Multi-site calibration and validation of a net ecosystem carbon exchange model for croplands
Introduction
Terrestrial ecosystems play an important role in the global carbon cycle. Photosynthesis by vegetation and respiration from autotrophic and heterotrophic organisms represent the two major carbon fluxes between atmosphere and terrestrial biosphere. Terrestrial ecosystems store large amounts of carbon, and especially soils contain about twice as much carbon as the atmosphere (Rustad et al., 2000). Over 37% of the world’s landmass is agricultural land (FAO Statistical Yearbook, 2014). Thus, carbon fluxes in agroecosystems constitute a significant part of the global carbon cycle. The quantification and prediction of terrestrial carbon sinks and sources and their dynamics, variabilities, and controls are of major importance for climate change research and the optimization of management strategies affecting the ecosystem’s carbon budget (e.g., Baldocchi, 2003, Kuzyakov, 2006, Subke et al., 2006). The net ecosystem exchange (NEE) of carbon dioxide and its two components, gross primary production (GPP) and terrestrial ecosystem respiration (TER), are of particular interest (Suleau et al., 2011, Sus et al., 2010). The total CO2 efflux from soils, one of the major compartments of TER (Moureaux et al., 2008, Suleau et al., 2011), derives from decomposition of soil organic matter and dead plant material by microorganisms, from direct root respiration, and from microbial respiration of root exudates and rhizodepositions (Kuzyakov, 2006, Kuzyakov and Domanski, 2000). In this study, we consider the last two CO2 sources as one sum, and refer to it as “rhizosphere respiration”.
NEE is increasingly being monitored using the eddy covariance (EC) technique, which provides information on net carbon fluxes for a relatively large area with a high temporal resolution (Baldocchi, 2003). This allows to investigate the relation between CO2 efflux and weather conditions or crop development stages (Sus et al., 2010). Due to methodological and technical constraints, significant gaps occur in high-quality EC data, which prohibits direct computation of annual NEE. Gap-filling methods (e.g., Reichstein et al., 2005) and their application with meteorological and EC data overcome this limitation, but e.g., they cannot be used for predictive modeling of carbon balances addressing climate change effects. Alternatively, terrestrial ecosystem models that provide a physical description of processes in the agroecosystem can be used to assess annual NEE sums. An additional advantage of such models is that they allow to quantify interrelations and feedbacks in biogeochemical processes and fluxes of agricultural systems. Mechanistic models like ORCHIDEE–STICS (de Noblet-Ducoudré et al., 2004), DNDC (Li et al., 2005), or SPAc (Sus et al., 2010) were developed for this purpose and have been successfully applied in a number of studies (e.g., Sus et al., 2010, Wattenbach et al., 2010, Wu et al., 2009, Yuan et al., 2012). In most of these studies, the carbon assimilation by plants was captured well by the models, but a significant bias in the simulation of the respiratory fluxes was observed. This inevitably causes systematic errors in the estimation of the overall carbon balance. An improved representation of processes linked to respiration may help to decrease systematic errors and in combination with soil respiration (Rsoil) measurements, it may help to reduce the uncertainty in the estimation of annual NEE. For this purpose, we coupled a one-dimensional soil water, heat, and CO2 flux model (SOILCO2; Šimůnek and Suarez, 1993), a pool concept of soil carbon turnover (RothC; Coleman and Jenkinson, 2008), and a crop growth module (SUCROS; Spitters et al., 1989). In addition, the coupled model, further referred to as AgroC, was extended with routines for root exudation, root decay, as well as for a managed grassland system. The main motivation for the coupling was a more detailed representation of sources and locations of CO2 production, the gas transport in the soil, and the fluxes in the ecosystem.
