We invite abstracts that address one of the following themes:
(1) Advances in representing responses of land ecosystem processes to climate variability, extremes or disturbances;
(2) Advances in accounting for species interactions, trait acclimation and adaptation, and vegetation demography and dynamics;
(3) Methods for improving the representation of biogeochemical processes and interactions in different ecosystems;
(4) site, regional and global studies taking advantage of in-situ measurements, Earth observing systems or laboratory experiments, improving ecosystem model processes.
Session Format
The session will include oral presentations and poster sessions to facilitate knowledge exchange and collaboration among participants.
We, as conveners, are a diverse group of scientists from five different universities who work with various ecosystem models. We hope to use this session to discuss model improvements and share knowledge between different models.
Orals: Tue, 5 May, 08:30–12:30 | Room 2.95
Terrestrial ecosystem models are key tools for understanding and predicting climate and CO2 impacts on vegetation. Despite decades of research, there remains considerable uncertainty as to how to simulate canopy photosynthesis and transpiration, respiration, growth, and turnover. This uncertainty includes the approach to use, its complexity, and its parameterisation, and covers all processes from light interception to soil water dynamics to demography. While simulated carbon fluxes are often compared with eddy-covariance measurements, simulated growth is more difficult to evaluate. Tree rings can provide a useful measure of interannual growth rates, and are therefore an important potential constraint on model representations. Here, we use a new central European oak ring width chronology, 1901-2015 CE, to evaluate a range of model assumptions regarding key processes. Capturing more than a limited amount of the interannual variability in the chronology is difficult (i.e. achieving Pearson’s r > 0.5, including the long-term trend). It is clear that soil hydrological dynamics are a key driver and need to be represented well, as do leaf phenology, respiration, and responses to temperature and CO2. Spatial variability in climate and other factors such as soil type and depth also plays an important role. Less clear is how to explain some years with very strong or very weak growth. Explanations are sort in extreme temperatures (e.g. frost damage), extreme drought stress (e.g. causing xylem cavitation), recovery from storm damage, carry-over between years, and separation of controls on growth and photosynthesis. Recommendations are made regarding requirements for process representations and for future work to better understand drivers of tree-ring variability in temperate forest ecosystems.
How to cite: Friend, A. and Büntgen, U.: Modelling central European oak growth, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17909, https://doi.org/10.5194/egusphere-egu26-17909, 2026.
Wood density is a key functional trait influencing tree growth, forest dynamics and carbon storage, yet dynamic global vegetation models (DGVMs) typically represent it as a fixed parameter for each plant functional type. This assumption neglects well-documented environmental and interannual variability in wood density and its potential feedbacks of forest dynamics.
In this study, we explore the consequences of environmentally dependent wood density for tree- and forest-level carbon storage by integrating a temperature-response function of wood density into the DGVM LPJ-GUESS. Using a well-documented relationship between temperature and latewood density extracted from tree ring data from 52 sites, we simulate forest recovery following stand-replacing disturbance and compare model behaviour with and without dynamic wood density.
We show that allowing wood density to vary with temperature alters tree growth and carbon content, with cascading effects on within-cohort competition, forest structure, and stand-level carbon storage. Warmer conditions produce higher wood density, leading to slower diameter and height growth and thus smaller trees relative to simulations with fixed wood density, while lower wood density promotes taller trees that overall capture more carbon. These effects depend strongly on tree life stage and forest recovery phase: before canopy closure, climate-driven variability in wood density induces large divergence in individual tree carbon content (up to 32%), whereas after canopy closure, competitive interactions dominate and climate effects stabilise. Overall, dynamic wood density alters the size distribution of the forest compared to constant wood density simulations. The implications are that by shifting carbon storage, from relatively more small trees to fewer large trees (or vice versa), other processes are influenced such as tree mortality and the resulting flux of carbon through deadwood to soil pools.
Our study demonstrates that environmentally driven variation in wood density constitutes an important, yet overlooked, mechanism shaping forest structure and carbon dynamics. Incorporating dynamic wood density into DGVMs may therefore be a useful avenue to explore for improving ecological feedbacks and realism of forest carbon storage predictions, particularly in young and regenerating forests under a changing climate.
How to cite: Eckes-Shephard, A. H., Voss, A. C., Zhou, H., Fonti, P., and Olin, S.: Environmentally dependent wood density reshapes forest structure and carbon storage in a demographic vegetation model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7417, https://doi.org/10.5194/egusphere-egu26-7417, 2026.
Over the past 20 years the understanding of carbon cycling within soils has radically advanced. However, the representation of soil carbon within Earth System Models has remained based on outdated models which do not reflect our current understanding of soil carbon processes. The Millennial version 2 soil carbon module was developed to allow Earth System Models to represent soil carbon processes such as microbial decomposition, aggregation, and mineral sorption. Millennial consists of five interacting soil carbon pools: particulate organic carbon, mineral-associated organic carbon, aggregate carbon, microbial biomass, and low molecular weight carbon, all of which can be measured in natural soils.
Here we have incorporated Millennial into the University of Victoria Earth System Climate model, an Earth system model of intermediate complexity, with a simplified atmosphere and full complexity land surface and ocean modules. Preliminary results suggest that Millennial generates soil carbon pools consistent with observations across regional climates and ecosystems. A series of model experiments have been devised to explore how soil carbon as simulated by Millennial behaves relative to traditional methods for simulating soil carbon. Our experiments will show how much this higher fidelity soil carbon model structure will change projections of future climate change and remaining carbon budgets.
How to cite: MacDougall, A. H., Nzotungicimpaye, C.-M., MacIsaac, A. J., and Abramoff, R. Z.: Incorporating a modern understanding of soil carbon dynamics into Earth system model simulations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10779, https://doi.org/10.5194/egusphere-egu26-10779, 2026.
Global tropical rainforests represent an important carbon sink and have significant potential for mitigating the effects of human-induced climate change. However, current estimates of carbon stocks and fluxes in tropical forests are highly uncertain due to spatial variation in structure, composition, and dynamics of tropical forests. Canopy structure metrics, including vertical leaf area index (LAI) profiles, canopy height and biomass have been identified as essential variables for predicting tropical forest canopy biomass dynamics and climate feedbacks in heterogeneous landscapes. In this study, we assimilate lidar-derived measurements of forest canopy height and vertical LAI profile from NASA’s Global Ecosystem Dynamics Investigation (GEDI), space-borne lidar into ED2, a cohort-based terrestrial biosphere model. We then use the GEDI-constrained model to analyze carbon dynamics across Southeast Asia under different scenarios and climate forcings. In addition, we carried out model simulations without GEDI observations to evaluate how different initialization methods would impact predictions of carbon fluxes and states. Our results show that GEDI constrained model has similar predictions with ground-data initialized simulations. In addition, regional analysis of long-term regional simulations suggests that initialization with GEDI as opposed to with output from a historical simulation has larger effects on regional-scale aboveground biomass predictions than in the effects of future CO2 emissions, land use scenarios, and climate forcings. Our study demonstrates how information on forest structure from GEDI retrievals can improve the accuracy of TBM predictions of carbon dynamics in tropical forests and thereby inform decisions about how tropical forests can be managed to promote forest carbon storage and uptake in context of ongoing changes in earth’s climate.
How to cite: Liu, S., Ouyang, W., Lin, C., Chang, L., Xu, X., Longo, M., Ordway, E., Armston, J., Tang, H., Dubayah, R., and Moorcroft, P.: GEDI-constrained Estimates of Terrestrial Carbon Dynamics in Southeast Asian Tropical Forests over the Coming Century, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8703, https://doi.org/10.5194/egusphere-egu26-8703, 2026.
