[1]徐剑文,刘剑光,李健,等.基于可解释性机器学习的影响长江流域棉花农艺性状的关键气象因子解析[J].江苏农业学报,2026,42(03):554-562.[doi:doi:10.3969/j.issn.1000-4440.2026.03.013]
　XU Jianwen,LIU Jianguang,LI Jian,et al.Analysis of key meteorological factors affecting cotton agronomic traits in the Yangtze River Basin based on interpretable machine learning[J].,2026,42(03):554-562.[doi:doi:10.3969/j.issn.1000-4440.2026.03.013]
点击复制

基于可解释性机器学习的影响长江流域棉花农艺性状的关键气象因子解析()

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
42
期数:
2026年03期
页码:
554-562
栏目:
耕作栽培·资源环境
出版日期:
2026-03-31

文章信息/Info

Title:
Analysis of key meteorological factors affecting cotton agronomic traits in the Yangtze River Basin based on interpretable machine learning
作者:
徐剑文; 刘剑光; 李健; 赵君; 许栩; 邵明灿
(江苏省农业科学院经济作物研究所/农业农村部长江中下游棉花和油菜重点实验室,江苏南京210014)
Author(s):
XU Jianwen; LIU Jianguang; LI Jian; ZHAO Jun; XU Xu; SHAO Mingcan
(Institute of Industrial Crops, Jiangsu Academy of Agricultural Sciences/Key Laboratory of Cotton and Rapeseed, Ministry of Agriculture and Rural Affairs, Nanjing 210014, China)
关键词:
棉花; 农艺性状; 机器学习; 气象因子
Keywords:
cotton; agronomic traits; machine learning; meteorological factors
分类号:
S562
DOI:
doi:10.3969/j.issn.1000-4440.2026.03.013
文献标志码:
A
摘要:
本研究基于2020-2024年长江流域国家棉花品种区域试验数据,系统分析生育期不同阶段气象条件对棉花生产的影响,并采用沙普利可加性解释(SHAP)方法解析气象因子的影响效应及其交互作用。结果表明,与线性回归模型和支持向量机（SVM）模型相比,随机森林模型和XGBoost模型对单株铃数、纤维长度、纤维比强度等棉花农艺性状的预测能力较好。积温、高温天数与晴天数是影响棉花性状的关键气象因子,其作用具有明显的生育期阶段性:吐絮期积温升高通过促进早熟缩短生育期,全生育期及花铃期高温对纤维长度、纤维伸长率、纤维整齐度等纤维品质产生不利影响,出苗期至开花期晴天有利于纤维长度和纤维比强度的提升。在棉花产量和纤维品质形成过程中,气象因子间存在协同或拮抗效应。本研究结果为气候变化背景下棉花栽培管理优化与育种目标调整提供了理论依据。
Abstract:
Based on the data from the national cotton variety regional trials in the Yangtze River Basin from 2020 to 2024, this study systematically analyzed the effects of meteorological conditions at different growth stages on cotton production, and used the Shapley Additive Explanations (SHAP) method to quantify the effects and interactions of meteorological factors. The results showed that, compared with the linear regression model and the support vector machine (SVM) model, the random forest model and the XGBoost model exhibited better predictive performance for cotton agronomic traits such as bolls per plant, fiber length, and fiber specific strength. Accumulated temperature, number of high-temperature days, and number of sunny days were the key meteorological factors affecting cotton traits, and their effects displayed obvious stage-specific characteristics during the growth period: increased accumulated temperature during the boll-opening stage promoted earliness and shortened the growth period; high temperature during the whole growth period and the flowering and boll-forming stage adversely affected fiber quality traits including fiber length, elongation, and uniformity; sunny days from seedling emergence to flowering were conducive to the improvement of fiber length and fiber specific strength. Synergistic or antagonistic effects existed among meteorological factors during the formation of cotton yield and fiber quality. This study provides a theoretical basis for the optimization of cotton cultivation management and the adjustment of breeding objectives under climate change.

