PM_(2.5)时空序列缺失数据的反距离权重插值方法补缺研究  

Inverse distance weight interpolation method for missing data of PM_(2.5) spatiotemporal series

作　　者：梁玉柔 伍红玲[2] 王伟鹏 程峰[1] 段平[1] LIANG Yurou;WU Hongling;WANG Weipeng;CHENG Feng;DUAN Ping(Faculty of Geography,Yunnan Normal University,Kunming,Yunnan 650500,China;Yunnan Provincial Basic Geographic Information Center,Kunming,Yunnan 650100,China;Yunnan Real Estate Registry Centre,Kunming,Yunnan 650051,China)

机构地区：[1]云南师范大学地理学部,云南昆明650500 [2]云南省基础地理信息中心,云南昆明650100 [3]云南省不动产登记中心,云南昆明650051

出　　处：《环境与职业医学》2025年第2期171-178,共8页Journal of Environmental and Occupational Medicine

基　　金：国家自然科学基金项目(41901336);云南省基础研究计划项目(202101AT070078);云南省高层次人才培养支持计划项目(YNWRQNBJ-2020-103)。

摘　　要：[背景]细颗粒物(PM_(2.5))监测站点因受各种因素的影响,导致在某一段时间的数据缺失。这种数据缺失会影响空气质量评估和污染治理决策的制定。[目的]提出一种基于粒子群算法(PSO)优化距离指数的反距离权重(IDW)时空插值方法,用于PM_(2.5)时空序列缺失数据的插值补缺,提升插值补缺精度。[方法]研究分为两部分:第一部分以长三角地区为研究区,基于2017年1月1日四个时刻的小时尺度PM_(2.5)观测数据进行插值补缺实验;第二部分以京津冀地区为研究区,基于2017年1月1日至10日共10 d的天尺度PM_(2.5)观测数据进行插值补缺实验。实验分别选取均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分比误差(MAPE)和平均相对误差(MRE)4种指标对补缺精度进行评估。[结果]基于PSO优化距离指数的IDW时空插值方法提升了PM_(2.5)时空序列缺失数据补缺的精度。在长三角地区的小时尺度实验中,相较于距离指数取2时,上述方法的RMSE、MAE、MAPE、MRE平均提升了0.17μg·m^(-3)、0.27μg·m^(-3)、0.17%、0.01%。基于该方法生成的4个时间段PM_(2.5)空间场图,清晰展示了长三角地区小时尺度PM_(2.5)浓度的时空分布特征。在京津冀地区的天尺度实验中,PSO优化的距离指数相比传统方法表现更优,插值精度提升约为0.215μg·m^(-3)、0.283μg·m^(-3)、0.174%、0.014%。此外,该方法生成的四个季节的PM_(2.5)空间场图展现了京津冀地区在不同季节的PM_(2.5)浓度时空分布特征,进一步验证了该方法的有效性和适用性。[结论]基于PSO优化的IDW时空插值补缺方法在长三角和京津冀地区的PM_(2.5)时空序列缺失数据应用中展现出较优的插值精度和可靠性,为大气污染防治和公众健康保护提供了重要参考。[Background]Fine particulate matter(PM_(2.5))monitoring stations may generate missing data for a certain period of time due to various factors.This data loss will adversely affect air quality assessment and pollution control decision-making.[Objective]To propose an inverse distance weighted(IDW)spatiotemporal interpolation method based on particle swarm optimization(PSO)to interpolate and fill missing PM_(2.5) spatiotemporal sequence data and increase interpolation accuracy.[Methods]An interpolation experiment was designed into two parts.The first part used hourly PM_(2.5) observational data from four moments on January 1,2017 in the Yangtze River Delta region.The second part employed daily PM_(2.5) observational data from the first 10 d of January 2017 in the Beijing-Tianjin-Hebei region.Interpolation accuracy was evaluated using four metrics:root mean square error(RMSE),mean absolute error(MAE),mean absolute percentage error(MAPE),and mean relative error(MRE).[Results]IDW spatiotemporal interpolation method optimized with PSO significantly improved the accuracy of filling missing PM_(2.5) spatiotemporal sequence data.In the hourly-scale experiment conducted in the Yangtze River Delta region,compared to a distance index of 2,the accuracy metrics RMSE,MAE,MAPE,and MRE generated by the proposed method improved on average by 0.17μg·m^(−3),0.27μg·m^(−3),0.17%,and 0.01%,respectively.The PM_(2.5) spatial field maps generated for four moments based on this method clearly illustrated the spatiotemporal distribution characteristics of hourly PM_(2.5) concentrations in the Yangtze River Delta region.In the daily-scale experiment conducted in the Beijing-Tianjin-Hebei region,the PSO-optimized distance index outperformed the traditional method,with interpolation accuracy improvements of approximately 0.215μg·m^(−3),0.283μg·m^(−3),0.174%,and 0.014%,respectively.Furthermore,the seasonal PM_(2.5) spatial field maps generated by this method revealed the spatiotemporal distribution characteristics of PM_(2.5) con

关 键 词：细颗粒物 时空序列 插值 反距离权重 粒子群算法 

分 类 号：R12[医药卫生—环境卫生学]