The heterogenous structure and diversity of forest stands shape resource availabilities to individual trees and lead to diversity in stress responses. Heterogeneity in size and/or traits thus strongly determines tree and ecosystem carbon balances. While short-term net ecosystem carbon and water exchange might be well approximated by the average behavior of the top canopy trees, we expect both longer-term structural shifts as well as stress responses to extreme conditions to be more strongly dependent on the demography of below-canopy trees. Dynamic vegetation models (DVM) resolve and track this heterogeneity. They simulate growth, mortality, competition, and carbon allocation strategies. Thereby they propagate changes in environmental conditions into changes in structure of forest ecosystems.
Here, we use daily ecosystem flux measurements of gross primary production (GPP) and multi-annual forest inventories (distribution of diameter at breast height, DBH) as observational constraints to calibrate BiomeEP (Weng et al., 2015). BiomeEP is a process-based DVM grounded in optimality principles for computational efficiency and parameter sparsity. It includes acclimation of photosynthetic capacity via P-model (Stocker et al., 2020), a perfect plasticity assumption for canopy layering (Strigul et al., 2008) and makes use of empirical allometric relationships.
Our model implementation shows single-core runtimes (including 2000 years of spin-up) on the order of seconds for individual sites and thus enables site-specific inference of physiological traits. As next step the PROFOUND data set (Reyer et al. 2020) containing GPP and DBH data from seven European sites will be used for model data fusion. We aim to reproduce observed GPP and DBH distributions (targeting both absolute numbers and relative size distribution) and estimate parameter identifiability as well as across-sites generalizability. Preliminary results demonstrated the sensitivity of targets on the species-specific rate parameters for growth and mortality. Across-site generalizable parameters could pave the way to inform large scale predictions under future environmental changes.
How to cite: Bernhard, F., Weng, E., and Stocker, B.: Model-data fusion of daily ecosystem fluxes (GPP) and forest inventories (DBH) with the process-based dynamic vegetation model BiomeEP, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21054, https://doi.org/10.5194/egusphere-egu26-21054, 2026.
Drought-deciduous phenology is a key adaptive strategy shaping growth, mortality, and competitive dynamics in tropical forests, yet it remains poorly represented in most vegetation models. We introduce a mechanistic drought-deciduous phenology module in FATES-Hydro that links canopy leaf loss and refoliation directly to plant hydraulic states, enabling trait-based representation of drought responses within a vegetation demographic framework.
Leaf shedding is driven by the fraction of the canopy experiencing critically low leaf water potential, while refoliation is permitted only after sustained hydraulic recovery above a second threshold for multiple consecutive wet days. This formulation captures both rapid drought-induced canopy loss and delayed recovery constrained by hydraulic and carbon availability, and operates consistently with size-structured demography and carbon allocation.
We evaluate the model across six tropical forest sites spanning a strong moisture gradient and focus on a sensitivity analysis to identify trait-mediated controls on phenology and demographic outcomes. At seasonal dry sites, phenological parameters dominate canopy dynamics. The leaf-off water-potential threshold exerts first-order control over both leaf shedding and refoliation timing by regulating dry-season soil moisture depletion. The recovery threshold further delays or accelerates leaf-on timing, while the assumed distribution of leaf water potential within the crown primarily controls whether canopy loss occurs gradually or abruptly.
These phenological controls generate pronounced growth–mortality trade-offs along the moisture gradient. At dry sites, delayed leaf shedding enhances carbon uptake during the wet season but increases drought exposure and mortality, whereas earlier shedding reduces productivity while maintaining hydraulic safety.
Hydraulic trait sensitivities are strongly site dependent. Root distribution is the dominant control at seasonal sites, with deeper roots delaying leaf-off, shortening leaf-off duration, and advancing refoliation. This effect weakens in wetter forests. Rooting depth has little influence on canopy phenology at wet sites but increases gross primary productivity and evapotranspiration. Stomatal sensitivity (P50gs) plays a secondary role, regulating carbon gain and mortality differently across climates: risky stomatal strategies increase wet-season carbon fixation but elevate leaf loss and mortality at dry sites, while conferring higher growth without increased mortality at wet sites. In contrast, maximum xylem conductance has negligible influence on phenology, growth, or mortality. The lack of sensitivity differences between two wet sites with contrasting rainfall seasonality further indicates that light, rather than water, constrains wet tropical forest dynamics. These results highlight how integrating hydraulics, phenology, and demography enables trait-based predictions of tropical forest responses to increasing drought stress.
How to cite: Ding, J., McDowell, N., Pivovaroff, A., Xu, C., Song, G., Wu, J., Mas, E., Fang, Y., and Koven, C.: Trait-based representation of drought-deciduous phenology in a demographic vegetation model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1911, https://doi.org/10.5194/egusphere-egu26-1911, 2026.
Global climate change has led to changes in plant phenology, potentially altering growing season length and the productivity of plants. Simulating phenological changes is fundamental to predicting changes in ecosystem function. However, the existing methods have not adequately represented the joint control of plant development by multiple environmental resources, including temperature, precipitation and photoperiod. In this study, we introduce the concept of an “environmental resource space” (ERS) and present generic algorithms for interpreting and predicting plant green-up and green-down. We found that the ERS-derived indices, including quantity (S) and synergistic efficiency (V) of resources, had a greater importance than other environmental variables in explaining variations in the green-up and green-down periods of natural ecosystems. Ground and satellite observations in the mid- and high latitudes of the Northern Hemisphere supported a significant positive relationship between phenological period length and the S:V ratio. An ERS-based model can predict the green-up and green-down periods of plants with an accuracy of 0.7-0.8 at a hemispheric scale. The ERS framework and algorithms could help predict the combined effects of multiple environmental changes on the phenology and function of natural ecosystems.
How to cite: Cen, X., He, N., Campioli, M., Marchand, L., Liu, D., Treat, C., Yu, K., Huang, Y., He, L., Li, J., Zhang, J., Jiao, C., Wang, S., and Butterbach-Bahl, K.: Environmental resource perspective on plant green-up and green-down, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7465, https://doi.org/10.5194/egusphere-egu26-7465, 2026.
Dwarf-shrubs are a fundamental component of boreal, Arctic, and alpine ecosystems, where they contribute substantially to ecosystem carbon sequestration and long-term storage, potentially influencing feedback mechanisms between terrestrial ecosystems and the global climate system. To date, dwarf-shrubs remain inadequately represented in most land surface models, while their interactions with climate are highly uncertain. As part of the DURIN project, we aim to develop and implement a new dwarf-shrub plant functional type in the Community Land Model (CLM) coupled with the Ecosystem Demography model FATES (Functionally Assembled Terrestrial Ecosystem Simulator). Combining information from field observations and vegetation modelling, we will provide new insights on the roles and contributions of dwarf-shrubs in climate-biosphere feedbacks, and ultimately contribute to an enhanced Earth system model performance in predicting future changes in the boreal and Arctic region.
This objective involves calibrating CLM-FATES using the extensive field observation data collected across four sites in Norway as part of the DURIN project, ranging from physiology to ecosystem fluxes, carbon allocation, below-ground interactions, and soil properties. These measurements capture habitat change (open and forested) as well as latitudinal and inland-coastal gradients, providing crucial information on environmental controls of dwarf-shrubs and ensuring robust model parameterization and calibration. The DURIN data will be used to calibrate key physiological and ecosystem parameters in the model, including those related to photosynthesis, carbon allocation, and plant-soil hydraulics. They will also support further model developments of a dwarf-shrub plant functional type, such as revised allometric relationships, improved biomass allocation schemes, or enhanced parameterization of cold and drought stresses. Here, we present the first results from integrating these field observations into CLM-FATES, outlining the emerging representation of dwarf-shrubs in the model and the next steps in its development.
