Attributing uncertainties in simulated biospheric carbon fluxes to different error sources

Abstract. Optimized biogenic carbon fluxes for Europe were estimated from high-resolution regional-scale inversions, utilizing atmospheric CO2 measurements at 16 stations for the year 2007. Additional sensitivity tests with different data-driven error structures were performed. As the atmospheric network is rather sparse and consequently contains large spatial gaps, we use a priori biospheric fluxes to further constrain the inversions. The biospheric fluxes were simulated by the Vegetation Photosynthesis and Respiration Model (VPRM) at a resolution of 0.1° and optimized against eddy covariance data. Overall we estimate an a priori uncertainty of 0.54 GtC yr−1 related to the poor spatial representation between the biospheric model and the ecosystem sites. The sink estimated from the atmospheric inversions for the area of Europe (as represented in the model domain) ranges between 0.23 and 0.38 GtC yr−1 (0.39 and 0.71 GtC yr−1 up-scaled to geographical Europe). This is within the range of posterior flux uncertainty estimates of previous studies using ground-based observations.

Estimating gross primary productivity (GPP) is important for simulating the subsequent carbon cycle elements and assessing the capacity of terrestrial ecosystems to support the sustainable development of human society. Light use efficiency (LUE) models were widely used to estimate GPP due to their concise model structures. However, quantifying LUEmax (maximum light use efficiency) and representing the responses of photosynthesis to environmental factors are still subject to large uncertainties, which lead to substantial errors in GPP simulations. In this study, we developed a hybrid model based on machine learning and a LUE model for GPP estimates. This hybrid model was built by targeting LUE with a machine learning approach, namely multi-layer perceptron (MLP), and then, estimating GPP within a LUE model framework with the MLP-based LUE and other required inputs. We trained the hybrid LUE (H-LUE) model and then, compared it against two conventional LUE models, the vegetation photosynthesis model (VPM) and vegetation photosynthesis and respiration model (VPRM), regarding GPP estimation, using tower-based daily-scale observations from 180 flux sites that cover nine different plant function types (PFTs). The results revealed better performance (R2 = 0.86 and RMSE = 1.79 gC m−2 d−1 on the test dataset) of the H-LUE model compared to the VPM and VPRM. Evaluations of the three models under four different extreme conditions consistently revealed better performance of the H-LUE model, indicating greater adaptability of the model to varied environments in the context of climate change. Furthermore, we also found that the H-LUE model can reasonably represent the responses of the LUE to meteorological variables. Our study revealed the reliable and robust performance of the developed hybrid LUE when simulating GPP across global biomes, providing references for developing better hybrid GPP models.

How and to what extent does precipitation on multi-temporal scales and soil moisture at different depths determine carbon flux responses in a water-limited grassland ecosystem?

Estimating biogenic CO 2 fluxes is essential to quantify urban anthropogenic emissions, yet urban vegetation heterogeneity presents a significant challenge to making accurate estimations. We have developed an hourly temporal, 10‐m spatial resolution biogenic CO 2 flux estimation framework based on the Vegetation Photosynthesis and Respiration Model (VPRM) and its variants (UrbanVPRM and VPRM‐modified). Unlike lower‐resolution models, our approach captures finer‐scale variability, particularly in fragmented urban green spaces like street trees and lawns. Results show that vegetation in Munich offsets 2.0%–2.8% of annual anthropogenic CO 2 emissions in the study domain, with tree‐covered areas as primary sinks and grasslands as net sources. During summer, daytime CO 2 uptake can match or exceed anthropogenic emissions. Evaluations employing city park field measurements and eddy covariance towers confirm strong performance of our models, while highlighting VPRM‐modified's advantage in grasslands and croplands, and UrbanVPRM's improvements in urban areas via impervious surface correction. These findings highlight the value of high‐resolution modeling in improving urban carbon flux assessments.

