Search

Mapping in situ eddy covariance measurements of terrestrial land–atmosphere fluxes to the globe is a key method for diagnosing the Earth system from a data-driven perspective. We describe the first global products (called X-BASE) from a newly implemented upscaling framework, FLUXCOM-X, representing an advancement from the previous generation of FLUXCOM products in terms of flexibility and technical capabilities. The X-BASE products are comprised of estimates of CO2 net ecosystem exchange (NEE), gross primary productivity (GPP), evapotranspiration (ET), and for the first time a novel, fully data-driven global transpiration product (ETT), at high spatial (0.05°) and temporal (hourly) resolution. X-BASE estimates the global NEE at −5.75 ± 0.33 Pg C yr−1 for the period 2001–2020, showing a much higher consistency with independent atmospheric carbon cycle constraints compared to the previous versions of FLUXCOM. The improvement of global NEE was likely only possible thanks to the international effort to increase the precision and consistency of eddy covariance collection and processing pipelines, as well as to the extension of the measurements to more site years resulting in a wider coverage of bioclimatic conditions. However, X-BASE global net ecosystem exchange shows a very low interannual variability, which is common to state-of-the-art data-driven flux products and remains a scientific challenge. With 125 ± 2.1 Pg C yr−1 for the same period, X-BASE GPP is slightly higher than previous FLUXCOM estimates, mostly in temperate and boreal areas. X-BASE evapotranspiration amounts to 74.7×103 ± 0.9×103 km3 globally for the years 2001–2020 but exceeds precipitation in many dry areas, likely indicating overestimation in these regions. On average 57 % of evapotranspiration is estimated to be transpiration, in good agreement with isotope-based approaches, but higher than estimates from many land surface models. Despite considerable improvements to the previous upscaling products, many further opportunities for development exist. Pathways of exploration include methodological choices in the selection and processing of eddy covariance and satellite observations, their ingestion into the framework, and the configuration of machine learning methods. For this, the new FLUXCOM-X framework was specifically designed to have the necessary flexibility to experiment, diagnose, and converge to more accurate global flux estimates.

Global collections of synthesized flux tower data such as FLUXNET have accelerated scientific progress beyond the eddy covariance community. However, remaining data issues in FLUXNET data pose challenges for users, particularly for multi-site synthesis and modelling activities. Here, we present complementary consistency flags (C2Fs) for flux tower data, which rely on multiple indications of inconsistency among variables, along with a methodology to detect discontinuities in time series. The C2F relates to carbon and energy fluxes, as well as to core meteorological variables, and consists of the following: (1) flags for daily data values, (2) flags for entire-site variables, and (3) flags at time stamps that mark large discontinuities in the time series. The flagging is primarily based on combining outlier scores from a set of predefined relationships among variables. The methodology to detect break points in the time series is based on a non-parametric test for the difference in distributions of model residuals. Applying C2F to the FLUXNET 2015 dataset reveals the following: (1) among the considered variables, gross primary productivity and ecosystem respiration data were flagged most frequently, in particular during rain pulses under dry and hot conditions. This information is useful for modelling and analysing ecohydrological responses. (2) There are elevated flagging frequencies for radiation variables (shortwave, photosynthetically active, and net). This information can improve the interpretation and modelling of ecosystem fluxes with respect to issues in the driver. (3) The majority of long-term sites show temporal discontinuities in the time series of latent energy, net ecosystem exchange, and radiation variables. This should be useful for carefully assessing the results in terms of interannual variations in and trends of ecosystem fluxes. The C2F methodology is flexible for customizing and allows for varying the desired strictness of consistency. We discuss the limitations of the approach that can present starting points for future developments.

Separating the components of ecosystem-scale carbon exchange is crucial in order to develop better models and future predictions of the terrestrial carbon cycle. However, there are several uncertainties and unknowns related to current photosynthesis estimates. In this study, we evaluate four different methods for estimating photosynthesis at a boreal forest at the ecosystem scale, of which two are based on carbon dioxide (CO2) flux measurements and two on carbonyl sulfide (COS) flux measurements. The CO2-based methods use traditional flux partitioning and artificial neural networks to separate the net CO2 flux into respiration and photosynthesis. The COS-based methods make use of a unique 5-year COS flux data set and involve two different approaches to determine the leaf-scale relative uptake ratio of COS and CO2 (LRU), of which one (LRUCAP) was developed in this study. LRUCAP was based on a previously tested stomatal optimization theory (CAP), while LRUPAR was based on an empirical relation to measured radiation. For the measurement period 2013–2017, the artificial neural network method gave a GPP estimate very close to that of traditional flux partitioning at all timescales. On average, the COS-based methods gave higher GPP estimates than the CO2-based estimates on daily (23 % and 7 % higher, using LRUPAR and LRUCAP, respectively) and monthly scales (20 % and 3 % higher), as well as a higher cumulative sum over 3 months in all years (on average 25 % and 3 % higher). LRUCAP was higher than LRU estimated from chamber measurements at high radiation, leading to underestimation of midday GPP relative to other GPP methods. In general, however, use of LRUCAP gave closer agreement with CO2-based estimates of GPP than use of LRUPAR. When extended to other sites, LRUCAP may be more robust than LRUPAR because it is based on a physiological model whose parameters can be estimated from simple measurements or obtained from the literature. In contrast, the empirical radiation relation in LRUPAR may be more site-specific. However, this requires further testing at other measurement sites.

Carbon (C) and greenhouse gas (GHG) research has traditionally required data collection and analysis using advanced and often expensive instruments, complex and proprietary software, and highly specialized research technicians. Partly as a result, relatively little C and GHG research has been conducted in resource-constrained developing countries. At the same time, these are often the same countries and regions in which climate change impacts will likely be strongest and in which major science uncertainties are centered, given the importance of dryland and tropical systems to the global C cycle. Increasingly, scientific communities have adopted appropriate technology and approach (AT&A) for C and GHG research, which focuses on low-cost and low-technology instruments, open-source software and data, and participatory and networking-based research approaches. Adopting AT&A can mean acquiring data with fewer technical constraints and lower economic burden and is thus a strategy for enhancing C and GHG research in developing countries. However, AT&A can have higher uncertainties; these can often be mitigated by carefully designing experiments, providing clear protocols for data collection, and monitoring and validating the quality of obtained data. For implementing this approach in developing countries, it is first necessary to recognize the scientific and moral importance of AT&A. At the same time, new AT&A techniques should be identified and further developed. All these processes should be promoted in collaboration with local researchers and through training local staff and encouraged for wide use and further innovation in developing countries.