How to cite: Rezsöhazy, J., Geange, S. R., Tang, H., Fisher, R. A., Birkeli, K., and Vandvik, V.: From field observations to improved land surface models: The case of dwarf-shrubs in Norway, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12153, https://doi.org/10.5194/egusphere-egu26-12153, 2026.
Terrestrial ecosystems currently absorb approximately one-third of anthropogenic CO₂ emissions, constituting a central component of the global carbon cycle. However, the persistence of terrestrial carbon uptake under future warming and extreme heat events remains highly uncertain, in part due to limited understanding of how photosynthetic processes respond to thermal stress. In current land surface models, the temperature sensitivity of photosynthesis is often represented using simplified empirical formulations that inadequately capture physiological failure under extreme conditions.
The widely applied Farquhar–von Caemmerer–Berry (FvCB) framework commonly employs Arrhenius-type temperature response functions with fixed parameters derived from empirical fitting, which perform reasonably well near moderate temperatures but struggle to represent rapid declines in photosynthetic capacity at high temperatures. Moreover, how these limitations differ between C₃ and C₄ plants, despite their contrasting photosynthetic pathways and thermal strategies, remains poorly constrained at the global scale.
Here, we integrate a global meta-analysis drawing on published and newly compiled datasets within a mechanistic framework to assess the thermal responses of photosynthetic processes across C₃ and C₄ species. Our results indicate that even C₄ plants, despite their comparatively high thermal tolerance, exhibit pronounced enzymatic constraints under extreme heat. We further identify a coherent pattern in which biochemical and photochemical processes respond over a similar temperature range; however, biochemical limitations consistently arise at lower temperatures than photochemical limitations, suggesting that heat stress leads to metabolic failure prior to photochemical impairment.
These findings suggest that current temperature response formulations in land surface models may systematically overestimate photosynthetic stability under extreme heat, underscoring the need for improved mechanistic representation of thermal sensitivity to better project terrestrial carbon uptake under future climate extremes.
How to cite: Zhao, J. and Prentice, I. C.: High-temperature responses of photosynthetic parameters, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5613, https://doi.org/10.5194/egusphere-egu26-5613, 2026.
Leaf photosynthesis and respiration respond to leaf temperature, which differs from air temperature depending on radiation load, transpiration and heat exchange rates. However, most terrestrial biosphere models (TBMs) do not use leaf temperature to compute photosynthesis and respiration. Instead, they use directly air temperature, or an average surface temperature that incorporate soil and non-green biomass compartments. While these two approaches are computationally efficient, the absence of explicit leaf temperature simulation potentially hinders the representation of extreme events (e.g. heat stress) and their repercussion on carbon and water fluxes. To predict leaf temperature, it is necessary to explicitly account for leaf energy budget, photosynthesis and transpiration feedbacks (E-P-T).
Here, we explored the merits of coupling E-P-T processes in TBMs to simulate explicit leaf temperature and its feedback on carbon assimilation. Sensitivity analysis of the coupled E-P-T processes using the big-leaf configuration of the ORCHIDEE v2.2 TBM (used in CMIP6) resulted into leaf-to-air temperature differences varying between 1 and 10 °C in natural conditions. This translated into a change in carbon assimilation ranging from -35 % to +110 % at the leaf level. A comparison of simulated leaf temperature with measured canopy surface temperature at eddy-covariance fluxes sites in various E-P-T configurations showed that adding ecological constraints on photosynthesis and transpiration through the photosynthesis coordination and the least-cost hypothesis (P-model) improved the representation of top canopy temperature compared to a classical fixed parameterization. The improvement in canopy temperature estimates was best for deciduous broadleaved forests with an average reduction of the error by 1.2 ± 0.8 °C (9 sites). More importantly, the improvement in leaf temperature estimates mainly occurs at elevated temperature (> 30°C).
Our results argue for the inclusion of an explicit representation of leaf temperature in TBMs to avoid biases in the carbon balance estimates. Fully coupling leaf processes through temperature will also be essential for accurately simulating and disentangling the effects of heat and drought stresses under future conditions. However, such implementation will only be possible if accompanied with space-time concomitant observations of leaf temperature and traits that are currently lacking. The ongoing deployment of digital cameras (e.g. thermal, multispectral, SIF etc.) on existing networks (e.g. ICOS) for tracking canopy temperature and trait variability, combined with punctual field observation campaigns, future remote sensing missions (e.g. TRISHNA), as well as new hybrid modelling methods are all timely and promising ways for improving our understanding and representation of leaf temperature in TBMs.
How to cite: Peaucelle, M., Meunier, F., Stocker, B., Archambeau, J., Ogée, J., Duflos, E., De Felix, L., Chipeaux, C., Irvine, M., Anderson, J., Viovy, N., and Verbeeck, H.: Accounting for photosynthetic traits acclimation improves the simulation of forest canopy temperature in Terrestrial Biosphere Models at the ecosystem level, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11088, https://doi.org/10.5194/egusphere-egu26-11088, 2026.
Land surface models (LSMs) often exhibit substantial biases in simulating vegetation photosynthesis and respiration, largely due to their reliance on numerous plant functional type (PFT)–specific parameters. Recent advances based on Eco-Evolutionary Optimality (EEO) theory suggest that many of these parameters can be reduced, as vegetation carbon fluxes can be represented using universal optimal light and carboxylation conditions rather than prescribed PFT-dependent traits. Studies have demonstrated that EEO-based approaches perform remarkably well across a wide range of FLUXNET sites. In this study, we implement an EEO-based photosynthesis scheme within the gridded Joint UK Land Environment Simulator (JULES) to evaluate the scalability and performance of the theory at the global scale. This is a critical step beyond site-level evaluation of the theory, enabling assessment of EEO under diverse climatic and ecological conditions worldwide. We compare simulations from the EEO-enabled JULES configuration (JULES-EEO) against two model variants: JULES-NoAdap_NoAcclim, and JULES-Acclim; both of which rely on PFT-specific parameterizations. JULES-NoAdap_NoAcclim assumes no vegetation adaptation or acclimation, while JULES-Acclimation incorporates thermal acclimation following the Kumarathunge scheme. Through this intercomparison, we assess whether EEO can robustly reduce biases in global carbon flux simulations relative to conventional pft-parameter formulations. Superior performance of the EEO-based model offers the potential for improved computational efficiency by eliminating iterative, PFT-specific calculations, thereby enhancing overall model speed. The results provide new insights into the applicability of eco-evolutionary optimality theory at global scales and help identify potential pathways for further refinement of vegetation process representations in Earth system models.
How to cite: Kangari, N. R., Gan, W., Vidale, P. L., and Best, M.: EEO theory for photosynthesis and respiration in gridded standalone JULES for simulating better carbon fluxes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12215, https://doi.org/10.5194/egusphere-egu26-12215, 2026.
Climate change increases the frequency and intensity of drought events, highlighting the need to accurately predict vegetation responses to water stress to reduce uncertainties in climate projections. Droughts can determine the competitive outcome between plant species, thereby affecting ecosystem carbon, water, and energy fluxes. Capturing species- or cohort-specific responses to drought (where cohorts group species with similar plant traits and water stress strategies) in land surface models requires representing structural and functional diversity, which determines how plants compete for water under stress.
