for climate adaptation and urban resilience. Here, we leverage over 80 million crowd-sensed plant observations from 326 European cities as “living sensors” to map high-resolution patterns of urban climate and soil properties. This approach builds on consolidated knowledge of plant ecological preferences, integrated through three new pan-European systems of ecological indicator values.Beyond the urban heat island, we identify additional consistent contrasts between built-up and green areas in moisture, light, soil pH, disturbance, and salinity. The magnitude of these within-city environmental gradients rivals those observed between cities thousands of kilometers apart across Europe. Environmental conditions in built-up areas are remarkably similar across cities, highlighting urban environmental homogenization. In contrast, urban forests maintain natural environmental diversity, contributing to cooling, moisture retention, and key ecosystem functions.Our new sensing approach, called mobile crowd sensing of environments (MCSE), supports participatory assessment of nature-based solutions and provides actionable insights for planners, policymakers, and local communities. It enables evidence-based decision-making for climate adaptation, sustainable urban development, and the promotion of human health and well-being under rapid urbanization and climate change.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessCrowd-sensed plants as living sensors of urban climate and soils: High-resolution insights from 326 European cities
Human populations are increasingly concentrated in cities, creating some of Earth’s most modified ecosystems. Yet, spatially explicit, observation-based assessments of urban climates and especially soils remain scarce. This limits evidence-based planning for climate adaptation and urban resilience. Here, we leverage over 80 million crowd-sensed plant observations from 326 European cities as “living sensors” to map high-resolution patterns of urban climate and soil properties. This approach builds on consolidated knowledge of plant ecological preferences, integrated through three new pan-European systems of ecological indicator values.Beyond the urban heat island, we identify additional consistent contrasts between built-up and green areas in moisture, light, soil pH, disturbance, and salinity. The magnitude of these within-city environmental gradients rivals those observed between cities thousands of kilometers apart across Europe. Environmental conditions in built-up areas are remarkably similar across cities, highlighting urban environmental homogenization. In contrast, urban forests maintain natural environmental diversity, contributing to cooling, moisture retention, and key ecosystem functions.Our new sensing approach, called mobile crowd sensing of environments (MCSE), supports participatory assessment of nature-based solutions and provides actionable insights for planners, policymakers, and local communities. It enables evidence-based decision-making for climate adaptation, sustainable urban development, and the promotion of human health and well-being under rapid urbanization and climate change.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Including microbial communities in soil carbon-nitrogen cycling modeling via a hybrid neural-mechanistic modeling approach
A recurring challenge in ecosystem science is modeling the variance of biogeochemical process rates in connection with local microbial community composition. Mechanistic models usually relies on fixed parameters that ignore such ecological variations. Purely statistical approaches require extensive data and, lacking process-based information, often overfit to training conditions, limiting their ability to generalize. We present here a hybrid modeling framework that combines these approaches, allowing mechanistic biogeochemical models to adapt their parameters based on local microbial community structure.Our approach uses neural networks to translate microbial community composition (bacterial and fungal taxonomic data) into site-specific key parameters in a mechanistic carbon-nitrogen cycling equations. Since these intermediate parameters likely capture multiple processes, we view them as functional parameters that allow the mechanistic model to flexibly incorporate the variance of decomposition rates due to local microbial communities, while still maintaining the interpretable structure of process-based equations and retaining the deterministic information for the processes we know how to model.The innovation lies in including community composition from sequencing directly as a driver of parameter variation within established biogeochemical theory, preserving information that would otherwise be lost (for example assembling the sequencing data into diversity indicators). Literature-derived constraints ensure parameters remain within physically plausible ranges, but the neural components learns how microbial community structure modulates these values locally to improve predictions.This methodological framework demonstrates that we can link communities with decomposition processes without requiring a complete mechanistic understanding (with consequent biases due to likely missing processes) of every intermediate step. This approach is broadly applicable, solving the difficulties coming from knowing that functional diversity influences biogeochemical processes but with an incomplete understanding of all the underlying mechanistic complexity, embedding the paradigm of soil decomposition kinetics as emergent ecological properties rather than as fixed intrinsic characteristics.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessIncluding microbial communities in soil carbon-nitrogen cycling modeling via a hybrid neural-mechanistic modeling approach
