Uncertainty analyses of the VegSyst model applied to greenhouse crops

Nowadays, roughly more than 70 and 20% of the Mexican greenhouses use low and medium technological levels, respectively.
In order to increase yield, together with water and energy efficiency, better control strategies of root and shoot environment are needed.
However, looking into better control strategies such as optimal control, reliable dynamic models of greenhouses crops are required.
VegSyst is a discrete-time dynamic model that predicts dry matter production (DMP), thermal time (TT), Nitrogen uptake (Nup) and crop transpiration (ETc) for greenhouse crops.
The model has been used in developing a decision support system for Mediterranean greenhouse crops.
Nevertheless, until now no uncertainty analysis (UA) of this model has been reported.
Since understanding the uncertainty could increase the reliability of using the VegSyst model, in the current research both, a frequentist and a Bayesian uncertainty analysis on the model parameters, were carried out.
In both approaches, firstly probability distribution functions for each model parameter were defined.
Secondly, a Latin Hypercube sampling procedure was applied in order to generate several thousands of parametersRSQUO values.
Thirdly, Monte Carlo simulation was used in generating the outputs of the VegSyst model.
Finally, basic statistic measures were calculated and several plots generated in case of the frequentist uncertainty analysis whereas in case of the Bayesian uncertainty analysis, the generalized likelihood uncertainty estimation (GLUE) methodology was applied.
In the second paradigm measurements of the predicted variables of the VegSyst model were also included.
Data collected in an experiment with a tomato crop grown in Chapingo, Mexico, were used to carry out both uncertainty analyses.
Main results from both uncertainty analysis methods were that VegSyst predicts DMP and Nup better than crop transpiration thus the last process could be improved further.

XXX International Horticultural Congress IHC2018: III International Symposium on Innovation and New Technologies in Protected Cultivation

I.L. López-Cruz, A. Martínez-Ruiz, A. Ruiz-García, M. Gallardo

frequentist uncertainty analysis, GLUE, probability distribution function, Monte Carlo simulation

https://doi.org/10.17660/ActaHortic.2020.1271.28