首页|期刊导航|气象与环境学报|天气气候事件危险性指数构建及其对农产品价格波动的影响研究

天气气候事件危险性指数构建及其对农产品价格波动的影响研究OA

Construction of weather and climate event hazard index and its impact on agricultural product price fluctuations

中文摘要英文摘要

基于1961-2023 年哈尔滨市13 个气象台站逐日气温、降水量资料,构建了雪灾、冷空气和强降水事件等天气气候事件危险性指数模型,并建立了数据集.分析了哈尔滨2021-2023 年各类农产品粮油类、肉蛋禽、水产品、蔬菜、水果等逐周价格数据波动特点,研究了各类农产品价格波动与前7、前14、前21 d 的天气气候事件危险性指数的相关性,并采用随机森林、支持向量机、BP 神经网络等机器学习方法,对比了是否包含天气气候事件危险性指数因子的农产品价格波动模型的性能.结果表明:雪灾事件危险性指数由强降雪过程总量、过程日最大降雪量和过程日均降雪量构建;冷空气事件危险性指数由冷空气过程持续天数、过程最大降温幅度和过程最低温度构建;强降水事件危险性指数以强降水过程降水总量表示.肉蛋禽、蔬菜和水果价格波动序列虽满足平稳性条件、但不存在显著线性自相关,未检出条件异方差特征,农产品价格预测中常用的基于时间序列自相关性的模型不适用本文的研究对象.农产品价格波动与天气气候事件危险性指数的相关系数未通过0.05 显著性检验,不能仅用天气气候事件危险性指数构建传统的统计回归模型对价格波动进行拟合和预测,但从相关性可以看出不同种类的农产品价格波动对天气气候事件的敏感性存在差异,肉蛋禽价格波动与前7d 天气气候事件指数相关性较好,水果与前14 d 气气候事件指数相关性最强、蔬菜价格波动与前21 d 天气气候事件指数相关性较好.在引入天气气候事件危险性指数因子后,发现随机森林模型对这三类农产品价格波动预测的 MSE 和 MAE 均显著降低,尤其是对肉蛋禽价格波动的预测 MSE 和 MAE 分别降低39.7%和21.6%.

Based on the daily temperature and precipitation data from 13 meteorological stations in Harbin from 1961 to2023,hazard index models were constructed for weather and climate events such as snow disasters,cold air,and heavy precipitation events,and a dataset was established.On this foundation,the paper analyzed the weekly price-fluctuation characteristics of various agricultural products in Harbin from 2021 to 2023,including grains,coarse grains,meat,eggs,poultry,aquatic products,vegetables,and fruits.This paper also delved into the correla-tion between the price fluctuations of various agricultural products and the hazard indices of weather and climate e-vents over the preceding 7-day,14-day,and 21-day periods.Furthermore,machine-learning techniques,namely ran-dom forest,support vector machine,and BP neural network,were employed to evaluate the performance of agricul-tural product price-fluctuation models,both with and without the weather and climate event hazard index factor.The results show that the hazard index for snow-disaster events is formulated by taking into account the total vol-ume of heavysnowfall during the heavy-snowfall event,the maximum daily snowfall during the heavy-snowfall pe-riod,and the average daily snowfall during the same period.For cold-air events,the hazard index is determined based on the duration of the cold-air event,the maximum temperature decline during the event,and the lowest tem-perature during the event.For heavy-precipitation events,the hazard index is denoted by the total precipitation a-mount during the heavy-precipitationevent.The price-fluctuation series for meat,eggs,poultry,vegetables,and fruits fail to display the characteristics of stationarity,autocorrelation,and heteroscedasticity.Models based on the auto-correlation of time series,which are commonly employed in predicting agricultural product prices,are inappro-priate for the research subjects of this study.The correlation coefficients between the price fluctuations of agricul-tural products and the hazard indices of weather and climate events are not significant at the 0.05 level.Conse-quently,it is not practicable to solely utilize the hazard indices of weather and climate events to develop traditional statistical regression models for fitting and forecasting price fluctuations.Nevertheless,the correlation analysis re-veals differences in the sensitivity of price fluctuations of various agricultural product categories to weather and cli-mate events.Specifically,the price fluctuations of meat,eggs,and poultry are relatively strongly correlated with the weather and climate event indices of the preceding 7 days;those of fruits exhibit the highest correlation with the indices of the preceding 14 days;and the price fluctuations of vegetables show a strong correlation with the indices of the preceding 21 days.After introducing the factor of the hazard index of weather and climate events,both the mean squared error(MSE)and mean absolute error(MAE)of the random forest model for predictingthe price fluctuations of these three types of agricultural products are significantly reduced.In particular,for the prediction of the price fluctuations of meat,eggs,and poultry,the MSE and MAE decrease by 39.7%and 21.6%,respectively.

周长明;陈莉;孙爽;徐永清;陈刚;张东;王兴刚

黑龙江省气候中心,黑龙江 哈尔滨 150030||五营国家气候观象台,黑龙江 伊春 153000黑龙江省气候中心,黑龙江 哈尔滨 150030||五营国家气候观象台,黑龙江 伊春 153000黑龙江省气候中心,黑龙江 哈尔滨 150030||五营国家气候观象台,黑龙江 伊春 153000黑龙江省气候中心,黑龙江 哈尔滨 150030||五营国家气候观象台,黑龙江 伊春 153000黑龙江省气象局,黑龙江 哈尔滨 150030黑龙江省发展和改革委员会,黑龙江 哈尔滨 150030黑龙江省价格监测中心,黑龙江 哈尔滨 150001

天文与地球科学

农产品价格波动天气气候事件危险性指数模型性能

Price fluctuation of agricultural productsWeather and climate eventsHazard indexModel perform-ance

《气象与环境学报》 2026 (3)

83-91,9

东北区域气象科技协同创新联合基金项目(2024GX007)、黑龙江省气象局科技创新发展项目(HQ2023047)和黑龙江省自然科学基金(PL2026D023)共同资助.

10.3969/j.issn.1673-503X.2026.03.009

评论