Severe droughts are expected to become more frequent and persistent. However, their effect on autumn leaf senescence, a key process for deciduous trees and ecosystem functioning, is currently unclear. We hypothesized that (I) severe drought advances the onset of autumn leaf senescence in temperate deciduous trees and (II) tree species show different dynamics of autumn leaf senescence under drought. We tested these hypotheses using a manipulative experiment on beech saplings and 3 years of monitoring mature beech, birch and oak trees in Belgium. The autumn leaf senescence was derived from the seasonal pattern of the chlorophyll content index and the loss of canopy greenness using generalized additive models and piecewise linear regressions. Drought and associated heat stress and increased atmospheric aridity did not affect the onset of autumn leaf senescence in both saplings and mature trees, even if the saplings showed a high mortality and the mature trees an advanced loss of canopy greenness. We did not observe major differences among species. To synthesize, the timing of autumn leaf senescence appears conservative across years and species and even independent of drought, heat and increased atmospheric aridity. Therefore, to study autumn senescence and avoid confusion among studies, seasonal chlorophyll dynamics and loss of canopy greenness should be considered separately.

In the last 20 years, the FLUXNET network provided unique measurements of CO2, energy and other greenhouse gas exchanges between ecosystems and atmosphere measured with the eddy covariance technique. These data have been widely used in different and heterogeneous applications, and FLUXNET became a reference source of information not only for ecological studies but also in modeling and remote sensing applications. The data are, in general, collected, processed and shared by regional networks or by single sites, and for this reason it is difficult for users interested in analyses involving multiple sites to easily access a coherent and standardized dataset. For this reason, periodic FLUXNET collections have been released in the last 15 years, every 5 to 10 years, with data standardized and shared under the same data use policy. However, the new tools available for data analysis and the need to constantly monitor the relations between ecosystem behavior and climate change require a reorganization of FLUXNET in order to increase the data interoperability, reduce the delay in the data sharing and facilitate the data use, all this while keeping in mind the great effort made by the site teams to collect these unique data and respecting the different regional and national network organizations and data policies. Here a proposal for a new organization of FLUXNET is presented with the aim of stimulating a discussion for the needed developments. In this new scheme, the regional and national networks become the pillars of the global initiative, organizing clusters and becoming responsible for the processing, preparation and distribution of datasets that users will be able to access in real time and with a machine-to-machine tool, obtaining always the most updated collection possible but keeping a high standardization and common data policy. This will also lead to an increase in the FAIRness (Findability, Accessibility, Interoperability and Reusability) of the FLUXNET data that will ensure a larger impact of the unique data produced and a proper data management and traceability.

The sources of systematic error responsible for introducing significant biases in the eddy covariance (EC) flux computation are manifold, and their correct identification is made difficult by the lack of reference values, by the complex stochastic dynamics, and by the high level of noise characterizing raw data. This work contributes to overcoming such challenges by introducing an innovative strategy for EC data cleaning. The proposed strategy includes a set of tests aimed at detecting the presence of specific sources of systematic error, as well as an outlier detection procedure aimed at identifying aberrant flux values. Results from tests and outlier detection are integrated in such a way as to leave a large degree of flexibility in the choice of tests and of test threshold values, ensuring scalability of the whole process. The selection of best performing tests was carried out by means of Monte Carlo experiments, whereas the impact on real data was evaluated on data distributed by the Integrated Carbon Observation System (ICOS) research infrastructure. Results evidenced that the proposed procedure leads to an effective cleaning of EC flux data, avoiding the use of subjective criteria in the decision rule that specifies whether to retain or reject flux data of dubious quality. We expect that the proposed data cleaning procedure can serve as a basis towards a unified quality control strategy for EC datasets, in particular in centralized data processing pipelines where the use of robust and automated routines ensuring results reproducibility constitutes an essential prerequisite.

FLUXNET comprises globally distributed eddy-covariance-based estimates of carbon fluxes between the biosphere and the atmosphere. Since eddy covariance flux towers have a relatively small footprint and are distributed unevenly across the world, upscaling the observations is necessary to obtain global-scale estimates of biosphere–atmosphere exchange. Based on cross-consistency checks with atmospheric inversions, sun-induced fluorescence (SIF) and dynamic global vegetation models (DGVMs), here we provide a systematic assessment of the latest upscaling efforts for gross primary production (GPP) and net ecosystem exchange (NEE) of the FLUXCOM initiative, where different machine learning methods, forcing data sets and sets of predictor variables were employed. Spatial patterns of mean GPP are consistent across FLUXCOM and DGVM ensembles (R2>0.94 at 1∘ spatial resolution) while the majority of DGVMs show, for 70 % of the land surface, values outside the FLUXCOM range. Global mean GPP magnitudes for 2008–2010 from FLUXCOM members vary within 106 and 130 PgC yr−1 with the largest uncertainty in the tropics. Seasonal variations in independent SIF estimates agree better with FLUXCOM GPP (mean global pixel-wise R2∼0.75) than with GPP from DGVMs (mean global pixel-wise R2∼0.6). Seasonal variations in FLUXCOM NEE show good consistency with atmospheric inversion-based net land carbon fluxes, particularly for temperate and boreal regions (R2>0.92). Interannual variability of global NEE in FLUXCOM is underestimated compared to inversions and DGVMs. The FLUXCOM version which also uses meteorological inputs shows a strong co-variation in interannual patterns with inversions (R2=0.87 for 2001–2010). Mean regional NEE from FLUXCOM shows larger uptake than inversion and DGVM-based estimates, particularly in the tropics with discrepancies of up to several hundred grammes of carbon per square metre per year. These discrepancies can only partly be reconciled by carbon loss pathways that are implicit in inversions but not captured by the flux tower measurements such as carbon emissions from fires and water bodies. We hypothesize that a combination of systematic biases in the underlying eddy covariance data, in particular in tall tropical forests, and a lack of site history effects on NEE in FLUXCOM are likely responsible for the too strong tropical carbon sink estimated by FLUXCOM. Furthermore, as FLUXCOM does not account for CO2 fertilization effects, carbon flux trends are not realistic. Overall, current FLUXCOM estimates of mean annual and seasonal cycles of GPP as well as seasonal NEE variations provide useful constraints of global carbon cycling, while interannual variability patterns from FLUXCOM are valuable but require cautious interpretation. Exploring the diversity of Earth observation data and of machine learning concepts along with improved quality and quantity of flux tower measurements will facilitate further improvements of the FLUXCOM approach overall.