A recurring challenge in ecosystem science is modeling the variance of biogeochemical process rates in connection with local microbial community composition. Mechanistic models usually relies on fixed parameters that ignore such ecological variations. Purely statistical approaches require extensive data and, lacking process-based information, often overfit to training conditions, limiting their ability to generalize. We present here a hybrid modeling framework that combines these approaches, allowing mechanistic biogeochemical models to adapt their parameters based on local microbial community structure.Our approach uses neural networks to translate microbial community composition (bacterial and fungal taxonomic data) into site-specific key parameters in a mechanistic carbon-nitrogen cycling equations. Since these intermediate parameters likely capture multiple processes, we view them as functional parameters that allow the mechanistic model to flexibly incorporate the variance of decomposition rates due to local microbial communities, while still maintaining the interpretable structure of process-based equations and retaining the deterministic information for the processes we know how to model.The innovation lies in including community composition from sequencing directly as a driver of parameter variation within established biogeochemical theory, preserving information that would otherwise be lost (for example assembling the sequencing data into diversity indicators). Literature-derived constraints ensure parameters remain within physically plausible ranges, but the neural components learns how microbial community structure modulates these values locally to improve predictions.This methodological framework demonstrates that we can link communities with decomposition processes without requiring a complete mechanistic understanding (with consequent biases due to likely missing processes) of every intermediate step. This approach is broadly applicable, solving the difficulties coming from knowing that functional diversity influences biogeochemical processes but with an incomplete understanding of all the underlying mechanistic complexity, embedding the paradigm of soil decomposition kinetics as emergent ecological properties rather than as fixed intrinsic characteristics.
- Open Access
- 23 February 2026
- Published by Copernicus GmbH
- Vol. 2026, 1-22
Comment on egusphere-2026-229
Soil organic carbon (SOC) density is a key variable for quantifying soil carbon stocks, yet its modelling is challenged by sparse and inconsistent measurements of bulk density and coarse fragments relative to SOC content. Conventional digital soil mapping approaches typically model SOC density as a single target variable, thereby underutilising abundant SOC content data and overlooking physical relationships among soil properties. This study evaluates a soil science-informed neural network for SOC density prediction that explicitly constrains the SOC–BD relationship, and compares it with univariate and multivariate neural network architectures. Across sparsely sampled target variables, including SOC density, bulk density, and coarse fragments, the soil science-informed model achieves comparable or slightly improved prediction accuracy relative to multivariate and univariate models. Although it yields lower accuracy for SOC content, the soil science-informed model better preserves physically plausible SOC–BD joint distributions and generates smoother, more temporally stable SOC density trajectories. Overall, the results demonstrate that incorporating soil physical constraints into machine learning models adds value beyond univariate accuracy, improving robustness, plausibility, and temporal coherence of SOC density predictions under sparse data conditions. Moreover, the latent parameters inferred by the soil science-informed model improve model interpretability and offer additional soil science relevant insights beyond predictive accuracy.
- 23 February 2026
- Published by Copernicus GmbH
- Vol. 2026, 1-22
Open AccessComment on egusphere-2026-229
Soil organic carbon (SOC) density is a key variable for quantifying soil carbon stocks, yet its modelling is challenged by sparse and inconsistent measurements of bulk density and coarse fragments relative to SOC content. Conventional digital soil mapping approaches typically model SOC density as a single target variable, thereby underutilising abundant SOC content data and overlooking physical relationships among soil properties. This study evaluates a soil science-informed neural network for SOC density prediction that explicitly constrains the SOC–BD relationship, and compares it with univariate and multivariate neural network architectures. Across sparsely sampled target variables, including SOC density, bulk density, and coarse fragments, the soil science-informed model achieves comparable or slightly improved prediction accuracy relative to multivariate and univariate models. Although it yields lower accuracy for SOC content, the soil science-informed model better preserves physically plausible SOC–BD joint distributions and generates smoother, more temporally stable SOC density trajectories. Overall, the results demonstrate that incorporating soil physical constraints into machine learning models adds value beyond univariate accuracy, improving robustness, plausibility, and temporal coherence of SOC density predictions under sparse data conditions. Moreover, the latent parameters inferred by the soil science-informed model improve model interpretability and offer additional soil science relevant insights beyond predictive accuracy.
