基于叶绿素荧光技术的李子干旱胁迫状态监测  

Drought Stress Status Monitoring of Plum Based on Chlorophyll Fluorescence Technology

作　　者：孙小添 王海超 张钧尧 孙凯 李颖[1] 王继璇 孙萌 裴志永[1] SUN Xiaotian;WANG Haichao;ZHANG Junyao;SUN Kai;LI Ying;WANG Jixuan;SUN Meng;PEI Zhiyong(College of Energy and Transportation Engineering,Inner Mongolia Agricultural University,Hohhot 010018,China)

机构地区：[1]内蒙古农业大学能源与交通工程学院,呼和浩特010018

出　　处：《内蒙古农业大学学报(自然科学版)》2025年第1期43-48,100,共7页Journal of Inner Mongolia Agricultural University(Natural Science Edition)

基　　金：内蒙古自治区科技计划项目(2023YFDZ0015);内蒙古自治区自然科学基金项目(2023QN03029);国家自然科学基金地区科学基金项目(32301665)。

摘　　要：为探索叶绿素荧光技术在监测作物干旱胁迫状态中的应用,本文以李子园为研究场所,采用连续投影算法(successive projections algorithm,SPA)和迭代保留信息变量法(iteratively retains informative variables,IRIV)进行参数优选,通过随机森林(random forest,RF)和BP神经网络(back propagation neural network,BNN)建立干旱胁迫状态识别模型,分析干旱胁迫对李子叶绿素荧光参数的影响,以期为及时发现李子干旱胁迫状态和胁迫等级划分提供科学依据。结果表明:利用SPA和IRIV算法筛选出5个关键叶绿素荧光参数,显著提升了BNN模型在李子树干旱胁迫状态识别上的准确率。BNN模型准确率提升至64%,而RF模型在全参数集和公共参数集下均达到84%。混淆矩阵分析进一步证实了公共参数集的RF模型在干旱状态识别上的整体最佳表现,特别是在中度干旱识别上的准确率提升。这些发现为及时发现李子干旱胁迫状态和胁迫等级划分提供了科学依据,也为大尺度作物干旱监测提供了有力数据支持。This study explored the application of chlorophyll fluorescence technology for assessing drought stress in plum orchards.By employing the Successive Projections Algorithm(SPA)and Iteratively Retains Informative Variables(IRIV)for refined parameter selection,the models for drought stress state identification were constructed using the Random Forest(RF)and Backpropagation Neural Network(BNN)methodologies.The analysis targeted the influence of drought stress on plum leaf chlorophyll fluorescence parameters,aiming to provide a scientific foundation for the prompt detection and classification of plum drought stress levels.The re-sults indicated that the application of SPA and IRIV algorithms in selecting five key chlorophyll fluorescence parameters markedly enhanced the accuracy of both the BNN and RF models in recognizing plum tree drought stress,with the BNN model achieving an accuracy of 64%and the RF model reaching 84%across both full and common parameter sets.Further analysis using a confusion ma-trix highlighted the superior performance of the RF model with the common parameter set,particularly in identifying moderate drought stress.These findings not only facilitate the timely identification and classification of plum drought stress but also provide robust data support for large-scale monitoring of crop drought conditions.

关 键 词：干旱胁迫 叶绿素荧光参数 李子 监测 

分 类 号：Q945[生物学—植物学] S7-9[农业科学—林学]