Most land surface models, including QUINCY, represent vegetation using a few fixed plant functional types, defined by shared photosynthetic pathway, phenology, structure, and climatic range, with little or no interaction between co-occurring species. To address this limitation, we introduced vegetation demography into QUINCY, focusing on grasslands. At the site scale, the model now represents multiple cohorts, defined by distinct plant trait combinations that influence water stress responses, which share the same soil resources and allow for explicit water competition between cohorts. By accounting for how plants compete for water at different soil depths, the model links cohort interactions to changes in transpiration driven by soil water availability and root distribution, supporting more process-based simulations of grassland drought responses. To ensure consistent coupling between water and carbon fluxes, physiological water stress is diagnosed based on the reduction in transpiration caused by competition for soil water. This approach maintains coherence between transpiration and carbon uptake under drought conditions.
Site-scale simulations with this cohort-based water competition scheme allow detailed analyses of water use strategies, transpiration partitioning, and drought responses in grassland communities. By incorporating in situ observations of species composition and plant traits to define cohorts, the framework directly connects field data with model simulations, supporting more accurate predictions of grassland responses to drought.
This framework provides a basis for assessing water competition outcomes in grassland communities under future climate scenarios and will be extended to include nutrient competition. Overall, the introduction of cohort-based water competition in QUINCY represents a step toward more realistic simulations of ecosystem responses to environmental stress, offering insights into the role of plant diversity and structure in modulating drought impacts.
How to cite: Abadie, C., Rolo, V., Seitz, J., Daly, L., Girish Nair, G., Papastefanou, P., and Caldararu, S.: Cohort-based water competition in a land surface model to assess grassland responses to drought, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13512, https://doi.org/10.5194/egusphere-egu26-13512, 2026.
Extreme weather poses a severe challenge to global food security, and timely, accurate agricultural disaster warnings can effectively mitigate crop yield losses. Traditional agricultural disaster warning models often rely on sparse meteorological station observations, making it difficult to capture micro-meteorological variations at the field level. Moreover, they frequently overlook the decisive role of crop stress tolerance in disaster occurrence. This study proposes a multi-spatio-temporal fusion network (MSTF-Net) featuring a unique dual-tower architecture. One tower utilizes high-dimensional remote sensing features to capture crop growth status and micro-topography on the forecast date, while the other tower employs long short term memory (LSTM) to process “past 30 days + future 7 days” meteorological time-series data, simulating the dynamic evolution of environmental stress. Within a unified framework using Sentinel-2 imagery, this approach simultaneously achieves field-level disaster warning, cause attribution, and loss assessment to mitigate yield losses from disasters. Results demonstrate that MSTF-Net achieves 12% higher accuracy compared to LSTM models using only meteorological data and multilayer perceptron (MLP) models using only remote sensing data. The model maintains high-precision early warnings (AUC > 0.85) and loss assessments within a 7-day window, meeting crop growth scheduling needs. In summary, the proposed MSTF-Net model delivers effective field-level agricultural disaster warnings, offering a feasible pathway to mitigate agricultural disaster losses.
How to cite: Wu, Y., Zeng, W., Ao, C., Ma, T., and Huang, J.: Field-Level Agricultural Disaster Warning and Loss Assessment Based on Multi-Spatio-Temporal Fusion Network, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8663, https://doi.org/10.5194/egusphere-egu26-8663, 2026.
A stable Earth system requires a healthy biosphere, but many ecosystems are being pushed beyond safe limits due to human activities such as land-use, resource extraction and climatic changes.
We map the “risk for ecosystem destabilization” (EcoRisk) due climate and land-use pressures based on simulations with the dynamic global vegetation model LPJmL. EcoRisk quantifies ecosystem dissimilarity relative to a preindustrial reference on a scale ranging from no change (0) to very strong change (1). It captures shifts in the vegetation structure (e.g., transition from forest to savanna), as well as relative (relevant for the local scale) and absolute shifts (relevant for the global scale) in soil and vegetation carbon stocks and fluxes, nitrogen stocks and fluxes, and changes in the water cycle [Stenzel et al. 2024].
Currently, almost 30% of the Earth’s global land area show severe changes (EcoRisk exceeds 0.55, a threshold derived from 10 independent indicators of biosphere integrity). These transgressions have steadily increased since 1600 and accelerated after 1925 [Stenzel et al. 2025]. A preliminary assessment of the future status according to simulations from ISIMIP3b scenarios indicates that EcoRisk continues to rise under SSP3-7.0, whereas under SSP1-2.6 it might plateau after mid-century, depending on the land-use scenario.
EcoRisk however is not only useful as a biosphere integrity indicator, but can also be used for model evaluation, because it can be computed separately for vegetation structure, water, carbon or nitrogen. It thus provides a diagnostic tool for model evaluation, benchmarking, and isolating the sources of change after major code revisions.
Data availability: https://biointegrity.pik-potsdam.de
Stenzel, F.; Braun, J.; Breier, J.; Erb, K.; Gerten, D.; Heinke, J.; Matej, S.; Ostberg, S.; Schaphoff, S. & Lucht, W.: biospheremetrics v1.0.2: an R package to calculate two complementary terrestrial biosphere integrity indicators -- human colonization of the biosphere (BioCol) and risk of ecosystem destabilization (EcoRisk), Geoscientific Model Development, 2024, 17, 3235-3258
Stenzel, F.; Ben Uri, L.; Braun, J.; Breier, J.; Erb, K.; Gerten, D.; Haberl, H.; Matej, S.; Milo, R.; Ostberg, S.; Rockström, J.; Roux, N.; Schaphoff, S. & Lucht, W.: Breaching planetary boundaries: Over half of global land area suffers critical losses in functional biosphere integrity, One Earth, 2025, 8, 101393
How to cite: Stenzel, F., Breier, J., Gerten, D., Ostberg, S., Schaphoff, S., and Lucht, W.: A vegetation model based indicator measuring the risk for ecosystem destabilization (EcoRisk), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17066, https://doi.org/10.5194/egusphere-egu26-17066, 2026.
Satellite imagery holds immense potential for crop monitoring due to its wide coverage and long-term stable historical data. However, frequent cloudy and rainy weather results in extremely fragmented optical remote sensing data in the temporal dimension, creating numerous observation gaps. This study proposes a temporal masked auto-encoder (T-MAE) framework that treats cloud occlusion as a natural mask. By performing self-supervised pre-training on large-scale unlabeled Sentinel-2 imagery, the model is forced to learn the intrinsic temporal dependencies and spectral evolution patterns of crop growth. Furthermore, the reliability of the reconstructed spectral data is evaluated by generating plot-level crop type maps using the reconstructed spectral time-series. The research results indicate that: T-MAE can reconstruct complete crop growth curves with high precision even under extreme conditions with only 20% valid observations. In downstream classification tasks, the classifier based on T-MAE pre-trained features achieved higher accuracy compared to bidirectional long short-term memory (Bi-LSTM) and Temporal Convolutional Neural Network (Temp-CNN) models (which rely on linear interpolation and Whittaker smoothing algorithms to handle cloud occlusion). Moreover, the model pre-trained with a 75% masking rate yielded higher classification accuracy than those pre-trained with 25% and 90% masking rates. In conclusion, T-MAE not only outperforms existing methods in crop classification accuracy but also demonstrates superior spatiotemporal generalization and robustness against interference. This work provides a new paradigm for addressing the dual challenges of label scarcity and cloud interference in agricultural remote sensing.