Modern eddy covariance (EC) systems collect high-frequency data (10–20 Hz) via digital outputs of instruments. This is an important evolution with respect to the traditional and widely used mixed analog/digital systems, as fully digital systems help overcome the traditional limitations of transmission reliability, data quality, and completeness of the datasets.However, fully digital acquisition introduces a new problem for guaranteeing data synchronicity when the clocks of the involved devices themselves cannot be synchronized, which is often the case with instruments providing data via serial or Ethernet connectivity in a streaming mode. In this paper, we suggest that, when assembling EC systems in-house, aspects related to timing issues need to be carefully considered to avoid significant flux biases.By means of a simulation study, we found that, in most cases, random timing errors can safely be neglected, as they do not impact fluxes significantly. At the same time, systematic timing errors potentially arising in asynchronous systems can effectively act as filters leading to significant flux underestimations, as large as 10 %, by means of attenuation of high-frequency flux contributions. We characterized the transfer function of such filters as a function of the error magnitude and found cutoff frequencies as low as 1 Hz, implying that synchronization errors can dominate high-frequency attenuations in open- and enclosed-path EC systems. In most cases, such timing errors neither be detected nor characterized a posteriori. Therefore, it is important to test the ability of traditional and prospective EC data logging systems to assure the required synchronicity and propose a procedure to implement such a test relying on readily available equipment.

Extreme hydrometeorological conditions typically impact ecophysiological processes on land. Satellite-based observations of the terrestrial biosphere provide an important reference for detecting and describing the spatiotemporal development of such events. However, in-depth investigations of ecological processes during extreme events require additional in situ observations. The question is whether the density of existing ecological in situ networks is sufficient for analysing the impact of extreme events, and what are expected event detection rates of ecological in situ networks of a given size. To assess these issues, we build a baseline of extreme reductions in the fraction of absorbed photosynthetically active radiation (FAPAR), identified by a new event detection method tailored to identify extremes of regional relevance. We then investigate the event detection success rates of hypothetical networks of varying sizes. Our results show that large extremes can be reliably detected with relatively small networks, but also reveal a linear decay of detection probabilities towards smaller extreme events in log–log space. For instance, networks with ≈ 100 randomly placed sites in Europe yield a ≥ 90 % chance of detecting the eight largest (typically very large) extreme events; but only a ≥ 50 % chance of capturing the 39 largest events. These findings are consistent with probability-theoretic considerations, but the slopes of the decay rates deviate due to temporal autocorrelation and the exact implementation of the extreme event detection algorithm. Using the examples of AmeriFlux and NEON, we then investigate to what degree ecological in situ networks can capture extreme events of a given size. Consistent with our theoretical considerations, we find that today's systematically designed networks (i.e. NEON) reliably detect the largest extremes, but that the extreme event detection rates are not higher than would be achieved by randomly designed networks. Spatio-temporal expansions of ecological in situ monitoring networks should carefully consider the size distribution characteristics of extreme events if the aim is also to monitor the impacts of such events in the terrestrial biosphere.

Understanding the global carbon (C) cycle is of crucial importance to map current and future climate dynamics relative to global environmental change. A full characterization of C cycling requires detailed information on spatiotemporal patterns of surface–atmosphere fluxes. However, relevant C cycle observations are highly variable in their coverage and reporting standards. Especially problematic is the lack of integration of the carbon dioxide (CO2) exchange of the ocean, inland freshwaters and the land surface with the atmosphere. Here we adopt a data-driven approach to synthesize a wide range of observation-based spatially explicit surface–atmosphere CO2 fluxes from 2001 to 2010, to identify the state of today's observational opportunities and data limitations. The considered fluxes include net exchange of open oceans, continental shelves, estuaries, rivers, and lakes, as well as CO2 fluxes related to net ecosystem productivity, fire emissions, loss of tropical aboveground C, harvested wood and crops, as well as fossil fuel and cement emissions. Spatially explicit CO2 fluxes are obtained through geostatistical and/or remote-sensing-based upscaling, thereby minimizing biophysical or biogeochemical assumptions encoded in process-based models. We estimate a bottom-up net C exchange (NCE) between the surface (land, ocean, and coastal areas) and the atmosphere. Though we provide also global estimates, the primary goal of this study is to identify key uncertainties and observational shortcomings that need to be prioritized in the expansion of in situ observatories. Uncertainties for NCE and its components are derived using resampling. In many regions, our NCE estimates agree well with independent estimates from other sources such as process-based models and atmospheric inversions. This holds for Europe (mean ± 1 SD: 0.8 ± 0.1 PgC yr−1, positive numbers are sources to the atmosphere), Russia (0.1 ± 0.4 PgC yr−1), East Asia (1.6 ± 0.3 PgC yr−1), South Asia (0.3 ± 0.1 PgC yr−1), Australia (0.2 ± 0.3 PgC yr−1), and most of the Ocean regions. Our NCE estimates give a likely too large CO2 sink in tropical areas such as the Amazon, Congo, and Indonesia. Overall, and because of the overestimated CO2 uptake in tropical lands, our global bottom-up NCE amounts to a net sink of −5.4 ± 2.0 PgC yr−1. By contrast, the accurately measured mean atmospheric growth rate of CO2 over 2001–2010 indicates that the true value of NCE is a net CO2 source of 4.3 ± 0.1 PgC yr−1. This mismatch of nearly 10 PgC yr−1 highlights observational gaps and limitations of data-driven models in tropical lands, but also in North America. Our uncertainty assessment provides the basis for setting priority regions where to increase carbon observations in the future. High on the priority list are tropical land regions, which suffer from a lack of in situ observations. Second, extensive pCO2 data are missing in the Southern Ocean. Third, we lack observations that could enable seasonal estimates of shelf, estuary, and inland water–atmosphere C exchange. Our consistent derivation of data uncertainties could serve as prior knowledge in multicriteria optimization such as the Carbon Cycle Data Assimilation System (CCDAS) and atmospheric inversions, without over- or under-stating bottom-up data credibility. In the future, NCE estimates of carbon sinks could be aggregated at national scale to compare with the official national inventories of CO2 fluxes in the land use, land use change, and forestry sector, upon which future emission reductions are proposed.

