Julien Minet | Université de Liège (original) (raw)
Phone: +32 63 23 08 50
Address: Université de Liège - Arlon Campus Environnement
185, Avenue de Longwy
B-6700 Arlon
Belgique
less
Uploads
Papers by Julien Minet
Biogeosciences, 2015
ABSTRACT Eddy covariance data from four European grassland sites are used to probabilistically in... more ABSTRACT Eddy covariance data from four European grassland sites are used to probabilistically invert the CARAIB dynamic vegetation model (DVM) with ten unknown parameters, using the DREAM(ZS) Markov chain Monte Carlo (MCMC) sampler. We compare model inversions considering both homoscedastic and heteroscedastic eddy covariance residual errors, with variances either fixed a~priori or jointly inferred with the model parameters. Agreements between measured and simulated data during calibration are comparable with previous studies, with root-mean-square error (RMSE) of simulated daily gross primary productivity (GPP), ecosystem respiration (RECO) and evapotranspiration (ET) ranging from 1.73 to 2.19 g C m−2 day−1, 1.04 to 1.56 g C m−2 day−1, and 0.50 to 1.28 mm day−1, respectively. In validation, mismatches between measured and simulated data are larger, but still with Nash–Sutcliffe efficiency scores above 0.5 for three out of the four sites. Although measurement errors associated with eddy covariance data are known to be heteroscedastic, we showed that assuming a classical linear heteroscedastic model of the residual errors in the inversion do not fully remove heteroscedasticity. Since the employed heteroscedastic error model allows for larger deviations between simulated and measured data as the magnitude of the measured data increases, this error model expectedly lead to poorer data fitting compared to inversions considering a constant variance of the residual errors. Furthermore, sampling the residual error variances along with model parameters results in overall similar model parameter posterior distributions as those obtained by fixing these variances beforehand, while slightly improving model performance. Despite the fact that the calibrated model is generally capable of fitting the data within measurement errors, systematic bias in the model simulations are observed. These are likely due to model inadequacies such as shortcomings in the photosynthesis modelling. Besides model behaviour, difference between model parameter posterior distributions among the four grassland sites are also investigated. It is shown that the marginal distributions of the specific leaf area and characteristic mortality time parameters can be explained by site-specific ecophysiological characteristics. Lastly, the possibility of finding a common set of parameters among the four experimental sites is discussed.
2012 14th International Conference on Ground Penetrating Radar (GPR), 2012
Hydrology and Earth System Sciences Discussions, 2013
Abstract Characterizing the spatial and temporal variability of soil moisture using geophysical m... more Abstract Characterizing the spatial and temporal variability of soil moisture using geophysical methods is an important issue in many hydrological researches and applications. In order to bridge the scale gap between large-scale remote sensing of soil ...
Biogeosciences, 2015
ABSTRACT Eddy covariance data from four European grassland sites are used to probabilistically in... more ABSTRACT Eddy covariance data from four European grassland sites are used to probabilistically invert the CARAIB dynamic vegetation model (DVM) with ten unknown parameters, using the DREAM(ZS) Markov chain Monte Carlo (MCMC) sampler. We compare model inversions considering both homoscedastic and heteroscedastic eddy covariance residual errors, with variances either fixed a~priori or jointly inferred with the model parameters. Agreements between measured and simulated data during calibration are comparable with previous studies, with root-mean-square error (RMSE) of simulated daily gross primary productivity (GPP), ecosystem respiration (RECO) and evapotranspiration (ET) ranging from 1.73 to 2.19 g C m−2 day−1, 1.04 to 1.56 g C m−2 day−1, and 0.50 to 1.28 mm day−1, respectively. In validation, mismatches between measured and simulated data are larger, but still with Nash–Sutcliffe efficiency scores above 0.5 for three out of the four sites. Although measurement errors associated with eddy covariance data are known to be heteroscedastic, we showed that assuming a classical linear heteroscedastic model of the residual errors in the inversion do not fully remove heteroscedasticity. Since the employed heteroscedastic error model allows for larger deviations between simulated and measured data as the magnitude of the measured data increases, this error model expectedly lead to poorer data fitting compared to inversions considering a constant variance of the residual errors. Furthermore, sampling the residual error variances along with model parameters results in overall similar model parameter posterior distributions as those obtained by fixing these variances beforehand, while slightly improving model performance. Despite the fact that the calibrated model is generally capable of fitting the data within measurement errors, systematic bias in the model simulations are observed. These are likely due to model inadequacies such as shortcomings in the photosynthesis modelling. Besides model behaviour, difference between model parameter posterior distributions among the four grassland sites are also investigated. It is shown that the marginal distributions of the specific leaf area and characteristic mortality time parameters can be explained by site-specific ecophysiological characteristics. Lastly, the possibility of finding a common set of parameters among the four experimental sites is discussed.
2012 14th International Conference on Ground Penetrating Radar (GPR), 2012
Hydrology and Earth System Sciences Discussions, 2013
Abstract Characterizing the spatial and temporal variability of soil moisture using geophysical m... more Abstract Characterizing the spatial and temporal variability of soil moisture using geophysical methods is an important issue in many hydrological researches and applications. In order to bridge the scale gap between large-scale remote sensing of soil ...