Modelling energy cane growth in contrasting environments: calibration and validation