How to cite: Liu, Z., Zeng, W., Ao, C., Ma, T., Zhang, H., Wu, Y., and Sun, Y.: Crop spectral data reconstruction and classification based on temporal masked autoencoder, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8646, https://doi.org/10.5194/egusphere-egu26-8646, 2026.
Dynamic Global Vegetation Models (DGVMs) play a central role in assessing ecosystem responses to climate change, yet their increasing complexity and technical requirements often limit their accessibility beyond specialist communities. In this presentation, I introduce SEIB-Explorer, a standalone platform that couples the execution of the individual-based SEIB-DGVM with interactive three-dimensional visualization and lightweight analysis tools. The software provides a graphical user interface that streamlines model execution and result exploration, lowering the barrier to entry for non-specialists.
SEIB-Explorer enables users to visualize forest structure at a single site as a virtual stand, with plant functional types clearly distinguished and temporal changes examined through an interactive time slider. Multiple information panels summarize model metadata, annual fluxes, and time series of carbon, water, and vegetation-related variables. A visually enhanced display mode further supports interpretability for educational use and public engagement.
By integrating model execution, visualization, and basic exploratory analysis within a single environment, SEIB-Explorer reduces technical and cognitive overhead while promoting reproducible workflows and interdisciplinary exchange. This approach demonstrates how improved accessibility and interpretability can broaden the impact of DGVMs in research, education, and outreach contexts.
Figure 1.
SEIB-Explorer execution view (standard display mode). This example shows a mixed conifer–broadleaf forest in northern Hokkaido, Japan, after 100 simulation years from bare ground under late-20th-century climate conditions (e.g., 1981–2000).
Figure 3. Eight example views from the left-hand information panel. All panels correspond to the same simulation year and output as Fig. 1: (a) legend, (b) annual flux summary, (c) seasonal cycle of carbon variables, (d) seasonal cycle of water variables, (e) seasonal cycle of atmospheric variables, and (f) seasonal cycle of radiative balance.
How to cite: Sato, H.: Making Dynamic Vegetation Models More Accessible: The SEIB-Explorer Run-and-View Environment, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8479, https://doi.org/10.5194/egusphere-egu26-8479, 2026.
Plant-pollinator interactions represent a mutualistic relationship of global importance, contributing to the reproduction of most of the world's vascular plants. However, a range of drivers such as climate change and increased land-use intensity are contributing to observed declines in global pollinator numbers. The risk of this decline to both ecosystems and human-wellbeing remains unclear, despite observed cases of pollen limitation in ecosystems and crop systems. Pollen-limited natural vegetation can lead to habitat degradation, impacting biodiversity and regulatory ecosystem services, whilst pollen-limited crops can result in crop shortages, threatening food security. Global, spatial models of pollinator-dependence in plants are required to identify where vegetation is most vulnerable to becoming pollen-limited if pollinators decline.
We take a dataset of measured pollinator-dependence values and use it to phylogenetically model the pollinator-dependence of 159,366 wild plant species, using and testing an assumption that pollinator-dependence can be inferred from species relatedness. We pair this with species distribution maps and pollinator-dependence data for 98 crops to generate a global, spatially explicit, dataset of pollinator-dependence in both wild and cultivated plants at a 20 km resolution.
We find that natural vegetation has a global average pollinator-dependence of 0.51 (0 = pollinator-independent, 1 = complete pollinator-dependence). This rises to approach 0.70 in the tropics, and tropical forests specifically. The Amazon Basin and the Indonesian Archipelago emerge as geographical hotspots of high pollinator-dependence in natural vegetation. We show that areas that have higher numbers of plant species, tend to have a higher average pollinator-dependence. Additionally, we see many pollinator-dependent crops, such as coffee and cocoa, being grown in areas where adjacent ecosystems are highly pollinator-dependent. We also find that approximately half of the global human population lives close to pollinator-dependent natural vegetation.
This study offers an insight into global trends of pollinator-dependence in wild and cultivated plants, and explores how the risk of a global pollinator decline to biodiversity and human well-being may be spatially uneven. Our dataset allows for pollinator-dependence to be incorporated into spatially explicit ecosystem models, allowing for further work aimed at understanding the relative importance of pollinators in limiting the growth of plant populations at a global scale.
How to cite: Rawden, J., Newbold, T., and Springmann, M.: Mapping the global distribution of pollinator-dependence in wild and cultivated plants, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13561, https://doi.org/10.5194/egusphere-egu26-13561, 2026.
How to cite: Kasparian, J., Allaman, H., Gaponenko, I., and Goyette, S.: A multi-parameter definition of climate change velocity, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15195, https://doi.org/10.5194/egusphere-egu26-15195, 2026.
Large-scale ecosystem models often exhibit substantial uncertainty when simulating topographically complex landscapes because fine-scale microclimate and hydrologic connectivity are poorly represented within coarse grid cells. This uncertainty can propagate into limitations in capturing climate responses in water, carbon, and nitrogen cycling, especially during seasonal transitions and high- and low-flow periods. Here, we extended the widely used dynamic ecosystem model LPJ-GUESS by linking sub-grid topographic heterogeneity to ecosystem functioning through (i) topography-conditioned microclimate and (ii) lateral hillslope water redistribution, while maintaining consistency in coupled nitrogen losses.
Within a standard 0.5° grid cell, we discretize the landscape into representative topographic types based on elevation and aspect. Temperature and incoming shortwave radiation are adjusted per type using elevation, slope, and aspect, allowing vegetation and soil processes to respond to locally resolved climate conditions per topographical type. Hydrology is extended with slope-dependent partitioning between infiltration and runoff, and a daily lateral transfer scheme that redistributes surface and subsurface water downslope through simple storage/retention, thereby introducing the time-lagged hydrologic response absent from the default vertical-only bucket structure. We find that enhanced downslope drainage can unrealistically intensify nitrogen leaching at low landscape positions, so we further implement a nitrogen constraint that limits leaching under weak percolation and represents stronger retention at wetter, low-elevation areas.
We evaluate stepwise model versions in the Krycklan catchment (northern Sweden) using multi-variable observations, including catchment outlet discharge and eddy-covariance measured ecosystem evapotranspiration and carbon fluxes. Introducing sub-grid heterogeneity into LPJ-GUESS reduces runoff seasonality biases and improves performance against observed water and carbon fluxes relative to the default LPJ-GUESS. Our model development within LPJ-GUESS offers a transferable scheme to improve sub-grid topographical process representation in heterogeneous landscapes, and contributes to better simulations of ecosystem responses to climate variability and extremes.
How to cite: Zhou, H., A. Miller, P., Tang, J., Lindeskog, M., Pilesjö, P., Ahlström, A., Tai, A., Wu, J., and Olin, S.: From grid-average to hillslopes: Adding subgrid topography and lateral water redistribution to the LPJ-GUESS terrestrial ecosystem model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12157, https://doi.org/10.5194/egusphere-egu26-12157, 2026.