Spatio-temporal fields of land–atmosphere fluxes derived from data-driven models can complement simulations by process-based land surface models. While a number of strategies for empirical models with eddy-covariance flux data have been applied, a systematic intercomparison of these methods has been missing so far. In this study, we performed a cross-validation experiment for predicting carbon dioxide, latent heat, sensible heat and net radiation fluxes across different ecosystem types with 11 machine learning (ML) methods from four different classes (kernel methods, neural networks, tree methods, and regression splines). We applied two complementary setups: (1) 8-day average fluxes based on remotely sensed data and (2) daily mean fluxes based on meteorological data and a mean seasonal cycle of remotely sensed variables. The patterns of predictions from different ML and experimental setups were highly consistent. There were systematic differences in performance among the fluxes, with the following ascending order: net ecosystem exchange (R2 R2 > 0.6), gross primary production (R2> 0.7), latent heat (R2 > 0.7), sensible heat (R2 > 0.7), and net radiation (R2 > 0.8). The ML methods predicted the across-site variability and the mean seasonal cycle of the observed fluxes very well (R2 > 0.7), while the 8-day deviations from the mean seasonal cycle were not well predicted (R2

The production of bioenergy in Europe is one of the strategies conceived to reduce greenhouse gas (GHG) emissions. The suitability of the land use change from a cropland (REF site) to a short-rotation coppice plantation of hybrid poplar (SRC site) was investigated by comparing the GHG budgets of these two systems over 24 months in Viterbo, Italy. This period corresponded to a single rotation of the SRC site. The REF site was a crop rotation between grassland and winter wheat, i.e. the same management of the SRC site before the conversion to short-rotation coppice. Eddy covariance measurements were carried out to quantify the net ecosystem exchange of CO2 (FCO2), whereas chambers were used to measure N2O and CH4 emissions from soil. The measurements began 2 years after the conversion of arable land to SRC so that an older poplar plantation was used to estimate the soil organic carbon (SOC) loss due to SRC establishment and to estimate SOC recovery over time. Emissions from tractors and from production and transport of agricultural inputs (FMAN) were modelled. A GHG emission offset, due to the substitution of natural gas with SRC biomass, was credited to the GHG budget of the SRC site. Emissions generated by the use of biomass (FEXP) were also considered. Suitability was finally assessed by comparing the GHG budgets of the two sites. CO2 uptake was 3512 ± 224 g CO2 m−2 at the SRC site in 2 years, and 1838 ± 107 g CO2 m−2 at the REF site. FEXP was equal to 1858 ± 240 g CO2 m−2 at the REF site, thus basically compensating for FCO2, while it was 1118 ± 521 g CO2 m−2 at the SRC site. The SRC site could offset 379.7 ± 175.1 g CO2eq m−2 from fossil fuel displacement. Soil CH4 and N2O fluxes were negligible. FMAN made up 2 and 4 % in the GHG budgets of SRC and REF sites respectively, while the SOC loss was 455 ± 524 g CO2 m−2 in 2 years. Overall, the REF site was close to neutrality from a GHG perspective (156 ± 264 g CO2eq m−2), while the SRC site was a net sink of 2202 ± 792 g CO2eq m−2. In conclusion the experiment led to a positive evaluation from a GHG viewpoint of the conversion of cropland to bioenergy SRC.

Plant phenological development is orchestrated through subtle changes in photoperiod, temperature, soil moisture and nutrient availability. Presently, the exact timing of plant development stages and their response to climate and management practices are crudely represented in land surface models. As visual observations of phenology are laborious, there is a need to supplement long-term observations with automated techniques such as those provided by digital repeat photography at high temporal and spatial resolution. We present the first synthesis from a growing observational network of digital cameras installed on towers across Europe above deciduous and evergreen forests, grasslands and croplands, where vegetation and atmosphere CO2 fluxes are measured continuously. Using colour indices from digital images and using piecewise regression analysis of time series, we explored whether key changes in canopy phenology could be detected automatically across different land use types in the network. The piecewise regression approach could capture the start and end of the growing season, in addition to identifying striking changes in colour signals caused by flowering and management practices such as mowing. Exploring the dates of green-up and senescence of deciduous forests extracted by the piecewise regression approach against dates estimated from visual observations, we found that these phenological events could be detected adequately (RMSE < 8 and 11 days for leaf out and leaf fall, respectively). We also investigated whether the seasonal patterns of red, green and blue colour fractions derived from digital images could be modelled mechanistically using the PROSAIL model parameterised with information of seasonal changes in canopy leaf area and leaf chlorophyll and carotenoid concentrations. From a model sensitivity analysis we found that variations in colour fractions, and in particular the late spring `green hump' observed repeatedly in deciduous broadleaf canopies across the network, are essentially dominated by changes in the respective pigment concentrations. Using the model we were able to explain why this spring maximum in green signal is often observed out of phase with the maximum period of canopy photosynthesis in ecosystems across Europe. Coupling such quasi-continuous digital records of canopy colours with co-located CO2 flux measurements will improve our understanding of how changes in growing season length are likely to shape the capacity of European ecosystems to sequester CO2 in the future.