参考文献/References:

[1]LI X, ZHANG Z G, PAN Z L, et al. Demonstrating almost half of cotton fiber quality variation is attributed to climate change using a hybrid machine learning-enabled approach[J]. European Journal of Agronomy,2025,162:127426.
[2]GAO M, XU B J, WANG Y H, et al. Quantifying individual and interactive effects of elevated temperature and drought stress on cotton yield and fiber quality[J]. Journal of Agronomy & Crop Science,2021,207(3):12462.
[3]陈柏青,张悦,王科,等. 基于APSIM的新疆棉花生长与产量动态预测方法[J]. 农业机械学报,2025,56(5):82-90.
[4]范振岐. 南疆阿拉尔垦区密植棉花株高模拟研究[J]. 棉花学报,2024,36(4):320-327.
[5]王园. 可解释人工智能的配棉模型与系统研究[D]. 武汉:武汉纺织大学,2023.
[6]KOUSHIK A, MANOJ M, NEZAMUDDIN N. Explaining deep learning-based activity schedule models using SHapley Additive exPlanations[J]. Transportation Letters:the International Journal of Transportation Research,2025,17(3):442-457.
[7]李晶晶,张钟莉莉,闫华,等. 基于多特征选择的鲜食玉米需水量预测及可解释性分析[J]. 农业工程学报,2025,41(10):101-108.
[8]朱志畅,葛焱,臧晶荣,等. 基于无人机图像和SHAP特征筛选的小麦田间产量预测方法研究[J]. 麦类作物学报,2025,45(2):264-274.
[9]LOBELL D B, BURKE M B. On the use of statistical models to predict crop yield responses to climate change[J]. Agricultural & Forest Meteorology,2010,150(11):1443-1452.
[10]INIYAN S, VARMA V A, NAIDU C T, et al. Crop yield prediction using machine learning techniques[J]. Advances in Enginee-ring Software,2023,175:103326.
[11]KAUSHIK S, SINGH K. AI-driven smart irrigation and resource optimization for sustainable precision agriculture[J]. Journal of Scientific Innovation and Advanced Research,2025,1(2):168-177.
[12]DAI Y S, WANG H H, YANG M F, et al. A machine learning workflow for classifying and predicting the annual climatic status of cotton in Xinjiang, China[J]. Industrial Crops & Products,2025,226:120623.
[13]SAROWAR M S, FALGUNY B E, ASURA K M, et al. Crop yield prediction using machine learning:an extensive and systematic literature review[J]. Smart Agricultural Technology,2025,10:100718.
