A Monte Carlo approach for estimating the uncertainty of predictions with the tomato plant growth model, Tomgro

The output of plant growth models is subject to different sources of uncertainty. In this research, the objective is to analyse the uncertainty of the predictions with a model that simulates the growth, development and yield of a tomato crop, Tomgro, due to the variability on model parameters. The uncertainty of an output variable of the model is defined as the variation caused, when the model parameter is varied in its measured or estimated distribution space. Monte Carlo methodologies were used to test the uncertainty of the estimation of some main output variables. For this purpose, the model was run repeatedly with different sets of parameters sampled from their estimated multivariate normal distributions. The uncertainty was quantified by the coefficients of variation (CV) of the slopes of the selected output variables. The total uncertainty of the estimation of mature fruit dry weight caused by the uncertainty of all model parameters is very high, with a CV of 46%. The light and CO(2) use efficiency are the dominant parameters. The stochastic approach of estimating the uncertainty of output variables proves to be a promising tool to assess uncertainties of model predictions. (c) 2006 TAgrE. All rights reserved Published by Elsevier Ltd