|Authors: ||A. Cooman, E. Schrevens|
|Keywords: ||tomato, growth model, variation, parameter distribution, deterministic model, stochastic model, uncertainty propagation.|
In former studies based on TOMGRO, the parameters were considered fixed, the model deterministic.
Here, the uncertainty on predictions of the modified TOMGRO model is analyzed, using a database that was compiled in the Bogotá Plateau.
The uncertainty on an output variable of the model is defined here as the variation caused on the output variable, quantified by a standard deviation, when the model parameter or input factor is varied in its measured or estimated distribution space.
A sampling-based methodology was used to test the uncertainty on the estimation of the total vegetative dry weight (VEGW), total leaf area (LA) and total fruit dry weight (DWF). For this purpose, the model was run repeatedly with different sets of parameters sampled from their multivariate normal distributions.
The uncertainty was quantified by the coefficient of variation of the slopes (CVo) of the output variables that appear when the model parameters were varied.
The highest uncertainty on the simulated DWF was caused by the light- and CO2 use efficiency. The uncertainty caused by the parameters related to dry matter distribution on the estimation of DWF was always below 8%. The development-parameters had nearly no effect on the uncertainty of the estimation of DWF. The total uncertainty of the estimation of mature fruit dry weight caused by the uncertainty of all model parameters was very high, with a CVo of 46%. The developed method has proven to be a promising tool to study uncertainties on model predictions and can be used for the evaluation of production systems when an assessment of the risk, caused by the error on parameter estimations, is desired.
Download Adobe Acrobat Reader (free software to read PDF files)
URL www.actahort.org Hosted by KU Leuven