Phosphorus (P) loss from agricultural lands is a major contributor to surface water quality deterioration. Elevated atmospheric CO2 concentrations (eCO2) may affect P loss directly by altering hydrological processes and indirectly by influencing soil P cycling. However, the combined effects of these two mechanisms on P loss remain considerably uncertain. This study employed a more physically-based SWAT-CO2 model, which incorporates a nonlinear gs-CO2 and a LAI-CO2 function, to project P loss from corn fields in the macro-scale watershed (~500,000 km2) of the Upper Mississippi River Basin (UMRB) under eCO2 and future climate change. Results showed that the modified SWAT-CO2 model predicted 7.9% less total phosphorus (TP) loss than the original SWAT at 825 ppm CO2 during the baseline (1985-2014). Future TP loss projections deviated between models compared to the baseline (30.3% increase by the modified SWAT-CO2 vs 40.1% increase by the original SWAT under high emission scenario in the 2071-2100 period). Moreover, different forms of P loss exhibited distinct change patterns over time for both models. Soluble phosphorus (SOLP) loss increased 16.5%-58.8%, while organic phosphorus (ORGP) loss changed from -11.1% to 38.8% across all SSP scenarios. As a result, economic costs for reducing TP loss to low risk were projected to rise, with costs during the 2071-2100 period exceeding those during both 2041-2070 and the baseline periods, particularly under SSP5-8.5 scenario. These findings highlight the importance of eCO2 in predicting P loss and underscore the need for increased economic investment to achieve P-related sustainable environmental development goals.
How to cite: Wen, N.: Improving hydrological modeling to close the gap between elevated CO2 concentration and crop response: Implications for water resources and water quality, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5245, https://doi.org/10.5194/egusphere-egu26-5245, 2026.
Climate change is expected to increase the frequency and severity of Multi-Year Droughts (MYDs), yet their impacts on vegetation remain poorly understood. While satellite records provide valuable insights, they span only recent decades, limiting the number of MYDs available for analysis. Dynamic global vegetation models (DGVMs), such as LPJmL-5 (von Bloh et al., 2018), can help overcome this limitation by simulating vegetation dynamics over longer timescales. However, their ability to capture drought impacts has not yet been systematically evaluated. In this study, we benchmarked LPJmL-5 against MODIS-derived Gross Primary Production (GPP) to assess how well it captures vegetation responses to (multi-year) droughts. We show that LPJmL-5 reproduces GPP reasonably well, but there is a performance decline in parts of the Southern Hemisphere and in regions with croplands. During MYDs, the model captures the main spatial and temporal patterns of GPP decline, yet it tends to overestimate vegetation resilience at drought onset and simulates rapid post-drought recovery, leading to muted overall drought impacts. These biases appear to arise from a simplified representation of vegetation mortality processes in the model. As a result, long-term losses in biomass and shifts in ecosystem structure are often underestimated. To improve this behaviour, we incorporated a drought mortality function into LPJmL-5. This links mortality to water stress and vapour pressure deficit, with vegetation specific parameterization. We calibrated and evaluated its performance across known drought events. With the extended drought mortality representation in LPJmL-5, we refine the process representation, which in turn leads to more realistic vegetation dynamics and ultimately greater confidence in predictions of ecosystem responses under a changing climate.
How to cite: Ruijsch, D., Hauswirth, S., Biemans, H., Billing, M., Müller, C., von Bloh, W., and Wanders, N.: Drought stress in LPJmL-5: benchmarking and model improvements, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6381, https://doi.org/10.5194/egusphere-egu26-6381, 2026.
Peatlands are crucial for the regulation of the land carbon cycle and for the atmospheric methane budget. In Earth system models, improving wetland biogeochemistry and the representation of peat-specific processes has been shown to strengthen the realism of carbon-methane feedbacks and hence, the response of high-latitude land to climate warming in future simulation scenarios. We aimed to implement an explicit peatland representation in the University of Victoria Earth System Climate Model (UVic ESCM) by enhancing the WETMETH wetland methane framework to better account for the unique characteristics of peat soils and their associated carbon and methane cycling. The peatland development is based on two main components: (i) the integration of a high-resolution peatland distribution dataset (GPM 2.0) aggregated to the UVic ESCM grid using sub-cell fractional coverage, and (ii) the introduction of peatland-specific parameterizations within peat-designated grid cells to constrain key biogeochemical rates controlling long-term carbon storage and methane emissions. A 5000-year spin-up was carried to establish stable peat carbon and methane baselines, and two key peat-specific parameters were calibrated against independent global constraints.
The peatland configuration reproduces an equilibrium global peat carbon stock of 599.7 Pg C (target ≈600 Pg C) and a pre-industrial CH₄ concentration of 809.2 ppbv (target ≈808 ppbv), with negligible long-term drift. The implementation of explicit peatlands resulted shows, relative to a 1995–2015 baseline, global peat carbon changes are ~0%, 0%, −1%, and −1% by 2100 and ~0%, −4%, −12%, and −37% by 2300, for SSPs 1-2.6, 2-4.5, 4-6.0 and 5-8.5 respectively, while peatland methane emissions increase by ~+6%, +15%, +19%, and +37% by 2100 and ~+4%, +10%, +16%, and +54% by 2300. In summary, the explicit peatland implementation has successfully constrained peat carbon and methane to observationally consistent baselines, which now yield a robust peatland carbon-methane climate sensitivity and a reduced capacity of peatlands to retain carbon while amplifying their contribution to atmospheric methane under sustained warming.
How to cite: Guo, X. and MacDougall, A.: Peat carbon persistence and methane amplification under warming: explicit peatlands in UVic ESCM–WETMETH constrained by global targets, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8165, https://doi.org/10.5194/egusphere-egu26-8165, 2026.
Nitrate leaching is a major form of agricultural non-point source pollution and a critical source of groundwater contamination. Clarifying its occurrence characteristics is a prerequisite for formulating targeted measures to control nutrient losses from cropland. This study employed and improved a hydrological model (Soil and Water Assessment Tool-Freeze-Thaw; SWAT-FT) to investigate the driving effect of freeze-thaw cycles on soil hydrothermal dynamics and nitrate vertical migration within the 0-2 m soil layer in the Upper Mississippi River Basin under future climate change. The results showed that the fully calibrated SWAT-FT model, which considered the insulating effect of winter snow cover and water-ice phase change processes, provided more physically meaningful simulations of soil hydrothermal processes in winter compared to the original SWAT model. In the future, SWAT-FT predicted higher winter surface soil temperatures and greater fluctuations in deeper soil temperatures, along with lower soil water content than SWAT, fully demonstrating the key role of a physically based process mechanism model in simulating soil hydrological processes. Moreover, nitrate leaching peaked at approximately 33.7 kg ha-1 in May following fertilization but was confined to the surface soil layer. Furthermore, nitrate leaching in November, March, and April, which the months associated with freeze-thaw cycles, was similarly elevated, with migration into deeper soil layers further influenced by legacy soil nitrogen. Particularly in April, the amount of nitrate finally leached out of the soil layer accounted for approximately 50% of the total annual leaching. These results reveal a “dual risk” in nitrate losses: the immediate risk post-fertilization and the long-term risk from legacy soil nitrogen. This highlights the vital role of freeze-thaw cycles on nitrate losses and underscores the necessity of developing targeted management measures.
How to cite: Zhang, Y. and Chen, Y.: Assessing the impact of freeze-thaw cycles on nitrate leaching in the Mollisol regions under global warming, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4706, https://doi.org/10.5194/egusphere-egu26-4706, 2026.
Earth System Models (ESMs) play a central role in understanding and projecting the global carbon cycle by explicitly coupling the atmosphere, land, and ocean components. In particular, a large fraction of the interannual variability (IAV) in atmospheric CO₂ growth rate is attributed to variability in terrestrial carbon uptake. Therefore, an adequate representation of interannual variability in terrestrial ecosystem processes, such as gross primary production (GPP), is essential for robust global carbon cycle assessments.
