Bayesian calibration of process-based forest models: bridging the gap between models and data.

Process-based forest models generally have many parameters, multiple outputs of interest and a small underlying empirical database. These characteristics hamper parameterization. Bayesian calibration offers a solution to the calibration problem because it applies to models of any type or size. It provides parameter estimates, with measures of uncertainty and correlation among the parameters. The procedure begins by quantifying the uncertainty about parameter values in the form of a prior probability distribution. Then data on the output variables are used to update the parameter distribution by means of Bayes' Theorem. This yields a posterior calibrated distribution for the parameters, which can be summarized in the form of a mean vector and variance matrix. The predictive uncertainty of the model can be quantified by running it with different parameter settings, sampled from the posterior distribution. In a further step, one may evaluate the posterior probability of the model itself (rather than that of the parameters) and compare that against the probability of other models, to aid in model selection or improvement. Bayesian calibration of process-based models cannot be performed analytically, so the posterior parameter distribution must be approximated in the form of a representative sample of parameter values. This can be achieved by means of Markov Chain Monte Carlo simulation, which is suitable for process-based models because of its simplicity and because it does not require advance knowledge of the shape of the posterior distribution. Despite the suitability of Bayesian calibration, the technique has rarely been used in forestry research. We introduce the method, using the example of a typical forest model. Further, we show that reductions in parameter uncertainty, and thus in output uncertainty, can be effected by increasing the variety of data, increasing the accuracy of measurements and increasing the length of time series.

[1]  Ernst-Detlef Schulze,et al.  Carbon and Nitrogen Cycling in European Forest Ecosystems , 2000, Ecological Studies.

[2]  Michael Goldstein,et al.  Probabilistic Formulations for Transferring Inferences from Mathematical Models to Physical Systems , 2005, SIAM J. Sci. Comput..

[3]  A. O'Hagan,et al.  Bayesian calibration of computer models , 2001 .

[4]  John Skilling,et al.  Data analysis : a Bayesian tutorial , 1996 .

[5]  Alan R. Ek,et al.  The consequences of hierarchy for modeling in forest ecosystems , 2000 .

[6]  E. Jaynes.  Probability theory : the logic of science , 2003 .

[7]  J. Ghazoul,et al.  Communicating complexity and uncertainty in decision making contexts: Bayesian approaches to forest research , 2003 .

[8]  M. J. W. Jansen.  Data use and Bayesian statistics for model calibration , 1999 .

[9]  N. Metropolis,et al.  Equation of State Calculations by Fast Computing Machines , 1953, Resonance.

[10]  Aaron M. Ellison,et al.  AN INTRODUCTION TO BAYESIAN INFERENCE FOR ECOLOGICAL RESEARCH AND ENVIRONMENTAL , 1996 .

[11]  Christian P. Robert,et al.  Monte Carlo Statistical Methods , 2005, Springer Texts in Statistics.

[12]  Andreas Fischlin,et al.  The response of the carbon cycle in undisturbed forest ecosystems to climate change: a review of plant-soil models , 1995 .

[13]  Thomas E. Burk,et al.  Fitting process-based models with stand growth data: problems and experiences , 1994 .

[14]  Peter Millard,et al.  The effect of nitrogen enrichment on the carbon sink in coniferous forests: Uncertainty and sensitivity analyses of three ecosystem models , 2004 .

[15]  Philip J. Radtke,et al.  Bayesian melding of a forest ecosystem model with correlated inputs , 2002 .

[16]  Harry T. Valentine,et al.  Assessing Uncertainty in a Stand Growth Model by Bayesian Synthesis , 1999, Forest Science.

[17]  Marcel Van Oijen,et al.  On the use of specific publication criteria for papers on process-based modelling in plant science , 2002 .

[18]  Michael Goldstein,et al.  Bayesian Forecasting for Complex Systems Using Computer Simulators , 2001 .

[19]  Aaron M. Ellison,et al.  Bayesian inference in ecology , 2004 .

[20]  S. Carpenter,et al.  Ecological forecasts: an emerging imperative. , 2001, Science.

[21]  Hubert Hasenauer,et al.  Biased predictions for tree height increment models developed from smoothed ‘data’ , 1997 .

[22]  Stephen P. Prisley,et al.  A synthesis of literature on evaluation of models for policy applications, with implications for forest carbon accounting , 2004 .

[23]  Kenneth H. Reckhow,et al.  Bayesian inference in non-replicated ecological studies , 1990 .

[24]  Shoufan Fang,et al.  A Bayesian approach for estimating the parameters of a forest process model based on long-term growth data , 1999 .

[25]  Belinda E. Medlyn,et al.  Design and use of a database of model parameters from elevated [CO2] experiments , 1999 .

[26]  Marvin H. J. Guber.  Bayesian Spectrum Analysis and Parameter Estimation , 1988 .

[27]  Dave Higdon,et al.  Combining Field Data and Computer Simulations for Calibration and Prediction , 2005, SIAM J. Sci. Comput..

[28]  Alan R. Ek,et al.  Process-based models for forest ecosystem management: current state of the art and challenges for practical implementation. , 2000, Tree physiology.

[29]  David J. Spiegelhalter,et al.  Introducing Markov chain Monte Carlo , 1995 .

[30]  Michiel J. W. Jansen,et al.  Calibration in a Bayesian modelling framework , 2004 .

[31]  Annikki Mäkelä,et al.  Performance analysis of a process‐based stand growth model using Monte Carlo techniques , 1988 .

[32]  Harry T. Valentine,et al.  Bayesian synthesis for quantifying uncertainty in predictions from process models. , 2000, Tree physiology.

[33]  R J Leipold,et al.  On the dangers of adjusting the parameters values of mechanism-based mathematical models. , 1996, Journal of theoretical biology.

[34]  Robert V. O'Neill,et al.  Error propagation and uncertainty in process modeling , 1988 .