India needs a high-resolution estimation of carbon sources and sinks to implement the country’s climate change action plans and mitigation strategy effectively. Current carbon estimates over the Indian region based on the “Bottom-up” approach suffer from significant uncertainty, which calls for more process-based models and atmospheric inverse modelling to obtain a more accurate budget. Inverse models constrain the carbon fluxes based on atmospheric observation of CO2 mole fractions. The unavailability of amble observations over the Indian domain critically impacts estimation accuracy. Fortunately, there are increasing efforts to improve the availability of CO2 observation over the domain. Along with the observations, the availability of a suitable transport model to simulate the CO2 distribution is essential to the accurate inverse estimation of carbon fluxes. The inability of coarse-resolution global models to simulate the fine-scale variability in CO2 distribution warrants developing a regional high-resolution modelling system. Here we evaluate the performance of a regional high-resolution modelling system which utilises meteorological fields from the Weather Research and Forecasting (WRF) model to simulate the CO2 transport over the Indian domain using a lagrangian particle dispersion model, Stochastic Time-Inverted Lagrangian Transport Model (STILT). Using lagrangian models enables us to study the CO2 distribution at very high resolution (even at sub-grid scale) with reduced cost. We use the vegetation photosynthesis and Respiration Model (VPRM), coupled with the modelling system, to simulate the biospheric fluxes. The anthropogenic and biomass burning fluxes are obtained from different available inventories. We use CO2 in-situ observations from different parts of the Indian domain, which utilises flask measurements and PICARRO CRDS instruments, to evaluate the modelling system. Our high-resolution modelling framework shows good skill in simulating the CO2 variability over the region. The results of the evaluation will be discussed in detail during the presentation.

Biogeochemical models must include a broad variety of biological and physical processes to test our understanding of the terrestrial carbon cycle and to predict ecosystem biomass and carbon fluxes. We combine the photosynthesis and biophysical calculations in the Simple Biosphere model, Version 2.5 (SiB2.5) with the biogeochemistry from the Carnegie‐Ames‐Stanford Approach (CASA) model to create SiBCASA, a hybrid capable of estimating terrestrial carbon fluxes and biomass from diurnal to decadal timescales. We add dynamic allocation of Gross Primary Productivity to the growth and maintenance of leaves, roots, and wood and explicit calculation of autotrophic respiration. We prescribe leaf biomass using Leaf Area Index (LAI) derived from remotely sensed Normalized Difference Vegetation Index. Simulated carbon fluxes and biomass are consistent with observations at selected eddy covariance flux towers in the AmeriFlux network. Major sources of error include the steady state assumption for initial pool sizes, the input weather data, and biases in the LAI.

Controls of carbon flux in a semi-arid grassland ecosystem experiencing wetland loss: Vegetation patterns and environmental variables

Understanding of regional greenhouse gas emissions into the atmosphere is a prerequisite to mitigate climate change. In this study, we investigated the regional contributions of carbon dioxide (CO2) at the location of the high Alpine observatory Jungfraujoch ("JFJ", Switzerland, 3580 m a.s.l.). To this purpose, we combined receptor-oriented atmospheric transport simulations for CO2 concentration in the period of 2009–2017 with stable carbon isotope (δ13C-CO2) information. We applied two Lagrangian particle dispersion models driven by output from two different numerical weather prediction systems (FLEXPART-COSMO and STILT-ECMWF) in order to simulate CO2 concentration at JFJ based on regional CO2 fluxes, to estimate atmospheric δ13C-CO2, and to obtain model-based estimates of the mixed source signatures (δ13Cm). Anthropogenic fluxes were taken from a fuel type-specific version of the EDGAR v4.3 inventory and ecosystem fluxes were based on the Vegetation Photosynthesis and Respiration Model (VPRM). The simulations of CO2, δ13C-CO2 and δ13Cm were then compared to observations performed by quantum cascade laser absorption spectroscopy. Around 40 % of the regional CO2 variability above or below the large-scale background was captured by the models, and up to 35 % of the regional variability in δ13C-CO2. This is remarkable considering the complex Alpine topography, the low intensity of regional signals at JFJ, and the challenging measurements. Best agreement between simulations and observations in terms of short-term variability and intensity of the signals for CO2 and δ13C-CO2 was found between late autumn and early spring. The agreement was inferior in the early autumn periods and during summer. This may be associated with the atmospheric transport representation in the models. In addition, the net ecosystem exchange fluxes are a possible source of error, either through inaccuracies in their representation in VPRM for the (Alpine) vegetation or through a day (uptake) vs. night (respiration) transport discrimination to JFJ. Furthermore, the simulations suggest that JFJ is subject to relatively small regional anthropogenic contributions, due to its remote location (elevated and far from major anthropogenic sources), and the limited planetary boundary layer-influence during winter. Instead, the station is primarily exposed to summer-time ecosystem CO2 contributions, which are dominated by rather nearby sources (within 100 km). Even during winter, simulated gross ecosystem respiration accounted for approximately 50 % of all contributions to the CO2 concentrations above the largescale background. The model-based monthly mean δ13Cm ranged from −22 ‰ in winter to −28 ‰ in summer and reached the most depleted values of −35 ‰ at higher fractions of natural gas combustion, and the most enriched values of −17 to −12 ‰ when impacted by cement production emissions. Observation-based δ13Cm values derived by a moving Keeling-plot approach were in good agreement with the model-based estimates. They exhibited a larger scatter, while model-based estimates spread in a more narrow range. Overall, observation-based δ13Cm were limited to a smaller number of data points compared to model-based estimates owing to the stringent analysis prerequisites in combination with the low regional signal at JFJ.