Recent studies have suggested the potential importance of abiotic degradation in arid ecosystems. In this study, the role of photo- and thermal degradation in ecosystem CO2 and CO exchange is assessed. A field experiment was performed in Italy using an FTIR-spectrometer (Fourier Transform Infrared) coupled to a flux gradient system and to flux chambers. In a laboratory experiment, field samples were exposed to different temperatures and radiation intensities. No photodegradation-induced CO2 and CO fluxes of in literature suggested magnitudes were found in the field nor in the laboratory study. In the laboratory, we measured CO2 and CO fluxes that were derived from thermal degradation. In the field experiment, CO uptake and emission have been measured and are proposed to be a result of biological uptake and abiotic thermal degradation-production. We suggest that previous studies, addressing direct photodegradation, have overestimated the role of photodegradation and observed fluxes might be due to thermal degradation, which is an indirect effect of radiation. The potential importance of abiotic decomposition in the form of thermal degradation, especially for arid regions, should be considered in future studies.

In this paper we explore the skill of hyperspectral reflectance measurements and vegetation indices (VIs) derived from these in estimating carbon dioxide (CO2) fluxes of grasslands. Hyperspectral reflectance data, CO2 fluxes and biophysical parameters were measured at three grassland sites located in European mountain regions using standardized protocols. The relationships between CO2 fluxes, ecophysiological variables, traditional VIs and VIs derived using all two-band combinations of wavelengths available from the whole hyperspectral data space were analysed. We found that VIs derived from hyperspectral data generally explained a large fraction of the variability in the investigated dependent variables but differed in their ability to estimate midday and daily average CO2 fluxes and various derived ecophysiological parameters. Relationships between VIs and CO2 fluxes and ecophysiological parameters were site-specific, likely due to differences in soils, vegetation parameters and environmental conditions. Chlorophyll and water-content-related VIs explained the largest fraction of variability in most of the dependent variables. Band selection based on a combination of a genetic algorithm with random forests (GA–rF) confirmed that it is difficult to select a universal band region suitable across the investigated ecosystems. Our findings have major implications for upscaling terrestrial CO2 fluxes to larger regions and for remote- and proximal-sensing sampling and analysis strategies and call for more cross-site synthesis studies linking ground-based spectral reflectance with ecosystem-scale CO2 fluxes.

This paper reports a comparison between large-scale simulations of three different land surface models (LSMs), ORCHIDEE, ISBA-A-gs and CTESSEL, forced with the same meteorological data, and compared with the carbon fluxes measured at 32 eddy covariance (EC) flux tower sites in Europe. The results show that the three simulations have the best performance for forest sites and the poorest performance for cropland and grassland sites. In addition, the three simulations have difficulties capturing the seasonality of Mediterranean and sub-tropical biomes, characterized by dry summers. This reduced simulation performance is also reflected in deficiencies in diagnosed light-use efficiency (LUE) and vapour pressure deficit (VPD) dependencies compared to observations. Shortcomings in the forcing data may also play a role. These results indicate that more research is needed on the LUE and VPD functions for Mediterranean and sub-tropical biomes. Finally, this study highlights the importance of correctly representing phenology (i.e. leaf area evolution) and management (i.e. rotation–irrigation for cropland, and grazing–harvesting for grassland) to simulate the carbon dynamics of European ecosystems and the importance of ecosystem-level observations in model development and validation.

Errors in gas concentration measurements by infrared gas analysers can occur during eddy-covariance campaigns, associated with actual or apparent instrumental drifts or biases due to thermal expansion, dirt contamination, aging of components or errors in field operations. If occurring on long timescales (hours to days), these errors are normally ignored during flux computation, under the assumption that errors in mean gas concentrations do not affect the estimation of turbulent fluctuations and, hence, of covariances. By analysing instrument theory of operation, and using numerical simulations and field data, we show that this is not the case for instruments with curvilinear calibrations; we further show that if not appropriately accounted for, concentration biases can lead to roughly proportional systematic flux errors, where the fractional errors in fluxes are about 30–40% the fractional errors in concentrations. We quantify these errors and characterize their dependency on main determinants. We then propose a correction procedure that largely – potentially completely – eliminates these errors. The correction, to be applied during flux computation, is based on knowledge of instrument calibration curves and on field or laboratory calibration data. Finally, we demonstrate the occurrence of such errors and validate the correction procedure by means of a field experiment, and accordingly provide recommendations for in situ operations.

This paper, developed under the framework of the RECCAP initiative, aims at providing improved estimates of the carbon and GHG (CO2, CH4 and N2O) balance of continental Africa. The various components and processes of the African carbon and GHG budget are considered, existing data reviewed, and new data from different methodologies (inventories, ecosystem flux measurements, models, and atmospheric inversions) presented. Uncertainties are quantified and current gaps and weaknesses in knowledge and monitoring systems described in order to guide future requirements. The majority of results agree that Africa is a small sink of carbon on an annual scale, with an average value of −0.61 ± 0.58 Pg C yr−1. Nevertheless, the emissions of CH4 and N2O may turn Africa into a net source of radiative forcing in CO2 equivalent terms. At sub-regional level, there is significant spatial variability in both sources and sinks, due to the diversity of biomes represented and differences in the degree of anthropic impacts. Southern Africa is the main source region; while central Africa, with its evergreen tropical forests, is the main sink. Emissions from land-use change in Africa are significant (around 0.32 ± 0.05 Pg C yr−1), even higher than the fossil fuel emissions: this is a unique feature among all the continents. There could be significant carbon losses from forest land even without deforestation, resulting from the impact of selective logging. Fires play a significant role in the African carbon cycle, with 1.03 ± 0.22 Pg C yr−1 of carbon emissions, and 90% originating in savannas and dry woodlands. A large portion of the wild fire emissions are compensated by CO2 uptake during the growing season, but an uncertain fraction of the emission from wood harvested for domestic use is not. Most of these fluxes have large interannual variability, on the order of ±0.5 Pg C yr−1 in standard deviation, accounting for around 25% of the year-to-year variation in the global carbon budget. Despite the high uncertainty, the estimates provided in this paper show the important role that Africa plays in the global carbon cycle, both in terms of absolute contribution, and as a key source of interannual variability.