Various sources of measured data are available for validation, calibration, evaluation, and structural improvement of terrestrial ecosystem models. In the last decade, substantial progress has been made in implementing model-data fusion techniques to make optimal use of available measurements (e.g., Richardson et al., 2010, Sus et al., 2010, Trudinger et al., 2007, Wu et al., 2009, Yuan et al., 2012). Such model-data fusion techniques, including calibration techniques, require the formulation and minimization of an objective function that quantifies the mismatch between model predictions and observations (Evans, 2003, Herbst et al., 2008, Wang et al., 2009). Detailed measurements of biotic and abiotic processes and fluxes allow to improve process models on various spatiotemporal scales, and to verify model assumptions, parameters, and performance (Richardson et al., 2010, Williams et al., 2009, Yuan et al., 2012). However, the use of multiple objective functions or constraints in model calibration may be challenging because of the need to combine measurements with variable spatial scale, temporal scale, magnitude, and uncertainty. For example, optimizing the simulation regarding one data source (e.g., NEE) can lead to a low model performance (trade-off) regarding another data source (e.g., heterotrophic soil respiration) (Richardson et al., 2010). Other important decisions to be made before model calibration include the selection and appropriate weighting of observations, the choice of an optimization algorithm (Trudinger et al., 2007), and the selection of model parameters being altered during calibration (Wu et al., 2009). These decisions differ between model studies, which will influence the results of NEE predictions (Evans, 2003, Trudinger et al., 2007).
The main goal of this study is to present the mechanistic model AgroC and to evaluate its model performance simulating biophysical processes and interactions in agroecosystems. In a first step, AgroC was calibrated with soil moisture, soil temperature, biometric, and soil CO2 flux measurements of three test sites in Germany cropped with winter wheat, barley, or grass. After calibration, it was evaluated how well AgroC simulates the hourly NEE through comparison with EC measurements. In the next step, we optimized the AgroC model using EC measurements by estimating plant and Rsoil parameters. In addition, we evaluated how joint use of EC and Rsoil measurements in the calibration affected the estimated cumulative NEE and model performance. Finally, we evaluated the effect of data-transformation (e.g., log-transformation) on the model results with a focus on estimated NEE.
Section snippets
The AgroC model
AgroC is a coupled model developed from the SOILCO2/RothC model (Herbst et al., 2008) and the SUCROS model for crop growth (Spitters et al., 1989). The SOILCO2/RothC model simulates vertical water, heat, and CO2 fluxes in a soil column, and the source term of heterotrophic respiration over soil depth and time, which is given by the turnover of depth-specific carbon pools (Coleman and Jenkinson, 2008; Šimůnek and Suarez, 1993; Šimůnek et al., 1996). The carbon turnover rates depend on the soil
Soil temperature and water content
All simulations described measured soil temperature very well using the default settings. The RMSE was below 1.0 °C and the ME larger than 0.93 when measurements for all depths and sites were considered (see Fig. 2).
After calibration, the soil moisture dynamics were reproduced well by the AgroC model (Fig. 3). Estimated soil hydraulic parameters are summarized in Table A1. The RMSE was below 0.020 cm3 cm−3, the ME above 0.74 and the r above 0.86 for all sites and profile depths. For Merzenhausen,
Conclusions
The present study demonstrates that a crop growth module coupled to a model of soil CO2 production, soil water and heat flux can be used to simulate hourly NEE in agricultural systems. After calibrating the model for soil moisture, crop development, and Rsoil, the simulation of hourly NEE agreed well to EC measurements. For further validation, the application of AgroC to cropping systems in different European climate regions would be interesting.
An additional calibration based on EC
Acknowledgements
This research was supported by FACCE MACSUR – Modelling European Agriculture with Climate Change for Food Security and by the German Federal Ministry of Education and Research BMBF, project IDAS–GHG [grant number 01LN1313A]. The measurement infrastructure providing observational data was supported by the German Research Foundation DFG through the Transregional Collaborative Research Centre 32 (TR 32) and Terrestrial Environmental Observatories (TERENO). We thank Axel Knaps (Sicherheit und
References (88)
- et al.
Winter wheat carbon exchange in Thuringia, Germany
Agric. For. Meteorol.
(2004) - et al.
Sensitivity of simulated soil heterotrophic respiration to temperature and moisture reduction functions
Geoderma
(2008) - et al.