However, in many coupled ESM simulations participating in CMIP, the phase of internal climate variability does not coincide with that of the real world, making direct time-series comparisons of interannual variability fundamentally difficult. This limitation has not yet been fully resolved even with the use of atmosphere–ocean data assimilation, and it has hindered robust evaluation of ecosystem processes governing IAV. As a result, most previous CMIP model evaluations of terrestrial carbon cycle processes have focused on mean states or climatological characteristics, while the representation of variability and ecosystem responses to extreme events has remained insufficiently assessed.
In this study, we propose a phase-insensitive model evaluation framework that is less sensitive to phase mismatches in internal variability. Within this framework, we focus on the general relationships between environmental variability—such as precipitation and temperature—and GPP responses, with particular emphasis on extreme and/or nonlinear variations that strongly contribute to interannual variability. Using multiple CMIP6 models, we evaluate the distributional properties (variability structure) of GPP and climate drivers, as well as their relationships, at monthly and seasonal timescales, through comparison with observational datasets. In addition, we examine whether biases in simulated GPP primarily arise from biases in environmental drivers or from differences in ecosystem response structures.
To capture ecosystem response diversity that is not evident in global-mean analyses, the evaluation is conducted at a regional scale across multiple climate zones. Regions exhibiting pronounced interannual variability in GPP are selected, and model behaviors are systematically compared in terms of seasonality and environmental responses. Because variability at annual timescales is strongly influenced by ecosystem functioning at seasonal scales, monthly and seasonal analyses provide an effective basis for diagnosing terrestrial ecosystem representations in ESMs.
This presentation highlights both common features and inter-model differences in the representation of GPP variability across CMIP models, and discusses implications for evaluating and improving terrestrial ecosystem processes in Earth system models.
How to cite: Satoh, Y. and Hajima, T.: Assessing Terrestrial GPP Responses to Climate Variability in CMIP6 Earth System Models: How Do Ecosystems Respond to Large Climate Fluctuations?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10218, https://doi.org/10.5194/egusphere-egu26-10218, 2026.
Temperate forests play a central role in Europe’s carbon and water cycles, yet process-based vegetation models remain weakly constrained at the species and daily timescale, particularly with respect to physiological and radiative controls on carbon–water coupling. In Wallonia (Belgium), forests cover 33% of the territory and represent about 80 % of the country’s forest area, with European beech (Fagus sylvatica), oaks (Quercus robur and Quercus petraea), Norway spruce (Picea abies) and Douglas fir (Pseudotsuga menziesii) among the dominant species.
Here, we present a species-specific, daily-scale sensitivity analysis of the CARAIB dynamic vegetation model at the FLUXNET/ICOS Vielsalm site (BE-Vie), a mixed forest site with European beech and Douglas fir as dominant species. Model performance is evaluated using eddy-covariance–derived gross primary production (GPP), total ecosystem respiration (TER), net ecosystem exchange (NEE) and actual evapotranspiration (AET) products for 1997–2021 processed with multiple friction-velocity (u*) filtering methods.
We systematically examine how forest management (year of plantation, thinning), nitrogen availability, plant functional traits, and radiative processes shape simulated GPP, TER, NEE and water-use efficiency (WUE, i.e., the ratio of GPP to transpiration). Physiological parameters constraining the model include slope of stomatal relationship (g₁), specific leaf area (SLA), carbon-to-nitrogen ratio (C:N) and fraction of sapwood (fsw), as well as the level of isohydricity of the tree species. Radiative sensitivity is assessed using diffuse radiation fraction at the top of canopy and leaf optical properties. Soil respiration sensitivity is also assessed through parameters controlling its dependence on temperature and soil water content.
We put a particular emphasis on understanding the way the studied parameters impact the response of GPP, TER and NEE to droughts, by comparing drought years (e.g., 2018, 2019 and 2020) to normal years. The findings demonstrate that species-specific, process-based calibration is essential for improving dynamic vegetation model reliability in European temperate forests under management, climate and environmental changes.
How to cite: Chaudhari, T., Verma, A., Hambuckers, A., Ghilain, N., Lecart, B., and François, L.: Species-Specific Controls on Carbon–Water Coupling in European Temperate Forests: A Process-Based Sensitivity Analysis Using the CARAIB Dynamic Vegetation Model , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14613, https://doi.org/10.5194/egusphere-egu26-14613, 2026.
Plant responses to dry environments are shaped by diverse adaptive strategies linked to plant hydraulic traits, as reflected by the coexistence of deciduous and evergreen tree species in tropical seasonally dry forests. Empirical evidence suggests that leaf shedding is associated with declining leaf water potential, while leaf flushing depends on xylem rehydration. However, these physiological mechanisms are rarely incorporated into Dynamic Global Vegetation Models (DGVMs), which typically represent drought deciduous phenology using highly-simplified, threshold-based schemes with fixed rates of leaf phenological change. Here, we develop a stress–gradient based phenology scheme in which leaf shedding and leaf flushing are driven by the temporal gradients of leaf water potential and xylem water potential, respectively. This gradient–driven phenology mechanism was implemented in the LPJ-GUESS DGVM and validated with in-situ observations of phenological responses along hydraulic gradients. The new model has successfully reproduced phenological dynamics for the 7 selected locations and substantially improves model performance in simulating plant transpiration. We provide evidence that plant hydraulics are key controls for the phenological dynamics of tropical dry forests. The proposed stress-gradient phenological mechanism, linking phenology to plant hydraulic status, is an efficient approach to represent landscape phenology and improve simulations of water and carbon cycling over the tropical drylands. It may also help improve our understanding of forest response to drought stress, which remains largely unknown under warming climates.
How to cite: Zhu, Y., Pugh, T. A. M., and Wu, M.: Modelling Tropical Dry Forest Phenology from a Plant Hydraulic Perspective, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12529, https://doi.org/10.5194/egusphere-egu26-12529, 2026.
Climate change accelerates the global hydrological cycle, which has escalating impacts on human health and the socioeconomic development. However, many existing Earth system models neglect the more complex processes of topography-driven vegetation-surface-groundwater interactions, thereby failing to accurately capture climate-hydrological responses. To address this gap, we integrate the three-dimensional surface-subsurface hydrological model ParFlow with the dynamic global vegetation model LPJ-GUESS to investigate how lateral groundwater flow and vegetation dynamics jointly regulate hydrological fluxes. The fully coupled ParFlow-LPJ-GUESS (PF-LPJG) model (with and without lateral flow) and stand-alone LPJ-GUESS model were used to run hydrological simulations over a 38-year period at a resolution of 10 km across the Danube River Basin. The results demonstrate that the PF-LPJG model substantially improves streamflow and surface soil moisture simulations without requiring parameter calibration compared to stand-alone LPJ-GUESS, mitigates the underestimation of summer low flows during dry years, increases the accuracy of peak flow timing in wet years, and achieves a Kling-Gupta Efficiency (KGE) > 0.5 and Spearman’s ρ > 0.80 at over 80 % of gauging stations. Seasonal soil moisture anomalies are better captured (R = 0.51) compared to satellite-based products. Additionally, the modelled WTD agrees well with in-situ monitoring-well data, as indicated by a low RSR value (~1.31). Notably, the coupled model improves the representation of bare-soil evaporation and reduces transpiration-to-evaporation (T/E) ratio fluctuations, aligning more closely with the GLEAM v4.2 product. Sensitivity analysis reveals that shallow-rooted vegetation exhibits strong decreases in LAI and AET when lateral flow is removed, while slightly increasing LAI and AET in deep-rooted regions. The coupled model PF-LPJG entails a mechanistic framework for capturing bidirectional interactions among surface-subsurface water, vegetation dynamics and ecosystem biogeochemical processes, which can be applied to other catchments or climatic conditions to deeply analyze climate-induced modification on vegetation-water-carbon interactions. Future work will focus on how the lateral flow affects vegetation greening under different climatic conditions.
