Accéder directement au contenu Accéder directement à la navigation
Direction d'ouvrage ou d'actes

Bayesian calibration of the Pasture Simulation Model (PaSim) to simulate European grasslands under climate extremes: case study at Stubai (Austria)

Abstract : The Bayesian method for the calibration of complex models, applying Markov Chain Monte Carlo (MCMC) algorithm, was used to quantify the uncertainty associated with the parameterization of the Pasture Simulation Model (PaSim) for the evaluation of climate change impacts. A case study of applying the method to calibrate 2008 and 2009 model parameters in a mountain grassland in the Central Eastern Alps (Stubai, Austria, 47° 05′ North, 11° 11′ East) is presented to assess the effects of changes in climate conditions (with or without temperature extremes). In the case study, the prior distributions of the 17 most influential parameters used to estimate vegetation and soil variables were based on both literature and previous studies, and subsequently updated using the likelihood distributions resulting from site-specific measurements. The posterior uncertainty distribution of the model results was generated using Monte Carlo simulations with posterior parameter probability distributions. It was shown that using site-specific information to update the prior uncertainty distribution, the resulting uncertainty associated with the model results could be reduced. As an example, for the parameter „maximum specific leaf area‟, the coefficient of variation was reduced of about 99% compared to the prior probability. Posterior estimations of output variables were closer to observations, e.g. the RMSE of the leaf area index was reduced to about the half of that of prior distribution. Similar results were also obtained with other parameters and model outputs. These results indicate that, thanks to the incorporation of quantitative uncertainty analysis into grassland simulation, a better informed decision can be made with respect to climate change where specific impacts and issues of adaptation arise.
Mots-clés : calibration Bayesianne
Type de document :
Direction d'ouvrage ou d'actes
Liste complète des métadonnées

https://hal.inrae.fr/hal-02806115
Déposant : Migration Prodinra <>
Soumis le : samedi 6 juin 2020 - 01:42:41
Dernière modification le : vendredi 19 juin 2020 - 17:04:03

Identifiants

  • HAL Id : hal-02806115, version 1
  • PRODINRA : 183763

Collections

Citation

Haythem Ben Touhami, Romain Lardy, Michael Bahn, Gianni Bellocchi, Vincent Barra. Bayesian calibration of the Pasture Simulation Model (PaSim) to simulate European grasslands under climate extremes: case study at Stubai (Austria). 11th International Probabilistic Safety Assessment and Management Conference and the Annual European Safety and Reliability Conference 2012, Jun 2012, Helsinki, Finland. 1, 10 p., 2012, 978-1-62276-436-5. ⟨hal-02806115⟩

Partager

Métriques

Consultations de la notice

10