[14]韩典辰. 集成过程模型和机器学习算法的夏玉米产量预报方法研究[D]. 南京:南京信息工程大学,2025.
[15]张业翔,陈奉献,张宇红,等. 基于自动机器学习模型预测作物籽粒重金属浓度[J]. 应用生态学报,2025,36(6):1889-1897.
[16]KASHYAP G R, SHANKARAPPA S, NAGARAJA M K, et al. Machine learning ensembles, neural network, hybrid and sparse regression approaches for weather based rainfed cotton yield forecast[J]. International Journal of Biometeorology,2024,68(6):1179-1197.
[17]CDRIC R, BHM K. Introducing Geo-Glocal explainable artificial intelligence[J]. IEEE Access,2025,13:30952-30964.
[18]李林,邰红忠,卢金宝,等. 不同化控方式对棉花农艺性状及产量的影响[J]. 塔里木大学学报,2016,28(2):86-92.
[19]胡杰,李忠婷,陈钰秦,等. 2018-2022年不同磷肥基施滴施对北疆土壤磷素供应及棉花产量的影响[J]. 农业工程学报,2025,41(12):107-113.
[20]韩迎春,王国平,范正义,等. 主要气候因子对麦棉两熟棉花产量的影响[J]. 生态学报,2013,33(10):3185-3191.
[21]王士红,杨中旭,李秋芝,等. 年际间气象因子变化对棉花产量的影响[J]. 中国农学通报,2014,30(24):162-166.
[22]宋桂成,王苗苗,陈全战,等. 陆地棉花器官耐高温性的评价指标研究[J]. 棉花学报,2015,27(6):495-505.
[23]韩慧君. 气候生态因素对棉花产量与纤维品质的影响[J]. 中国农业科学,1991,24(5):23-29.
[24]常舒冬,石美亮,马明君. 2020-2023年冀中地区气候特征对棉花产量和纤维品质的影响[J]. 中国棉花,2025,52(10):54-60.
[25]熊宗伟,王雪姣,顾生浩,等. 中国主产棉区气象因子和纤维品质的相关性研究[J]. 棉花学报,2014,26(2):95-104.
[26]郝宏飞,辜永强,郝宏蕾,等. 2024年新疆巴楚县棉花品质及气象条件影响分析[J]. 中国棉花,2025,52(12):49-52.
[27]孙帅,李顺澳,彭冬梅,等. 基于多种机器学习模型的石河子棉区棉花产量预测研究[J]. 沙漠与绿洲气象,2024,18(6):166-172.
[28]王秀娟,沙达开提·依斯坎旦尔,袁兵年,等. 哈密区域棉花生育期气候变化对棉花品质影响研究[J]. 中国纤检,2024（10）:102-107.
[29]罗义腾. 基于卷积长短期记忆神经网络的棉花产量预测研究与应用[D]. 泰安:山东农业大学,2024.
[30]YU T X, ZHANG H, CHEN S K, et al. EXGEP:a framework for predicting genotype-by-environment interactions using ensembles of explainable machine-learning models[J]. Briefings in Bioinformatics,2025,26(4):bbaf414.
[31]CELIK M F, ISIK M S, TASKIN G, et al. Explainable artificial intelligence for cotton yield prediction with multisource data[J]. IEEE Geoscience and Remote Sensing Letters. 2023,20:8500905.