How to cite: Jia, Z., Fu, Y. H., Chen, S., and Tang, J.: Advancing Ecohydrological Modelling with Coupled ParFlow-LPJ-GUESS: Role of Lateral Flow in Vegetation and Hydrology Simulations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16516, https://doi.org/10.5194/egusphere-egu26-16516, 2026.
The mechanism of community assembly has always been a core aspect of community ecology. Exploring the interspecific interactions and the distribution characteristics of different organ traits, as well as their main influencing factors, during community development is of great significance for community assembly, succession, and adaptation to climate change. However, artificial forests cannot compare with natural forests in terms of biodiversity conservation and ecosystem multifunctionality. Therefore, an in-depth exploration of the assembly mechanisms of natural forest communities is beneficial for maximizing their ecological, economic, and social benefits. Current research mostly focuses on single-organ, single-functional groups and their responses to single environmental variables. There is little exploration of the diversity of species and the distribution patterns of functional traits in different functional groups and organs during the assembly process of natural communities, as well as their environmental explanations. This study takes the Quercus wutaishanica as community in the warm temperate deciduous forest of northern China as an example. Community surveys were conducted in Quercus wutaishanica communities spanning five provinces from west to east in northern China, including different functional groups such as trees, shrubs, and herbs. Samples of leaves, twigs, and fine roots were collected, and their morphological and chemical functional traits were measured. Environmental factors including soil and topography were also measured. This study aims to deeply explore the interactions between functional traits and species in natural forest communities and their environmental explanations, elucidate the ecological processes involved in the assembly process of warm temperate deciduous broad-leaved forest communities in northern China, and reveal their community assembly mechanisms. The goal is to provide a theoretical basis for biodiversity conservation, the restoration and reconstruction of natural secondary forests, and responding to climate change.
How to cite: Xing, S., Han, H., Quan, Z. J., Sun, Y., and Wang, H.: Research on the construction mechanism of warm temperate deciduous broad-leaved forest communities based on plant functional traits in northern China，, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21178, https://doi.org/10.5194/egusphere-egu26-21178, 2026.
Understanding plant biophysical and biochemical responses to drought is essential for predicting ecosystem carbon–water exchange and gross primary productivity (GPP) under a changing climate. Here, we investigate stomatal and non-stomatal responses to natural drought events by integrating carbonyl sulfide (COS) fluxes as a novel observational constraint into the Soil–Canopy Observation of Photosynthesis and Energy fluxes (SCOPE) model. Because plants take up COS in parallel with CO2 via practically similar pathways but do not re-emit COS, it can be used as promising proxy for GPP and stomatal conductance. SCOPE is a physically based land-surface model that links plant processes to spectrally resolved within-canopy radiative transfer; adding COS provides an independent constraint on canopy conductance and carbon uptake. We leverage multiple years of concurrent COS and CO2 flux as well as hyperspectral reflectance measurements from a Scots pine dominated montane forest in Austria, including naturally occurring drought periods, to refine model representations of stomatal regulation, internal conductance, and water-stress responses at the ecosystem scale. This model–data integration framework improves detection and prediction of drought impacts on canopy function and enhances constraints on ecosystem carbon–water dynamics.
How to cite: de Vries, A., Spielmann, F. M., Platter, A., Hammerle, A., van der Tol, C., and Wohlfahrt, G.: Seeing through the canopy: COS-constrained SCOPE modelling to investigate plant drought responses , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6727, https://doi.org/10.5194/egusphere-egu26-6727, 2026.
Process-based crop models are extensively used to assess the impacts of climate change, environmental variations, and management practices on crop yields. However, parameters sourced from the literature are often not universally applicable, necessitating calibration to enhance the model performance. The in-situ observed crop yield (hereafter referred to as “observed crop yield”) is commonly used for model calibration. Satellite-based data, such as evapotranspiration (ET), offers additional insights into plant growth and holds significant potential for enhancing calibration efforts. The LPJ-GUESS (Lund-Potsdam-Jena General Ecosystem Simulator) model has been applied extensively to simulate crop yields across various scales, but it has not been calibrated for the regional-scale yield simulation. This study aims to enhance the performance of LPJ-GUESS in simulating crop yields in southern Sweden through calibration using observed crop yield and satellite-based ET data. Results demonstrated that calibrating with observed crop yield substantially improved simulation accuracy for both spring barley and winter wheat, reducing the normalized root mean square error (nRMSE) from 50.3% to 12.8% and from 15.5% to 12.2%, respectively. Sensitivity analysis identified four key parameters influencing yield simulations: minimum C:N ratio (CNmin), N demand reduction by leaves (Ndred), and the retranslocation of nitrogen and carbon (Nret and Cret). Calibration using the Penman-Monteith-Leuning Version 2 (PML-V2) ET product moderately enhanced yield simulation accuracy, particularly for winter wheat, achieving an nRMSE of 14.2%, demonstrating its potential as an alternative when especially when the long-term and continuous observed crop yield is not available. The calibrated LPJ-GUESS model effectively simulated crop yield for both crop types under drought and normal conditions, highlighting its robustness across varying environmental scenarios.
How to cite: Li, X., Zhang, W., Wu, M., Olin, S., Zhou, H., Huang, X., Thapa, S., Bouras, E. H., and Duan, Z.: Calibrating the dynamic vegetation model LPJ-GUESS for crop yield simulation in southern Sweden using observed crop yield and satellite-based evapotranspiration data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4744, https://doi.org/10.5194/egusphere-egu26-4744, 2026.
Due to the coupling of gas exchange in terrestrial ecosystems, transpiration (T) estimates are a key insight into global climate, water, and carbon patterns. While extensive datasets of evapotranspiration (ET) are abundant, largely due to global eddy covariance networks, independent measurements of T are relatively sparse. As such, there is a need to partition T out of eddy covariance-measured ET so that we can better understand how the individual components of the terrestrial water vapor flux are contributing to the global water cycle and changing under a warming climate. Physics-informed machine learning (PI-ML) presents a novel way to partition eddy covariance-measured ET even without extensive T datasets for validation. PI-ML works to constrain the model to obey underlying governing equations driving the system dynamics, giving unique insights into model estimates. Here, PI-ML is introduced to estimate the ecosystem transpiration ratio (T/ET) using eddy covariance data collected from a soybean field in Ontario, Canada. The model, founded on the principle that transpiration follows a sine curve over a 24hr period, constrains both the upper and lower bounds of T/ET by assuming transpiration is negligible overnight and that the agricultural site has periods with negligible soil evaporation. The PI-ML model was validated against leaf-level transpiration measurements collected over the 2025 growing season as well as compared to results from other eddy covariance-based ET partitioning methods. Preliminary results for the 2019 growing season showed the PI-ML estimated a daytime average T/ET of 0.584 in the soybean field, compared to 0.468 estimated from an underlying water use efficiency-based method and 0.653 estimated from a method using data-driven machine learning. The physically realistic T estimates produced by the PI-ML show the model’s ability to accurately represent the ecosystem dynamics of the soybean site where it was applied. Accurate T estimates give way for better management of our limited water resources, leading to increased water quality and food security when used in agricultural settings.
How to cite: Cochran, E. and Eichelmann, E.: Using Physics-Informed Machine Learning to Partition Eddy Covariance based Evapotranspiration, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7019, https://doi.org/10.5194/egusphere-egu26-7019, 2026.