Estimating shoot to root ratios and annual carbon inputs in soils for cereal crops
Agric. Ecosyst. Environ.
(1997) - et al.
Spatio-temporal drivers of soil and ecosystem carbon fluxes at field scale in an upland grassland in Germany
Agric. Ecosyst. Environ.
(2015) Defining misfit between biogeochemical models and data sets
J. Mar. Syst.
(2003)- et al.
Estimating the size of the inert organic matter pool from total soil organic carbon content for use in the Rothamsted carbon model
Soil Biol. Biochem.
(1998) - et al.
Validation of a minimum microclimate disturbance chamber for net ecosystem flux measurements
Agric. For. Meteorol.
(2013) - et al.
Multiyear heterotrophic soil respiration: evaluation of a coupled CO2 transport and carbon turnover model
Ecol. Model.
(2008) - et al.
Improved procedure for obtaining statistically valid parameters estimates from soil respiration data
Soil Biol. Biochem.
(1995) - et al.
Gas-phase diffusivity and tortuosity of structured soils
J. Contam. Hydrol.
(2010)
Sources of CO2 efflux from soil and review of partitioning methods
Soil Biol. Biochem.
Eddy covariance flux corrections and uncertainties in long-term studies of carbon and energy exchanges
Agric. For. Meteorol.
A strategy for quality and uncertainty assessment of long-term eddy-covariance measurements
Agric. For. Meteorol.
River flow forecasting through conceptual models. Part I – a discussion of principles
J. Hydrol.
A multi-model comparison of soil carbon assessment of a coniferous forest stand
Environ. Modell. Softw.
On the spatial variation of soil rhizospheric and heterotrophic respiration in a winter wheat stand
Agric. For. Meteorol.
Carbohydrate content, fructan and sucrose enzyme activities in roots, stubble and leaves of ryegrass (Lolium perenne L.) as affected by source/sink modification after cutting
J. Plant Physiol.
A predictive model of rainfall interception in forests, 1. Derivation of the model from observations in a plantation of Corsican pine
Agric. Meteorol.
Colloidal properties and potential release of water-dispersible colloids in an agricultural soil depth profile
Geoderma
LINGRA, a sink/source model to simulate grassland productivity in Europe
Eur. J. Agron.
The carbon budget of a winter wheat field: an eddy covariance analysis of seasonal and inter-annual variability
Agric. For. Meteorol.
Full accounting of the greenhouse gas (CO2, N2O, CH4) budget of nine European grassland sites
Agric. Ecosyst. Environ.
Quantifying the effects of soil variability on crop growth using apparent soil electrical conductivity measurements
Eur. J. Agron.
Respiration of three Belgian crops: partitioning of total ecosystem respiration in its heterotrophic, above- and below-ground autotrophic components
Agric. For. Meteorol.
A linked carbon cycle and crop developmental model: description and evaluation against measurements of carbon fluxes and carbon stocks at several European agricultural sites
Agric. Ecosyst. Environ.
Carbon fluxes in the rhizosphere of winter wheat and spring barley with conventional vs integrated farming
Soil Biol. Biochem.
Correcting eddy-covariance flux underestimates over a grassland
Agric. For. Meteorol.
Soil respiration and human effects on global grasslands
Global Planet. Change
A review of applications of model-data fusion to studies of terrestrial carbon fluxes at different scales
Agric. For. Meteorol.
The carbon balance of European croplands: a cross-site comparison of simulation models
Agric. Ecosyst. Environ.
Crop Evapotranspiration. Guidelines for computing crop water requirements
FAO Irrigation and Drainage Paper No. 56
The seasonal growth of pasture grasses
J. Agric. Sci.
Assessing the eddy covariance technique for evaluating carbon dioxide exchange rates of ecosystems: past, present and future
Global Change Biol.