Abstract. Urban areas are significant contributors to global CO2 emissions, requiring detailed monitoring to support climate neutrality goals. This study presents a high-resolution modeling framework using GRAMM/GRAL, adapted for simulating atmospheric CO2 concentrations from anthropogenic and biospheric sources and sinks in Zurich, Switzerland. The framework resolves atmospheric concentrations at the building scale, and it employs a detailed inventory of anthropogenic emissions as well as biospheric fluxes, which were calculated using the Vegetation Photosynthesis and Respiration Model (VPRM). Instead of simulating the full dynamics of meteorology and atmospheric transport, the dispersion of CO2 is precomputed for more than 1000 static weather situations, from which the best match is selected for any point in time based on the simulated and measured meteorology in and around the city. In this way, time series over multiple years can be produced with minimal computational cost. Measurements from a dense network of mid-cost CO2 sensors are used to validate the model, demonstrating its capability to capture spatial and temporal CO2 variability. Applications to other cities are discussed, emphasizing the need for high-quality input data and tailored solutions for diverse urban environments. The work contributes to advancing urban CO2 monitoring strategies and their integration with policy frameworks.

<p>Evaluating atmospheric transport simulations against observations helps refining bottom-up estimates of greenhouse gas fluxes and identifying gaps in our understanding of regional and category-specific contributions to atmospheric mole fractions. This insight is critical in the efforts to mitigate anthropogenic environmental impact. Beside total mole fractions, stable isotope ratios provide further constraints on source-sink processes [1-3].</p><p>Here, we present two receptor-oriented model simulations for carbon dioxide (CO<sub>2</sub>) mole fraction and δ<sup>13</sup>C-CO<sub>2</sub> stable isotope ratios for a nine year period (2009-2017) at the High Altitude Research Station Jungfraujoch (Switzerland, 3580 m asl). The model simulations of CO<sub>2</sub> were performed on a 3-hourly time-resolution with two backward Lagrangian particle dispersion models driven by two different numerical weather forecast fields: FLEXPART-COSMO and STILT-ECMWF. Anthropogenic CO<sub>2</sub> fluxes were based on the EDGAR v4.3 emissions inventory aggregated into 14 source categories representing fossil and biogenic fuel uses as well as emissions from cement production. Biospheric CO<sub>2</sub> fluxes representing the photosynthetic uptake and respiration of 8 plant functional types were based on the Vegetation Photosynthesis and Respiration Model (VPRM). The simulated CO<sub>2</sub> emissions per source and sink category were weighted with category-specific δ<sup>13</sup>C-CO<sub>2</sub> signatures from published experimental studies. Background CO<sub>2</sub> values at the boundaries of both model domains were taken from global model simulations and the corresponding δ<sup>13</sup>C-CO<sub>2</sub> values were constructed as suggested in Ref. [3]. We compare the simulations to a unique data set of continuous in-situ observations of CO<sub>2</sub> mole fractions and δ<sup>13</sup>C-CO<sub>2</sub> stable isotope ratios by quantum cascade laser absorption spectroscopy as described in previous work [1, 4-5], available for the whole nine year period at the site.</p><p>The simulated atmospheric CO<sub>2</sub> and δ<sup>13</sup>C-CO<sub>2</sub> time-series are in good agreement with the observations and capture the observed variability at the models' 3-hourly time-resolution. This allows for an in-depth evaluation of the contribution of different CO<sub>2</sub> emission sources and for an allocation of source regions when Jungfraujoch is influenced by air masses from the planetary boundary layer. In brief, the receptor-oriented model simulations suggest that anthropogenic CO<sub>2</sub> contributions are primarily of fossil origin (90%). Anthropogenic emissions contribute between 60% in February, and 20% in July/August, to the CO<sub>2</sub> enhancements observed at Jungfraujoch. The remaining fraction is due to biosphere respiration, which thus largely dominates emissions during the summer season. However, intense photosynthetic CO<sub>2</sub> uptake during June, July and August roughly outweighs CO<sub>2</sub> contributions from anthropogenic activities and biosphere respiration at JFJ.</p><p> </p><p> </p><p>REFERENCES</p><p>[1] Tuzson et al., 2011. ACP, 11, 1685</p><p>[2] Röckmann et al., 2016. ACP, 16, 10469</p><p>[3] Vardag et al., 2016. Biogeosciences, 13, 4237</p><p>[4] Tuzson et al., 2008. Appl. Phys. B, 92, 451</p><p>[5] Sturm et al., 2013. AMT 6, 1659</p>

