Improving growth period prediction of greenhouse cut chrysanthemum: Incorporating precise photo-thermal effect and multi-model ensemble approaches

Accurate prediction of the cut chrysanthemum growth cycle is essential for precise market scheduling and quality control. While existing models primarily target the flowering date, key developmental stages remain poorly quantified. Through controlled experiments with varying planting dates and light levels, this study systematically analyzed the stage-specific influence of photosynthetic photon flux (PPF). We found that reduced PPF significantly delays both vegetative growth and harvest timing but does not affect floral bud differentiation. To address these stage-dependent responses, we developed a multi-model ensemble (MME) framework that integrates the most accurate models for each critical phase: the accumulated photo-thermal product for initiating short-day treatment, the triangular-function-based relative thermal effect for bud emergence, and the chrysanthemum clock model for the optimal harvest date. Validation results demonstrate that this integrated approach achieves significantly higher predictive accuracy than any single model. This research not only provides a reliable tool for the year-round precision production management of cut chrysanthemum but also offers a physiologically-based MME methodology reference for modeling the growth of horticultural crops.