Networks that merge and harmonise eddy-covariance measurements from many different parts of the world have become an important observational resource for ecosystem science. Empirical algorithms have been developed which combine direct observations of the net ecosystem exchange of carbon dioxide with simple empirical models to disentangle photosynthetic (GPP) and respiratory fluxes (Reco). The increasing use of these estimates for the analysis of climate sensitivities, model evaluation and calibration demands a thorough understanding of assumptions in the analysis process and the resulting uncertainties of the partitioned fluxes. The semi-empirical models used in flux partitioning algorithms require temperature observations as input, but as respiration takes place in many parts of an ecosystem, it is unclear which temperature input – air, surface, bole, or soil at a specific depth – should be used. This choice is a source of uncertainty and potential biases. In this study, we analysed the correlation between different temperature observations and nighttime NEE (which equals nighttime respiration) across FLUXNET sites to understand the potential of the different temperature observations as input for the flux partitioning model. We found that the differences in the correlation between different temperature data streams and nighttime NEE are small and depend on the selection of sites. We investigated the effects of the choice of the temperature data by running two flux partitioning algorithms with air and soil temperature. We found the time lag (phase shift) between air and soil temperatures explains the differences in the GPP and Reco estimates when using either air or soil temperatures for flux partitioning. The impact of the source of temperature data on other derived ecosystem parameters was estimated, and the strongest impact was found for the temperature sensitivity. Overall, this study suggests that the choice between soil or air temperature must be made on site-by-site basis by analysing the correlation between temperature and nighttime NEE. We recommend using an ensemble of estimates based on different temperature observations to account for the uncertainty due to the choice of temperature and to assure the robustness of the temporal patterns of the derived variables.

Globally, terrestrial ecosystems have absorbed about 30% of anthropogenic greenhouse gas emissions over the period 2000–2007 and inter-hemispheric gradients indicate that a significant fraction of terrestrial carbon sequestration must be north of the Equator. We present a compilation of the CO2, CO, CH4 and N2O balances of Europe following a dual constraint approach in which (1) a land-based balance derived mainly from ecosystem carbon inventories and (2) a land-based balance derived from flux measurements are compared to (3) the atmospheric data-based balance derived from inversions constrained by measurements of atmospheric GHG (greenhouse gas) concentrations. Good agreement between the GHG balances based on fluxes (1294 ± 545 Tg C in CO2-eq yr−1), inventories (1299 ± 200 Tg C in CO2-eq yr−1) and inversions (1210 ± 405 Tg C in CO2-eq yr−1) increases our confidence that the processes underlying the European GHG budget are well understood and reasonably sampled. However, the uncertainty remains large and largely lacks formal estimates. Given that European net land to atmosphere exchanges are determined by a few dominant fluxes, the uncertainty of these key components needs to be formally estimated before efforts could be made to reduce the overall uncertainty. The net land-to-atmosphere flux is a net source for CO2, CO, CH4 and N2O, because the anthropogenic emissions by far exceed the biogenic sink strength. The dual-constraint approach confirmed that the European biogenic sink removes as much as 205 ± 72 Tg C yr−1 from fossil fuel burning from the atmosphere. However, This C is being sequestered in both terrestrial and inland aquatic ecosystems. If the C-cost for ecosystem management is taken into account, the net uptake of ecosystems is estimated to decrease by 45% but still indicates substantial C-sequestration. However, when the balance is extended from CO2 towards the main GHGs, C-uptake by terrestrial and aquatic ecosystems is offset by emissions of non-CO2 GHGs. As such, the European ecosystems are unlikely to contribute to mitigating the effects of climate change.

The allocation of carbon (C) taken up by the tree canopy for respiration and production of tree organs with different construction and maintenance costs, life span and decomposition rate, crucially affects the residence time of C in forests and their C cycling rate. The carbon-use efficiency, or ratio between net primary production (NPP) and gross primary production (GPP), represents a convenient way to analyse the C allocation at the stand level. In this study, we extend the current knowledge on the NPP-GPP ratio in forests by assessing the temporal variability of the NPP-GPP ratio at interannual (for 8 years) and seasonal (for 1 year) scales for a young temperate beech stand, reporting dynamics for both leaves and woody organs, in particular stems. NPP was determined with biometric methods/litter traps, whereas the GPP was estimated via the eddy covariance micrometeorological technique. The interannual variability of the proportion of C allocated to leaf NPP, wood NPP and leaf plus wood NPP (on average 11% yr−1, 29% yr−1 and 39% yr−1, respectively) was significant among years with up to 12% yr−1 variation in NPP-GPP ratio. Studies focusing on the comparison of NPP-GPP ratio among forests and models using fixed allocation schemes should take into account the possibility of such relevant interannual variability. Multiple linear regressions indicated that the NPP-GPP ratio of leaves and wood significantly correlated with environmental conditions. Previous year drought and air temperature explained about half of the NPP-GPP variability of leaves and wood, respectively, whereas the NPP-GPP ratio was not decreased by severe drought, with large NPP-GPP ratio on 2003 due mainly to low GPP. During the period between early May and mid June, the majority of GPP was allocated to leaf and stem NPP, whereas these sinks were of little importance later on. Improved estimation of seasonal GPP and of the contribution of previous-year reserves to stem growth, as well as reduction of data uncertainty, will be of relevance to increase the accuracy of the seasonal assessment of the NPP-GPP ratio in forests.

Several studies sustained the possibility that a photochemical reflectance index (PRI) directly obtained from satellite data can be used as a proxy for ecosystem light use efficiency (LUE) in diagnostic models of gross primary productivity. This modelling approach would avoid the complications that are involved in using meteorological data as constraints for a fixed maximum LUE. However, no unifying model predicting LUE across climate zones and time based on MODIS PRI has been published to date. In this study, we evaluate the effectiveness with which MODIS-based PRI can be used to estimate ecosystem light use efficiency at study sites of different plant functional types and vegetation densities. Our objective is to examine if known limitations such as dependence on viewing and illumination geometry can be overcome and a single PRI-based model of LUE (i.e. based on the same reference band) can be applied under a wide range of conditions. Furthermore, we were interested in the effect of using different faPAR (fraction of absorbed photosynthetically active radiation) products on the in-situ LUE used as ground truth and thus on the whole evaluation exercise. We found that estimating LUE at site-level based on PRI reduces uncertainty compared to the approaches relying on a maximum LUE reduced by minimum temperature and vapour pressure deficit. Despite the advantages of using PRI to estimate LUE at site-level, we could not establish an universally applicable light use efficiency model based on MODIS PRI. Models that were optimised for a pool of data from several sites did not perform well.

