基于可见/近红外光谱的柑橘叶片不同形态氮含量无损监测研究

Citrus cultivation area and yield in China have ranked first over the world. However, the fruit grower always applied excessively nitrogen fertilizer in the citrus orchards. Hence, the scientific method of nitrogen application is urgently needed. The visible/near infrared spectroscopy has the characteristics of real time and non-destructive, which can be used to realize the non-destructive monitoring and recommendation of nitrogen by coupling modeling with the total leaf nitrogen concentration. However, the optimum value of leaf nitrogen concentration is wide so the accuracy of nitrogen application needs to be improved. Concentrations and distribution of different nitrogen forms (assimilable nitrogen, functional nitrogen and structural nitrogen) are the formation and transport of photosynthate. It is theoretically more reliable by monitoring the concentrations of different nitrogen forms to make fertilization guidance, than that of total leaf nitrogen concentration. Therefore, the project will be conducted on the 4-year old citrus trees. Changes of different nitrogen forms by the different nitrogen application rates will be investigated, illustrating the relationship between absorption and distribution characteristic of different leaf nitrogen forms and the yield as well as the fruit quality; the visible/near infrared spectroscopy technique will be used to construct a nondestructive model to monitoring the concentrations of different nitrogen forms in the crucial growth stages; Finally, based on the nondestructive monitoring model and nitrogen absorption characteristics, the effect of nitrogen reduction was evaluated by combining soil nitrogen and fertilizer utilization. Our study will provide a theoretical basis and technical to support the non-destructive monitoring the concentrations of different nitrogen forms and nitrogen management in citrus orchards.

我国柑橘栽培面积和产量已跃居世界首位，然而橘园氮肥施用过量问题突出，迫切需要科学的施氮方法。可见/近红外光谱具有实时、无损的特点，能通过与叶片全氮含量的耦合建模实现其无损监测与推荐施氮。然而叶片全氮含量适宜值较宽，指导施氮准确性有待提高。叶片不同形态氮（营养性氮、功能性氮和结构性氮）含量及分配是形成和转运同化物的基础，监测叶片不同形态氮含量并指导施肥，理论上比叶片全氮含量更可靠。为此，本项目以4年生柑橘为研究对象，探究叶片不同形态氮含量的时空变化特征及其与当年果实产量、品质的关系，结合不同施氮处理建立基于叶片不同形态氮的调控施氮回归方程；利用可见/近红外光谱技术，研究叶片不同形态氮与光谱之间的耦合机制，构建其无损监测模型；以无损监测模型和回归方程为基础，结合土壤供氮与肥料利用率实时推荐施氮，评价其“减氮增效”效果。为实现柑橘叶片不同形态氮素含量无损监测和调控施氮提供理论依据和技术支持。

我国柑橘栽培面积和产量已跃居世界首位，然而橘园氮肥施用过量问题突出，迫切需要科学的施氮方法。叶片不同形态氮（营养性氮、功能性氮和结构性氮）含量及分配是形成和转运同化物的基础，监测叶片不同形态氮含量并指导施肥，理论上比叶片全氮含量更可靠。为此，本项目以4年生（初果期树）与10年生（盛果期树）柑橘为研究对象，探究叶片不同形态氮含量的时空变化特征及其与当年果实产量、品质的关系。结果表明，柑橘叶片不同形态氮含量大小顺序为功能性氮＞结构性氮＞营养性氮含量，分别占总氮含量的65-70%，20-30%，5-10%，随生育期和不同氮肥施用量变化而变化，膨大期叶片功能性氮含量与当年产量相关性最好，R2可达0.65。利用可见/近红外光谱技术，研究叶片不同形态氮与光谱之间的耦合机制，分析了全波段及敏感波段结合植被指数、化学计量法和机器学习法的建模效果，分别构建基于全波段的反向传播神经网络（BPNN）和基于敏感波段的随机森林（RF）模型，两者对于膨大期叶片功能性氮含量无损估测效果最好，测试集R2为0.74-0.78，RMSE为0.83 g kg-1。以无损监测模型和调控施氮回归方程为基础，结合土壤供氮与肥料利用率实时推荐施氮。结果表明，相较于对照施氮处理，调控施氮能在氮肥施用不足时补充氮营养，减少产量品质的损失；也能在前期过量施氮的情况下节肥41%，总计增产幅度为35-47.5%，提高氮肥偏生产力和氮肥农学效率。本研究结果为实现柑橘叶片不同形态氮素含量无损监测和调控施氮提供理论依据和技术支持。