基于作物模型与机器学习的水稻障碍型冷害脆弱性研究  

The Rice Vulnerability to Sterile-type Chilling Disaster in China Based on Crop Model and Machine Learning

作　　者：张静[1,2,3] 张朝[1,2] 张亮亮[1,2] 曹娟 骆玉川[1,2] 韩继冲 陶福禄 ZHANG Jing;ZHANG Zhao;ZHANG Liang-liang;CAO Juan;LUO Yu-chuan;HAN Ji-chong;TAO Fu-lu(Joint International Research Laboratory of Catastrophe Simulation and Systemic Risk Governance,Beijing Normal University at Zhuhai,Zhuhai 519087,China;School of National Safety and Emergency Management,Beijing Normal University,Beijing 100875;Guangzhou Institute of Geography,Guangdong Academy of Sciences,Guangzhou 510070;Key Laboratory of Land Surface Pattern and Simulation,Institute of Geographical Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101)

机构地区：[1]北京师范大学教育部巨灾模拟与系统性风险应对国际合作联合实验室,珠海519087 [2]北京师范大学国家安全与应急管理学院,北京100875 [3]广东省科学院广州地理研究所,广州510070 [4]中国科学院地理科学与资源研究所陆地表层格局与模拟重点实验室,北京100101

出　　处：《中国农业气象》2024年第9期1053-1066,共14页Chinese Journal of Agrometeorology

基　　金：国家重点研发计划项目(2020YFA0608201);国家自然科学基金项目(42101095,41977405,42061144003);博士后创新人才支持计划(BX20200064)。

摘　　要：以水稻障碍型冷害为例,提出一种突破数据限制并兼顾承灾体反应机理的新方法,构建农业气象灾害脆弱性曲线。基于1990-2010年气象数据设计水稻抽穗开花期的障碍型冷害情景,通过MCWLA-Rice作物模型和机器学习混合建模法计算对应的水稻单产损失,最终建立不同水稻主产区的障碍型冷害脆弱性曲线,并对1961-2010年水稻障碍型冷害单产损失快速评估。结果表明:(1)机器学习可有效重现作物模型模拟精度(RRMSE<6%,R^(2)>0.93)。(2)水稻障碍型冷害脆弱性具有明显的空间和种植制度差异,高纬地区到低纬地区水稻脆弱性整体呈下降趋势,且晚稻脆弱性普遍低于早稻。(3)一季稻抽穗开花期障碍型冷害的年均单产损失(1224kg·hm^(−2))高于双季稻(早稻:868kg·hm^(−2);晚稻:807kg·hm^(−2))。Using the case study of a sterile-type chilling disaster during the head-flowering phase of rice,this study presents a novel approach to constructing vulnerability curves that can overcome data limitations while also taking into account crop growth mechanisms.Meteorological data during 1990-2010 was used to generate sterile-type chilling scenarios at county scale,estimated rice yield losses through combining one crop model(MCWLA)and machine learning(XGBoost)method,finally developed sterile-type chilling vulnerability curves for each main rice-planting zone in China and estimated long-term historical(1961-2010)yield loss caused by sterile-type chilling disasters.The results showed that:(1)Machine learning could effectively reproduce the estimation ability of crop model(RRMSE<6%,R^(2)>0.93).(2)The sterile-type vulnerability decreased with decreasing latitude,and was weaker in growing seasons for late rice than that in early rice.(3)The historical yield loss was higher for single rice(1224kg·ha^(−1))than for double rice(early rice:868kg·ha^(−1);late rice:807kg·ha^(−1)).

关 键 词：水稻 低温冷害 脆弱性 作物模型 机器学习 

分 类 号：S426[农业科学—植物保护] S511[自动化与计算机技术—控制理论与控制工程] TP181[自动化与计算机技术—控制科学与工程]