The characteristics of sorbitol and sucrose export out of mature leaves in seedlings of peach (Prunus persica L. Batsch cv GF 305) were investigated by simulating carbon fluxes through the leaf. Three treatments were employed: a control treatment and two treatments modifying leaf export, the latter using either shading or girdling. Photosynthesis and 14C partitioning into sorbitol and sucrose were measured during carbohydrate pool buildup at the beginning of the photoperiod, and the export rate of sorbitol and sucrose was modeled using a PSPICE (Simulation Program with Integrated Circuit Emphasis) simulator. The simulation allowed prediction of the resulting sorbitol and sucrose contents, which were compared to experimental carbohydrate contents. The apparent Km for sorbitol and sucrose phloem loading, estimated by carbon flux modeling, was 6.6 and 4 mol m-3, respectively. The predicted export capacity of the leaf, characterized by the estimated Vmax values for phloem loading of sorbitol and sucrose, was similar to the photosynthetic carbon flux measured under the leaf growth conditions. This export capacity was enhanced in plants in which all leaves except those studied were shaded. The mature leaf had a higher storage capacity for sorbitol than for sucrose in control plants, especially in the girdled treatment. Sucrose content appears to be tightly regulated.

Contrasting responses of grassland water and carbon exchanges to climate change between Tibetan Plateau and Inner Mongolia

