农业工程学报2025，Vol.41Issue(14)：91-101,11.DOI:10.11975/j.issn.1002-6819.202412055
基于熵权法与网格搜索优化的水稻延迟型冷害指标构建
Delayed-type chilling injury index for rice using entropy weight method and grid search optimization
摘要
Abstract
Chilling injury is one of the major meteorological hazards in recent years.It has remarkably constrained the stable and high-yield production of rice,leading to the low resilience of regional agriculture.The overall frequency of chilling events has decreased as a consequence of global warming.The extremely low-temperature events have become a risk in the high-latitude regions,due to an increasing variability of climate.Therefore,it is essential to accurately assess a chilling injury,especially the delayed type.The effective prevention and mitigation can be developed in the stable and sustainable production of rice.However,the existing indices of chilling injury are mostly limited to singular indicators,static baseline thresholds,and homogeneous criteria for the classification of the severity.This study aimed to introduce an evaluation index for the delayed-type chilling injury in rice.Data-driven techniques were integrated with the hyperparameter optimisation in order to improve the accuracy and adaptability of the assessments.The daily anomalies were dynamically calculated according to the heat indices and 30-year moving averages.The entropy weight method was employed to quantify the relative importance of each disaster-inducing factor.The key variables were taken as the cumulative negative anomaly and the maximum number of consecutive days with negative anomalies.Accordingly,a comprehensive cold intensity index(CCII)was constructed to determine the criteria.The severity of the chilling injury was classified after optimization.The coarse-and fine-grained grid search techniques were used along with cross-validation.The precise classification of the chilling injury severity was realized to establish the differentiated grading thresholds.The chilling events were assessed in the three northeastern provinces of China.Furthermore,the spatiotemporal analysis was made on the chilling injury frequency.The results revealed that the CCII-based classification agreed well with the historical disaster records,which was an accuracy rate of 80.77%.The index was accurately identified in the representative years,in which the remarkable chilling injury occurred(i.e.,1969,1992,and 2009),as well as years with negligible impact(e.g.,2007).There was consistency with the historical documentation.The spatiotemporal analysis of the chilling frequency revealed that the severe chilling events were mainly concentrated in the northern Heilongjiang and eastern Jilin.From the 1960s to the 2010s,the overall frequency of rice chilling injury also exhibited a markedly declining trend,with the 1970s symbolised as the peak period.Local outbreaks sometimes occurred in some regions,indicating the potential risk of extreme chilling events.The delayed-type chilling injury index can offer a precise and quantitative way to assess the frequency,intensity,and duration of the low-temperature damage to rice.The conventional indices can also provide a robust approach to accurately assess the chilling injury.Moreover,it can offer scientific support to the decision-making on rice production in Northeast China.关键词
水稻/延迟型冷害/熵权法/网格搜索/冷害等级/冷害频率/时空分布Key words
rice/delayed-type chilling injury/entropy weight method/grid search/chilling injury severity level/chilling injury frequency/spatiotemporal distribution分类
农业科技引用本文复制引用
武晋雯,纪瑞鹏,孙龙彧,冯锐,姜丽霞,于成龙,于文颖,陈妮娜..基于熵权法与网格搜索优化的水稻延迟型冷害指标构建[J].农业工程学报,2025,41(14):91-101,11.基金项目
国家重点研发计划项目(2022YFD2300201) （2022YFD2300201）
辽宁省应用基础研究计划项目(2022JH2/101300193、2023JH2/101300090) （2022JH2/101300193、2023JH2/101300090）
辽宁省农业气象灾害重点实验室项目(2024SYIAEKFZD08) （2024SYIAEKFZD08）
国家自然科学基金项目(31671575) （31671575）
黑龙江省自然科学基金项目(LH2024D021) （LH2024D021）