- Open Access
- 28 January 2026
- Published by Copernicus GmbH in EGUsphere
- Vol. 2026, 1-22
Soil science-informed neural networks for soil organic carbon density modelling under scarce bulk density data
Soil organic carbon (SOC) density is a key variable for quantifying soil carbon stocks, yet its modelling is challenged by sparse and inconsistent measurements of bulk density and coarse fragments relative to SOC content. Conventional digital soil mapping approaches typically model SOC density as a single target variable, thereby underutilising abundant SOC content data and overlooking physical relationships among soil properties. This study evaluates a soil science-informed neural network for SOC density prediction that explicitly constrains the SOC–BD relationship, and compares it with univariate and multivariate neural network architectures. Across sparsely sampled target variables, including SOC density, bulk density, and coarse fragments, the soil science-informed model achieves comparable or slightly improved prediction accuracy relative to multivariate and univariate models. Although it yields lower accuracy for SOC content, the soil science-informed model better preserves physically plausible SOC–BD joint distributions and generates smoother, more temporally stable SOC density trajectories. Overall, the results demonstrate that incorporating soil physical constraints into machine learning models adds value beyond univariate accuracy, improving robustness, plausibility, and temporal coherence of SOC density predictions under sparse data conditions. Moreover, the latent parameters inferred by the soil science-informed model improve model interpretability and offer additional soil science relevant insights beyond predictive accuracy.
- 28 January 2026
- Published by Copernicus GmbH in EGUsphere
- Vol. 2026, 1-22
Open AccessSoil science-informed neural networks for soil organic carbon density modelling under scarce bulk density data
Soil organic carbon (SOC) density is a key variable for quantifying soil carbon stocks, yet its modelling is challenged by sparse and inconsistent measurements of bulk density and coarse fragments relative to SOC content. Conventional digital soil mapping approaches typically model SOC density as a single target variable, thereby underutilising abundant SOC content data and overlooking physical relationships among soil properties. This study evaluates a soil science-informed neural network for SOC density prediction that explicitly constrains the SOC–BD relationship, and compares it with univariate and multivariate neural network architectures. Across sparsely sampled target variables, including SOC density, bulk density, and coarse fragments, the soil science-informed model achieves comparable or slightly improved prediction accuracy relative to multivariate and univariate models. Although it yields lower accuracy for SOC content, the soil science-informed model better preserves physically plausible SOC–BD joint distributions and generates smoother, more temporally stable SOC density trajectories. Overall, the results demonstrate that incorporating soil physical constraints into machine learning models adds value beyond univariate accuracy, improving robustness, plausibility, and temporal coherence of SOC density predictions under sparse data conditions. Moreover, the latent parameters inferred by the soil science-informed model improve model interpretability and offer additional soil science relevant insights beyond predictive accuracy.
- Top Cited PapersOpen Access
- 15 January 2026
- Published by Springer Nature in Nature Cities
- Vol. 3 (2) , 126-135
Urbanization signatures on climate and soils uncovered by crowd-sensed plants
Human populations are increasingly concentrated in cities, creating some of Earth’s most modified ecosystems. However, we lack concepts for assessing heterogenous urban environments, especially their soils, at large spatial scales. Here we uncover fine-scale urban climate and soil patterns across 326 European cities by using more than 80 million crowd-sensed plant observations as living sensors of the environment. In addition to the urban heat island, we identify similar contrasts between the built-up and green areas for moisture, soil pH, salinity and soil disturbance. These within-city environmental contrasts correspond to differences between cities that are about 1,500–3,000 km apart. Climate, especially soil, conditions are more similar between cities for built-up areas than for forests, indicating urban homogenization tendencies. Urban forests serve as a source of environmental diversity, cooling and moisture retention. The crowd sensing of urban environments is facilitated by their citizens, which can support science, policy and help guide urban planning toward livable cities.