相似文献/References:

[1]何虎翼,谭冠宁,何新民,等.63 份马铃薯品种(系)资源农艺性状的主成分与聚类分析[J].江苏农业学报,2017,(01):27.[doi:10.3969/j.issn.1000-4440.2017.01.005 ]
　HE Hu-yi,TAN Guan-ning,HE Xin-min,et al.Principal component and cluster analysis for agronomic traits of 63 potato varieties or clones[J].,2017,(03):27.[doi:10.3969/j.issn.1000-4440.2017.01.005 ]
[2]赵亮,狄佳春,陈旭升.棉花基因组数据库中ＣＰＳ＆ＫＳ 基因的查找与分析[J].江苏农业学报,2016,(01):27.[doi:10.3969/j.issn.1000-4440.2016.01.004 ]
　ZHAO Liang,DI Jia-chun,CEHN Xu-sheng.Analysis of ent-copalyl diphosphate aynthase and ent-kaurene synthase (CPS&KS) gene family in cotton genome databases[J].,2016,(03):27.[doi:10.3969/j.issn.1000-4440.2016.01.004 ]
[3]赵君,刘剑光,吴巧娟,等.棉花种质种仁含油量测定及其遗传多样性分析[J].江苏农业学报,2015,(05):975.[doi:doi:10.3969/j.issn.1000-4440.2015.05.006]
　ZHAO Jun,LIU Jian-guang,WU Qiao-juan,et al.Kernel oil content and genetic diversity of upland cotton germplasm[J].,2015,(03):975.[doi:doi:10.3969/j.issn.1000-4440.2015.05.006]
[4]邵初阳,何晓兰,徐照龙,等.甜高粱种质资源多样性及主要农艺参数聚类分析[J].江苏农业学报,2015,(05):984.[doi:doi:10.3969/j.issn.1000-4440.2015.05.007]
　SHAO Chu-yang,HE Xiao-lan,XU Zhao-long,et al.Genetic diversity of sweet sorghum germplasm resources and clustering of major agronomical traits[J].,2015,(03):984.[doi:doi:10.3969/j.issn.1000-4440.2015.05.007]
[5]杨长琴,刘瑞显,张国伟,等.花铃期干旱对棉纤维素累积及纤维比强度的影响[J].江苏农业学报,2015,(06):1218.[doi:doi:10.3969/j.issn.1000-4440.2015.06.005]
　YANG Chang-qin,LIU Rui-xian,ZHANG Guo-wei,et al.Cellulose accumulation and fiber strength affected by drought during flowering and bolling stage in cotton[J].,2015,(03):1218.[doi:doi:10.3969/j.issn.1000-4440.2015.06.005]
[6]杨长琴,刘瑞显,张国伟,等.花铃期渍水对棉铃对位叶光合速率、物质累积及产量的影响[J].江苏农业学报,2015,(04):732.[doi:10.3969/j.issn.1000-4440.2015.04.004]
　YANG Chang-qin,LIU Rui-xian,ZHANG Guo-wei,et al.Photosynthesis of subtending leaves of bolls, dry matter accumulation and cotton yield in response to waterlogging during flowering and boll-forming stage[J].,2015,(03):732.[doi:10.3969/j.issn.1000-4440.2015.04.004]
[7]刘雅辉,王秀萍,鲁雪林,等.棉花耐盐相关序列扩增多态性（SRAP）分子标记筛选[J].江苏农业学报,2015,(03):484.[doi:10.3969/j.issn.1000-4440.2015.03.003]
　LIU Ya-hui,WANG Xiu-ping,LU Xue-lin,et al.Selection of sequence-related amplified polymorphism molecular marker associated with salt tolerance of cotton[J].,2015,(03):484.[doi:10.3969/j.issn.1000-4440.2015.03.003]
[8]王为,叶泗洪,潘宗瑾,等.棉花分子标记冗余性检测与评价的方法[J].江苏农业学报,2015,(02):247.[doi:10.3969/j.issn.1000-4440.2015.02.004]
　WANG Wei,YE Si-hong,PAN Zong-jin,et al.An approach to detecting and evaluating molecular marker redundancy in cotton[J].,2015,(03):247.[doi:10.3969/j.issn.1000-4440.2015.02.004]
[9]王小虎,钟卫国,李标,等.江苏省杂交粳稻与常规粳稻新品种主要性状比较[J].江苏农业学报,2017,(04):729.[doi:doi:10.3969/j.issn.1000-4440.2017.04.002]
　WANG Xiao-hu,ZHONG Wei-guo,LI Biao,et al.Comparative analysis of major traits of new varieties between japonica hybrid rice and conventional japonica rice in Jiangsu province[J].,2017,(03):729.[doi:doi:10.3969/j.issn.1000-4440.2017.04.002]
[10]郭琪,徐珍珍,黄芳,等.棉花HKT基因家族的全基因组分析[J].江苏农业学报,2017,(05):975.[doi:doi:10.3969/j.issn.1000-4440.2017.05.003]
　GUO Qi,XU Zhen-zhen,HUANG Fang,et al.Genome-wide analysis of high-affinity potassium transporter gene family in cotton[J].,2017,(03):975.[doi:doi:10.3969/j.issn.1000-4440.2017.05.003]

备注/Memo

备注/Memo:
收稿日期：2026-01-12基金项目：国家自然科学基金项目(32572366)；农业生物育种国家科技重大专项(2022ZD04019)；江苏省农业科技自主创新基金项目[CX(24)3120]作者简介：徐剑文(1984-),男,江苏南京人,博士,副研究员,主要从事棉花区域试验、农业数据分析和棉花遗传育种研究。(E-mail)xujianwen@jaas.ac.cn通讯作者：邵明灿,(E-mail)shaomingcan@jaas.ac.cn
更新日期/Last Update: 2026-04-17