Plant growth modeling has attracted a lot of attention due to its potential applications. Many scientific disciplines are involved, and a lot of research effort and intensive computer methods were needed to understand better the complex mechanisms underlying plant evolution. Among the numerous challenges, one can cite mathematical modeling, parameterization, estimation and prediction. One of the most promising models that have been proposed in the literature is the GreenLab functional–structural plant growth model. In this study, we focus only on one of its versions, named GreenLab-1, particularly adapted to a certain class of plants with known organogenesis, such as sugar beet, maize, rapeseed and other crop plants. The parameters of the model are related to plant functioning, and the vector of observations consists of organ masses measured only once at a given observation time. Previous efforts for parameter estimation in GreenLab-1 include Kalman-type filters, stochastic variants of EM and/or ECM algorithms, and hybrid sequential importance sampling algorithms with Bayesian estimation only for the functional parameters of the model. In this paper, the first purely Bayesian approach for parameter estimation of the GreenLab-1 model is proposed. This approach has much more flexibility in handling complex structures, thus providing a useful tool for analyzing such types of models. In order to sample from the posterior distribution an MCMC algorithm is used and its implementation issues are also discussed. The performance of this method is illustrated on a simulated and a real dataset from the sugar beet plant, and a comparison is made with the MLE approach.

植物生长建模因其潜在的应用而引起了很多关注。涉及许多科学学科，需要大量的研究工作和密集的计算机方法来更好地理解植物进化背后的复杂机制。在众多挑战中，可以举出数学建模、参数化、估计和预测。文献中提出的最有前途的模型之一是 GreenLab 功能结构植物生长模型。在这项研究中，我们只关注其中一个版本，名为 GreenLab-1，特别适用于具有已知器官发生的某一类植物，如甜菜、玉米、油菜籽和其他作物植物。该模型的参数与工厂功能有关，观察向量由在给定观察时间仅测量一次的器官质量组成。GreenLab-1 中先前的参数估计工作包括卡尔曼型滤波器、EM 和/或 ECM 算法的随机变体，以及仅对模型的功能参数进行贝叶斯估计的混合顺序重要性采样算法。在本文中，提出了第一个用于 GreenLab-1 模型参数估计的纯贝叶斯方法。这种方法在处理复杂结构方面具有更大的灵活性，从而为分析此类模型提供了有用的工具。为了从后验分布中采样，使用了 MCMC 算法，并且还讨论了它的实现问题。该方法的性能在来自甜菜厂的模拟和真实数据集上进行了说明，