- 15 January 2026
- Published by Springer Nature in Nature Cities
- Vol. 3 (2) , 126-135
Top Cited PapersOpen AccessUrbanization signatures on climate and soils uncovered by crowd-sensed plants
Human populations are increasingly concentrated in cities, creating some of Earth’s most modified ecosystems. However, we lack concepts for assessing heterogenous urban environments, especially their soils, at large spatial scales. Here we uncover fine-scale urban climate and soil patterns across 326 European cities by using more than 80 million crowd-sensed plant observations as living sensors of the environment. In addition to the urban heat island, we identify similar contrasts between the built-up and green areas for moisture, soil pH, salinity and soil disturbance. These within-city environmental contrasts correspond to differences between cities that are about 1,500–3,000 km apart. Climate, especially soil, conditions are more similar between cities for built-up areas than for forests, indicating urban homogenization tendencies. Urban forests serve as a source of environmental diversity, cooling and moisture retention. The crowd sensing of urban environments is facilitated by their citizens, which can support science, policy and help guide urban planning toward livable cities.
- Open Access
- 11 August 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-33
Comment on egusphere-2025-3123
We present the Hybrid Hydrological Carbon Cycle Model (H2CM)—a global model that couples the terrestrial water and carbon cycles by integrating a process-informed deep learning approach with observational constraints for the water and carbon cycles. H2CM extends the hybrid hydrological model with vegetation (H2MV) to represent key terrestrial carbon fluxes, including gross primary productivity (GPP), autotrophic and heterotrophic respiration at daily resolution and 1-degree spatial scale. H2CM uses neural networks to learn and predict ecosystem properties governing water and carbon fluxes, such as carbon and water use efficiencies and basal respiration rate. H2CM uniquely combines multiple observational constraints synergistically: on top of hydrological and vegetation data constraints on terrestrial water storage variations, snow water equivalent, evapotranspiration, runoff and fraction of photosynthetically active radiation, the carbon cycle is informed by an observation-based GPP product, and net ecosystem exchange (NEE) from satellite and in-situ based atmospheric CO2 inversion datasets. H2CM reproduces the seasonal and interannual dynamics of carbon fluxes well. H2CM outperforms both purely data-driven models as well as state-of-the-art process-based model ensembles in capturing NEE seasonality, especially in challenging regions such as the South American tropics and Southern Africa. Moreover, H2CM reveals emergent spatial patterns in precipitation use efficiency, light use efficiency, and water-carbon coupling, consistent with empirical ecological understanding. Notably, we show that H2CM learns to represent the rain pulse effect on respiration in dry regions, which is often not well reproduced by global models. H2CM represents a key step toward a new generation of hybrid land surface models, with planned extensions to include the energy cycle.
- 11 August 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-33
Open AccessComment on egusphere-2025-3123
We present the Hybrid Hydrological Carbon Cycle Model (H2CM)—a global model that couples the terrestrial water and carbon cycles by integrating a process-informed deep learning approach with observational constraints for the water and carbon cycles. H2CM extends the hybrid hydrological model with vegetation (H2MV) to represent key terrestrial carbon fluxes, including gross primary productivity (GPP), autotrophic and heterotrophic respiration at daily resolution and 1-degree spatial scale. H2CM uses neural networks to learn and predict ecosystem properties governing water and carbon fluxes, such as carbon and water use efficiencies and basal respiration rate. H2CM uniquely combines multiple observational constraints synergistically: on top of hydrological and vegetation data constraints on terrestrial water storage variations, snow water equivalent, evapotranspiration, runoff and fraction of photosynthetically active radiation, the carbon cycle is informed by an observation-based GPP product, and net ecosystem exchange (NEE) from satellite and in-situ based atmospheric CO2 inversion datasets. H2CM reproduces the seasonal and interannual dynamics of carbon fluxes well. H2CM outperforms both purely data-driven models as well as state-of-the-art process-based model ensembles in capturing NEE seasonality, especially in challenging regions such as the South American tropics and Southern Africa. Moreover, H2CM reveals emergent spatial patterns in precipitation use efficiency, light use efficiency, and water-carbon coupling, consistent with empirical ecological understanding. Notably, we show that H2CM learns to represent the rain pulse effect on respiration in dry regions, which is often not well reproduced by global models. H2CM represents a key step toward a new generation of hybrid land surface models, with planned extensions to include the energy cycle.
