Fitting a functional structural plant model based on global sensitivity analysis

Calibration of a functional structural plant model is a challenging task because of the complexity of model structure. Parameter estimation through gradient-based optimization technique was highly dependent on initial parameter values. This motivated the use of global sensitivity analysis technique to choose parameter subset in fitting the data sequence. Global sensitivity indices were computed using the source sink ratio as the output of interest, which regulates all organ growth. By fitting on chrysanthemum data from nine sampling dates, it is shown that sensitivity analysis method helps to identify the influential parameters for a given sampling date. As a result, fitting process is less dependent on the initial parameter values. Current work provides a new method of calibrating a plant growth model with multiple outputs.