Quantification of ecosystem carbon pools is a fundamental requirement for estimating carbon fluxes and for addressing the dynamics and responses of the terrestrial carbon cycle to environmental drivers. The initial estimates of carbon pools in terrestrial carbon cycle models often rely on the ecosystem steady state assumption, leading to initial equilibrium conditions. In this study, we investigate how trends and inter-annual variability of net ecosystem fluxes are affected by initial non-steady state conditions. Further, we examine how modeled ecosystem responses induced exclusively by the model drivers can be separated from the initial conditions. For this, the Carnegie-Ames-Stanford Approach (CASA) model is optimized at set of European eddy covariance sites, which support the parameterization of regional simulations of ecosystem fluxes for the Iberian Peninsula, between 1982 and 2006. The presented analysis stands on a credible model performance for a set of sites, that represent generally well the plant functional types and selected descriptors of climate and phenology present in the Iberian region – except for a limited Northwestern area. The effects of initial conditions on inter-annual variability and on trends, results mostly from the recovery of pools to equilibrium conditions; which control most of the inter-annual variability (IAV) and both the magnitude and sign of most of the trends. However, by removing the time series of pure model recovery from the time series of the overall fluxes, we are able to retrieve estimates of inter-annual variability and trends in net ecosystem fluxes that are quasi-independent from the initial conditions. This approach reduced the sensitivity of the net fluxes to initial conditions from 47% and 174% to −3% and 7%, for strong initial sink and source conditions, respectively. With the aim to identify and improve understanding of the component fluxes that drive the observed trends, the net ecosystem production (NEP) trends are decomposed into net primary production (NPP) and heterotrophic respiration (RH) trends. The majority (~97%) of the positive trends in NEP is observed in regions where both NPP and RH fluxes show significant increases, although the magnitude of NPP trends is higher. Analogously, ~83% of the negative trends in NEP are also associated with negative trends in NPP. The spatial patterns of NPP trends are mainly explained by the trends in fAPAR (r=0.79) and are only marginally explained by trends in temperature and water stress scalars (r=0.10 and r=0.25, respectively). Further, we observe the significant role of substrate availability (r=0.25) and temperature (r=0.23) in explaining the spatial patterns of trends in RH. These results highlight the role of primary production in driving ecosystem fluxes. Overall, our study illustrates an approach for removing the confounding effects of initial conditions and emphasizes the need to decompose the ecosystem fluxes into its components and drivers for more mechanistic interpretations of modeling results. We expect that our results are not only specific for the CASA model since it incorporates concepts of ecosystem functioning and modeling assumptions common to biogeochemical models. A direct implication of these results is the ability of this approach to detect climate and phenology induced trends regardless of the initial conditions.

Eddy covariance data are increasingly used to estimate parameters of ecosystem models. For proper maximum likelihood parameter estimates the error structure in the observed data has to be fully characterized. In this study we propose a method to characterize the random error of the eddy covariance flux data, and analyse error distribution, standard deviation, cross- and autocorrelation of CO2 and H2O flux errors at four different European eddy covariance flux sites. Moreover, we examine how the treatment of those errors and additional systematic errors influence statistical estimates of parameters and their associated uncertainties with three models of increasing complexity – a hyperbolic light response curve, a light response curve coupled to water fluxes and the SVAT scheme BETHY. In agreement with previous studies we find that the error standard deviation scales with the flux magnitude. The previously found strongly leptokurtic error distribution is revealed to be largely due to a superposition of almost Gaussian distributions with standard deviations varying by flux magnitude. The crosscorrelations of CO2 and H2O fluxes were in all cases negligible (R2 below 0.2), while the autocorrelation is usually below 0.6 at a lag of 0.5 h and decays rapidly at larger time lags. This implies that in these cases the weighted least squares criterion yields maximum likelihood estimates. To study the influence of the observation errors on model parameter estimates we used synthetic datasets, based on observations of two different sites. We first fitted the respective models to observations and then added the random error estimates described above and the systematic error, respectively, to the model output. This strategy enables us to compare the estimated parameters with true parameters. We illustrate that the correct implementation of the random error standard deviation scaling with flux magnitude significantly reduces the parameter uncertainty and often yields parameter retrievals that are closer to the true value, than by using ordinary least squares. The systematic error leads to systematically biased parameter estimates, but its impact varies by parameter. The parameter uncertainty slightly increases, but the true parameter is not within the uncertainty range of the estimate. This means that the uncertainty is underestimated with current approaches that neglect selective systematic errors in flux data. Hence, we conclude that potential systematic errors in flux data need to be addressed more thoroughly in data assimilation approaches since otherwise uncertainties will be vastly underestimated.

Globally, the year 2003 is associated with one of the largest atmospheric CO2 rises on record. In the same year, Europe experienced an anomalously strong flux of CO2 from the land to the atmosphere associated with an exceptionally dry and hot summer in Western and Central Europe. In this study we analyze the magnitude of this carbon flux anomaly and key driving ecosystem processes using simulations of seven terrestrial ecosystem models of different complexity and types (process-oriented and diagnostic). We address the following questions: (1) how large were deviations in the net European carbon flux in 2003 relative to a short-term baseline (1998–2002) and to longer-term variations in annual fluxes (1980 to 2005), (2) which European regions exhibited the largest changes in carbon fluxes during the growing season 2003, and (3) which ecosystem processes controlled the carbon balance anomaly . In most models the prominence of 2003 anomaly in carbon fluxes declined with lengthening of the reference period from one year to 16 years. The 2003 anomaly for annual net carbon fluxes ranged between 0.35 and –0.63 Pg C for a reference period of one year and between 0.17 and –0.37 Pg C for a reference period of 16 years for the whole Europe. In Western and Central Europe, the anomaly in simulated net ecosystem productivity (NEP) over the growing season in 2003 was outside the 1σ variance bound of the carbon flux anomalies for 1980–2005 in all models. The estimated anomaly in net carbon flux ranged between –42 and –158 Tg C for Western Europe and between 24 and –129 Tg C for Central Europe depending on the model used. All models responded to a dipole pattern of the climate anomaly in 2003. In Western and Central Europe NEP was reduced due to heat and drought. In contrast, lower than normal temperatures and higher air humidity decreased NEP over Northeastern Europe. While models agree on the sign of changes in simulated NEP and gross primary productivity in 2003 over Western and Central Europe, models diverge in the estimates of anomalies in ecosystem respiration. Except for two process models which simulate respiration increase, most models simulated a decrease in ecosystem respiration in 2003. The diagnostic models showed a weaker decrease in ecosystem respiration than the process-oriented models. Based on the multi-model simulations we estimated the total carbon flux anomaly over the 2003 growing season in Europe to range between –0.02 and –0.27 Pg C relative to the net carbon flux in 1998–2002.

