Methodology for modeling and parameter estimation of the growth process of Lobesia botrana

Abstract

Lobesia botrana (L. botrana), is a quarantine pest that causes damage to grapevines and generates economic losses for the region of Cuyo in Argentina. Different researchers have sought to safeguard the integrity of the vineyards, generating alert systems based on models that allow detecting the peaks of occurrence of the pest, and knowing the growth process of the moth, according to the environmental conditions of each region. In this work, a methodology for estimating unknown parameters in semi-physical models based on first principles (MSBPP) is proposed, with a particular application in the growth model of L. botrana under laboratory conditions. The main contribution consists of a methodology for parameter estimation of an MSBPP, which considers a mathematical model developed by the authors in previous work, the structural identifiability analysis of the model in question, and the estimation of the set of unknown parameters that meet the structural identifiability property. In this work, the non-linear least squares algorithm and an Extended Kalman Filter are considered the main estimation tools. An improvement in the adjustment of the mathematical model to the experimental data was evidenced, in relation to those previously obtained. In addition, the degree of affinity of each growth stage for its limiting factor was established, and new mortality profiles were presented.