An evaluation of selected perennial ryegrass growth models for development and integration into a pasture management decision support system
J. Agric. Sci.
Inverse determination of heterotrophic soil respiration response to temperature and water content under field conditions
Biogeochemistry
Root biomass and shoot to root ratios of perennial forage crops in eastern Canada
Can. J. Plant Sci.
Crop-specific simulation parameters for yield forecasting across the European Community
Simulation Reports CABO-TT, No 32
RothC-26.3. A model for the turnover of carbon in soil
Model Description and Windows Users Guide
Coupling the Soil-Vegetation-Atmosphere-Transfer Scheme ORCHIDEE to the agronomy model STICS to study the influence of croplands on the European carbon and water budgets
Agronomie
Shuffled complex evolution approach for effective and efficient global minimization
J. Optimiz. Theory App.
Mesoscale eddies affect near-surface turbulent exchange: evidence from Lidar and tower measurements
J. Appl. Meteor. Climatol.
FAO Statistical Yearbook 2014
Simulation of field water use and crop yield
Actual evapotranspiration and precipitation measured by lysimeters: a comparison with eddy covariance and tipping bucket
Hydrol. Earth Syst. Sci.
Cited by (21)
A comprehensive analysis of CO<inf>2</inf> exchanges in agro-ecosystems based on a generic soil-crop model-derived methodology
2023, Agricultural and Forest MeteorologyResearch progress on carbon sources and sinks of farmland ecosystems
2023, Resources, Environment and SustainabilityYield potential and factor influencing yield gap in industrial hemp cultivation under nemoral climate conditions
2022, European Journal of AgronomyCitation Excerpt :In this study, we used the mechanistic AgroC model because of its physical representation for soil water and heat transport, as well as the ability to parameterize any crop to be investigated. The AgroC model was already applied to several field crops like winter wheat (Klosterhalfen et al., 2017), maize (Žydelis et al., 2018, 2021a), sugar beet (Herbst et al., 2021), or oat and barley (Groh et al., 2020; Groh et al., 2022). However, so far the AgroC model has not been parameterised for hemp.
Constraining water limitation of photosynthesis in a crop growth model with sun-induced chlorophyll fluorescence
2021, Remote Sensing of EnvironmentCitation Excerpt :Fig. 8 compares the NEE estimates at noon with the two different AgroC parametrizations. It shows that estimating the h3 parameter from SIF measurements resulted in a significant improvement in the mid-season NEE estimates, as the default parametrization from (Klosterhalfen et al., 2017) underestimates the water stress in the mid-season, explaining the difference in NSE between both model runs (Table 7). As the sugar beets experienced less water stress in the late season, calculating h3 based on SIF observations did not improve the NEE estimates there.
Revisiting the Feddes reduction function for modeling root water uptake and crop transpiration
2021, Journal of HydrologyCitation Excerpt :In most of them, the values of the Feddes parameters (h1, h2, h3h, h3l, and h4) were not obtained experimentally nor optimized in some way. In such cases, the authors used values retrieved from literature, sometimes with adjustments but without giving further details (e.g., Brunetti et al., 2019; Busoni et al., 1983; Dash et al., 2015; Durigon et al., 2012; Gabiri et al., 2018; Kambale et al., 2017; Karandish and Šimůnek, 2018, 2017; Klosterhalfen et al., 2017; Luo et al., 2003; Lv et al., 2014; Patle et al., 2018; Phogat et al., 2019, 2018, 2017, 2010; Pinheiro et al., 2019; Salehi et al., 2017; Wang et al., 2020; Wine et al., 2015; Yang et al., 2017; Yu et al., 2010; Žydelis et al., 2018). In contrast, only a few attempts at the optimization of the Feddes parameters were made (e.g., Cai et al., 2017; Dainese and Tarantino, 2020; dos Santos et al., 2017; Homaee et al., 2002; Hopmans and Guttièrez-Ravé, 1988; Massoud et al., 2019; Sela et al., 2015; Vandoorne et al., 2012; Xu et al., 2013; Yang et al., 2013).
Future climate change will accelerate maize phenological development and increase yield in the Nemoral climate
2021, Science of the Total Environment