We applied a site evaluation approach combining Lagrangian Stochastic footprint modeling with a quality assessment approach for eddy-covariance data to 25 forested sites of the CarboEurope-IP network. The analysis addresses the spatial representativeness of the flux measurements, instrumental effects on data quality, spatial patterns in the data quality, and the performance of the coordinate rotation method. Our findings demonstrate that application of a footprint filter could strengthen the CarboEurope-IP flux database, since only one third of the sites is situated in truly homogeneous terrain. Almost half of the sites experience a significant reduction in eddy-covariance data quality under certain conditions, though these effects are mostly constricted to a small portion of the dataset. Reductions in data quality of the sensible heat flux are mostly induced by characteristics of the surrounding terrain, while the latent heat flux is subject to instrumentation-related problems. The Planar-Fit coordinate rotation proved to be a reliable tool for the majority of the sites using only a single set of rotation angles. Overall, we found a high average data quality for the CarboEurope-IP network, with good representativeness of the measurement data for the specified target land cover types.

Characterizing ecosystem-atmosphere interactions in terms of carbon and water exchange on different time scales is considered a major challenge in terrestrial biogeochemical cycle research. The respective time series currently comprise an observation period of up to one decade. In this study, we explored whether the observation period is already sufficient to detect cross-relationships between the variables beyond the annual cycle, as they are expected from comparable studies in climatology. We investigated the potential of Singular System Analysis (SSA) to extract arbitrary kinds of oscillatory patterns. The method is completely data adaptive and performs an effective signal to noise separation. We found that most observations (Net Ecosystem Exchange, NEE, Gross Primary Productivity, GPP, Ecosystem Respiration, Reco, Vapor Pressure Deficit, VPD, Latent Heat, LE, Sensible Heat, H, Wind Speed, u, and Precipitation, P) were influenced significantly by low-frequency components (interannual variability). Furthermore, we extracted a set of nontrivial relationships and found clear seasonal hysteresis effects except for the interrelation of NEE with Global Radiation (Rg). SSA provides a new tool for the investigation of these phenomena explicitly on different time scales. Furthermore, we showed that SSA has great potential for eddy covariance data processing, since it can be applied as a novel gap filling approach relying on the temporal correlation structure of the time series structure only.

Steppe ecosystems represent an interesting case in which the assessment of carbon balance may be performed through a cross validation of the eddy covariance measurements against ecological inventory estimates of carbon exchanges (Ehman et al., 2002; Curtis et al., 2002). Indeed, the widespread presence of ideal conditions for the applicability of the eddy covariance technique, as vast and homogeneous grass vegetation cover over flat terrains (Baldocchi, 2003), make steppes a suitable ground to ensure a constrain to flux estimates with independent methodological approaches. We report about the analysis of the carbon cycle of a true steppe ecosystem in southern Siberia during the growing season of 2004 in the framework of the TCOS-Siberia project activities performed by continuous monitoring of CO2 fluxes at ecosystem scale by the eddy covariance method, fortnightly samplings of phytomass, and ingrowth cores extractions for NPP assessment, and weekly measurements of heterotrophic component of soil CO2 effluxes obtained by an experiment of root exclusion. The carbon balance of the monitored natural steppe was, according to micrometeorological measurements, a sink of carbon of 151.7±36.9 g C m−2, cumulated during the growing season from May to September. This result was in agreement with the independent estimate through ecological inventory which yielded a sink of 150.1 g C m−2 although this method was characterized by a large uncertainty (±130%) considering the 95% confidence interval of the estimate. Uncertainties in belowground process estimates account for a large part of the error. Thus, in particular efforts to better quantify the dynamics of root biomass (growth and turnover) have to be undertaken in order to reduce the uncertainties in the assessment of NPP. This assessment should be preferably based on the application of multiple methods, each one characterized by its own merits and flaws.

Eddy covariance technique to measure CO2, water and energy fluxes between biosphere and atmosphere is widely spread and used in various regional networks. Currently more than 250 eddy covariance sites are active around the world measuring carbon exchange at high temporal resolution for different biomes and climatic conditions. In this paper a new standardized set of corrections is introduced and the uncertainties associated with these corrections are assessed for eight different forest sites in Europe with a total of 12 yearly datasets. The uncertainties introduced on the two components GPP (Gross Primary Production) and TER (Terrestrial Ecosystem Respiration) are also discussed and a quantitative analysis presented. Through a factorial analysis we find that generally, uncertainties by different corrections are additive without interactions and that the heuristic u*-correction introduces the largest uncertainty. The results show that a standardized data processing is needed for an effective comparison across biomes and for underpinning inter-annual variability. The methodology presented in this paper has also been integrated in the European database of the eddy covariance measurements.

Knorr et al. (2005) concluded that soil organic carbon pools with longer turnover times are more sensitive to temperature. We show that this conclusion is equivocal, largely dependent on their specific selection of data and does not persist when the data set of Kätterer et al. (1998) is analysed in a more appropriate way. Further, we analyse how statistical properties of the model parameters may interfere with correlative analyses that relate the Q10 of soil respiration with the basal rate, where the latter is taken as a proxy for soil organic matter quality. We demonstrate that negative parameter correlations between Q10-values and base respiration rates are statistically expected and not necessarily provide evidence for a higher temperature sensitivity of low quality soil organic matter. Consequently, we propose it is premature to conclude that stable soil carbon is more sensitive to temperature than labile carbon.