- Open Access
- 8 August 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-33
Comment on egusphere-2025-3123
We present the Hybrid Hydrological Carbon Cycle Model (H2CM)—a global model that couples the terrestrial water and carbon cycles by integrating a process-informed deep learning approach with observational constraints for the water and carbon cycles. H2CM extends the hybrid hydrological model with vegetation (H2MV) to represent key terrestrial carbon fluxes, including gross primary productivity (GPP), autotrophic and heterotrophic respiration at daily resolution and 1-degree spatial scale. H2CM uses neural networks to learn and predict ecosystem properties governing water and carbon fluxes, such as carbon and water use efficiencies and basal respiration rate. H2CM uniquely combines multiple observational constraints synergistically: on top of hydrological and vegetation data constraints on terrestrial water storage variations, snow water equivalent, evapotranspiration, runoff and fraction of photosynthetically active radiation, the carbon cycle is informed by an observation-based GPP product, and net ecosystem exchange (NEE) from satellite and in-situ based atmospheric CO2 inversion datasets. H2CM reproduces the seasonal and interannual dynamics of carbon fluxes well. H2CM outperforms both purely data-driven models as well as state-of-the-art process-based model ensembles in capturing NEE seasonality, especially in challenging regions such as the South American tropics and Southern Africa. Moreover, H2CM reveals emergent spatial patterns in precipitation use efficiency, light use efficiency, and water-carbon coupling, consistent with empirical ecological understanding. Notably, we show that H2CM learns to represent the rain pulse effect on respiration in dry regions, which is often not well reproduced by global models. H2CM represents a key step toward a new generation of hybrid land surface models, with planned extensions to include the energy cycle.
- 8 August 2025
- Published by Copernicus GmbH
- Vol. 2025, 1-33
Open AccessComment on egusphere-2025-3123
We present the Hybrid Hydrological Carbon Cycle Model (H2CM)—a global model that couples the terrestrial water and carbon cycles by integrating a process-informed deep learning approach with observational constraints for the water and carbon cycles. H2CM extends the hybrid hydrological model with vegetation (H2MV) to represent key terrestrial carbon fluxes, including gross primary productivity (GPP), autotrophic and heterotrophic respiration at daily resolution and 1-degree spatial scale. H2CM uses neural networks to learn and predict ecosystem properties governing water and carbon fluxes, such as carbon and water use efficiencies and basal respiration rate. H2CM uniquely combines multiple observational constraints synergistically: on top of hydrological and vegetation data constraints on terrestrial water storage variations, snow water equivalent, evapotranspiration, runoff and fraction of photosynthetically active radiation, the carbon cycle is informed by an observation-based GPP product, and net ecosystem exchange (NEE) from satellite and in-situ based atmospheric CO2 inversion datasets. H2CM reproduces the seasonal and interannual dynamics of carbon fluxes well. H2CM outperforms both purely data-driven models as well as state-of-the-art process-based model ensembles in capturing NEE seasonality, especially in challenging regions such as the South American tropics and Southern Africa. Moreover, H2CM reveals emergent spatial patterns in precipitation use efficiency, light use efficiency, and water-carbon coupling, consistent with empirical ecological understanding. Notably, we show that H2CM learns to represent the rain pulse effect on respiration in dry regions, which is often not well reproduced by global models. H2CM represents a key step toward a new generation of hybrid land surface models, with planned extensions to include the energy cycle.