Precision agriculture for banana crops has been little investigated so far. The main difficulty to implement precision agriculture methods lies in the asynchronicity of this crop: after a few cycles, each plant has its own development stage in the field. It results in a diversity of the phenological stages within a field and also a continuous production of bananas over time. Therefore, maps of agronomic interest derived from plant responses are difficult to produce using existing methods. This study proposes a mapping approach that handles the diversity of phenological stages and the temporal continuity of production. It explores the feasibility of applying this general approach to a plant response parameter which is the time between flowering and maturity (time to harvest) of banana denoted tfm . The tfm gives an insight into the spatial distribution of vigour. The study was conducted using a large database (more than 395 000 observations) generated by two commercial farms in 2015 and 2016 in Cameroon. The temporal variability of tfm , which is induced by meteorological and operational constraints, and the spatial variability, which is assumed to be due to environmental factors, was assessed by decomposing the tfm variance. This method allowed mapping of the effect of the temporal variability as well as the effect of agri-environmental variables on tfm using a block kriging method. Spatial structures highlighted by this decomposition either at the farm level or at the field level, suggest that the map of the effect of environmental factors on tfm could be used to support agronomic decisions. This idea is reinforced by the identification of factors explaining the environmental variability of tfm and by the temporal stability of the spatial structures.

中文翻译：

---

香蕉作物特性的空间分析和制图：香蕉生产的异步性问题和考虑它的统计方法的提议

迄今为止，对香蕉作物的精准农业的研究很少。实施精准农业方法的主要困难在于这种作物的不同步性：经过几个周期后，每种植物在田间都有自己的发育阶段。它导致一个领域内物候阶段的多样性，并且随着时间的推移持续生产香蕉。因此，使用现有方法难以产生源自植物反应的农学利益图。本研究提出了一种映射方法来处理物候阶段的多样性和生产的时间连续性。它探讨了将这种通用方法应用于植物响应参数的可行性，该参数是用 tfm 表示的香蕉开花和成熟之间的时间（收获时间）。tfm 可以洞察活力的空间分布。该研究是使用 2015 年和 2016 年喀麦隆的两个商业农场生成的大型数据库（超过 395 000 次观测）进行的。通过分解 tfm 方差来评估由气象和操作限制引起的 tfm 的时间变异性以及假定由环境因素引起的空间变异性。这种方法允许使用块克里金法绘制时间可变性的影响以及农业环境变量对 tfm 的影响。这种分解在农场层面或田间层面突出显示的空间结构表明，环境因素对 tfm 的影响图可用于支持农艺决策。