Carbon and water cycle dynamics of vegetation are controlled primarily by photosynthesis and stomatal conductance (gs). Our goal is to improve the representation of these key physiological processes within the JULES land surface model, with a particular focus on refining the temperature sensitivity of photosynthesis, impacting modelled carbon, energy and water fluxes. We test (1) an implementation of the Farquhar et al. (1980) photosynthesis scheme and associated plant functional type-dependent photosynthetic temperature response functions, (2) the optimality-based gs scheme from Medlyn et al. (2011), and (3) the Kattge and Knorr (2007) photosynthetic capacity thermal acclimation scheme. New parameters for each model configuration are adopted from recent large observational datasets that synthesise global experimental data. These developments to JULES incorporate current physiological understanding of vegetation behaviour into the model, and enable users to derive direct links between model parameters and on-going measurement campaigns that refine such parameter values. Replacement of the original Collatz et al. (1991) C3 photosynthesis model with the Farquhar scheme results in large changes in GPP for current-day, with ~10 % reduction in seasonal (June–August; JJA and December–February; DJF) mean GPP in tropical forests, and ~20 % increase in the northern high latitude forests in JJA. The optimality-based gs model decreases the latent heat flux for the present-day (~10 %, with an associated increase in sensible heat flux) across regions dominated by needleleaf evergreen forest in the northern hemisphere summer. Thermal acclimation of photosynthesis coupled with the Medlyn gs scheme reduced tropical forest GPP by up to 5 %, and increased GPP in the high northern latitude forests by between 2 to 5 %. Evaluation of simulated carbon and water fluxes by each model configuration against global data products show this latter configuration generates improvements in these key areas. Thermal acclimation of photosynthesis coupled with the Medlyn gs scheme improved modelled carbon fluxes in tropical and high northern latitude forests in JJA, and improved the simulation of evapotranspiration across much of the northern hemisphere in JJA. Having established good model performance for the contemporary period, we force this new version of JULES offline with a future climate scenario corresponding to rising atmospheric greenhouse gases (SSP5 RCP8.5). In particular, these calculations allow understanding of the effects of long-term warming. We find that the impact of thermal acclimation coupled with the optimality-based gs model on simulated fluxes increases latent heat flux (+50 %) by year 2050 compared to the JULES model configuration without acclimation. This new JULES configuration also projects increased GPP across tropical (+10 %) and northern latitude regions (+30 %) by 2050. We conclude that thermal acclimation of photosynthesis with the Farquhar photosynthesis scheme and the new optimality-based gs scheme together improve the simulation of carbon and water fluxes for current-day, and has a large impact on modelled future carbon cycle dynamics in a warming world.

The biospheric water and carbon cycles are intimately coupled, so simulating carbon fluxes by vegetation also requires modelling of the water fluxes, with each component influencing the other. Observations of river streamflow integrate information at the catchment scale and are widely available over a long period; they therefore provide an important source of information for validating or calibrating vegetation models. In this paper, we analyse the performance of the Sheffield dynamic global vegetation model (SDGVM) for predicting river streamflow and quantifying how this information helps to constrain carbon flux predictions. The SDGVM is run for 29 large catchments in the United Kingdom. Annual streamflow estimates are compared with long time-series observations. In 23 out of the 29 catchments, the bias between model and observations is less than 50 mm, equivalent to less than 10% of precipitation. In the remaining catchments, larger errors are because of combinations of unpredictable causes, in particular various human activities and measurement issues and, in two cases, unidentified causes. In one of the catchments, we assess to what extent a knowledge of annual streamflow can constrain model parameters and in turn constrain estimates of gross primary production (GPP). For this purpose, we assume the model parameters are uncertain and constrain them by the streamflow observations using the generalized likelihood uncertainty estimation method. Comparing the probability density function of GPP with and without constraint shows that streamflow effectively constrains GPP, mainly by setting a low probability to GPP values below about 1100 g C-1 m2 yr-1 . In other words, streamflow observations allow the rejection of low values of GPP, so that the potential range of possible GPP values is almost halved.

ABSTRACTErrors are quite large in the simulated carbon and water fluxes obtained by global models used for the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, and reducing those errors is important for improving our confidence about these models and their projections. Errors in model parameter values are a major cause of those large modelling errors but can be significantly reduced if model parameter values are optimized. While parameter optimizations have been carried out at local sites or regional scales, parameter optimizations have been rarely conducted at the global scale because of the high computing costs required to optimize a large (>100) number of model parameters. In this study, we used an adaptive surrogate modelling based optimization (ASMO) method to maximize the match between simulated monthly global gross primary production (GPP) and latent heat flux (LE) derived by two global land surface models (LSMs) and the model‐data products for global GPP and LE from the 1982–2008 period generated by the Max Plank Institute. The ASMO method only required a few hundred model runs to find the optimal values of all optimized parameters for the two global LSMs [the Australian Community Atmosphere‐Biosphere‐Land Exchange (CABLE) and joint UK land environment simulator (JULES)]. Our results show that up to 65% of the model errors can be reduced by parameter optimization for most of the plant functional types (PFTs) and that the model performances of CABLE and JULES are significantly improved at 72 and 93% of the land points, respectively. At last, we discuss the limitations of this work and recommend that parameter optimization based on surrogate modelling using various observational data sets and acceptable prior information of uncertainties in model structure and observations should be considered as a key step in improving the performance of global LSMs or model intercomparisons.